Professional Documents
Culture Documents
RADIOTRON
DESIGNER'S
HANDBOOK
Edited By
F. LANGFORD -SMITH
B.Sc., B.E. (1st class honours)
Senior Member I.R.E. (U.S.A.)
A.M.I.E. (Aust.)
FOURTH EDITION
under direction of
Commercial Engineering
RCA Victor Division
RADIO CORPORATION OF AMERICA
Harrison, New Jersey
April, 1953
IL . +
RADIOTRON DESIGNER'S
HANDBOOK
SALES TERRITORIES
Enquiries, from the listed countries, regarding supplies of the English
Language edition should be addressed as under :
PREFACE
This book has been written as a comprehensive, self-explanatory reference hand-
book for the benefit of all who have an interest in the design and application of radio
receivers or audio amplifiers. Everything outside this field -television, radio trans-
mission, radar, industrial electronics, test equipment and so on -has been excluded
to limit the book to a reasonable size.
An effort has been made to produce a handbook which, in its own sphere, is as self-
contained as possible. Extensive references to other sources of information have
been included for the reader who might require additional detail.
The success of the previous edition, of which over 280,000 copies have been sold
throughout the world, encouraged Amalgamated Wireless Valve Company Pty. Ltd.,
to undertake the compilation of the present up-to -date edition which is more com-
plete, great pains having been taken to fill in many of the gaps in the data published
hitherto. This has involved a considerable amount of both experimental and an-
alytical work by the editor and by engineers assisting in the project. Some original
work, previously unpublished, has been included.
Although the various chapters have been written by individual authors, all are truly
the result of team work, each having been carefully and critically examined by several
other engineers specializing in that particular field. In this way the accuracy of
mathematical calculations and of individual statements was checked and re- checked
to ensure that the reader would not be misinformed.
I wish to express my grateful thanks to Mr. R. Lambie, Manager, Mr. D. M.
Sutherland, Works Manager, and to the following collaborating engineers-
J. E. Bailey, Dr. W. G. Baker, D. Barnett, J. D. G. Barrett, Dr. G. Builder,
N. V. C. Cansick, W. N. Christiansen, Dr. E. R. Dalziel, K. G. Dean, H. L.
Downing, J. Gilchrist, I. C Hansen, R. Herbert, F. Holloway, I. Hood, D. G.
Lindsay, W. S. McGuire, E. J. Packer, J. Pritchard, R. J. Rawlings, B. Sandel,
J. Stacey, R. D. Stewart, J. E. Telfer, R. Vine, E. Watkinson.
I also wish to express my thanks to Mr. R. Ainsworth for his invaluable work in
sub -editing and assistance in indexing and to Mr. R. H. Aston for compiling data.
F. LANGFORD -SMITH
NOTES
References to valve types are to the prototype (e.g. 6J7) and include all equivalent
types (e.g. 6J7 -G, 6J7 -GT etc.) unless otherwise indicated.
If any errors are noted, please write to the Editor, Radiotron Designer's Handbook,
Amalgamated Wireless Valve Co. Pty. Ltd., 47 York Street, Sydney, Australia.
(y)
ACKNOWLEDGEMENTS
The editor wishes to acknowledge his indebtedness to the publishers of the
following periodicals for permission to reproduce the diagrams listed below. In
some cases the diagrams are based largely on the originals, although modified.
PERIODICALS
Audio Engineering 13. 50D, 50E. 14. 1A, B, C, D, E, F, G, H. 15. 15, 16,
57A, 57B, 60. 16. 15. 17. 7, 8, 15A, 15B, 15C, 18, 19A, 19B, 19C, 20, 22, 24B,
24C, 27A, 27B, 27C, 31, 37A. 19. 6A. 20. 10, 13, 14.
A.W.A. Technical Review 25. 33, 34, 35, 36, 37, 38, 39. 26. 6. 27. 44.
Communications 7. 72, 73. 16. 16. 18. 17, 18. 23. 17, 18, 19, 20, 21, 22, 23,
24, 25, 26.
Electronic Engineering 5. 13E, 13F, 13G, 13H, 13J, 13K, 13L, 13M, 13N, 13P,
13Q, 13R, 13S, 13T, 13U, Table 5. 7. 51D. 11. 7, 8, 9, 10, 11, 12, 13. 12. 3B,
3C, 11A, 11B. 15. 4, 63. 17. 21, 26. 18. 6A, 6B.
Electronic Industries 17. 6.
Electronics 1. 8. 4. 34. 5. 21, 22. 7. 51. 9. 17, 18. 12. 47A, 47B, 48.
15. 17, 18A, 18B, 38, 39, 45, 46, 52, 53, 61. 16. 4, 10, 12, 17, 18, 19, 20, 21, 22, 23,
24, 25, 26, 27, 28. 21. 9. 35. 24, 25. 36. 3, 5.
FM -TV Radio Communication 17. 35. 20. 25, 26, 27.
Journal Acoustical Society of America 20. 28.
Journal British I.R.E. 5. 13C, 13D.
Proceedings I.R.E. (U.S.A.) 7. 58, 59. 12. 7C. 13. 9B. 14. 6. 20. 24. 25.
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21. 30. 4, 5, 6, 7, 8, 9.
37. 2, 3, 4.
Proceedings I.E.E. 7. 60, 65, 69, 70.
Proceedings I.R.E. Australia 35. 19, 20, 21.
Q.S.T. 16. 13.
Radio 13. 50D, 50E. 15. 47A, 47B.
Radio and Hobbies (Australia) 15. 14B, 35B, 37B, 37C.
Radio and TV News 15. 43A, 43B. 16. 14. 18. 7A, 7B.
Radio Craft 15. 42, 44A, 49, 54. 17. 33.
Radio Electronics 15. 64.
R.C.A. Review 17. 37A. 27. 20, 21, 25, 27.
Tele -Tech 15. 14A. 36. 2. 38. Chart 38.4.
T.V. Engineering 10. 10.
Wireless Engineer 5. 13B. 7. 75, 76, 77, 78, 79, 80, 81, 82, 83. 11. 1, 6A.
Wireless World 4. 10, 26. 7. 51A, 54B, 54C, 55D, 55E, 55F, 59C. 12. 34, 35,
39. 15. 5, 13, 41, 50, 51, 55, 56A, 56B, 58A. 16. 8, 9, 11. 17. 24A, 28, 32, 35B,
35C, 35D, 35E, 35F, 35G. 18. 26. 20. 5, 9, 11, 12, 16, 17. 31. 6.
Reference numbers are given either in the titles or in the text. Lists of refer-
ences giving authors' names and details of articles are given at the end of each chapter.
ACKNOWLEDGEMENTS (Continued)
OTHER PUBLICATIONS
Aerovox Research Worker, Vol. 11, Nos. 1 and 2, January- February 1939, Figs. 5
and 6. Published by permission of Aerovox Corporation. Tables 38. 43, 44.
Charts 38. 1, 2.
Allegheny Ludlum Steel Corporation-curves 3.6% silicon steel. Fig. 5. 20.
American Standards Association " American Standard for Noise Measurement "
Z24.22 -1942. Figs. 19. 7, 8.
Australian Radio Technical Services and Patents Co. Pty. Ltd.-Technical Bulletins.
Figs. 15. 22, 23, 24, 25, 26, 27, 43. 31. 3, 4, 5.
Amalgamated Wireless Australasia Ltd. Figs. 18. 7. 28. 2.
Electrical and Musical Industries. Fig. 17. 25.
Jensen Manufacturing Company-Jensen Technical Monograph No. 3. Figs.
14. 2, 3, 4, 5.
P. R. Mallory & Co. Inc. " Fundamental Principles of Vibrator Power Supply "
Figs. 32. 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13.
National Bureau of Standards Circular C74, Fig. 208 redrawn and published by per-
mission of National Bureau of Standards. Fig. 10. 3.
Standard Transformer Corporation " Engineering a transformer." Figs. 5. 18C,
18D. Table 38. 70.
General Electric Company Ltd. " Radio receiver for use with high fidelity ampli-
fiers." Fig. 15. 62A.
Radio and Television Manufacturers Association (U.S.A.)- Extracts from, or sum-
maries of, the following Standards :
ET-107, ET-109, ET-110, REC -11, REC -101, REC -103, REC -104, REC -105,
REC -106 -A, REC -107, REC -108, REC -113, REC -114, REC- 115 -A, REC -116,
REC -117, REC -118, REC -119, REC -121, REC -128, S -410, S -416, S -417, S -418,
S -504, SE- 101 -A, SE -103, SE -105, TR- 105 -A, TR- 113 -A.
Note:- References in the text to R.M.A. Standards should be taken as having
been issued by RTMA.
I.R.E. Standards, Figs. 3. 11B, 15. 37. 1, 2, 3, 4.
Radio Industry Council- Information from standards on resistors and capacitors.
Chapter 38 Sections 2 and 3.
Radio Corporation of America. 1. 6. 2. 32A, 33, 34, 35, 47. 3. 16, 17. 10. 6.
13. 26, 30, 48. 17. 5A. 23. 9, 10, 11, 12, 13, 14, 15, 16. 30. 2A, 2B, 2C, 11.
33. 14. 35. 17.
CHAPTER HEADINGS
PART 1 : THE RADIO VALVE
CHAPTER PAGE
1 INTRODUCTION TO THE RADIO VALVE 1
...
481
544
603
15 TONE COMPENSATION AND TONE CONTROL ... ... 635
16 VOLUME EXPANSION, COMPRESSION AND LIMITING ... ... 679
17 REPRODUCTION FROM RECORDS... ... ... ... ... 701
18 MICROPHONES, PRE -AMPLIFIERS, ATTENUATORS AND MIXERS... 775
19 UNITS FOR THE MEASUREMENT OF GAIN AND NOISE ... 806
20 LOUDSPEAKERS ... ... ... ... ... ... ... ... ... 831
21 THE NETWORK BETWEEN THE POWER OUTPUT STAGE AND THE LOUDSPEAKER 880
(x)
CONTENTS
SECTION PAGE
(v) Cathode loadlines ... 29
(vi) With reactive loads... 30
5. TRIODE OPERATION OF PENTODES ... ... 34
(i) Triode operation of pentodes ... 34
(ii) Examples of transconductance calculation .. 34
(iii) Triode amplification factor ... ... 35
(iv) Plate resistance ... ... ... 36
(v) Connection of suppressor grid ... ... ... ... 36
6. CONVERSION FACTORS, AND THE CALCULATION OF CHARACTERISTICS OTHER
THAN THOSE PUBLISHED ... ... ... ... 36
(i) The basis of valve conversion factors 36
(ii) The use of valve conversion factors ... ... ... 37
(iii) The calculation of valve characteristics other than those pub-
lished ... ... ... ... ... 40
(iv) The effect of changes in operating conditions 42
7. VALVE EQUIVALENT CIRCUITS AND VECTORS ... ... ... 45
(i) Constant voltage equivalent circuit ... 45
(ii) Constant current equivalent circuit ... 46
(iii) Valve vectors ... ... ... ... 47
8. VALVE ADMITTANCES ... ... ... ... ... ... 49
(i) Grid input impedance and admittance 49
(ii) Admittance coefficients 50
(iii) The components of grid admittance -input resistance-
input capacitance -grid input admittance ... ... 51
(iv) Typical values of short-circuit input conductance ... 55
(v) Change of short- circuit -input capacitance with transconduct-
ance ... ... ... ... 55
(vi) Grid -cathode capacitance ... ... ... ... 56
(vii) Input capacitances of pentodes (published values) ... 56
(viii) Grid-plate capacitance ... ... 57
9. MATHEMATICAL RELATIONSHIPS ... ... 57
(i) General ... ... 57
(ii) Resistance load ... 58
(iii)Power and efficiency ... ... 59
(iv) Series expansion ; resistance load 61
(v) Series expansion ; general case ... ... 63
(vi)
(vii) -
The equivalent plate circuit theorem
Dynamic load line general case
(viii) Valve networks -general case ...
...
...
63
64
64
(ix) Valve coefficients as partial differentials 64
(x) Valve characteristics at low plate currents 65
10. REFERENCES ... ... 66
CHAPTER 3. THE TESTING OF OXIDE -COATED CATHODE
HIGH -VACUUM RECEIVING VALVES
1. BASIS OF TESTING PRACTICE ... ... ... 68
(i) Fundamental physical properties 69
(ii) Basic functional characteristics 70
(iii) Fundamental characteristic tests ... ... ... 73
(iv) Valve ratings and their limiting effect on operation 75
(A) Limiting ratings ... ... ... 75
(B) Characteristics usually rated 75
(C) Rating systems ... ... 77
(D) Interpretation of maximum ratings 77
(E) Operating conditions ... 80
(v) Recommended practice and operation .. 80
(a) Mounting ... ... 80
(b) Ventilation 81
CONTENTS
SECTION PAGE
(c) Heater -cathode insulation ... 81
(d) Control grid circuit resistance 82
(e) Operation at low screen voltages .. 84
(f) Microphony ... ... ... .. 84
(g) Hum ... ... 84
(h) Stand-by operation ... ... 84
2. CONTROL OF CHARACTERISTICS DURING MANUFACTURE 85
(i) Importance of control over characteristics ... 85
(ii) Basic manufacturing test specification 85
(iii) Systematic testing ... ... ... 86
(iv) Tolerances on characteristics ... 88
3. METHODS OF TESTING CHARACTERISTICS ... 89
(i) General conventions ... 90
(ii) General characteristics ... 91
(a) Physical dimensions 91
(b) Shorts and continuity ... 91
(c) Heater (or filament) current 93
(d) Heater to cathode leakage ... 94
(e) Inter -electrode insulation ... 94
(f) Emission ... ... ... .. 94
(g) Direct inter -electrode capacitances ... ... 95
(iii) Specific diode characteristics ...
(a) Rectification test ... .. 99
(b) Sputter and arcing 100
(c) Back emission ... ... ... 100
(d) Zero signal or standing diode current ... ... 101
(iv) Specific triode, pentode and beam tetrode characteristics... 101
(A) Reverse grid current ... ... ... 101
(B) Grid current commencement voltage .. 102
(C) Positive grid current ... ... ... 102
(D) Positive voltage electrode currents ... ... 103
(E) Transconductance or mutual conductance... 103
(F) Amplification factor ... ... 104
(G) Plate resistance ... ... 104
(H) A.C. amplification ... 105
(I) Power output ... ... ... 105
(J) Distortion ... ... 106
(K) Microphony ... ... ... 107
(L) Audio frequency noise ... 107
(M) Radio frequency noise ... 107
(N) Blocking ... ... ... 107
(0) Stage gain testing ... ... ... 108
(P) Electrode dissipation ... ... 108
(v) Specific converter characteristics ... 108
(A) Methods of operation including oscillator excitation 108
(1) Oscillator self-excited 108
(2) Oscillator driven ... 109
(3) Static operation ... ... 109
(B) Specific characteristics ... ... 109
(a) Reverse signal grid current 109
(b) Signal -grid current commencement ... 109
(c) Mixer positive voltage electrode currents 109
(d) Mixer conversion transconductance 109
(e) Mixer plate resistance 111
(f) Mixer transconductance ... 111
(g) Oscillator grid current 111
CONTENTS PAGE
SECTION
(h)Oscillator plate current ... 113
(i)Oscillator transconductance 113
(j)Oscillator amplification factor 113
(k)Oscillator plate resistance ... 113
(1) Signal -grid blocking... ... 113
(m) Microphony ... ... ... 113
(n) R -F noise ... ... ... ... 113
(vi) Tests for special characteristics ... ... 113
(A) Short- circuit input admittance ... 113
(B) Short- circuit feedback admittance ... 117
(C) Short- circuit output admittance ... 117
(D) Short- circuit forward admittance ... 117
(E) Perveance ... ... ... ... 117
(vii) Characteristics by pulse methods -point by point ... 118
(viii) Characteristics by curve tracer methods ... ... 119
4. ACCEPTANCE TESTING ... ... ... 120
(i) Relevant characteristics ... ... 120
(ii) Valve specifications 120
(iii) Testing procedure 120
5. SERVICE TESTING AND SERVICE TESTER PRACTICE ... ... ... ... 121
(i) Purpose and scope of service testing and discussion of as-
sociated problems ... ... ... ... ... ... 121
(ii) Fundamental characteristics which should be tested ... 122
(iii) Types of commercial testers ... ... ... ... ... 122
(iv) Methods of testing characteristics in commercial service testers 123
(a) Shorts testing ... ... ... ... ... ... 123
(b) Continuity testing ... ... 123
(c) Heater to cathode leakage ... ... 123
(d) Emission testing ... ... ... 123
(e) Mutual conductance testing ... 123
(f) Plate conductance testing ... 124
(g) Reverse grid current testing ... 124
(h) Power output testing ... ... ... 124
(i) Conversion conductance testing ... ... 124
(j) Oscillator mutual conductance testing ... 124
(k) Noise testing ... ... ... ... 124
(1) A.C. amplification testing ... 124
(v) A.C. versus d.c. electrode voltages in testers ... 124
(vi) Pre -heating ... 125
(vii) Testing procedure ... 125
6. REFERENCES ... ... ... ... 125
(xiv)
_
CONTENTS PAGE
SECTION
9. PRACTICAL RESISTORS, CONDENSERS AND INDUCTORS 186
(i) Practical resistors ... 186
(ii) Practical condensers ... ... 191
Combination units 197
(iii)
(iv) Practical inductors ... .. ... ... ... ... 197
(v) References to practical resistors and condensers ... 198
206
3. AUDIO-FREQUENCY TRANSFORMERS
(i) General considerations .. 206
(ii) Core materials ... ... ... ... 207
(iii) Frequency response and distortion
(a) Interstage transformers 209
(b) Low level transformers 210
(c) Output transformers ... ... 211
(iv) Designing for low leakage inductance 217
(v) Winding capacitance ... ... ... ... 219
(vi) Tests on output transformers 227
(vii) Specifications for a -f transformers 228
(xv)
CONTENTS
SECTION
PAGE
CHAPTER 6. MATHEMATICS
I. ARITHMETIC AND THE SLIDE RULE ... ... 255
(i) Figures ... ... ... ... 255
(ii)Powers and roots ... 255
(iii) Logarithms ... 255
(iv) The slide rule ... ... ... ... 257
(v) Short cuts in arithmetic 258
2. ALGEBRA ... ... ... 259
(i) Addition ... 260
(ii) Subtraction 260
(iii) Multiplication 260
(iv) Division 260
(v) Powers 260
(vi) Roots ... ... ... ... 261
(vii) Brackets and simple manipulations ... 261
(viii) Factoring ... ... 262
(ix) Proportion ... 262
(x) Variation ... 262
(xi) Inequalities 262
(xii) Functions 263
(xiii) Equations ... ... 263
(xiv) Formulae and laws ... 265
(xv) Continuity and limits ... ... ... 265
(xvi) Progressions, sequences and series ... ... 266
(xvii) Logarithmic and exponential functions ... 267
(xviii) Infinite series ... ... 268
(xix) Hyperbolic functions ... 269
(xx) General approximations 271
3. GEOMETRY AND TRIGONOMETRY 272
(i) Plane figures ... ... ... 272
(ii) Surfaces and volumes of solids 275
(iii) Trigonometrical relationships ... 275
4. PERIODIC PHENOMENA ... ... .. 278
5. GRAPHICAL REPRESENTATION AND j NOTATION ... 279
(i) Graphs .. ... ... ... 279
(ii) Finding the equation to a curve 281
(iii) Three variables ... ... ... 281
(iv) Vectors and j notation ... ... 282
6. COMPLEX ALGEBRA AND DE MOIVRE'S THEOREM ... 285
(i) Complex algebra with regular coordinates 285
(ii) Complex algebra with polar coordinates 286
(iii) De Moivre's Theorem ... 287
7. DIFFERENTIAL AND INTEGRAL CALCULUS 289
(i) Slope and rate of change ... ... 289
(ii) Differentiation ... ... ... 291
(iii) Integration ... ... 294
(iv) Taylor's Series ... ... 298
(v) Maclaurin's Series ... 299
8. FOURIER SERIES AND HARMONICS ... ... ... 299
(i) Periodic waves and the Fourier Series 299
(ii) Other applications of the Fourier Series 302
(iii) Graphical Harmonic Analysis ... 302
9. REFERENCES ... 304
CONTENTS
PAGE
SECTION
. 481
481
482
2. RESISTANCE- CAPACITANCE COUPLED TRIODES
(i) Choice of operating conditions ... 482
(ii) Coupling condenser ... 483
(iii) Cathode bias ... ... .. 484
(iv) Fixed bias ... 487
(v) Grid leak bias ... ... 489
(vi) Plate voltage and current ... 489
(vii) Gain and distortion at the mid- frequency 490
(viii) Dynamic characteristics ... ... ... ... 491
(ix) Maximum voltage output and distortion 491
(x) Conversion factors with r.c.c. triodes 493
(xi) Input impedance and Miller effect ... 493
(xii) Equivalent circuit of r.c.c. triode ... 494
(xiii) Voltage gain and phase s ... ... ... ... ... 494
(xiv) Comments on tabulated characteristics of resistance-
coupled triodes ... ... ... ... ... ... 495
CONTENTS
SECTION AGE
3. RESISTANCE- CAPACITANCE COUPLED PENTODES 496
(i) Choice of operating conditions 496
(ii) Coupling condenser 496
(iii) Screen by -pass 496
(iv) Cathode bias 499
(y) Fixed bias ... ... ... ... 501
(vi) Dynamic characteristics of pentodes ... 504
(vii) Gain at the mid -frequency ... ... ... ... ... 506
!viii) Dynamic characteristics of pentodes and comparison with
triodes ... ... ... ... ... 508
(ix) Maximum voltage output and distortion
(x) Conversion factors with r.c.c. pentodes ... 511
(xi) Equivalent circuit of r.c.c. pentode ... 512
(xii) Voltage gain and phase shift ... 512
(xiii) Screen loadlines ... ... ... ... ... ... 513
(xiv) Combined screen and cathode loadlines and the effect of toler-
ances ... ... ... ... ... ... ... 515
(xv) Remote cut -off pentodes as r.c.c. amplifiers ... ... 516
(xvi) Multigrid valves as r.c.c. amplifiers ... ... ... ... 516
(xvii) Special applications ... ... ... ... ... ... 516
(xviii) Comments on tabulated characteristics of resistance-
coupled pentodes ... ... 517
4. TRANSFORMER- COUPLED VOLTAGE AMPLIFIERS 517
(i) Introduction ... ... ... 517
(ii) Gain at the mid- frequency ... 517
(iii) Gain at low frequencies ... 517
(iv) Desirable valve characteristics ... 517
(v) Equivalent circuits ... ... ... 518
(vi) Gain and phase shift at all frequencies 518
(vii) Transformer characteristics ... ... 518
(viii) Fidelity ... ... ... 518
(ix) Valve loadlines ... ... 518
(x) Maximum peak output voltage 518
(xi) Transformer loading ... 518
(xii) Parallel feed ... .. 518
(xiii) Auto -transformer coupling 520
Applications
(xiv) ... ... 520
Special applications
(xv) 520
5. CHOKE-COUPLED AMPLIFIERS ... 521
(i) Performance ... ... 521
(ii) Application ... ... ... 521
6. METHODS OF EXCITING PUSH-PULL AMPLIFIERS ... ... 521
(i) Methods involving iron -cored inductors 521
(ii)Phase splitter ... ... ... ... ... 522
(iii) Phase inverter ... ... ... ... 524
(iv) Self-balancing phase inverter ... ... ... 524
(v) Self- balancing paraphase inverter ... ... ... 524
(vi) Common cathode impedance self-balancing inverters 526
(vii) Balanced output amplifiers with highly accurate balance ... 527
(viii) Cross coupled phase inverter ... ... ... ... ... 527
7. PUSH -PULL VOLTAGE AMPLIFIERS ... ... ... ... ... 527
(i) Introduction ... ... ... ... ... ... ... 527
(ii) Cathode resistors ... 527
(iii) Output circuit ... ... ... ... ... ... 527
(iv) Push -pull impedance -coupled amplifiers-mathematical treat-
ment ... ... ... ... ... ... ... 528
(v) Phase compressor 528
CONTENTS
SECTION PAGE
8. IN-PHASE AMPLIFIERS ... ... ... ... 529
(i) Cathode- coupled amplifiers ... 529
(ii) Grounded -grid amplifiers ... 529
(iii) Inverted input amplifiers ... 529
(iv) Other forms of in -phase amplifiers ... 529
9. DIRECT -COUPLED AMPLIFIERS... ... ... 529
(i) Elementary d -c amplifiers ... 529
(ii) Bridge circuit ... ... ... 530
(iii) Cathode- coupled ... 531
(iv) Cathode follower ... ... ... 531
(v) Phase inverter ... ... ... 532
(vi) Screen coupled ... ... 532
(vii) Gas tube coupled ... 532
(viii) Modulation systems ... 532
(ix) Compensated d.c. amplifiers ... ... ... 533
(x) Bridge -balanced direct current amplifiers ... 533
(xi) Cascode amplifiers ... ... ... ... 533
10. STABILITY, DECOUPLING AND Hum ... ... ... ... 535
(i) Effect of common impedance in power supply ... 535
(ii) Plate supply by- passing ... ... ... ... 535
(iii) Plate circuit decoupling ... 535
(iv) Screen circuit decoupling ... 537
(v) Grid circuit decoupling ... 538
(vi) Hum in voltage amplifiers ... ... 538
11. TRANSIENTS AND PULSES IN AUDIO FREQUENCY AMPLIFIERS 540
(i) Transient distortion ... 540
(ii) Rectangular pulses 540
12. MULTISTAGE VOLTAGE AMPLIFIERS ... 541
(i) Single -channel amplifiers ... 541
(ii) Multi- channel amplifiers 541
13. REFERENCES 542
CONTENTS
SECTION PAGE
(vi) Step -type tone control using negative feedback ... 662
(vii) Continuously- variable controls -general ... ... 662
(viii) Single control continuously-variable tone controls 662
(ix) Ganged continuously-variable tone controls ... 664
(x) Dual control continuously -variable tone controls 664
9. FEEDBACK TO PROVIDE TONE CONTROL ... 669
(i) Introduction ... ... ... ... ... 669
(ii) Amplifiers with feedback providing tone control 669
(iii) Whistle filters using feedback ... ... ... 672
10. AUTOMATIC FREQUENCY-COMPENSATED TONE CONTROL 672
(i) Introduction 672
(ü) Methods incorporating a tapped potentiometer 672
(iii) Methods incorporating step -type controls 673
(iv) Method incorporating inverse volume expansion with
multi-channel amplifier ... 673
11. WHISTLE FILTERS 673
(i) Resonant circuit filters 673
(ii) Narrow band rejection filter ... 675
(iii) Crystal filters ... 675
(iv) Parallel -T network 675
(v) Filters incorporating L and C 676
12. OTHER METHODS OF TONE CONTROL ... 676
(i) Multiple -channel amplifiers ... 676
(ii) Synthetic bass ... ... ... 676
13. THE LISTENER AND TONE CONTROL ... 677
14. EQUALIZER NETWORKS ... ... ... 677
15. REFERENCES ... 677
...
.
... 732
737
738
CONTENTS
SECTION PAGE
(vi) Equalizers for crystal pickups ... ... ... 741
(vii)Equalizers applying negative feedback to the pickup 743
Miscellaneous details regarding equalizing amplifiers
(viii) 743
(ix) Complete amplifiers ... ... ... ... 744
(x) Pickups for connection to radio receivers ... 751
(xi) Frequency test records ... ... ... 752
6. DISTORTION AND UNDESIRABLE EFFECTS ... ... 757
(i) Tracing distortion and pinch effect ... 757
(ii) Playback loss ... ... ... ... 760
(iii) Wow, and the effects of record warping 760
(iv) Distortion due to stylus wear 761
(v) Noise modulation ... ... 761
(vi) Pickup distortion 762
(vii) Acoustical radiation ... 762
(viii) Distortion in recording 762
7. NOISE REDUCTION ... ... 763
(i) Analysis of noise ... 763
(ii) High -frequency attenuation ... ... 763
(iii) High-frequency pre- emphasis and de- emphasis 763
(iv) Volume expansion ... ... 763
(v) Olson noise suppressor ... ... 763
(vi) Scott dynamic noise suppressor ... 764
(vii) Price balanced clipper noise suppressor 765
8. LACQUER DISC HOME RECORDING (DIRECT PLAYBACK) ... 766
(i) General description ... ... 766
(ii) Recording characteristic ... 766
(iii) Cutting stylus ... ... 767
(iv) Cutter head ... ... 767
(v) Equalization of cutter 767
(vi) Motor and turntable ... 767
(vii) Amplifier ... ... 767
(viii) Pickups for use on lacquer discs 768
(ix) Recording with embossed groove 768
9. REPRODUCTION FROM TRANSCRIPTION DISCS ... 769
(i) Introduction ... ... ... ... ... ... 769
(ii) Characteristics of record material, wear and noise 769
(iii) Sound track ... ... ... ... 770
(iv) Recording characteristics and equalization 770
(v) Translation loss and radius compensation 770
10. REFERENCES TO LATERAL DISC RECORDING ... ... 771
(xxviii)
CONTENTS
SECTION PAGE
(w)
CONTENTS
SECTION PAGE
3. AERIALS AND POWER TRANSFER 892
(i) Introduction 892
(ii) Power transfer ... 892
4. CHARACTERISTICS OF AERIALS ... ... ... 893
(i) Effective area of a receiving aerial 893
(ii) The power gain of an aerial 894
(iii) The beam -width of an aerial ... ... 894
5. EFFECTS OF THE EARTH ON THE PERFORMANCE OF AN AERIAL ... ... 894
(i) Introduction ... ... ... ... ... 894
(ii) A perfectly- conducting earth ... ... 894
(iii) An imperfectly conducting earth ... ... ... ... 896
(iv) The attenuation of radio waves in the presence of an imper-
fectly- reflecting earth ... ... ... ... ... 896
6. THE EFFECT OF THE IONOSPHERE ON THE RECEPTION OF RADIO SIGNALS ... 901
7. THE IMPEDANCE OF AN AERIAL ... ... ... 901
(i) Introduction ... 901
(ii) Resistive component of impedance 902
Reactive component of impedance
(iii) ... 903
Characteristic impedance of aerial
(iv) ... 903
(v)Examples of calculations ... 903
Dipoles ...
(vi) ... 904
Loop aerials
(vii) ... 905
8. DUMMY AERIALS ... ... ... ... ... 906
9. TYPES OF AERIAL USED FOR BROADCAST RECEPTION 907
(i) Introduction ... ... ... ... 907
(ii) Medium-frequency receiving aerials 907
(iii) Short-wave receiving aerials ... 908
(iv) V-H -F aerials ... ... ... 909
10. REFERENCES ... ... ... 911
(xxxiii)
CONTENTS
SECTION PAGE
(E) Crystal constants ... 1056
(F) Position of filter in circuit ... 1057
(G) Other types of crystal filters 1057
(iv) Design example ... 1057
7. DETUNING DUE TO A.V.C. ... 1061
(i) Causes of detuning 1061
(ii) Reduction of detuning effects 1062
8. STABILITY 1065
(i)Design data ... 1065
Neutralizing circuits
(ii) 1065
9. DISTORTION ... ... ... ... .. 1067
(i) Amplitude modulation i -f stages 1067
(ii) Frequency modulation i -f stages 1068
10. REFERENCES ... 1069
(xxxvii)
CONTENTS
SECTION
PAGE
5. A.C. ...
OPERATED RECEIVERS
1256
(i) Four valve receivers ... ... 1256
(ii) Five valve receivers ... ... 1259
(iii) Larger receivers 1260
(iv) Communication receivers ... 1260
6. A.C. /D.C. RECEIVERS ... ... ... ... ... 1264
(i) Series -resistor operation... ... 1264
(ii) Barretter operation ... ... 1266
(iii) Dial lamps ... ... 1266
(iv) Miscellaneous features 1267
7. BATTERY OPERATED RECEIVERS ... 1268
(i) General features ... 1268
(ii) Vibrator-operated receivers ... 1270
(iii)
8. CAR RADIO
(i)
Characteristics of dry batteries
...
Interference suppression
...
...
. .
...
1272
1275
1275
(ii) Circuit considerations ... ... 1276
(iii) Valve operating conditions ... .., 1277
9. MISCELLANEOUS FEATURES ... ... ... ... 1278
(i) Spurious responses ... .. , 1278
(ii) Reduction of interference 1279
(iii) Contact potential biasing ... 1280
(iv) Fuses ... ... ... ... 1281
(v) Tropic proofing ... . . . ... 1282
(vi) Parasitic oscillations .. 1283
(vii) Printed circuits ... 1283
(viii) Other miscellaneous features ... 1285
10. REFERENCES ... ... ... ... ... ... 1285
(xxxviii)
CONTENTS PAGE
SECTION
(iii) Test procedures and operating conditions .. ... 1315
(iv) Receiver adjustments 1316
(v) Performance tests .. 1317
3. AUDIO FREQUENCY AMPLIFIERS ... ... 1321
(i) Equipment and measurements ... 1321
(ii) Tests ... ... 1321
4. MEASUREMENTS ON COILS ... ... ... ... ... 1325
(i) Measurement of coefficient of coupling ... ... ... 1325
(ii) Measurement of primary resonant frequencies of aerial and r -f
coils ... ... ... ... ... ... ... ... 1325
(iii) Measurement of distributed capacitance across coils ... 1325
5. REFERENCES ... 1327
Section Page
1. Electricity and emission ... ... 1
It is interesting to note that the total current flow, equivalent to the total move-
ments of all the free electrons, irrespective of their directions, is very much greater
than that which occurs under any normal conditions of electric current flow. The
directions of movement are such that the external effects of one are generally cancelled
by those of another. Thus, in a metal, the oft- quoted picture of a flow of electrons
from the negative to the positive terminal is only a partial truth and apt to be mis-
leading. The velocity of the free electrons is very much less than that of the electric
current being of the order of only a few centimetres per second. The electron
current may be pictured as the successive impacts between one electron and another
in the direction of the current. In an insulator the number of free electrons is
practically zero, so that electric conduction does not take place. In a partial insulator
the number of free electrons is quite small.
The nucleus is a very complex body, including one or more protons which may
be combined with a number of neutrons *. The proton has a positive charge equal
and opposite to the charge on an electron but its mass is very much greater than that
of an electron. The simplest possible atom consists of one proton forming the nucleus
with one electron in an orbit around it-this is the hydrogen atom. Helium consists
of two protons and two neutrons in the nucleus, with two electrons rotating in orbits.
The neutron has a mass slightly greater than that of a proton, but the neutron has
no electric charge. An example of a more complicated atom is that of potassium
which has 19 protons and 20 neutrons in the nucleus, thus having a positive charge
of 19 units. The number of electrons in the orbits is 19, thus giving zero charge
for the atom as a whole, this being the normal condition of any atom. The common
form of uranium has 92 protons and 146 neutrons in the nucleus, with 92 electrons
rotating in orbits.
Under normal circumstances no electrons leave the surface or a substance since
the forces of attraction towards the centre of the body are too great. As the tem-
perature of the substance is raised, the velocity of the free electrons increases and
eventually, at a temperature which varies from one substance to another, some of the
free electrons leave the surface and may be attracted to a positive electrode in a vacuum.
This phenomenon is known as thermionic emission since its emission takes place
under the influence of heating. There are other types of emission such as photo
emission that occur when the surface of the substance is influenced by light, or
secondary emission when the surface is bombarded by electrons.
The radio valve makes use of thermionic emission in conjunction with associated
circuits for the purpose of producing amplification or oscillation. The most common
types of radio valves have hot cathodes, either in the form of a filament or an indirectly -
heated cathode. Many transmitting valves have filaments such as tungsten or
thoriated-tungsten, but nearly all receiving valves have what is known as an oxide
coated filament or cathode. The filament, or cathode sleeve, is usually made of
nickel or an alloy containing a large percentage of nickel and this is coated with a
mixture of barium and strontium carbonates that, during the manufacture of the
valve, are turned into oxides. A valve having an oxide- coated cathode has a very
high degree of emission as compared with other forms of emitters but requires very
great care during manufacture since it is readily poisoned by certain impurities which
may be present in the cathode itself or which may be driven out in the form of gas
from the bulb or the other electrodes.
Oxide -coated cathodes are generally operated at an average temperature of about
1050° Kelvin (777° C) which looks a dull red. Temperatures much above 1100 °K
generally cause a short life, while those below 960°K are very susceptible to poisoning
of the emission, and require careful attention to maintain a very high vacuum.
The thermionic valve is normally operated with its anodet current considerably
less than the maximum emission produced by its cathode. In the case of one having
a pure tungsten filament no damage is done to the filament if all the electrons emitted
*This is in accordance with the theory generally held at the time of writing ; it is, however, subject
to later modification.
tThe anode (also called the plate) is the positive electrode ; the cathode is the negative electrode.
1.1 ELECTRICITY AND EMISSION 3
are drawn away immediately to the anode. This is not so, however, with oxide coated
cathodes and these, for a long life and satisfactory service, require a total emission
very much greater than that drawn under operating conditions. In such a case a
cloud of electrons accumulates a short distance from the surface of the cathode and
supplies the electrons that go to the anode. This space charge as it is called, is
like a reservoir of water that supplies varying requirements but is itself replenished
at an average rate. The space charge forms a protection to the cathode coating against
bombardment and high electrostatic fields, while it also limits the current which would
otherwise be drawn by a positive voltage on the anode. If the electron emission
from the cathode is insufficient to build up this " space charge," the cathode coating
is called upon to supply Ingh peak currents that may do permanent injury to the coat-
ing and in extreme cases may even cause sputtering or arcing.
In multi-grid valves, if one grid has a positive potential and the next succeeding
grid (proceeding from cathode to plate) has a negative potential, there tends to be
formed an additional space charge. This outer space charge behaves as a source of
electrons for the outer electrodes, and is known as a virtual cathode.
An oxide -coated cathode, operated under proper conditions, is self -rejuvenating
and may have an extremely long working life. The life is, therefore, largely governed
by the excess emission over the peak current required in normal operation.
A valve having a large cathode area and small cathode current may have, under
ideal conditions, a life of the order of 50 000 hours, whereas one having extremely
limited surface area, such as a tiny battery valve, may have a working life of less than
1000 hours.
Under normal conditions a valve should be operated with its filament or heater
at the recommended voltage ; in the case of an oxide coated valve it is possible to
have fluctuations of the order of 10% up or down without seriously affecting the
life or characteristics of the valve [see Chapter 3 Sect. 1(iv)D] . The average voltage
should, however, be maintained at the correct value. If the filament or cathode is
operated continuously with a higher voltage than that recommended, some of the
coating material is evaporated and permanently lost, thus reducing the life of the valve.
Moreover, some of this vapour tends to deposit on the grid and give rise to what is
known as grid emission when the grid itself emits electrons and draws current
commonly known as negative grid current [for measurement see Chapter 3 Sect. 3
(iv)A] .
If the filament or heater is operated for long periods at reduced voltages, the effect
is a reduction in emission, but no damage is generally done to the valve unless the
cathode currents are sufficient to exhaust the " space charge." Low cathode tem-
perature is, therefore, permissible provided that the anode current is reduced in the
proper proportion.
During the working life of the valve, its emission usually increases over the early
period, reaches a maximum at an age which varies from valve to valve and from one
manufacturer to another, and then begins to fall. The user does not generally suffer
any detriment until the emission is insufficient to provide peak currents without
distortion.
Tests for the measurement of the emission of an oxide -coated cathode are described
in Chapter 3 Sect. 3(ii)f.
If a slight amount of gas is present some of the electrons will collide with atoms
of the gas and may knock off one or more electrons, which will serve to increase the
anode current, leaving atoms deficient in electrons. These are known as positive
ions since they carry a positive charge (brought about by the loss of electrons), and
the process is known as ionization. The positive ions are attracted by the negative
cathode, and being comparatively massive, they tend to bombard the cathode coating
in spite of the protection formed by the space charge.
Some types of rectifiers (e.g. OZ4) have no heaters, and the oxide-coated cathode
is initially heated by ion bombardment ; this flow of current is sufficient to raise the
cathode temperature so as to enable it to emit electrons in the usual manner. The
gas is an inert variety at reduced pressure. Although some types of gaseous thermionic
4 ELECTRICITY AND EMISSION 1.1
rectifiers will operate (once they have been thoroughly heated) without any filament
or heater voltage, this is likely to cause early failure through loss of emission.
Most thermionic valves are vacuum types and operate under a very high degree
of vacuum. This is produced during manufacture by a combination of vacuum
pumps and is made permanent by the flashing of a small amount of " getter " which
remains in the bulb ready to combine with any impurities which may be driven off
during life. Valves coming through on the production line are all tested for gas by
measuring the negative grid current under operating conditions ; methods of testing
are described in Chapter 3 Sect. 3(iv)A, where some values of maximum negative
grid current are also given. If a valve has been on the shelf for a long time, it fre-
quently shows a higher gas current, but this may usually be reduced to normal by
operating the valve under normal conditions, with a low resistance connection between
grid and cathode, for a short period. When a valve is slightly gassy, it usually shows
a blue glow (ionization) between cathode and anode. In extreme cases this may
extend outside the ends of the electrodes, but a valve in such condition should be
regarded with suspicion and tested before being used in any equipment, as it might
do serious damage. A slight crack may permit a very small amount of air to enter
the bulb, giving rise to a pink/violet glow which may readily be identified by any one
familiar with it ; this is a sign of immediate end of life.
The anode current of a thermionic valve is not perfectly steady, since it is brought
about by a flow of electrons from the cathode. When a valve is followed by very
high gain amplifiers, the rushing noise heard in the loudspeaker is partly caused by
the electrons in the valve, and partly by a somewhat similar effect (referred to as the
" thermal agitation " or " Johnson noise ") principally in the resistance in the grid
circuit of the first valve -see Chapter 4 Sect. 9(i)1, and Chapter 18 Sect. 2(ü). This
question of valve noise is dealt with in Chapter 18 Sect. 2(ü)c and Chapter 23 Sect. 6.
Some valves show a fluorescence on the inside of the bulb, which may fluctuate
when the valve is operating. This is perfectly harmless and may be distinguished
from blue glow by its position in the valve. In occasional cases fluorescence may also
be observed on the surfaces of the mica supports inside the valve.
The
Indirectly- heated cathodes consist of a cathode sleeve surrounding a heater. and
cathode sleeve may have a variety of shapes, including round (D), elliptical (E) for
rectangular (F) cross section. They are usually fitted with a light ribbon tag
connection to the lead going to the base pin.
D E F G H
A B C
FIG. 1.1
In an indirectly- heated valve, the function of the heater is solely to heat the cathode.
No emission should take place from the heater and the insulation
between heater
and cathode should be good. The heater is generally made of tungsten or a tungsten
alloy wire coated wit a substance capable of providing the necessary insulation at
to be
high temperature, such as alundum. In all applications where hum is likely 1.1.
troublesome, the heater is preferably of the double helical type, as G in
Fig.
often
Power valves and other types having elliptical or rectangular cathode sleeves, low
heater as in H. These are not generally suitable for use in very
employ a folded
level amplifiers whether for radio or audio frequencies.
(ii) Grids
Grids are constructed of very line wire wound around one, two or four side
common. Some valves have two, three,
rods
four or five grids
-
two being by far the most
inside one another, but all of these are similar in general form
although different in
dimensions.
In the case of some grids it is necessary to take precautions to limit the grid tem-
limit the grid
perature either to avoid g :id emission, in the case of control grids, or to These may,
temperature to prevent the formation of gas, in the case of screen grids.
for better heat radiation, be fitted with copper side rods and blackened
radiators either
the other electrodes. Grids are numbered in order from the cathode
above or below 2 grid the one
outwards, s.) that No. 1 grid will be the one closest to the cathode, No.
adjacent to it, and No. 3 the one further out again.
(iii) Plates
The plate of a receiving valve is the anode or positive electrode. It may
of
be in one
the valve.
of a great number of shapes, dependent on the particular application
their
The plates of power valves and rectifiers are frequently blackened to increase
heat radiation and thereby reduce their temperature.
(iv) Bulbs
the effects
The inside surfaces of glass bulbs are frequently blackened. This has to develop
of making them more or less conductive, thereby reducing the tendency
from the bulb.
static charges, and reducing the tendency towards secondary emission
(v) Voltages with valve operation
All voltages in radio valves are taken with respect to the
cathode, in the case of
heated
indirectly-heated valves, and the negative end of the filament with directly-
6 COMPONENT PARTS OF VALVES 1.2
FIG. 1.2
also been manufactured for special purposes. Fig. 1.2 shows the circuit of a diode
valve in which battery A is used to heat the filament or heater, and battery B to apply
a positive potential to the anode through the load resistor. The plate current is
measured by a milliammeter connected as shown, and the direction of current flow
isfrom the positive end of battery B towards the anode, this being the opposite of the
electron current flow. It should be noted that the negative end of battery B is re-
turned to the negative end of battery A in accordance with the usual convention. It
would be quite permissible to connect the negative end of the battery B to the positive
end of battery A so as to get the benefit of the voltage A applied to the anode, but
in this case, the total voltage applied to the anode would be A + B. If voltage of
battery B is reversed, it will be noted that the plate current is zero, thus indicating
the rectification that takes place in a diode. If an alternating voltage is applied,
current will only flow during the half-cycles when the anode is positive. This is
called a half-wave rectifier since it is only capable of rectifying one half of the cycle.
1.3 TYPES OF RADIO VALVES 7
Full wave* rectifiers are manufactured with two anodes and a common cathode and
these are arranged in the circuit so that one diode conducts during one half-cycle
and the other during the other half-cycle.
(ii) Triodes
A triode is a three electrode valve, the electrodes being the cathode, grid and
anode
(or plate). The grid serves to control the plate current flow, and if the grid is made
sufficiently negative the plate current is reduced to zero. The voltage on the grid is
controlled by battery C in Fig. 1.3, the other part of the circuit being as for the diode
in Fig. 1.2. When the grid is negative with respect to the cathode, it does not draw
appreciable current ; this is the normal condition as a class A1 amplifier. Although
an indirectly heated cathode has been shown in this instance, a directly heated valve
could equally well have been used. The heater in Fig. 1.3 may be supplied either
from an a.c. or d.c. source, which should preferably be connected to the cathode or
as close as possible to cathode potential.
As the grid is made more negative, so the plate current is decreased and when
the
grid is made more positive the plate current is increased. A triode is, therefore,
capable of converting a voltage change at the grid into a change of power in the load
resistor. It may also be used as a voltage amplifier or oscillator.
FIG. 1.3
Tetrodes
The capacitance between the grid and plate can be reduced by mounting an addi-
tional electrode, generally called the screen or screen grid, between the grid
and plate.
electrodes, hence the name tetrode. The function of the
The valve thus has four the
screen is to act as an electrostatic shield between grid and plate, thus reducing
grid-to -plate capacitance. The screen is connected to a positive potential (although
which it would
less than that of the plate) in order to counteract the blocking effect
otherwise have on the plate current -see Fig. 1.4. Owing to the comparatively
large spaces between the wires in the screen, most of the electrons from
the cathode
pass through the screen to the plate. So long as the plate voltage is higher
than the
and only
screen voltage, the plate current depends primarily on the screen voltage a much
to a slight extent on the plate voltage. This construction makes possible makes
higher amplification than with a triode, and the lower grid-to -plate capacitance
the high gain practicable at radio frequencies without instability.
//ENO DE
(PLATE)
RENT
GRID O
Fig. 1.4. Fundamental circuit r
LOAD
RESISTOR
including indirectly- heated tet- GRID
rode, B and C batteries, and
/
N PUT)
RESISTOR
load resistor in plate circuit. e
+ - 111111 1''+ FIG. 1.4
(iv) Pentodes
Electrons striking the plate with sufficient velocities may dislodge other electrons
when
and so cause what is known as " secondary emission." In the case of tetrodes,
*These are sometimes called biphase half-wave rectifiers.
8 TYPES OF RADIO VALVES 1.3
the plate voltage swings down to a low value under working conditions, the screen
may be instantaneously at a higher positive potential than the plate, and hence the
secondary electrons are attracted to the screen. This has the effect of lowering the
plate current over the region of low plate voltage and thus limits the permissible
plate voltage swing. This effect is avoided when a suppressor is inserted between
screen and plate. The suppressor is normally connected to the cathode as in Fig. 1.5.
Owing to its negative potential with respect to the plate, the suppressor retards the
movements of secondary electrons and diverts them back to the plate.
A valve with three grids is known as a pentode because it has five electrodes. Pen-
todes are commonly used as radio frequency amplifiers and as power amplifiers.
Pentode r -f amplifiers are of two main varieties, those having a sharp cut -off* charac-
teristic and those having a remote cut -off*. Valves having sharp cut -off charac-
teristics are generally used as audio frequency voltage amplifiers and anode bend
detectors, while remote cut -off amplifiers are used as r -f and i -f amplifiers. The
ANODE
(PLAT
SUPPR S PLATE CURRENT
SCREEN. LOAD Fig. 1.5. Fundamental circuit
GRID
RESISTOR
including indirectly- heated pen-
(INPUT) EATER tode, B and C batteries, and
RESISTOR
load resistor in plate circuit,
111111 11,11
FIG. 1.5
remote cut -off characteristic permits the application of automatic volume control
with a minimum of distortion ; this subject is treated in detail in Chapter 27 Sect. 3.
(y) Pentode power amplifiers
Pentode power amplifiers are commonly used in receiving sets to produce a -f power
outputs from about 1 watt up to about 5 watts. They differ from r -f amplifiers in
that no particular precautions are made to provide screening, and they are designed
for handling higher plate currents and screen voltages. In principle, however, both
types are identical and any r -f pentode may be used as a low-power a -f amplifier.
Beam power valves with " aligned " grids do not require a third grid to give
characteristics resembling those of a power pentode ; a typical structure is shown in
Fig. 1.6. Some " kinkless " tetrodes are also used as r -f and i -f amplifiers. All of
these may be treated as being, in most respects, equivalent to pentodes.
(vi) Combined valves
Many combinations of valves have been made. Two triodes are frequently mounted
in one envelope to form a " twin triode." One or more diodes are frequently com-
bined with triodes and pentodes to form second detectors. A combination of triode
and pentode in one envelope is also fairly common, one application being as a fre-
quency changer. Other combinations are triode -hexodes and triode -heptodes, all
of which are primarily intended for use as frequency changers or " converters." In
these, the triode grid is generally connected internally to No. 3 grid in the hexode or
heptode to provide the necessary mixing of the oscillator and signal voltages. A
hexode has four grids while the heptode has five, the outermost of which is a sup-
pressor functioning in the same manner as in a pentode.
In addition to this wide range of combinations, entirely different valves may be
combined in one envelope to save space in very small receivers. This is a practice
which appears to be dying out, particularly as the envelope size becomes smaller.
(vii) Pentagrid converters
Pentagrids are valves having 5 grids, so that they are really heptodes, but the name
pentagrid appears to make a convenient distinction between valves in this group
Sharp cut -off indicates that the plate -current characteristic is as straight as it can be made. A re-
mote cut -off characteristic indicates that the plate current does not become zero until the grid voltage
is made very much negative (usually over 30 volts).
1.3 TYPES OF RADIO VALVES 9
BEAM -
CONFINING
ELECTRODE
CATHODE
GRID
SCREEN
©C RCA
PLATE
Pig. 1.6. Internal structure of type 6L6 or 807 aligned grid beam power valve (diagram
by courtesy of R.C.A.).
(which do not normally require external oscillators) and those of the hexode or heptode
" mixer " type which are used with separate oscillators. Pentagrid converters are
of two main groups, the first of these being the 6A8 type of construction which in-
corporates an oscillator grid and oscillator anode (" anode grid ") as part of the main
cathode stream. The other group comprises the 6SA7, 6BE6 and 1R5 type of con-
struction which has no separate oscillator anode, the screen grid serving a dual pur-
pose. The various types of pentagrid converters are described in detail in Chapter 25.
screen voltage dropping resistor is undesirable with such valves and if the screen is
required to be operated at a lower voltage than the plate, it should be supplied from
a voltage divider having a bleed current of preferably 5 times the nominal screen
current. Alternatively, the screen voltage should be determined for the extreme cases
of zero and twice nominal screen current.
S6
34
33
Is 14 IS 16 IT
FILAMENT VOLTS
FIG. I.7
The heater voltage should be maintained at an average voltage equal to the recom-
mended voltage, thus leaving a margin of plus or minus 10% for line fluctuations
under normal conditions -see Chapter 3 Sect. 1(iv)D. If any wider variation is
required, this will involve decreased maximum grid circuit resistance for a higher
heater voltage and decreased plate current for lower heater voltage.
the normal operating voltages remain at 6.3 volts. In the case of tapped filaments
or heaters the first figure indicates the total voltage with both sections in series.
The second symbol is a letter which is allotted in sequence commencing with A,
except that I and O are not used ; rectifiérs follow the sequence backwards com-
mencing at Z. When all the single letters of a group are exhausted, the system then
proceeds with two lettters commencing with AB ; combinations of identical letters
are not normally used. The single -ended a.c. range has a first letter S while the second
letter may be that of the nearest equivalent in the double-ended range -e.g. type
6SK7 is the nearest single -ended equivalent to type 6K7. Another special case is the
first letter L which is used for lock-in types in the battery range.
mmmmmu
wwwwwwwmw
mwwmwmwímo
ummiiiiiiiiiii
mwmr.E=wwwwommmwn
wNiiiiiiiiiii
wwmwwowwow
w
.....
wwwmimeur.am
OtrwN
O
wwo wo owwwi°trï
wIwNw:ii ww
iiiiiiiiiiiiiiiiiiiiii;iiiiiiiiiiiii=iiiiiiiiiiii
iwO
wwwwwwNNwwwwwEw!iiiiiiiii
wwwwwwww
iiiiii
Nwww
wmww
MN=Iw,
MwwwwwNwr,wIwN
w
mawona w
iMMW
NE
í'wwww
owwwwawwwwwwwwwwr.ww MMil
wrlm
wIstw
:wu
le
u
C.
.í
.
iiiii/pis7
wwwwwtl:ww..íwwl_i . Míww ...
r.
'M!s...iGEmumwow
mw!mmmmu
wmzin
owwwwwwwwwMwommiiíiiiiiiiiiiiiii
wwwm!zmm.r.wwmwowo
iímiAr.aw
_=
mmt.mmm
Uwmuw
..wMn A
NLS7
wilmziwwomm. :
,. iw
.tr.muwwww.wwM
MwwMi
pt..wwur.Mmwwwww
NÌwE
mm ni
wwwwwwww..
,,r.wwwwMMwwMw
tttEME
1www2
e
70
Fig. 1.8.
I/w
MÏwwwNw
NWANwww
iiïii'iïiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
mw.
.Ww'MNwwww
MIA
e0 90
PER CENT
100
OF BASIC VOLTAGE
110
NNE
The final figure denotes the number of " useful elements " brought out to an
external connection.
The envelope of a metal valve, the metal base of a lock -in valve, and internal shield-
ing having its separate and exclusive terminal(s) are counted as useful elements.
A filament or heater counts as one useful element, except that a tapped filament or
heater of two or more sections of unequal rated section voltages or currents counts
as two useful elements. An octal -based glass valve having n useful elements exclusive
of those connected to Pin No. 1 is counted as having n -}- 1 useful elements. Elements
connected to terminals identified as " internal connection, do not use " do not count
as useful elements. Combinations of one or more elements connected to the same
terminal or terminals are counted as one useful element. For example a directly
heated triode with a non -octal base is denoted by 3 ; an indirectly- heated triode,
with a non -octal base is designated by 4 ; a directly- heated tetrode with a non -octal
base is designated by 4. A pentode with the suppressor internally connected to
filament or cathode is numbered as a tetrode. A metal envelope or octal -based glass
triode with an indirectly- heated cathode is designated by 5, a tetrode (or pentode with
the suppressor internally connected) by 6, and a triode -hexode converter usually by 8.
The suffix after the hyphen denotes the type of construction used. In general,
metal valves, lock -in types and miniature types have no suffixes, but octal -based glass
valves types are given the suffix G for the larger glass bulb or GT for the smaller
parallel -sided T9 bulb. The letter M indicates a metal- coated glass envelope and
octal base. X indicates a " low loss " base composed of material having a loss factor
of 0.035 maximum (determination of loss factor to be in accordance with ASTM
Designation D- 150 -41T). The letter Y indicates an intermediate -less base com-
posed of material having a loss factor of 0.1 maximum. The letter W indicates
a military type. The letters, A,B,C,D,E and F assigned in that order indicate a
later and modified version which can be substituted for any previous version but not
vice versa.
SECTION 7 : REFERENCES*
(A) GENERAL READING
1. Henderson, F. E. (book) " An Introduction to Valves, including reference to Cathode Ray Tubes "
(published by Wireless World, London on behalf of The General Electric Co. of England, 1942).
2. Spreadbury, F. G. (book) " Electronics " (Sir Isaac Pitman & Sons Ltd., London, 1947) Chapters
1 -4, 8-10.
3. Reich, H. J. (book) " Theory and Applications of Electron Tubes " (McGraw -Hill Book Co.,
New York & London, 2nd ed., 1944).
4. Chaffee, E. L. (book) " Theory of Thermionic Vacuum Tubes " (McGraw -Hill Book Co., New
York and London, 1933).
5. Emergency British Standard 1106: 1945 " Code of practice relating to the use of electronic valves
other than cathode ray tubes," British Standards Institution, London.
6. " Radio Valve Practice " (The British Radio Valve Manufacturers' Association, London, 1948).
7. Haller, C. E. " Filament and heater characteristics " Elect. 17.7 (July, 1944) 126.
8. " R.M.A. -NEMA Standards for designation system for receiving tubes " (Radio Manufacturers'
Association, U.S.A. -Standard ET -110).
9. Couch, W. " Oxide- coated cathodes " Elect. 22.10 (Oct. 1949) 164.
10. Couch, W. " Oxide cathode theory " Elect. 22.10 (Oct. 1949) 190.
1l. Koller, L. R. (book) " Physics of Electron Tubes " (McGraw-Hill Book Co. 1934) pp. 41 -51.
Additional references will be found in the Supplement commencing on page 1475.
(B) ATOMIC STRUCTURE
12. Stranathan, J. D. " Elementary particles of physics " Elect. 16.8 (Aug. 1943) 122.
13. Darrow, K. K. " Beginnings of nuclear physics " E.E. (Sept. 1945) 315.
14. Stranathan, J. D. (book) " The Particles of Modern Physics " (The Blakiston Co., Philadelphia,
U.S.A. 1943).
15. Cork, J. M. (book) " Radioactivity and Nuclear Physics " (D. van Nostrand Co. Inc., New York,
1947) pp. 13 -18.
16. Smith, J. J. " The atom and its nucleus " E.E. 66.12 (Dec. 1947) 1165.
17. Shoupp, W. E., and H. Odishaw " The nucleus -its structure and reactions " E.E. 67.2 (Feb. 1948)
125.
18. Lapp, R. E., and H. L. Andrews " Atomic structure " Proc. I.R.E. 36.9 (Sept. 1948) 1068, 1070.
Extract from book " Nuclear Radiation Physics " (Prentice -Hall Inc.).
and many others.
For abbreviations of titles of periodicals and references to periodicals see pages 1367-1369.
CHAPTER 2.
VALVE CHARACTERISTICS
Section Page
1. Valve coefficients ... ... 13
2. Characteristic curves ... ... 15
3. Resistance-loaded amplifiers 24
4. Transformer-coupled amplifiers ... 27
5. Triode operation of pentodes ... 34
6. Conversion factors and the calculation of characteristics other
than those published ... ... ... ... 36
7. Valve equivalent circuits and vectors 45
8. Valve admittances ... ... 49
9. Mathematical relationships 57
10. References ... 66
The triode or multigrid radio valve is a device which allows, under certain operating
conditions, an amplified replica of a voltage applied between grid and cathode to
appear across an impedance placed between plate and cathode.
A valve, in itself, does not provide amplification of the applied grid-to- cathode
voltage. The amplified voltage across the load impedance is due to the action of
the valve in controlling the power available from the power supply. The amount
of power which can be so controlled is determined by the operating conditions and
the characteristics of the valve and of its associated circuits.
The maximum voltage amplification which a valve is capable of giving under ideal
conditions is called the amplification factor, generally designated by the Greek symbol
µ (mu). This is not truly constant under all conditions (except for an imaginary
" ideal valve ") and varies slightly with grid bias and plate voltage in the case of a
triode, and is very faz from being constant with most multi-electrode valves.
The amplification factor (µ) is the ratio of the incremental* change in plate
voltage to the incremental change in control grid voltage in the opposite direction,
under the conditions that the plate current remains unchanged, and all other electrode
voltages are maintained constant.
There are two other principal Valve Coefficients, known as the mutual conductance
and the plate resistance (or anode resistance), the values of these also being somewhat
dependent upon the applied voltages.
An incremental change of voltage applied to an electrode may be taken as indicating a change so
small that the curvature of the characteristics may be neglected. For the mathematical treatment of
rate of change, see Chapter 6 Sect. 7(i) and (ii). For treatment of valve coefficients as partial differentials
see Chapter 2 Sect. 9(ix).
13
14 VALVE COEFFICIENTS 2.1
r,,
g , where g, is called the Plate Conductance (see also below).
Other valve coefficients are described below :-
The Mu- Factor, of which the amplification factor is a special case, is the ratio
of the incremental change in any one electrode voltage to the incremental change in
any other electrode voltage, under the conditions that a specified current remains
unchanged and that all other electrode voltages are maintained constant. Examples
are
lL171.02, R02v
The Conductance (g) is the incremental change in current to any electrode divided
by the incremental change in voltage to the same electrode, all other voltages remain-
ing unchanged.
Examples are grid conductance (g,), plate conductance (g,).
Transconductance is the incremental change in current to any electrode divided
by the incremental change in voltage to another electrode, under the condition that
all other voltages remain unchanged. A special case is the grid -plate transconductance
which is known as the mutual conductance. Another example is the plate -grid
transconductance (g).
Conversion transconductance (S ,) is associated with mixer (frequency changing )
valves, and is the incremental change in intermediate -frequency plate current divided
by the incremental change in radio -frequency signal -grid voltage producing it.
The Resistance (r) of any electrode is the reciprocal of the conductance ; for
example plate resistance is the reciprocal of plate conductance,
r, = 1/g.
Perveance (G) is the relation between the space- charge -limited cathode current
and the three-halves power of the anode voltage. It is independent of the electrode
voltages and currents, so long as the three- halves law holds :
rk
G = eb1la
The measurement of perveance is covered in Chapter 3 Sect. 3(vi)E.
tThis is strictly the "variational plate resistance" and must be distinguished from the d.c. plate
resistance.
2.2 CHARACTERISTIC CURVES 15
The method of measuring the plate and grid currents of a triode valve is shown in
Fig. 2.1 in which a tapping on the grid bias battery is returned to the cathode so as
to permit either positive or negative voltages to be connected to the grid. The grid
microammeter and vcltmeter should be of the centre -zero type, or provision made
for reversal of polarity. For more elaborate testing see Chapter 3 Sect. 3.
(i) Plate characteristics
The Plate Characteristic may be drawn by maintaining the grid at some constant
voltage, varying the plate voltage step -by-step from zero up to the maximum available,
and noting the plate current for each step of plate voltage. These readings may then
be plotted on graph paper with the plate voltage horizontal and plate current vertical.
This procedure may be repeated for other values of grid voltage to complete the Plate
Characteristic Family.
The Plate Characteristic Family for a typical triode is shown in Fig. 2.2. It
is assumed that the plate voltage has been selected as 180 volts, and the grid bias
-4 volts. By drawing a vertical line from 180 volts on the Eb axis (point K), the
quiescent operating point Q will be determined by its intersection with the
" E -4= " curve. By referring Q to the vertical scale (Ib) the plate current is
found to be 6mA. The plate resistance at the point Q is found by drawing a tangent
(EF) to the curve for E, _ -4 so that it touches the curve at Q.
The plate resistance (r,) at the point Q is then EK in volts (65) divided by QK
in amperes (6 mA = 0.006 A) or 10 800 ohms.
The amplification factor (µ) is the change of plate voltage divided by the change
of grid voltage for constant plate current. Line CD is drawn horizontally through Q,
and represents a line of constant plate current. Points C and D represent grid volt-
ages of -2 and -6, and correspond to plate voltages of 142 and 218 respectively.
The value of µ* is therefore (218 -
142) plate volts divided by a change of 4 grid
volts, this being 76/4 or 19.
The mutual conductance (g) is the change of plate current divided by the change
of grid voltage for constant plate voltage. Line AB, which is drawn vertically through
Q, represents constant plate voltage. Point A corresponds to 9.6 mA, while point B
corresponds to 2.6 mA, giving a difference of 7 mA. Since points A and B also
differ by 4 volts grid bias, the mutual conductance* is 7 mA divided by 4 volts, which
is 1.75 mA /volt or 1750 micromhos.
The value so determined is not exactly the value which would be obtained with a very small swing,
but is sufficiently accurate for most practical purposes.
16 (i) PLATE CHARACTERISTICS 2.2
--- ®IIIIII-e
----iiiiii1I111 Fig. 2.2. Plate charact-
2
EIWA IINAll eristic family of curves
WAIMIMINIIA for a typical triode.
riireliffirail
4
3
o
FIG.2.2
o
MINIVISMI
1o0
VOLT>'(Cb)
200 300
The plate characteristics of a pentode for one fixed screen voltage are shown
in Fig. 2.3. Owing to the high plate resistance of a pentode the slope of the portion
of the curves above the " knee" is frequently so fiat that it is necessary to draw ex-
tended tangents to the curves as at A, B and Q. A horizontal line may be drawn
through Q to intersect the tangents at A and B at points C and D. As with a triode,
points A and B are vertically above and below Q. The mutual conductance is AB
(4.1 mA) divided by 4 volts change of grid bias, that is 1.025 mA/V or 1025 micromhos.
The amplification factor is the change of plate voltage (CD = 447 volts) divided by
the change of grid voltage (4 volts) or 111.7. The plate resistance is EK /QK, i.e.
180/0.001 65 or 109 000 ohms.
The plate characteristics of a beam tetrode are somewhat similar to those
of a pentode except that the " knee " tends to be more pronounced at high values
of plate current.
The plate characteristics of a screen -grid or tetrode are in the upper portion
similar to a pentode, but the " knee " occurs at a plate voltage slightly greater than
the screen voltage and operation below the " knee " is normally inadvisable due to
instability.
The plate and screen characteristics of a pentode are shown in Fig. 2.4,
from which it will be seen that the total cathode (plate + screen) current for any
fixed grid bias is nearly constant, except at low plate voltages, and that the plate
current increases at the expense of the screen, and vice versa. A pentode is frequently
described as a " constant- current device," but the plate current is not so nearly con-
2.2 (ii) MUTUAL CHARACTERISTICS 17
stain as the combination of plate and screen currents, with fixed grid bias and screen
voltage.
(ii) Mutual characteristicst
The Mutual Characteristics may be drawn by maintaining the plate voltage con-
stant, and varying the grid from the extreme negative to the extreme positive voltage
desired. For any particular plate voltage, there is a negative grid voltage at which
the plate current becomes zero ; this is called the point of plate current cut-off, and
any increase of grid voltage in the negative direction has no effect on the plate current,
which remains zero. If the mutual characteristic were perfectly straight, the point
of plate current cut-off would be at a gr;d voltage of E0/ µ ; in reality, it occurs at
a point slightly more negative, owing to the curved foot of the characteristic.
150
CATHODE CURRE ECIa OV. Ecz= 285V.
ECI =OV.
M PLATE CURRENT
Fig. 2.4. Plate and screen char-
acteristics for a pentode, with
fixed screen and grid voltages,
U
showing also the cathode current
curve which is the sum of the
plate and screen currents at all
SCREEN CURRENT
EC1.O plate voltages.
200 300 400 500
FIG. 24 PLATE VOLTS (Eb)
lb
Eb+ 250
-EC 12 -1 0 -e -4 +EC
f1G. 2.5
The Mutual Characteristics of a triode are shown in Fig. 2.5. Each curve
corresponds to a constant plate voltage. Let P be a point on the Es = 250 curve,
and let us endeavour to find out what information is available from the curves. The
bias corresponding to P is given by R ( -6 volts) and the plate current is given by
S (6 mA). Let now a triangle ABC be constructed so that AP = PC, AB is vertical,
CB is horizontal and point B comes on the Eb = 200 curve.
The mutual conductance is given by AB /BC or 2.32 mA /4 volts, which is 0.580
mA /volt or 580 micromhos. Thus the slope of the characteristic is the mutual
conductance.**
tAlso known as Transfer Characteristics.
**This simple construction assumes that A P C is a straight line. In practice it is slightly curved but
the construction gives a very close approximation to the slope at point P because the slope of the tangent
at P is approximately the slope of the chord joining A and C.
18 (ii) MUTUAI. CHARACTERISTICS 2.2
The amplification factor is given by the change of plate voltage divided by the
change of grid voltage for constant plate current, that is
-
- Ebb Ebi Eby 250 200 12.5.
Ebb - -
CB AE, 4
The plate resistance is given by the change of plate voltage divided by the change
of plate current for constant grid voltage ; that is
ra
Eb, Eby Ebl - Eby 250- 200 -
21 600 ohms. -
AB AIb 2.32 x 10_3
These curves hold only if there is no series resistance in the plate circuit. They
could therefore be used for a transformer-coupled amplifier provided that the primary
of the transformer had negligible resistance. In the present form they could not be
used to predict the operation under dynamic conditions. The static operation point
P may, however, be located by their use.
The mutual characteristics of a pentode, for a fixed screen voltage, are very
similar to those of a triode except that each curve applies to a different value of screen
(instead of plate) voltage. The plate voltage of pentodes having high plate resistance
has only a very minor effect on the plate current, provided that it does not come
below the screen voltage.
Era e3VOLTs
PLATE VOLTS.. 100
SUPPRESSOR VOLTS ..0
t
s
W
EG2=I25=.
''C2 .125Y
-10
FIG. 2.7
-e -e -4 -2
CONTROL GRID VOLTS (Ea)
-e
fIG. 2e
e -4
CONTROL GRID VOLTS
,
EC)
Fig. 2.7. Mutual characteristics of a pentode, with constant plate voltage, and five
fixed screen voltages.
Fig. 2.8. Screen current mutual characteristics of a pentode (same as for Fig. 2.7).
voltages are changed. The grid current commencement point in perfectly hard
battery valves is usually slightly positive, so that they may be operated at zero bias
with negligible positive grid current (Curve B).
A typical valve at its normal negative bias will have negative (or reverse) grid current
which is the sum of gas (ionization) current, leakage current and grid primary emission
current. If the two latter are negligibly small, negative grid current (i.e. gas current)
will be roughly proportional to the plate current, and will increase with the pressure
of gas in the valve. If the plate current is maintained constant, the gas current varies
approximately as the square of the plate voltage ; reduced cathode temperature has
little effect on this relationship (Ref. Al2). See Chapter 1 Sect. 1 for general in-
formation regarding gas current and Chapter 3 Sect. 3(iv)A for the measurement of
reverse grid current.
References to grid current characteristics-Al2, Hl, H2.
-Cc
RIG 2 9
Contact potential is only one of several effects acting on the grid to change the
cross -over point (X) -the others include gas current, grid (primary) emission, leakage,
and the internal electron velocity of emission.
The grid variational conductance is equal to the slope of the grid characteristic
at the operating point. The conductance increases rapidly as the grid voltage is made
less than that corresponding to point Y, irrespective of the value of ionization current,
so that input circuit damping due to the flow of electrons from cathode to grid (i.e.
the positive component of the grid current) occurs in a typical valve even when the
grid current is zero or negative (grid voltages between X and Y in Fig. 2.9). It is
possible for the damping on the positive peaks of applied input voltage to be
quite serious even when the microammeter reads zero. This point is applied
in connection with r.c. triodes in Chapter 12 Sect. 2(iv).
If the valve has a leakage path between grid and cathode, the leakage current
is given by the line OF, which must be added to the gas current to give the grid current
characteristic G. If it has leakage between grid and plate (or screen) the leakage
current is given by the line HJ, which intersects the horizontal axis at a positive voltage
equal to the plate (or screen) voltage ; this also must be added to the other components
to provide the grid current characteristic. The combined leakage currents may be
measured by biasing the grid beyond the point of plate current cut -off provided that
the grid emission is negligibly small-otherwise see below.
Grid emission with a negative grid is the primary emission
of electrons due to grid heating from both cathode and plate
(or screen) ; it gradually increases as the valve becomes
warmer during operation. It increases the total negative grid
current and is included with leakage currents in the
total negative grid current indicated by a valve tester.
For methods of testing to discriminate between the
various components of negative grid current, see
Chapter 3 Sect. 3(iv)A.
If valve is operated with its grid completely open -circuited, the operating point
will be at D, since this is the only point corresponding to zero grid current, unless
the grid characteristic has a second point of zero grid current at a positive grid voltage
(see under grid blocking).
If the valve is operated with zero bias, that is with the grid resistor returned to
cathode, the grid static operating point will be at E, the intersection of the grid current
characteristic and the grid loadline through O. If the valve is one with positive
cross-over point, operating at zero bias, the grid static operating point will occur at F.
In all cases considered above, the operating points are for static conditions, and
any large signal vcltages applied to the grid may have an effect in shifting the operating
point. If the signal voltage swings the grid sufficiently to draw positive grid current,
the operating point will shift as the result of rectified current flowing through R,.
The etlect of negative grid current on the maximum grid circuit resistance and the
operation of a -f amplifiers is described in Chapter 12 Sect. 2(iii) and (iv) ; Sect. 3(iv)C
and (v) ; also Chapter 13 Sect. 10(i).
FIG. 2 ii
Ip
If the suppressor grid has the same
bias control voltage as the control grid,
Ep.250V the control characteristic will be as
2 Ecs100V
Eco.-3V shown by the curve marked " = E
Er" but in this case the plate resist-
ance, although initially slightly lower
for Er, = -3 than for E,, = 0, rises
rapidly as E,I = E,, is made more
negative.
-4o -20 o 20 40
Suppressor Grid
V°4"94 FIG. 2.13
(v) Constant current curves
Fig. 2.13. Suppressor characteristics of The third principal type of valve
a pentode (6SJ7) for fixed control grid, characteristic is known as the " Con-
screen and plate voltages. stant Current " Characteristic. A
2.2 (v) CONSTANT CURRENT CURVES 23
typical family of Constant Current Curves is shown in Fig. 2.15, these being
for a typical triode (type 801). The slope of the curves indicates the ampli-
fication factor, and the slope of the loadline indicates the stage voltage gain.
The operating point is fixed definitely by a knowledge of plate and grid supply volt-
ages, but the loadline is only straight when both plate and grid voltages follow the
same law (e.g., both sine wave). Distortion results in curved characteristics, so that
this form of representation is not very useful except for tuned -grid tuned -plate or
" tank- circuit " coupled r-f amplifiers. Constant Current Curves may be drawn
by transferring points from the other published characteristics. For a full treatment
the reader is referred to FIG. 2.14
CURVE CHARACTERISTIC
operation of vacuum tubes as Class I
general direction showed rapid changes in the rate of change, and usually at least
one temporary reversal of direction.
It was found that minimum drift occurred for plate currents between 0.1 and 1.0 mA
for indirectly -heated types, or between 10 and 100 µA for small filament types.
This drift occurs in diodes and all types of amplifying valves, being particularly
noticeable in its effects on high -mu triodes (on account of the short grid base) and on
power amplifiers (on account of the decrease in maximum power output). In direct -
coupled amplifiers this drift becomes serious, the first stage being the one most affected.
Most of the drift usually occurs during the first hundred hours of operation. If
stability is required it is advisable to age the valves for at least 2 days, but in some
cases this does not cure the rapid drift. Reference Al2, pp. 730 -733.
N:lz
120
N4á9= 440
-=`\
/:V\\Ní1C
/r!ISM111:21
ÌZIO\:9\\'2._
1
'M"ir,,\- 1 :.
SLAP" STAGE GAIN
. .
o .
a===a=_:ZZAí
IMISIRMMIBIFIVIM
-MIMI
Izo
Inn
400
l
000
E
000
0
0 100 200
0 200 FIG .2 1E PLATE VOLTS (,Eb
FIG. 215 PLATE VOLTS
Fig. 2.15. Constant current characteristics for a typical small transmitting triode (801).
Fig. 2.16. Triode plate characteristics with loadlines for five va.'ues of load resistance.
(i) Triodes
When there is a resistance load in the plate circuit, the voltage actually on the plate
is less than that of the supply voltage by the drop in the load resistor,
Ee = E55 RLIS -
This equation may be represented by what is known as a Load Line on the plate
rharacteristics. Since the load is a pure resistance it will obey Ohm's Law, and the
relationship between current and voltage will be a straight line ; the loadline will
therefore be a straight line.
2.3 RESISTANCE- LOADED AMPLIFIERS 25
The static operating point is the intersection of the loadline and the appropriate
characteristic curve. Fig. 2.16 shows several loadlines, corresponding to different
load resistors, drawn on a plate characteristic family. Zero load resistance is indicated
by a vertical loadline, while a horizontal line indicates infinite resistance.
A loadline may be drawn quite independently of the plate characteristics, as in Fig.
2.17. The point B is the plate supply voltage Ebb (in this case 300 V); the slopes of
the loadline AB = -1/R L and therefore AO = Ebb /R t (in this case 300/50 000 =
0.006 A = 6 mA). The voltage actually on the plate can only be equal to Ebb when
the current is zero (point B). At point A the voltage across the valve is zero and the
whole supply voltage is across R L; this is what happens when the valve is short-
circuited from plate to cathode. The plate current (Eb 5/R L) which flows under these
conditions is used as a reference basis for the correct operation of a resistance coupled
amplifier (Chapter 12). As the plate voltage, under high level dynamic conditions,
must swing about the operating point, the latter must be somewhere in the region of
the middle of AB ; the plate current would then be in the region of 0.5 Eb., /R L and
the plate voltage 0-5 Ebb -in other words, the supply voltage is roughly divided equally
between the valve and the load resistance. Actually, the operating point may be
anywhere within the limits 0.4 and 0.85 times Ebb /R ¿see Chapter 12 Sect. 2(vi)
and Sect. 3(vi).
In most resistance-loaded amplifiers, the plate is coupled by a capacitor to the grid
of the following valve, which has a grid resistor R, to earth. This resistor acts as a
load on the previous valve, but only under dynamic conditions. In Fig. 2.18 the
loadline AB is drawn, as in Fig. 2.17, and the operating point Q is fixed by selecting
the grid bias (here -6 volts). Through Q is then drawn another line CD having a
slope of -(l /R L + 1/R9); this is the dynamic loadline, and is used for determining
the voltage gain, maximum output voltage and distortion (Chapter 12).
12
"IS 217
IOO
Eb
200 300 VOLTS 0
FIG. 2 -IS
100 200
b)
PLATE MOLTS (E
The dynamic characteristic is the effective mutual characteristic when the valve
has a resistive load in the plate circuit t. While the slope of the mutual characteristic
is g,a or s /r
(r, + R,) being more nearly constant than r
the slope of the dynamic characteristic is µ /(r, + R L). Owing to
the dynamic characteristic is more
nearly straight than the mutual characteristic.
*The slope of AB is negative since the plate voltage is the difference between the supply voltage and
the voltage drop in R L, and the inverted form (1/R i) is due to the way in which the volve characteristics
are drawn wait current vertically and voltage horizontally. The slop., of AB is often loosely spoken
of as being the ees stanre of R L ,the negative sign and inverted form being understood.
+It does no make allowance for the fulkwing grid resistor, and does not therefore correspond to
the dynamic loadline. It is, of course, possible to derive from the dynamic loadline a modifica dynamic
characteristic which does make allowance for the grid resistor.
25 (i) TRIODES 2.3
DYNAMIC CHARACTERISTIC /
FOR Eb a 250.
RL IMil, // /
U
/ // / / /
SLOPE
Eb 3.250
/
/ 20O/ /
/ / 150 / //
/._
' -.2'5M
2
A
Fig. 2.19. Triode dynamic characteristic (solid line) for resistance loading.
Fig. 2.20. Illustrating power dissipation in a resistance -loaded triode.
(ii) Pentodes
Pentodes with resistive loads are treated in the same manner as triodes, the only
complication being the screen voltage which must be selected at some suitable value
2.3 (ii) PENTODES 27
and maintained constant (Fig. 2.21). The operating point as an amplifier will norm-
ally, as with a triode, be in the region of the middle of the loadline so that the voltage
across the valve and that across R L will be approximately the same. The only
special case is with very low values of R L (e.g. 20 000 ohms) where grid current
occurs at approximately Ei,, = 0, thereby limiting the useful part of the loadline.
ES e-a VOLTS
SCREEN VOLTS.mo
VOLTS.O
CONTROL
A
tCRIO VOLTS -4
NA ECO
z e
-10
4 -12
u 8
e
4
iy
0
0 100 SOO 0
FIG. 2.21 PLATE VOLTS go FIG. 2.22
100 200
E
PLATE VOLTS.
400
With any value of screen voltage, and any value of load resistance, it is possible
to select a grid bias voltage which will give normal operation as an amplifier. With
load resistance of 0.1 megohm and above, pentodes give dynamic characteristics
which closely resemble the shape of triode dynamic characteristics with slightly
greater curvature at the lower end ; at the upper end, provided that the screen voltage
is not too low, the pentode has a curved portion where the triode runs into grid current.
The top bend of the pentode dynamic characteristic is often used in preference to
the bottom bend for plate detection -see Chapter 27 Sect. 1(ii)C.
For further information on resistance coupled pentode amplifiers, reference should
be made to Chapter 12 Sect. 3.
-e F 0 -So
-f -I
CONTROL GRID VOLTS EC, A -4.S
-e-o
-TS
0
r s! -e0
12
Fig. 2.23. Triode plate characteristics and loadlines with transformer- coupled load,
allowing for the resistance of the primary winding.
Fig. 2.24. Plate characteristics of typical remote cut-off pentode with fixed screen and
suppressor voltages.
Remote cut -off r-f pentodes are similar, except that the mutual characteristics are
curved, and the distortion is greater. Fig. 2.24 shows the plate characteristics of a
typical remote cut -off pentode, with Et, = 250 volts. Two loadlines (AQB, A' Q' B')
have been drawn for grid bias voltages of -3 and -12 volts respectively, with a slope
corresponding to a load resistance of 200 000 ohms, as for an i-f amplifier. This
application of the loadline is not entirely valid, although it gives some useful informa-
tion, since the tuned plate circuit acts as a " flywheel " to improve the linearity and
reduce the distortion. This is a case in which constant current curves could be
used with advantage. However, the ordinary plate characteristics at least indicate
the importance of a high Q (high dynamic resistance) second i-f transformer if it
is desired to obtain high output voltages at even moderately high negative bias volt-
ages ; a steeper loadline would reach plate current cut -off at the high voltage peak.
6 BA6
El,. 250 V.
o
-40 -30 -20
-20 -10 0
FIG. 2 25
Fig. 2.25. Plate current characteristic of remote cut-off pentode with " sliding screen."
The straight lines indicate the mutual conductances at several points.
resistor designed to provide the correct screen voltage for the normal (minimum bias)
operating condition. The screen requires to be by- passed to the cathode.
The same method may be used with a sharp cut-off pentode to provide a longer
grid base. This does not make it possible to obtain the same results as with a pro-
perly designed remote cut -off pentode, although it does increase the maximum input
voltage which can be handled with a limited distortion. It is important to remember
that the extended plate current characteristic curve obtained by this method cannot
be used to determine the dynamic slope, since the latter is higher than would be cal-
culated from the characteristic. This is demonstrated in Fig. 2.25 which shows the
" sliding screen " plate current characteristic, with straight lines drawn to indicate
the mutual conductance at several points.
The procedure for deriving the " sliding screen " plate current characteris-
tic from the fixed voltage data is as follows -
Let plate and screen current curves be available for screen voltages of 50, 75, 100
and 125 volts (Fig. 2.26) and take the case with a series screen resistor (R,) of 250 000
ohms from a supply voltage of 300.
E5, Earoa* Iss Point Eat Point lal
50 V 250 V 1.0 mA A -0.1 E 3.7 mA
75 225 0-9 B -1.7 F 3-15
100 200 0-8 C -3.3 G 2-6
125 175 0.7 D -5.2 H 2.1
*The voltage drop in the screen resistance = 300 -Ea.
"/n m Edrov /Rs
¡Derived from the screen characteristics and transferred to the plate characteristics.
:Derived from the plate characteristics.
(p) Cathode loadlines
The static operating point with cathode self bias may be determined graphically
by the use of the mutual characteristic. The mutual characteristic of a triode shown
30 (v) CATHODE LOADLINES 2.4
in Fig. 2.27 applies to the voltage between plate and cathode -the total supply voltage
will be greater by the drop in the cathode resistor Rk.
Through O should be drawn a straight line OD, having a slope of -1 /R
The point P where OD intersects the curve corresponding to the plate -to- cathode
ohms.
voltage (here 250 V) will be the static operating point, with a bias -E,1 and plate
current I,.,.
In the case of pentodes, with equal plate and screen voltages, the " triode " mutual
characteristic should be used, if available. With the plate voltage higher than the
screen voltage, the triode mutual characteristic may be used as a fairly close approxi-
mation, provided that the triode curve selected is for a voltage the same as the screen
voltage.
Alternatively, pentodes may be treated as for triodes, except that the slope of OD
should be
1 I6
Rk Ia -I- I'2
where Ib and Ie, may be taken to a sufficient degree of accuracy as being the values
under published conditions. The plate current (Iá) may then be read from the curve,
and the screen current calculated from the ratio of screen to plate currents.
For the use of cathode loadlines with resistance coupled triodes and pentodes,
refer to Chapter 12.
PLATE VOLTS -
SOO
SUPPRESSOR VOLTS -o
ECI
-IO -P -1 -4 Z
CONTROL-GRID VOLTS(ECI
FIG.2.2C FIG. 2.27
Fig. 2.26. Plate and screen current characteristics of pentode illustrating procedure
for deriving " sliding screen" characteristics.
Fig. 2.27. Triode mutual characteristics with cathode bias loadline OD.
For convenience in application, the horizontal and vertical scales should be the
same as in the valve characteristics to which the loadline is to be applied. For example,
if on the plate characteristics one square represents 1 mA in the vertical direction and
25 V in the horizontal direction, the sanie proportion should be maintained for the
elliptical loadline. Having drawn the ellipse for any convenient value of E0, it may
be expanded or contracted in size, without changing its shape (that is the ratio of the
major to the minor axis when both are measured in inches).
Cf"
FIG. 2 29
Fig. 2.29. Resistive loadline (R L) ; inductive loadline (wL); and elliptical resultant
for R L in series with wL (dashed curve).
FJ = FG + FH. With similar procedure in the other three quadrants, the com-
bined loadline is shown to be an ellipse CJADB which is tilted, or rotated in the
clockwise direction as compared with the original ellipse. The maximum voltage
drop is greater than that across either R L or L alone, as would be expected.
32 (vi) WITH REACTIVE LOADS 2.4
If an elliptical loadline is known, as for example the dashed ellipse of Fig. 2.29,
its series components may readily be determined. Mark points A and B where the
ellipse reaches its maximum and minimum current values, then draw the line AB ;
the slope of AB gives R L. Mark O as the centre of the line AB ; draw COD hori-
zontally to cut the ellipse at points C and D.
Then wL = E,/I, ohms,
where E, = voltage corresponding to length OD
and I. = current (in amperes) corresponding to max. vertical height of
ellipse above line COD.
Alternatively
Length of horizontal chord of ellipse through O, in volts
coL
Maximum vertical extent of ellipse, in amperes
(b) Resistance and inductance in parallel
When the load is a resistance R L in parallel with an inductive reactance cL, the
maximum voltage across both will be E and the resistive loadline and reactive ellipse
may be drawn as for the series connection. In this case, however, the currents have
to be added. In Fig 2.30 the maximum current through RL is CK (corresponding
to +E,), while the maximum current through L is 0E. When the voltage is zero
and increasing, the current through RL is zero, and that through L is the
minimum value OP ; point P is therefore on the desired loadline. When
the voltage is its positive maximum (OC), the current through RL is CK and
that through L is zero ; point K is therefore on the desired loadline. Simil-
arly with points E and M. At an intermediate value, when the voltage is negative
and approaching zero (OR), the current through R L is RS, and that through L is
RT ; the total current is therefore RT -+ RS = RW, and W is on the desired loadline.
The loadline is therefore the ellipse PKEMW.
If an elliptical loadline is known, as for example the dashed ellipse of Fig. 2.30,
its parallel components may readily be determined. Mark points K and M where
the ellipse reaches its maximum and minimum voltage values, then draw the line KM ;
the slope of KM gives R L. Mark O as the centre of the line KM ; draw EOP vertic-
ally to cut the ellipse at points E and P.
Then cuL = E, /I, ohms,
where E, = voltage difference between points O and K,
and I, = current corresponding to length 0E, in amperes.
Alternatively
Maximum horizontal length of ellipse, in volts
aL =
Length of vertical chord of ellipse through 0, in amperes
(c) Resistance and capacitance
A similar shape of loadline is obtained when the inductance is replaced by a capacit-
ance of equal reactance, except that the direction of rotation is opposite.
2.4 (vi) WITH REACTIVE LOADS 33
d4
2
24
100 200 300 400
F10.2.51 PLATE VOLT!
Fig. 2.31. Triode plate characteristics with elliptical loadlines corresponding to resistance
25 000 ohms in series with reactance of 18 000 ohms.
in each direction. The examples taken have all been based on an arbitrary current
(I,) or voltage (ET), which may be made larger or smaller as desired. In Fig. 2.31
there is shown the elliptical loadline corresponding to a resistance of 25 000 ohms in
series with a reactance of 18 000 ohms, on triode plate characteristics with E6 250
volts, E _
-10 volts and peak grid amplitude ED m = 8 volts.
100
-15
-20
Fig. 2.32. Beam power amplifier plate characteristics with elliptical loadline corres-
ponding to a resistance of 4750 ohms in parallel with a reactance of 23 000 ohms.
Fig. 2.32 shows a typical beam power amplifier with an elliptical loadline with a
resistive load of 4750 ohms shunted by a reactance of 23 000 ohms. The plate volt-
age is 250 volts, grid bias -
12.5 volts, and grid swing from 0 to -25 volts..
34 (i) TRIODE OPERATION OF PENTODES 2.5
The assumption (or approximation) which must be made is -That the plate current
of a pentode valve does not change as the plate voltage is increased from the same voltage
as that of the screen up to the voltage for pentode operation.
This assumption means, in essence, that the plate resistance is considered to be
-a
infinite reasonable approximation for most r-f pentodes, and not seriously in
error for power pentodes and beam power valves.
In this typical example we can take the published characteristics, and assume that
the plate current and transconductance are the same for 100 as for 250 volts on the
plate. From then on the procedure is exactly as in the previous example. It is
important to note that the calculated triode characteristics only apply for a triode
plate voltage of 100 volts and a grid bias of -3 volts.
Example 3 : To find the screen transconductance under the conditions of Example 1.
From eqn. (2) we may derive the expression
-
g, = g, g,,, = 1482
-- 1185
= 297 tnicromhos.
This could equally well have been derived from eqn. (4) or (5).
-
The amplification factor of the plate of a pentode valve with respect to its screen
grid may be calculated from the expression
-
F'ui.p Pat92F,02v
where µ,,., = pentode amplification factor
(7)
The same result may be obtained with pentodes, provided that both the grid and
screen voltages are varied in the same proportion as the plate voltage. This result
is the basis of Valve Conversion Factors, so that we must always remember that
their
use is restricted to cases in which all the electrode voltages are changed in the same
proportion.
Let F, be the factor by which all the voltages are changed (i.e. grid, screen, and
plate), and let Ib' be the new plate current.
Then Ib' CC (F,.Eb)9 z (5)
But b' = F;.Ib
where F, is the factor by which the plate current is changed.
(6)
Therefore F,.Ib CC (F,.Eb)aiz
From the combination of (4) and (6) it will be seen that
F = F (7)
Now the power output is proportional to the product of plate voltage and plate
current on that
P, cc EbI b
(8)
and P.; cc (F,.Eb) (F +I b) (9)
so that P,, cc F,.F, (Eb.Ib) (10)
cc F,.F; (P,). (11)
We may therefore say that the power conversion factor F,, is given by the expression
F, = F, F; (12)
Therefore F, = F,5iz. (13)
The mutual conductance is given by
change of plate current
change of grid voltage
Therefore F,,,, = F; /F, = F,'1z /F, = F i. (14)
The Plate Resistance is given by
change of plate voltage
rn
change of plate current
Therefore F, = F, /Fi = F, /F,a"z = F,-á. (15)
This also applies similarly to the load resistance and cathode bias resistance.
We may therefore summarize our results so far :-
F; = F,'1z (7)
F = F,51' (13)
F,T = F,}-1 (14)
F, = F, (15)
These are shown in graphical form on the Conversion Factor Chart (Fig. 2.32A).
i
04 0.6 0.8 1.0 ao 25
10
/m
C==C.-S=r --.
iOw
.
=..w.C: °. 3
C:
i'i iiiii
.w:
M .MM Cw
M=..,_=_r=W_:::
.
=
111:11'3111:: . =..NM..=
M.MMP MMMW .
2
1..' =WAMISm=.CC
1
..2
11111
101111111 1111==M
111111=II=II=
._
°_--
1i1!.111118IIIIIIIIIIIIIIIIIIII=/,=II
lÜÌ iÚiii:l:iiÜii Ìi ÌIÌÌt ÌÌÌÌÌ
1111111111111111111.0111111111 I111111111J/..M1
1111111 11.11.1111111ì:11111111111I/.MiMM
,.011111111111111111111111=:.!111!/i.101
- _
_'=______:
0.e ÿ13.=-=-@__--__e! e:e: s' f:
rp;-r.i
_- __=--__-
--°e_=°
a=-=a to
_- _?.
- rEco o-a
x---jç-==3
-'...'i
__--
___= _--_-Z-
i1= °C _
° 6 :=6:ieË M ....... -_=-=_=
ww.ww
0.6
°"' .ii
@
w----------- -- w.M w. MI
:ce. F
: 2
04 -- --'= :£_==-=o>=s__ 004
o _ =
E
0-3
..
c_ - APPLIES TO PLATE, SCREEN =03
& CONTROL GRID VOLTAGE .
F¿ - APPLIES TO PLATES SCREEN u
M.mf
.I CURRENT. 1¡
FP - =,2
pZ
IID.I
APPLIES TO POWER OUTPUT ;
11111 1.11
IIII1111111.
1111111111UP
11 FT - APPLIES TO PLATE RESIS- IN
11 1 /IIIIIIIIIIn11 TANCE A LOAD RESISTANCE in
IIIIIIP./1IIIIIIIUIuI11111 Fg-
IIIIV'AniuIIIIIIIIIIIIIIIIIII
III r IInl IIIiIIII11111111111n-
APPLIES TO MUTUAL
CONOUTANCE.
-
J
IN
J.
111'd11IIMIIII
11:1111111 1 111IIIII
IInI1111111111 RE..
111111111111111111111M1111.1u......
d1nnu1I uu111u11111111IIIIIIIIII1111M1
0.1 0.1
0.4 06 0.8 1.0 1.5 20 2.5
VOLTAGE CONVERSION FACTOR (Fe]
The example given below is a straightforward case of a pentode valve whose charac-
teristics are given for certain voltages and which it is desired to operate at a lower
plate voltage.
Plate and screen voltage 250 volts
Control grid voltage -15 volts
Plate current 30 mA
2.6 (ii) USE OF VALVE CONVERSION FACTORS 39
Screen current 6 mA
Mutual conductance 2,000 pathos
Power Output 2.5 watts.
It is required to determine the optimum operating conditions for a plate voltage
of 200 volts.
The Voltage Conversion Factor (F,) = 200/250 = 0.8.
The new screen voltage will be 0.8 x 250 = 200 volts.
The new control grid voltage will be -(0.8 x 15) = -12 volts.
Reference to the chart then gives the following :
Current Conversion Factor (Fi) 0.72
Mutual Conductance Conversion Factor (F,,,,) 0.89
Power Output Conversion Factor (F2) 0.57
The new plate current will be 0.72 x 30 = 21.6 mA.
The new screen current will be 0.72 x 6 = 4.3 mA.
The new mutual conductance will be 0.89 x 2000 = 1780 µmhos.
The new power output will be 0.57 x 2.5 = 1.42 watts.
There are two effects not taken into account by conversion factors. The first
is contact potential, but its effects only become serious for small grid bias voltages.
The second is secondary emission, which occurs with the old type of tetrode at low
plate voltages ; in such a case the use of conversion factors should be limited to regions
of the plate characteristic in which the plate voltage is greater than the screen voltage.
With beam power amplifiers the region of both low plate currents and low plate
voltages should also be avoided for similar reasons.
The application of conversion factors to resistance -capacitance -coupled triodes
and pentodes is covered in Chapter 12 Sect. 2(x) and Sect. 3(x) respectively.
FIG. 2 .33
35
300
fi 25
e
r 200
s
9
IS
100
50
Fig. 2.33. Zero bias plate characteristics for type 807 beam power amplifier with six
values of screen voltage (Ref. E2).
Greater accuracy in the use of conversion factors over a wide range of screen voltages
may be obtained, if curves are available for zero bias at a number of different screen
voltages as in Fig. 2.33 (Ref. E2).
When the plate, screen, and grid voltages of a pentode or beam power amplifier
are multiplied by the same voltage conversion factor, the ratio of the plate current
at a given grid bias to that at zero bias does not change. In order to convert a given
family of plate characteristics to a new screen voltage condition, it is therefore only
necessary to have a zero-bias plate characteristic for the screen voltage of interest.
40 (ii) USE OF VALVE CONVERSION FACTORS 2.6
Example
Suppose that the family of plate characteristics shown in Fig. 2.34, which obtains
for a screen voltage of 250 volts, is to be converted for a screen voltage of 300 volts.
The zero-bias plate characteristic for E,_ = 300 volts, which is shown in Fig. 2.33,
is replotted as the upper curve in Fig. 2.35.
Since all bias values shown in Fig. 2.34 must be multiplied by 300/250 = 1.2,
corresponding plate characteristics for the new family obtain for bias values that are
20 per cent. higher than those shown in Fig. 2.34. Consider the conversion of
-10-volt characteristic of Fig. 2.34. At a plate voltage (Eb of 250 volts in Fig. 2.34,
AB /AC = 100/187 = 0.535. On the new characteristic in Fig. 2.35 which corres-
ponds to a bias of -12 volts, A'B' /A'C' must also equal 0535 at Eb = 300 volts.
Therefore, A'B' = 0.535 x A'C'. From the given zero -bias characteristic of Fig.
2.35, A'C' = 244 at Eb = 300 volts ; hence A'B' = 131 milliamperes. At Eb = 200
volts in Fig. 2.34 DE /DF = 98/183 = 0.535. Therefore, at Eb = 200 x 1.2 =
240 volts in Fig. 2.35, D'E' = 0.535 x 238 = 127 milliamperes. This process
is repeated for a number of plate voltages and a smooth curve is drawn through the
points on the new characteristic.
FIG. 2.34
PLATE VOLTS
Fig. 2.34. Plate characteristics for type 807 with fixed screen voltage and eight values
of grid voltage (Ref. E2).
The factor 0.535 cari be used for the -10 -volt characteristic at plate voltages
greater than that at which the knee on the zero -bias characteristic of Fig. 2.34 occurs ;
for plate voltages in the immediate region of the knee, a new factor should be deter-
mined for each point. The plate characteristics of Fig. 2.34 should not be converted
to the left of the dashed line of Fig. 2.34 because of space -charge effects. This
limitation is not a serious one, however, because the region over which the valve
usually operates can be converted with sufficient accuracy for most applications. The
converted plate characteristic of Fig. 2.35 for E,1 = -30 volts was obtained in a
similar manner to that for E,, = -12 volts.
The curves of Fig. 2.35 were checked under dynamic conditions by means of a
cathode -ray tube and the dotted portions show regions where measured results de-
parted from calculated results.
(iii) The calculation of valve characteristics other than those pub-
lished
It is frequently desired to make minor modifications in the operating conditions
of a valve, such as by a slight increase or decrease of the plate voltage, change in grid
2.6 (iii) CALCULATION OF CHARACTERISTICS 41
bias or load resistance. It is proposed to describe the effects which these changes
will have on the other characteristics of the valve.
The procedure to be adopted is summarized below :-
FIG. 2.35
4
PLATE VOLTS (Eb)
Fig. 2.35. Derived plate characteristics for type 807 with different screen voltage,
making use of Figs. 2.33 and 2.34 with conversion factors (Ref. E2).
lished characteristics to find the maximum plate dissipation ; calculate the maximum
plate current which can be permitted at the desired new plate voltage ; select a suit-
able plate current for the particular application (which must not exceed the maxi-
mum) ; and refer to the curves to find the grid bias to give the desired plate current.
If the valve is a power amplifier, the load resistance may be determined by one of
the methods described in Chapter 13 [e.g. triodes Sect. 2(iii) ; pentodes Sect. 3(iii)A].
Triode with resistor in plate circuit : Use conversion factors, with adjustments
as required in accordance with the method given in (iv) below.
Pentode or beam power amplifier If curves are available for the published
:
value of screen voltage, use the method in (iv) below to obtain the characteristics
for a plate voltage such that, when conversion factors are applied, the plate voltage
is the desired value. For example, if curves and characteristics are available for
plate and screen voltages of 250 volts, and it is desired to determine the characteristics
for a plate voltage of 360 volts and screen voltage of 300 volts firstly determine the
:
characteristics for a plate voltage of 300 and screen voltage of 250 ; then apply voltage
42 (iv) EFFECT OF CHANGES IN OPERATING CONDITIONS 2.6
conversion factors of 1.2 to the plate, screen and grid voltages so as to provide the
desired conditions.
If curves are available for the new value of screen voltage, use conversion factors
to bring the screen voltage to the desired value, then apply the method below to adjust
the plate voltage, load resistance and grid bias.
Fig. 2.36. Plate characteristics of power pentode illustrating effect of change of plate
voltage.
Since the power output is proportional to the area of the triangle under the loadline,
it is also proportional to the value of the load resistance, all triangles having ML as a
common side. It may readily be shown that
Eby - Emin
L Iman - Iby
R= Eb2 - Emin
and
Therefore
Imax - 162
-
R¿ Ebz Emin Imax Ib, - (17)
RL E61 -Emin Imax 162 -
which is also the ratio of the output powers. If 162 = I6y or the rise of plate current
is neglected as an approximation, then
R¿ E6z Emin - (18)
RL Eby Emin-
As an example, apply this to type 6V6 -GT under the following conditions- -
Published Desired
Condition Condition
Plate voltage 250 300 V
Screen voltage 250 250 V
Grid voltage -12.5 -12.5 V
Load resistance 5000 (see below) ohms
Plate current (161) 47 48* mA
Peak plate current (Imax) 90* 90* mA
Min. plate current (1,,,i,,) 8* 8* mA
Min. plate voltage (Emin) 35 35 V
Power output 4.5 (see below) W
From curve.
Using equation (17) --
RL -
300 35 90 -47 265 43
1.26
R 250 -35 90 -48 21542
whence Rí = 126 x 5000 = 6300 ohms.
The increase of power output is in proportion to the increase in load resistance.
i.e. P = 4.5 x 126 == 5.66 watts.
This method is remarkably accurate when there is very small rectification in the
plate circuit, as is usu,lly the case with power pentodes. With beam power amplifiers
of the 6L6 and 807 class, in which the rectification is considerable (strong second
values of I,
harmonic component), the " corrected " loadline should be used as a basis, and the
¡
l6, and Em should be those corresponding to the corrected loadline.
If the rise in plate current (Alb) is considerable, the point P' will be above the
centre point of the loadline MH, and there will be an appreciable amount of second
harmonic distortion ; this may be reduced to zero (if desired) by increasing the load
resistance slightly.
The voltage gain is also affected by the plate resistance, but this is quite a secondary effect unless
the plate resistance is less than 0.5 megohm. In most remote cut-off pentodes the plate resistance falls
rapidly as the bias is decreased towards the minimum bias, but this is more than counterbalanced by the
rise in mutual conductance.
2.7 (i) CONSTANT VOLTAGE EQUIVALENT CIRCUIT 45
i IG. 237
µR L
and -EL = voltage gain r, +RL
If the load
F.,
is made up of a resistor R L and an inductor X L in series
Scalar Values
-
Complex Values
ZL = RL +jXL 1/R L' + X L'
0E, = (r,, RL -{- jX L) I, V(r, + R L)' f- X L'.I
RL' + XL'
EL µ(RL + jXL)
E, -- r, + R +1XL v(r, + R L)' + X L'
and similarly for any other type of load.
46 (i) CONSTANT VOLTAGE EQUIVALENT CIRCUIT 2.7
The interelectrode capacitances are shown in the equivalent circuit of Fig. 2.38,
and may be taken as including the stray circuit capacitances. This circui may be
applied at frequencies up to nearly 10 Mc /s, beyond which the inductances of the
leads and electrodes become appreciable. It may also be applied to a screen grid
(tetrode) or pentode, provided that the screen is completely by- passed to the cathode ;
in this case C, k becomes the input capacitance (C,,,. k + C,,.,,) and C, k becomes
the output capacitance (C, to all other electrodes).
FIG. 2.3e
FIG. 2.39.
FIG. 2.40
In the constant voltage generator equivalent circuit, the current varies with load
impedance and plate resistance ; in the constant current equivalent circuit, the voltage
across the load and plate resistance varies with load impedance and plate resistance.
A constant current generator equivalent circuit, in which account is taken of
capacitances, is shown in Fig. 2.40. This circuit may be applied at frequencies
up to nearly 10 Mc/s, beyond which the inductances of the leads become appreciable.
It will be seen that C,, (which may be taken to include all capacitances from plate
to cathode, and the output capacitance of a pentode) is shunted across both r, and ZL.
In the case of a resistance-capacitance coupled stage, ZL would be the resultant of
R t and R, (following grid resistor) in parallel.
Maximum power output is obtained when the valve works into a load resistance
equal to its plate resistance provided that the valve is linear and completely distortion -
less over the whole range of its working, and also that it is unlimited by maximum
electrode dissipations or grid current. In practice, of course, these conditions do
not hold and the load resistance is made greater than the plate resistance.
At frequencies of 10 Mc /s and above, the effects of the inductance of connecting
leads (both internal and external to the valve) become appreciable. Although it
is possible to draw an equivalent circuit for frequencies up to 100 Mc /s, in which
2.7 (ii) CONSTANT CURRENT EQUIVALENT CIRCUIT 47
each capacitance is split into an electrode part and a circuit part (Ref. B21 Fig. 38)
the circuit is too complicated for analysis, and the new circuit elements that have
been introduced cannot be measured directly from the external terminals alone.
At frequencies above about 50 Mc /s, transit time effects also become appreciable.
The circuit which is commonly used for frequencies above 50 Mc /s is Fig. 2.47 in
which the valve is treated as a four terminal network with two input and two output
terminals. This is described in Sect. 8(iii)e.
riG.2 41
FIG. 2.42
FIG. 2.43
FIG. 2.44
Fig. 2.42. Equivalent circuit and vector diagram of resistance -loaded valve.
Fig. 2.43. Vector diagram of valve with resistance load and capacitance from grid to
plate.
Fig. 2.44. Vector diagram of valve with partially inductive load and capacitance from
grid to plate.
When the equivalent circuit includes more than one mesh, it is usual to proceed
around each mesh in turn, using some impedance, common to both, as the link
between each pair of meshes. For example Fig. 2.43 shows a valve with a capacitor
C,,,, from grid to plate, and a resistive load. Firstly, set down EL in any convenient
direction (here taken horizontally to the right) and IL in the same direction; then
draw I, leading by approximately 90° (actually I, leads E, by 90 °) and complete the
parellelogram to find the resultant current I, ; then draw I,r, in the same direction
as I, and complete the parallelogram to find the resultant Ep this completes the
first mesh. Finally take E° along µE, and complete the parallelogram to find the
resultant of E0 and EL, which will be E0.
It will be noted that this procedure gives different vector relationships from those
derived for the simpler case, this being the result of the different approach. With
respect to E0, both E, and EL are nearly in phase ; with respect to µE0, both I,r,
and EL are nearly in phase.
2.7 (iii) VALVE VECTORS 49
If the load is partially inductive (Fig. 2.44) the plate current and Irr, lag behind
EL and the resultant .iE, is determined by the parallelogram; E, and EL combine
to give the resultant E,; I, leads E, by 90 °, and IL. is determined by completing
the parallelogram of which I is one side and Ír the resultant.
FIG. 2.45
Fig. 2.45. Vector diagram of valve with partially capacitive load and capacitance
from grid to plate.
With a partially capacitive load (Fig. 2.45) the plate current and /Dr, lead EL, and
the resultant µE, is determined by the parallelogram; E, and EL combine to give
the resultant E,; I, leads E, by 90 °, and IL is determined by completing the parallelo-
gram of which I, is one side and Ir the resultant.
Component
: -
As with any other impedance (see Chapter 4 Sect. 6) it may be divided into its
various components
Normal Reciprocal
Resistive Grid input resistance (r,) conductance (g,)
Reactive Grid input reactance (X,) susceptance (B9)
Resultant Grid input impedance (Z,) admittance (Y,)
Normal values are treasured in ohms, while reciprocal values are measured in
reciprocal ohms (mhos). It is interesting to note that
a resistance of a conductance of
a reactance of is equivalent to a susceptance of
an impedance of an admittance of
1 megohm 1 micromho
B < BL, c =
+ JB L = 1/Z L, B p = 1 /X
G L, and
o 1 /2rrfC . (16)
re = 1
gv
- g,(o2 1
Cep
when the load is inductive (17)
Sturley, K. R. Radio Receiver Design, Part 1 (Chapman & Hall, London, 1943) p. 37 et tog
52 (iii) THE COMPONENTS OF GRID ADMITTANCE 2.8
gm GL
=ConL +GL2 t"
. 19)
+ Gr.)
r, (min) ± 2(g,, (20)
gm B°n
and these occur at B L = + (g,, + G L).
Fig. 2.46. Conditions for deriving input admittance (A) with plate -grid capacitance
coupling (B) general case including cathode circuit impedance.
where B,D and gp are neglected in comparison with the other components, and
(B LG k -BkG L) is very small. Thus the reflected resistance is increased, and the
damping decreased, as the result of the insertion of Zk.
The input capacitance under these conditions is given by
C, = C,n (GL +gm)GL +
BL' gm[gm2GL +gm (GkGL BkBL)] -
(23)
GL2 + BL' (G 12 + BL') [(Gk +gm)2 + Bi2]
which is less than with Z = O.
If the screen is by-passed to the cathode,
o
-1 GL2 +BL'r1 +g,(g, +2Gk)1
(24)
24
g, gmB°,BL Gk2 + Bk2
+BL2 +g,(GkGL-
C° - °DL (GL +gm)GL
-C GL2 + BLZ
gm[g,sGL
(G L8 + BL2) [(Gk +g,)2 +
BkBL)]
Bk2]
(25)
2.8 (iii) THE COMPONENTS OF GRID ADMITTANCE 53
IL
2
FIG. 24 7
Fig. 2.47. Alternative form of equivalent circuit for deriving input admittance
Phase angle of added component = phase angle of voltage gain + phase angle
of feedback admittance. (36)
Grid input admittance (Y9) = Y;,, + AY,,
Yfo. Y,b
= Y1,ß + (37)
Yout + YL
See also Ref. B21.
The measurement of the four short-circuit admittances is covered in Chapter 3
Sect. 3(vi) A, B, C and D and also Refs. B17, B21.
(39)
1 + Rkgm (Ib + I,2) //b
where R k cathode resistor for complete compensation of input capacitance
change with bias
AC
ti Cvxg,a (Ib + I,z) /Ib
AC = change in input capacitance in farads from normal operating con-
dition to cut -off,
C9k = grid -to- cathode capacitance in farads measured with valve cold,
Sm = mutual conductance in mhos at normal operating condition,
Ib = direct plate current in amperes
and I,z = direct screen current in amperes.
(vi) Grid-cathode capacitance
The mathematical treatment of the effects of grid -cathode capacitance has been
given above. Methods of neutralization are described in Chapter 26 Sect. 8.
The published grid-plate capacitances are usually in the form of a maximum value,
without any indication of the minimum or average value. In some cases the average
is fairly close to the maximum, while in others it may be considerably less. The
average value is likely to vary from one batch to another, and from one manufacturer
to another. Equipment should be designed to avoid instability with the maximum
value, although fixed neutralization should be adjusted on an average value, determined
by a test on a representative quantity of valves.
Effect of electrode voltages on grid -cathode capacitance -see Ref. B23.
-
resistance load (v) Series expansion ; general rase (vi) The equivalent plate circuit
theorem (vii) Dynamic load line general case (viii) Valve networks -general case
(ix) Valve coefficients as partial differentials (x) Valve characteristics at low plate
currents.
(i) General
Valve characteristics may be represented mathematically as well as graphically
(see Chapter 6 for mathematical theory).
The plate (or space) current is a function (F) of the plate and grid voltages and may
be expressed exactly as
fib= F(eb-{ µec+el) (1)
where e, is the equivalent voltage which would produce the same effect on the plate
current as the combined effects of the initial electron velocity of emission together
with the contact potentials. The amplification factor µ is not necessarily constant.
There will be a small current flow due to e, when eb and e, are both zero.
As an approximation, when e b and µe, are large, e, may be neglected. The func-
tion in eqn. (1) may also be expressed approximately in the form
fib N K(eb + µe,)" (2)
in which K is a constant. The value of n varies from about 1.5 to 2.5 o. cr the usual
operating range of electrode voltages, but is often assumed to be 1.5 (e.g. Conversion
Factors) over the region of nearly- straight characteristics, and 2.0 in the region of
the bottom bend (e.g. detection). We may take the total differential* of eqn. (2),
dib = áepdeb 1e de, (3)
which expresses the change in i b which occurs when e b and e, change simultaneously.
Now see (ix) below-provided that the valve is being operated entirely in the region
in which µ, g,,, and r,, are constant,
Vtb 1 and ai, =gm,
aeb rfl óec
_ deb (5)
and thus (it, constant)
de,
whence g,,r, = µ [see (ix) below]. (6)
The treatment so far has been on the basis of the total instantaneous voltages and
currents, e b, e ,, i b ; it is now necessary to distinguish more precisely between the
steady (d.c.) and varying (signal) voltages and currents.
For total differentiation see Chapter 6 Sect. 7(ii).
58 (i) GENERAL MATHEMATICAL RELATIONSHIPS 2.9
fb = Ibo +
ee
e,
=
= E -fe,e,
Eao (7)
i, -
e, + F. eu (9)
or
r,
This only holds under the condition that µ, g,,, and r, are constant over the operating
region.
-
g,,, e,, passed through in the opposite direction to i
This voltage et across the valve and the load can be developed by means of a current
r so that the total current through
r, is (g e, i,). Eqn. (18) is the basis of the constant current equivalent
circuit as in Sect. 7(ii).
The voltage gain (A) of an amplifying stage with a load resistance Et is
A = et = GRL =
RL . I
(19)
e, e,
Rz ! r, +
When the load is an impedance ZL, the voltage gain may be shown to be
A-- r, LZLZL
1
u
/Zz
- +r, (20)
where r, and Z are complex values (see Chapter 6).
If Z, = RL + jXL, the scalar value of A is given by
A = µVR L' +XLXL'
(21)
s/(r, + R L)' + XL'
l'he voltage gain may also be put into the alternative form
A - I go,
r, r ZLZL 22
k22)
Plate current
Power input from plate supply Pb; = Ebb loo
D.C. power absorbed in load
Quiescent plate dissipation
Pac = ¡boa RL = ELc
Ps. = Ebb lbe.
I.
But Ebb = Eb. + EL,.
Therefore P b = Ebo Ibo + EL* Ibu.
= Pso + Pa c (24)
Signal Condition :
Average value of total input = P,b = Ebb Ibo (25)
which is constant irrespective of the signal voltage.
*After book by M.I.T. Staff " Applied Electronics " (John Wiley & Sons Inc. New York, 1943)
pp. 419 -425.
60 (iii) POWER AND EFFICIENCY 2.9
(' 2a
1
(lau f iv)' RLd(wt)
277 0
¡`2rr 1 1 2rr
r2a y;
J Iau'RLd(wt) 277 21ouivRyd( 't) +
2I iv'RLd(wt) (27)
2,r
o J n
= Iao'RL+O-FPa (28)
i.e. PL = Pao + Pao
2r
_ 1
:,'RLd(wt) (29)
where P a o
2ar
o
2n
1
Plate dissipation P, = e ai ad( wt) (30)
2n
0
From eqns. (28) and (29) it will be seen that the power absorbed by the load increases
when the signal voltage increases, but the power input remains steady ; the plate
dissipation therefore decreases as the power output increases,
With sinusoidal grid excitation, linear Class A, valve operation and resistive load,
P., = E,l a = I,'R L.
Applying equation (16),
Pay
µ'E°'RL (36)
(r, + R L)'
Differentiating with respect to R L and equating to zero in order to find the condition
for maximum power output,
dP a [(r, + R L)' 2R L(r, + R L) 0 - l
dR L L (r, + R L) ` J
i.e. when (r, -}- R L)' -2RL(r, + R L) 0
or when R L = r, (37)
and the maximum power output is
2.9 (iii) POWER AND EFFICIENCY 61
µ2E22
E.2
Paco, = = 4 Ngm (38)
4r 9
The factor µgm is a figure of merit for power triodes.
If the load is an impedance (ZL = R L + jX L) the condition for maximum power
output is when
RL = ßr2 + XL2. (39)
In the general case, with a resistive load, the power output is given by eqn. (36) which
may be put into the form
µ2E.2 1
pay
t, r, (40)
If R L /r, = 2,
r RL
{ 2
11 %, while if R L /r, _- 4
the loss of power below the maximum is only
the loss of power is 36 %, so that " matching " of the load is not at all critical.
The treatment above is correct for both triodes and pentodes provided that both
are operated completely within the linear region, that is with limited grid swing. A
pentode is normally operated with a load resistance much less than the plate resistance
on account of the flattening of the output voltage characteristic which would other-
wise occur at low plate voltages.
This subject is considered further in Chapter 13, under practical instead of under
ideal conditions.
lz'rD
aa
Or, \\ 2
a22
a,
- R L) deb - r (r + R L) Debt
(2r (44)
6(r, + R L)5
If it is assumed to be constant (this is only approximately true for triodes and not
for pentodes)
e = e, +v
iL
The first term ale = p.e /(r, + R L) is similar to eqn. (16) above, which was re-
garded as approximately correct for small voltage inputs ; that is to say for negligible
distortion.
The first and second terms
n
+ aaea
express the plate current of a " square law detector " which is closely approached
by a triode operating as a grid or plate (" anode -bend ") detector with limited ex-
citation voltage.
The second and higher terms are associated with the production of components
of alternating plate current having frequencies differing from that of the applied
signal-i.e. harmonics and (if more than one signal frequency is applied) intermodula-
tion frequencies.
For example, with a single frequency input,
e = E,,, sin wt (45)
Therefore
e' = E,,,' sin' wt = j-Em' - j-E,,2 cos 2wt (46)
and
e' = I-Em' sin at }Em' sin 3wt.- (47)
The second term (e') includes a d.c. component (kEm') and a second harmonic com-
ponent. The third term includes a fundamental frequency component (k-Er' sin wt)
and a third harmonic component.
If the input voltage contains two frequencies (f, and f,) it may be shown that the
second term of eqn. (41) produces
a d.c. component
a fundamental f, component
a second harmonic of f,
a fundamental f, component
a second harmonic of f,
a difference frequency component (f, fa) -
a sum frequency component (f, + fa)
The third term of equation (41) produces
a fundamental f, component
a third harmonic of f,
a fundamental 12 component
a third harmonic of f,
a difference frequency component (2f, f,) -
a difference frequency component (2f, f,) -
a sum frequency component (2f, + f,)
a sum frequency component (2f, + f1)
In the case of an A-M mixer valve, f, may be the signal frequency andf, the oscillator
-
frequency. The normal i-f output frequency is (f, fa) while there are spurious
output frequencies of (f, + f,), (2f, + f,), (2f, + fi), (2f, f,) and (2f, f,).-
Even though no oscillator harmonics are injected into the mixer, components with
-
frequencies (2f, + f,) and (2f, -f,) are present in the output, thus demonstrating
mixing at a harmonic of the oscillator frequency.
If the input voltage contains more than two frequencies, or if the terms higher
than the third are appreciable, there will be greater numbers of frequencies in the
output. The effect of this on distortion is treated in Chapter 14.
It may be shown that the effect of the load resistance, particularly when it is greater
than the plate resistance, is to decrease the ratio of the harmonics and of the inter-
modulation components to the fundamental. This confirms the graphical treatment
in Chapter 12.
2.9 (vi) EQUIVALENT PLATE CIRCUIT THEOREM 63
Ne (` 8)
r, ZL
where e = e, (for amplifier use)
and ZL = impedance of the plate load at the frequency of the applied voltage.
For amplifier use this may be put into the form
µE0 (49)
l' r, +ZL.
This is the same as eqn. (16), except that RL has been replaced by ZL.
This is the basis of the Equivalent Plate Circuit Theorem which states$'that
the a.c. components of the currents and voltages in the plate (load) circuit
of a valve may be determined from an equivalent plate circuit in one of two forms
(1) a fictitious constant -voltage generator (µE,) in series with the plate resistance
-
of the valve, or
(2) a fictitious constant- current generator (I = g.,E,) in parallel with the plate
resistance of the valve.
These are applied in Sect. 7 of this chapter.
If a distortionless Class A amplifier or its equivalent circuit is excited with an alter-
nating grid voltage, the a.c. power in the load resistor R L (i.e. the output power) is
I,' R L.
The d.c. input from the plate supply to the valve and load (in the actual case) is
Pee = Ibo Ebb. Under ideal Class At conditions the d.c. current Ie, remains con-
stant, since the a.c. current is symmetrical and has no d.c. component.
Now the a.c. power input from the generator is
P, = uE,I, (50)
But I, - (r, + RL)
µE°
Therefore µE, = r,!, R LI,
and P, = pE,I, = r,I,' + R LI,'. (51)
In this equation
P, = a.c. power input from generator
r,!,' = a.c..power heating plate
R LI,' = a.c. power output = P,
The a.c. power P, can only come from the d.c. power P
dissipated in the valve,
which decreases to the lower value P, when the grid is excited.
The. total plate dissipation (P,) is therefore
where P= P, =
d.c. plate dissipation.
P - P, + r,I,' (52)
This
P, = P --
may be put into the form
(P, r,I,')
= P,° P°,
-
where P = R I,' = (P, r,I,') = - a.c. power output.
(53)
The statement may therefore be made, that the plate dissipation is equal to the d.c.
plate dissipation minus the a.c. power output.
A more general statement covering all types of valve amplifiers and oscillators is
that the plate input power is equal to the plate dissipation plus the power output.
This analysis, based on the equivalent plate circuit, reaches a conclusion in eqn. (46)
which is identical with eqn. (33) derived from a direct mathematical approach. It is,
however, helpful in clarifying the conditions of operation of a distortionless Class At
amplifier.
The preceding treatment only applies to amplifiers (e = e, in eqn. 41), but it
may be extended to cover cases where the load impedance contains other e.m.f's,
by using the principle of superposition-see Chapter 4 Sect. 7(viii).
It is possible to adopt a somewhat similar procedure to develop the Equivalent
Grid Circuit, or that for any other electrode in a multi-electrode valve.
Then µ - aeo
(i = constant), (55)
or more completely +
ai /be,
a:D /aeD
atD
gm = T (eD = constant). (56)
aCo
rD = - e (e =
D
constant), (57)
(58)
1
or more correctly*
at /ae
In a corresponding manner the gain (A) and load resistance (R L) of a resistance -
loaded amplifier may be given in the form of total differentials-
de,
A f
deo
and RL
de (59)
di,
Particular care should be taken with the signs in all cases, since otherwise serious
errors may be introduced in certain calculations.
duces a 10 -fold incrt.ase of plate current. If the logo of current is plotted against
plate voltage, the result should approximate to a straight line with a slope of 1/0.21.
In the case of triodes operating as triodes the relationship of plate current to grid -
cathode voltage is still approximately logarithmic, up to a value of plate current which
varies from type to type, but the slope of the curve is decreased by the plate -grid
voltage. The decrease in slope is approximately proportional to the grid bias and
therefore to 1/µ times the plate voltage. The curve at a given plate voltage is, in
general, steeper for a high -mu than for a low-mu triode. Over the region in which
the logarithmic relationship holds, the mutual conductance is proportional to the
plate current. For a given plate voltage and plate current, the g,,, in the low- current
region is greater for high-mu than for low-mu triodes, regardless of ratings. Also,
for a given triode at a given plate current, g,,, is greater than at lower plate voltages.
Maximum voltage gain in a d-c amplifier is obtained if the valve is operated at as
low a plate voltage as possible, and at a plate current corresponding to the top of the
straight portion of the characteristic when log, o of current is plotted against the grid
voltage.
With pentodes at low plate currents, the maximum gain is obtained when the screen
voltage is as low as is permissible without resulting in the flow of positive grid current.
Reference Al2 pp. 414 -418.
SECTION 10 : REFERENCES
(A) The following text books will be found helpful for general reading :-
Al. Reich, H. J. " Theory and Applications of Electron Tubes." (McGraw -Hill Book Company,
New York and London, 2nd edit. 1944) Chapters 1, 2, 3, 4 : Appendix pp. 671 -673.
A2. Massachusetts Institute of Technology " Applied Electronics " (John Wiley and Sons, New York ;
Chapman and Hall, London, 1943) Chapters 4, 8, 12.
A3. Sturley, K. R. " Radio Receiver Design " (Chapman and Ilall, London, 1943 and 1945) Part 1,
Chapter 2 ; Part 2, Chapters 9, 10.
A4. Everitt, W. L. " Communication Engineering " (McGraw -Hill Book Company, New York
and London, 2nd edit. 1937) Chapters 12, 13, 14, 15.
A5. Chaffee, E. L. " Theory of Thermionic Vacuum Tubes " (McGraw -Hill Book Company, New
York and London, 1933) Chapters 4, 5, 6, 7, 8, 10, 11, 12, 15, 19, 20, 21, 23.
A6. Terman, F. E. " Radio Engineering " (McGraw -Hill Book Company, New York and London
3rd edit. 1947).
A7. Terman, F. E. " Radio Engineers' Handbook " (McGraw -Hill Book Company, New York and
London, 1943) Sections 4, 5, 7.
A8. Everitt, W. L. (Editor) " Fundamentals of Radio " (Prentice -Hall Inc. New York, 1943) Chapters
4, 7.
A9. Henney, K. (Editor) " The Radio Engineering Handbook " (McGraw-Hill Book Company, New
York and London 3rd edit. 1941) Section 8.
A10. Albert, A. L. " Fundamental Electronics and Vacuum Tubes " (Macmillan Company, New
York, 1938) Chapters 4, 5, 6, 12.
All. Preisman, A. " Graphical Constructions for Vacuum Tube Circuits " (McGraw -Hill Book
Company, New York and London, 1943).
Al2. Valley, G. E., & H. Wallman. " Vacuum Tube Amplifiers " (M.I.T. Radiation Laboratory
Series, McGraw -Hill Book Co., New York and London, 1948).
(B) GENERAL
Bl. Chaffee, E. L. " Variational characteristics of triodes " (Book " Theory of Thermionic Vacuum
Tubes " p. 164 Sect. 79).
B2. R.C.A. " Application Note on receiver design- Battery operated receivers " (grid blocking)
No. 75 (May 28, 1937).
B3. Jonker, J. L. H. " Pentode and tetrode output valves " Philips Tec. Corn. 75 (July, 1940).
B4. Application of sliding screen -grid voltage to variable -mu tubes " Philips Tec. Corn. 81 (April
1941).
B5. Chaffee, E. L. " Characteristic Curves of Triodes " Proc. I.R.E. 30.8 (Aug. 1942) 383 -gives
met lod for obtaining static characteristic curves from one experimental curve.
B6. Thompson, B. J. " Space current flow in vacuum tube structures " Proc. I.R.E. 31.9 (Sept.
1943) 485.
B7. Pockman, L. T. " The dependence of inter-electrode capacitance on shielding " Proc. I.R.E.
32.2 (Feb. 1944) 91.
B8. Zabel, L. W. " Grid- current characteristics of typical tubes " Elect. 17.10 (Oct. 1944) 236. (Curves
of 6J5, 6SJ7-GT, 6K6 -G triode, 6V6 -GT triode, 38).
B9. Thurston, J. N. " Determination of the quiescent operating point of amplifiers employing cathode
bias " Proc. I.R.E. 33.2 (Feb. 1945) 135.
B10. Radio Design Worksheet No. 49 "Perveance" Radio (June 1946) 29.
B11. Haefner, S. J. " Dynamic characteristics of pentodes " Comm. 26.7 (July 1946) 14.
B12. Warner, J. C., & A. V. Loughren. " The output characteristics of amplifier tubes " Proc. I.R.E.
14.6 (Dec. 1926) 735.
B13. Zepler, E. E. " Triode interelectrode capacitances " W.E. 26.305 (Feb. 1949) 53.
B14. Pullen, K. A. " The use of G curves in the analysis of electron -tube circuits " Proc. I.R.E. 37.2
(Feb. 1949) 210.
Pullen, K. A. " G curves in tube circuit design " Tele -Tech. 8.7 (July 1949) 34 ; 8.8 (Aug. 1949)
33.
B15. Jones, T. I. " The dependence of the inter -electrode capacitances of valves upon the working
conditions" Jour. I.E.E. 81 (1937) 658.
B16. Humphreys, B. L., & B. G. James. "Interelectrode capacitance of valves- change with operating
conditions" W.F. 26.304 (Jan. 1949) 26.
B17. R.C.A. " Application Note on input admittance of receiving tubes " No. 118 (April 15, 1947).
B18. R.C.A. Application Note " Use of miniature tubes in stagger -tuned video intermediate frequency
systems" No. 126 (Dec. 15, 1947). Reprinted in Radiotronics No. 134.
B19. R.C.A. Application Note " A tube complement for ac/dc AM /FM receivers " (Jan. 2, 1948).
B20. R.C.A. Application Note " Use of sharp cut -off miniature pentode RCA -6CB6 in television re-
ceivers " No. 143 (March 31, 1950).
B21. " Standards on electron tubes- methods of testing 1950 " Proc. I.R.E. Part 1 38.8 (Aug. 1950) ;
Part 2 38.9 (Sept. 1950).
B22. Pullen, K. A. ' Use of conductance, or G, curves for pentode circuit design " Tele -Tech. 9.11
(Nov. 1950) 38.
Additional references will be found in the Supplement commencing on page 1475.
(C) INPUT IMPEDANCE ; HIGH FREQUENCY OPERATION
Cl. North, D. O. " Analysis of the effects of space- charge on grid impedance " Proc. I.R.E. 24.1
(Jan. 1936) 108.
C2. Bakker, C. J. " Some characteristics of receiving valves in short -wave reception" Philips Tec.
Rev. 1.6 (June 1936) 171.
C3. Strutt, M. J. O., & A. Van Der Ziel. " The behaviour of amplifier valves at very high frequencies "
Philips Tec. Rev. 3.4 (Apr. 1938) 103.
C4. R.C.A. " Application Note on input loading of receiving tubes at radio frequencies " No. 101
(Jan. 25, 1939).
C5. Hudson, P. K. " Input admittance of vacuum tubess " Comm. 23.8 (Aug. 1943) 54.
C6. R.C.A. " Application Note on use of miniature tubes in stagger -tuned video intermediate-frequency
systems " No. 126 (Dec. 15, 1947).
C7. Zepler, E. E., & S. S. Srivastava 'Interelectrode impedances in triodes and pentodes" W.E. 28.332
(May 1951) 146.
See also B17, B18, B19, B20, 1321 and Chapter 23 Sect. 5.
2.10 REFERENCES 67
Section Page
1. Basis of testing practice ... 68
2. Control of characteristics during manufacture 85
3. Methods of testing characteristics 89
4. Acceptance testing ... 120
5. Service testing and service tester practice 121
6. References ... 125
(a) Mounting.
(b) Ventilation.
(c) Heater -cathode insulation.
(d) Control grid circuit resistance.
(e) Operation at low screen voltages.
(f) Microphony.
(g) Hum.
(h) Stand-by operation.
Senior Engineer, in charge of Quality Control and Valve Application Laboratory, Amalgamated
Wireless Valve Co. Pty. L. td.
68
3.1 (i) FUNDAMENTAL PHYSICAL PROPERTIES 69
electrodes, with the cathode emitting, are termed " static characteristics ". These
characteristics consist of-
(a) the voltage -current relationship for each electrode, when constant voltages are
e.g., Diode,
Triode,
- -E-E
applied to all remaining electrodes.
-I Ib
b b,
b,
E ,1 constant
, -E
I E b constant
and similarly in the case of other types.
(b) the mutual voltage-current relationships between electrodes, when constant
voltages are applied to all electrodes except the voltage- varying electrode.
e.g., Triode, -- /1 -E
-
Pentode, //-E
-E
I t3
rl
,
«
E,, constant
E,,, E <3, E c2 constant
E b, E e3, E ,z constant
Er, Eb, E E ,2 constant
and similarly in the case of other types.
Derived Characteristic Parameters -By considering infinitesimal changes of
voltage and current of the static characteristics, described under (a) and (b) above, the
characteristic parameters-amplification factor, transconductance and variational plate
resistance are obtained, which are related as follows,
Amplification Factor
Transconductance - Plate Resistance
These characteristic parameters are derived by the following relations- -
transconductance
_alb ; E,,, E r3, E,2 constant (3)
SEb
amplification factor = ; E ray E rz constant (4)
SE
The amplification factor is a function of the electrode geometry only, but becomes also
dependent on the control grid voltage as the cut-off condition is approached. The
transconductance, as defined by eqn. (3) may be derived from the general space current
eqn. (1), and is a function of the electrode geometry, the total cathode space current
and uniformity of activation of the cathode surface.
The characteristic parameters normally used to describe the electrode geometry,
and in terms of which performance is interpreted, are -the amplification factor,
transconductance of the control grid with respect to the plate and the (a.c.) plate to
cathode resistance.
Dynamic Characteristics-By superimposing alternating voltages on the direct
voltage(s) of the control electrode(s), with suitable impedances in series with the
output electrode(s), dynamic characteristics are obtained which depend directly on
the static characteristics and characteristic parameters. These dynamic characteristics
include rectification, frequency conversion, voltage and power gain, oscillation, and
impedance transformation characteristics.
The static characteristics, characteristic parameters and the fundamental properties
previously described comprise the basic mechanical and electrical characteristics,
which determine the serviceability and application of a valve and its performance under
given conditions.
In order to appreciate the significance and limitations of the various tests, which
normally are applied to determine the condition of a valve and the probability that it
will continue to operate satisfactorily, the performances and individual tests must be
properly interpreted in terms of all the relevant characteristics and corresponding
fundamental physical properties. The manner and extent of the dependence and inter-
dependence of the various characteristics and fundamental physical properties must also
be recognized and understood.
FUNCTIONAL CHARACTERISTICS 73
3.1 (ii) BASIC
Notwithstanding the ratings shown against each type, the actual ratings of an indi-
vidual type depend entirely on its intended applications and may not include all those
listed. The absence of a rating for a particular characteristic may be taken to indicate
either that it has not previously been necessary to specify a rating for this characteristic,
or that the type in question was not intended originally for this application and has not
been processed accordingly. In the latter case, operation may not be satisfactory, and
reference should be made to the valve manufacturer for specific information.
(C) Rating systems
Valves are rated by either of two systems--the " absolute maximum " system, or the
" design-centre maximum " system. The absolute maximum system originated in the
early days of valve development and was based on the voltage characteristics of battery
supplies. Battery voltages could fall below their nominal values but seldom appreciably
exceeded them, so that valve maximum ratings set on the basis of specified battery
voltages were absolute maximum ratings. This system is still widely used by British
and European valve manufacturers and is the system of ratings used in the British
Services' Electronic Valve Specification K1001 and the U.S.A. Services' Specification
for Electron Tubes, JAN-1A (Ref. S2). With the introduction of power line and
car-radio operated receivers and the tendency of many designers to interpret absolute
maximum ratings as nominal values, it became necessary to re -rate valves according to
a system in which allowance was made in the ratings for the variations which occur
under both conditions of operation. Accordingly, the design-centre system was
adopted in U.S.A. by the Radio Manufacturers Association in 1939 for the rating of
receiving valves and since then has become the standard system for rating most
receiver types of American design, manufactured both in U.S.A. and elsewhere.
With either system, each maximum rating for a given valve type must be considered
in relation to all other maximum ratings for that type, so that no one maximum rating
will be exceeded in utilizing any other maximum rating.
Thus it will often happen that one rating alone will determine the limiting operating
conditions while other characteristics are below their maximum ratings.
(D) Interpretation of maximum ratings
In the absolute maximum or " absolute " system (except for filament or heater
voltage) the maximum ratings are limiting values above which the serviceability of the
valve may be impaired from the viewpoint of life and satisfactory performance.
Therefore, in order not to exceed these absolute ratings, the equipment designer
has the responsibility of determining an average design value for each rating below
the absolute value of that rating by an amount such that the absolute values will
never be exceeded under any usual condition of supply-voltage variation, load varia-
tion, or manufacturing variation in the equipment itself.
The equipment should be designed to operate the filament or heater of each valve
type at rated normal value for full-load operating conditions under average voltage-
supply conditions. Variations from this normal value due to voltage -supply fluctuation
or other causes, should not exceed ± 5 per cent. unless otherwise specified by the
valve manufacturer.
Under the " British Standard Code of Practice " B.S.1106, 1943 (Ref. 49) and the
British Radio Valve Manufacturers' Association's publication " Radio Valve Practice,"
August, 1948 (Ref. 48), recommended British practice is that in general it is not
permissible that the heater voltage should vary more than 7 per cent. from the rated
value.
It is a matter of experience, however, that the heaters and filaments of most modern
receiving valve oxide- coated cathode and filament types may be operated at voltages
whose maximum fluctuations do not exceed ± 10 per cent. from their rated values,
without serious effect on life or marked reduction in performance, provided that the
maximum ratings of the other electrodes are not exceeded. In cases where the
heater or filament voltage variations exceed, or are likely to exceed, f10 per cent., the
maximum ratings should be reduced and recommendations obtained from the valve
manufacturer as to the maximum ratings permissible under the particular conditions.
In the design- centre maximum or " design- centre " system, the maximum
ratings are working design-centre maxima. The basic purpose underlying this system
78 (iv) VALVE RATINGS : LIMITING EFFECT ON OPERATION 3.1
permitted. In such cases it may be inferred that the grid has not been suitably treated
to permit dissipation.
(E) Operating conditions
Typical operation-In addition to maximum ratings, information is published
on typical operating conditions for most of the various types, when used in particular
applications. These typical operating conditions are intended to provide guiding infor-
mation for the use of each type. They must not be considered as ratings, because each
type can, in general, be used under any suitable conditions within its rating limitations.
Datum point for electrode potentials -In published data, it is standard practice
for the values of grid bias and positive-potential -electrode voltages to be given with
reference to a specified datum point, as follows : -For types having filaments heated
with direct current, the negative hlai :ent terminal is taken as the datum point to which
other electrode voltages are referred. v: iÿincs having filaments heated with alter-
nating current, the mid-point (i.e., the centre tap on the filament transformer secondary,
or the mid -point on a resistor shunting the filament) is taken as the datum point. For
types having equi- potential cathodes (indirectly heated) the cathode is taken as the
datum point.
Grid bias for a.c. or d.c. filament excitation -If the filament of any type whose
data are given for a d.c. filament voltage is to be operated from an a.c. supply, the grid
bias given for d.c. filament operation should be increased by an amount approximately
equal to one half the rated filament voltage and be referred to the filament mid-point.
Conversely, if it is required to use d.c. filament excitation on any filament type whose
data are given for an a.c. filament voltage, the value of grid bias given should be de-
creased by an amount approximately equal to one -half the rated filament voltage, and be
referred to the negative filament terminal, instead of the mid -point as in a.c. operation.
This rule is only approximate and does not, in general, provide identical currents for
both types of filament excitation.
(v) Recommended practice and operation
The following additional limitations on valve practice and operation are based partly
on the recommendations of the British Standard Code of Practice (Ref. 49) also on the
B.V.A. Radio Valve Practice (Ref. 48), and on established design practice.
(a) Mounting -(1) Unless otherwise stated it is desirable that valves should be
mounted base down and in a vertical position. Where it is necessary to depart
from vertical mounting, the plane of the filament of directly heated valves should be
vertical. Similarly, the plane of the grid side rods (or major axis of the control grid)
of indirectly heated valves having a high transconductance and /or a long unsupported
cathode, should be vertical. This reduces the possibility of filament- and cathode-to-
grid short circuits and microphony in filament valves.
It is particularly undesirable that valves having high plate dissipation ratings should
be mounted base upwards without agreement from the valve manufacturer, as this
method of mounting seriously affects the flow of air around the bulb and may result
in the limiting temperature being exceeded.
Depending on the distribution of the bulb temperature, gas may be released from
the getter deposit under these conditions and the vacuum and emission affected.
(2) It is particularly important that the connections to floating contacts of sockets
for glass based valves should be as flexible as possible and that the contacts themselves
should float properly and not become rigidly locked in position. The use of a wiring
jig, having the nominal specified dimensions for the base type, inserted in the valve
socket during wiring, is desirable in order to locate the socket contacts correctly, so
that strain on the base pin seals is minimized when the valve is inserted. Prior to
insertion, the base pins of miniature valves should be straightened by means of a
pin -straightener, if misaligned. The pin -straightener may consist simply of a steel
block drilled with countersunk holes of the correct diameter and location, to slightly
larger tolerances than those specified for the pins of the base type. For specific
information and design, reference snould be made to the valve manufacturer.
(3) It is undesirable to use socket contacts as connecting tags in circuit wiring because
of circumstances which may arise if the valve is subsequen*.iy .placed by another
3.1 (v) RECOMMENDED PRACTICE AND OPERATION 81
exist across the insulation, there is a risk of modulation hum, particularly in cathode-
coupled oscillators and the like.
(d) Control grid circuit resistance
In all cases where published recommendations are available from the valve manu-
facturer, they should be followed. The maximum value of grid circuit resistance in the
case of a particular type of valve cannot be specified without knowledge of the valve
characteristics and conditions of operation. In no circumstances should valves be
operated without a d.c. connection between each electrode and cathode.
The maximum value of control grid circuit resistance which may be used under any
condition of operation depends initially upon the stability of performance required but
ultimately upon the transconductance under that condition, the increase in cathode
current and electrode dissipations which may occur due to the cumulative effects of
reverse grid current (gas, grid emission and leakage) permitted by the specification,
the maximum ratings for these characteristics, and the limiting effect of resistances in
series with the cathode and in the electrode circuits in series with the various electrode
supply voltages. The following formulae, expressing the relationships between the
various resistances, direct currents and electrode parameters in typical grid controlled
triodes, tetrodes and pentode circuits hold providing the control grid current is negli-
gible in comparison with the total cathode current.
For pentodes and tetrodes (Ref. 41)
dIk 1 R °2
Ilvi= dI _-f-Rk
gk
1- +P -11 (6)
i
C µ°i.°2 µ°102
For triodes (Ref. 41)
z1 I
dli[gm +Rkl 1+- ) +RL] (7)
where d I k = change in cathode current permitted by maximum ratings of plate dissipa-
tion and /or cathode current ; or change in performance which can be
tolerated.
d1ó =change in plate current permitted by maximum ratings of plate dissipa-
tion and/or cathode current ; or change in performance which can be
tolerated.
zJI 1= change in control grid (No. 1) current which is likely to occur or is
permitted by the valve specification.
R -grid resistor (ohms).
R,, = series screen resistor (ohms).
Rk= cathode resistor (ohms).
RL= series plate load resistor (ohms).
µO,.°4 =triode amplification factor of pentode or tetrode.
µ= amplification factor of triode.
P =ratio of screen current (1a2) to cathode current
Ia
1,2±1b
g,,, =grid-plate transconductance (mhos).
gk= grid -cathode transconductance (mhos)
Ik
=pm ló
To determine the limiting value of grid resistor which may be used, 41,, should be
taken as the maximum value of reverse grid current permitted by the specification,
under maximum electrode dissipation and cathode current conditions ; 41ó and
d1k are the changes in plate and cathode currents which result in maximum rated
plate and /or screen dissipations and /or maximum cathode current.
The maximum rating which is reached first determines the limiting value of the grid
resistor.
If the specified value of d1,, is not available, the highest typical value of reverse
grid current given in Sect. 3(iv)A for each major group of valve types may be used with
3.1 (v) RECOMMENDED PRACTICE AND OPERATION 83
discretion. In cases where the valve is being operated below maximum rated condi-
tions and at a total cathode current less than that for the grid current test conditions,
the value of 41 may be taken as varying approximately as the cathode current.
It should be clearly understood that equations (6) and (7) are fundamental, and, in
cases where no maximum grid circuit resistance values are published, may be used to
calculate maximum safe values. These equations are based on the margin (dlb)
between the operating plate current and that value of plate current which gives
maximum plate dissipation or maximum cathode current or which gives the maximum
tolerable change in performance, the limiting condition of operation being determined
by the rating which is reached first. It may happen that the published typical operating
conditions give maximum rated plate dissipation -in this case dlb or dlk is zero, and
the equations cannot therefore be applied to calculate R01 directly. In such a case the
procedure is to assume reasonable values of R°, and d101, and then to use the equations
to derive dlb or dlk. The operating plate or cathode current would then be made less
than the permissible maximum by the margin dlb or Ail,. Alternatively a lower value
of Rv7 may be chosen and the conditions of operation recalculated.
If a maximum value of Rol is published for fixed bias operation, then the value for
cathode bias may be determined from equations (8) to (13) :
For pentodes and tetrodes- general case
R°, for cathode bias I 1 PR g2 (8)
R0, for fixed bias L
[Rkl \ i+ 1201.02) 14°tc2 J
or approximately, +gk[Rk± PR °2 1 (9)
.
1
/101.02 J
(or when R02 =0) 1 +gkRk (10)
For triodes- -general case
R°, for cathode bias
g, r Rk ¡ l }
it + R LlJ (11)
R°, for fixed bias -1
{
µ µ
R
or approximately ,----, 1 R (12)
.
µ
(or when RL =O) 1 +g,,,Rk (13)
In the case of resistance -loaded triodes and pentodes with series screen resistors, the
values of Rol as derived from eqn. (6) or (7) respectively will be very high. In such
cases the limiting factor is the effect of the reverse grid current on the operating
bias. This effect is considered in detail in Chapter 12, Sect. 2(iv) for triodes with
fixed bias and Sect. 2(iii) with cathode bias also Sect. 3(v) for pentodes with fixed bias,
and Sect. 3(iv)C with cathode bias.
A high value of resistance between the control grid and the cathode should be avoided
wherever possible. In B.S.1106, 1943 (Ref. 49) it is recommended that, in general,
1 megohm with self-bias and 0.5 megohm with fixed bias are suitable
maxima and that
with receiving valves having an anode dissipation exceeding 10 watts the grid -circuit
resistance should not exceed 0.5 megohm when self -bias is used and 0.1 megohm with
fixed bias. These values must be reduced when one resistor is common to more than
one control grid circuit.
The maximum values of grid circuit resistance with r-f pentodes controlled by
a.v.c. is covered in Chapter 27 Sect. 3(iv)b.
Some valve types have an inherent negative resistance region in the positive portion
of the grid characteristic, due to secondary emission from the grid as a result of the
deposition or evaporation of active material from the cathode or deposition from the
getter during manufacture. In such cases the resultant grid current may change sign
from positive to negative as the grid voltage increases from zero in a positive direction,
ultimately becoming positive again. The maximum value of grid circuit resistance
*This assumes that the grid leakage current is small compared with the ionization current, and that the
latter does not increase appreciably during operation.
84 (v) RECOMMENDED PRACTICE AND OPERATION 3.1
which may be used under these conditions without grid blocking occurring, is that
value of resistance represented by the line drawn from the operating bias point on the
abscissa, tangential to the cross -over loop of the grid current characteristic below the
abscissa as shown in Fig. 2.11.
If the grid circuit resistance is such that the line cuts the loop of the characteristic
below the abscissa, blocking will occur if the instantaneous values of grid voltage
exceed the voltage of the point at which the line crosses the grid characteristic nearer
to the origin. See also Chapter 2 Sect. 2(iii).
Because of the difficulty of controlling the grid current commencement point of
oxide- coated cathode and filament valves during manufacture, due mainly to contact
potential variations caused by variable control grid surface conditions, it is desirable
that operating conditions should be chosen, such that the control grid bias is always suffi-
ciently large in relation to the contact potential, so that variations of the latter produce
only minor effects on performance. In general, the grid current commencement
voltage of indirectly heated cathode types is negative and may have a value up to
-1.0 volt but as a rule varies during life. The grid current commencement voltage
of filament types is usually positive and seldom exceeds about +0.5 volt, but usually
becomes less positive during life and may even become negative. The plate current
commencement voltage of diodes due to contact potential behaves in a similar manner.
(e) Operation at low screen voltages
As the grid current commencement voltage may in certain cases be dependent on the
value of screen voltage and may become more negative as the screen voltage is reduced,
it is in general undesirable to operate valves having low or zero control grid -bias at low
screen voltages. In the case of zero-bias operated filament types, operation at low
screen voltages may result in the grid current commencement voltage becoming nega-
tive and in the flow of positive current, causing either a change in operating conditions
due to additional negative bias developed across the grid resistor or damping of tuned
grid circuits due to lowered input resistance. In either case undesirable and often
wide variations in performance may result.
(f) Microphony
Small variations of electrode spacing cause corresponding variations in the output of
the valve, and it is desirable to ensure that little or no vibration reaches the valve.
Such vibration may reach the valve by way of the valve socket or acoustically, and it
should be noted that microphonic trouble may originate in the converter or i -f stages
through modulation of the signal or i -f carrier at audio frequencies. It is recommended
that, if possible, the position of the valve in relation to the source of vibration be so
chosen that microphony effects will be at a minimum. Methods which may be used to
minimize microphony effects are described in Chapter 35 Sect. 3(iv).
(g) Hum
Hum due to conditions within the valves is analysed in Chapter 31 Sect. 4(i). See
also Ref. 91 on heater -cathode leakage as a source of hum.
(h) Stand -by operation
Where it is necessary to maintain cathodes and filaments at normal operating
temperatures during stand-by periods, better life in general will be obtained
when the equipment is so designed that some, rather than no, cathode
current flows during such periods. It is also essential that the heater or filament
voltage be maintained as close to the rated value as possible. Continuous operation
of heaters or filaments at voltages exceeding their rated values by more than 10 per
cent., without cathode current flowing, will result, generally, in progressively short
life being obtained.
In equipments which are designed so that no cathode current flows during stand-by
periods for periods of less than 15 minutes, the filament voltage of quick -heating fila-
ment types should be reduced to 80 per cent. of normal, while the heater voltage of
indirectly heated cathode types should be maintained at normal rated value. For
longer periods, both filament and heater power should be turned off.
3.2 (i) IMPORTANCE OF CONTROL OVER CHARACTERISTICS 85
production -sampling test to control characteristics, 100 per cent. testing of which
may not be warranted or necessary.
Subject to the product having passed the design and life test criteria to be described,
it is now usually transferred to the warehouse for shipping and distribution to manufac-
turers and wholesalers.
(B) Design tests
All important secondary characteristics are controlled by regularly testing and
recording the characteristics of a relatively small sample selected at random, either fror
each production batch, or each day's production in the case of continuous production,
after all the first -production tests have been performed and the defectives for these
tests screened out. The sample size is determined by the production batch size on the
daily production level, provided characteristics are reasonably uniform. Typical
sample sizes used are 1 per cent. (minimum 5 valves) of each batch, or a fixed sample
size of 5 to 10 valves per type per day for types in continuous production.
The tests performed on the characteristics included in this group, on the above basis,
are called Design Tests and the usual procedure is that failure of more than a stipulated
perccitage of the valves in the sample to meet the prescribed test limits for a particular
characteristic, or failure of more than a stipulated percentage of all the characteristics
to meet their prescribed limits, constitute cause, initially for retesting a larger sample
usually 100 valves -for the characteristic(s) failed, and in the event of the failure
-
being confirmed, for 100 per cent. testing of the complete batch for those characteristics
in which failure occurred. Failure of any valve of the batch to meet the specification
for the characteristic(s) in question is then cause for rejection.
In the case of continuous production, the usual practice is for the failed design tests
to be made production tests temporarily until satisfactory control of the characteristics
in question has been re-established and the design test criteria for these characteristics
can again be met consistently.
Typical design test acceptance criteria require that not more than 10 per cent. of the
valves of the sample fail for any one design test or that all the design tests failed do not
exceed 20 per cent.
In addition to essential secondary characteristics, other characteristics which are
often controlled on a design test basis include those characteristics which are not
independent parameters but have their values determined by characteristics which
are tested 100 per cent., also characteristics which require to be controlled for particular
applications.
Design tests are performed only on valves which meet the prescribed production
tests. To this end, the latter normally are repeated and recorded when design tests are
performed.
(C) Recorded readings
Tertiary characteristics are controlled, in general, on the basis of criticisms and correc-
tion from recorded readings of all important characteristics obtained from the design
test samples, taken either daily, weekly or monthly, depending on the importance of
the characteristic and its liability to variation.
Failure of characteristics tested on this basis to meet the requirements of the specifi-
cation (or, where not specified, to meet the average range of variation usually main-
tained) in respect of the design test criteria is cause for criticism only, and not rejection ;
individual valves having the characteristics indicated are not screened from the
product, unless the effect of their variation on performance is likely to be serious.
Where a recorded reading test indicates continuing unsatisfactory control of any
particular characteristic, it is usual for that recorded reading to be made a design test
temporarily, if considered warranted, until satisfactory control has been re- established.
(D) Life tests
Due to the varying extent to which the mechanical and electrical characteristics may
be affected by undetected and consequently uncontrolled variations of manufacturing
processes, it is not possible to predict reliably by any schedule of instantaneous tests,
88 (iii) SYSTEMATIC TESTING 3.2
however comprehensive, the performance obtainable from any particular valve under
operating conditions, and it is necessary to rely on recorded observations of characteris-
tic variation during continuous or intermittent operation under controlled test condi-
tions, to determine ultimate performance. Such tests are known as Life Tests.
Provided maximum ratings have not been exceeded, the life of most oxide-coated
cathode high vacuum valves, assuming that the initial emission and electrode insulation
are satisfactory, is determined almost entirely by the extent to which the initial vacuum
is maintained during storage and during periods of operation and non -operation, as in
general the supply of emission- producing material available on the average filament or
cathode greatly exceeds that required for the lives normally obtained in practice.
The average rate and uniformity of deterioration of the vacuum- and emission -
dependent characteristics which normally occur slowly and uniformly in all valves
during operation, and which vary from valve to valve, and any excessive deterioration of
mechanical and electrical characteristics due to defective manufacture are controlled by
regular life tests of small samples of each type. These life tests are operated normally
under maximum rating conditions, for a specified period during which the performance
determining characteristic(s) may not deteriorate beyond prescribed values, or for the
period (which may not be less than a prescribed minimum) required for such deteriora-
tion to occur.
Owing to limitations of equipment and cost, regular control life tests are run usually
for periods of 500 to 1000 hours on small samples of the order of 5 valves per type per
week for types in continuous production. As variations in characteristics are most
liable to occur early in life, readings are usually spaced at increasing intervals to give
an indication of the complete life characteristic over the control periods.
A typical life test acceptance criterion of satisfactory life is that the average life
obtained per valve, considering all valves of the sample, must be not less than 80 per
cent. of the specified duration of the life test. The valves used must also be selected at
random and comply with the production and design test acceptance criteria.
In the event of the completed product failing to comply with the life test criteria,
delivery of the product is then withheld from she warehouse, until satisfactory life has
been re- established.
(E) Warehouse tests
When quality is uniform, the controls normally imposed prior to packing are retests
of electrode mechanical stability and continuity, and maintenance of the vacuum, by
means of a suitable interelectrode hot " shorts and continuity " test during which each
valve is tapped lightly with a specified mallet. Each valve is also re- inspected for any
deterioration of the envelope in the case of glass valves, cementing of phenolic bases
and top caps, dry or badly soldered pin and top cap connections, type designation and
general mechanical finish. Pins of miniature valves are also straightened, if necessary.
65%
1. of rated transconductance for r-f and i -f amplifiers.
50%
2. of rated conversion transconductance and 65% of rated oscillator grid
current for converter and mixer types.
3. 50% of rated transconductance for general purpose triode types.
4. 50% of rated power output for power output types.
5. 40% of minimum rated direct current for diode types.
6. 80% of rated current or voltage for rectifier types.
7. 70% of normal alternating output voltage for resistance -coupled amplifier
types.
Note. -Rated values are those referred to on R.M.A. Electron Tube Characteristic
Sheets under maximum typical operating conditions.
The assistance of A. H. Wardale, Member I.R.E. (U.S.A.) and D. H. Connolly, A.S.T.C., is ack-
nowledged.
90 METHODS OF TESTING CHARACTERISTICS 3.3
+ PLATE
HEATER O
VOLTAGE
SOURCE HEATER
-VVI/Vs
O O
FIG. 3.1
Pre-heating
Pre-heating with heater voltage applied to the heater only is generally adequate for
all ordinary purposes where it is necessary to conserve time, except where full electrode
dissipation pre -heating is required by particular valve specifications under acceptance
testing conditions, or when testing for reverse grid current (gas, grid emission and hot
leakage). When full electrode dissipation is applied, the time of pre -heating is normally
2 minutes (JAN -1A) to 5 minutes (R.M.A.). Maximum valve ratings should not be
exceeded during pre -heating.
See References S2, S7.
(ii) General characteristics
(a)Physical dimensions- Valves may be checked for overall length and maximum
diameter either by the use of " go /not go " gauges or by an adjustable length gauge
and outside calipers respectvely.
Standard ring gauges for checking maximum and minimum base and sleeve dia-
meters, also pin alignment gauges, are specified by R.M.A. Standard ET -106 (Ref. S6).
(b) Shorts and continuity -It
is important to ensure that electrodes such as control
and signal grids and suppressors, which do not normally dissipate any power, should
not be subjected to any appreciable power dissipation during the shorts and continuity
test, otherwise the valve may be damaged as the degassing treatment of these grids
during manufacture is normally much lighter than that of electrodes which are required
to dissipate power. In addition, the cathode current drawn, particularly in filament
types, should not be excessive. It is also important that shorts be checked between the
various electrode pin connections and pins designated as no connection, also that the
continuity of internal jumper connections between base pins be checked.
The circuit arrangement accepted as the most generally satisfactory for both shorts
and continuity with regard to sensitivity and speed of testing is that known as the
hexaphase shorts and continuity test. This circuit is now widely used, particularly by
valve manufacturers (e.g., Ref. 1, Fig. 54). A modification is shown in Fig. 3.2 which
uses 6 110 volt ;ç watt neon lamps and is entirely satisfactory for valves not requiring
more than 6 independent electrodes. The values shown for this circuit have been
92 (ii) GENERAL CHARACTERISTICS 3.3
chosen to suit the majority of receiving type valves in that the dissipations of the normal-
ly negative electrodes are kept low for reasons previously stated -see (b) above -and
may not be satisfactory for all valve types. In the case of a semi -universal tester it is
essential that the requirements above be observed, and also that all electrode dissipa-
tions and peak and average currents be kept within their maximum ratings, otherwise
valves may be damaged. It is also essential as in any shorts tester that the cathode -
to -grid voltage, and battery -type filament -to -grid voltage particularly, be
kept to a value not exceeding approximately 100 volts, owing to the small
spacing between these electrodes resulting in unreliable " flick " indications of the
neon lamps in normally good valves. The test is carried out with the cathode hot, and
the resistor R, is for the purpose of limiting the peak cathode current to a safe value
(3 mA per 50 milliamp. filament strand for 1.4 volt battery valves). The tester uses
split anode neon lamps and continuity of each valve electrode is indicated by the
glowing of one half of the split-anode neon lamp connected to that electrode. In the
case of a short -circuit to cathode, one lamp will light on both electrodes, while in the
case of a short-circuit between two or more other electrodes, two or more lamps will
so light. There is an optimum arrangement of connections from the terminals
A, B, C, D and E to give fairly low voltages between cathode and control grid or other
low- dissipation electrodes and to give uniform illumination on all lamps. In some cases,
experimenting may be necessary in changing over connections to obtain fairly uniform
illumination on all lamps. It is necessary that the insulation of the sockets and
wiring be sufficiently good to avoid residual glows on the neon lamps when a valve is not
in the socket. This is particularly important in humid climates.
3 PHASE 6 PHASE OUTPUT
INPUT
Fig. 3.2. Hexaphase shorts and continuity tester. Suitable values are R =5000 ohms,
R3=12,-7500 ohms, R, = 1500 ohms, R, -5000 ohms, R, =50 000 ohms for battery
valves other than power output, 33 000 ohms for normal a.c. types, or 15 000 ohms for
power output valves. Lamps (N) are 110 volt watt split-anode neon.
If more than six lamps are required, additional lamps may be incorporated at suitable
points in the " ring " network, but all such arrangements have the limitation that one
or more conditions exist where no shorts indication is possible on one or more pairs
of lamps.
In addition to the neon shorts and continuity test, filament type lamps may be used
to indicate continuity from shield to cathode or along a " jumper " linking two pins.
Continuity of heaters and filaments is indicated by means of a current meter of
suitable scale in series with the heater or filament, an open- circuit being indicated by
J. . 1 ....
TOP CAP
SOCKETS VIEWED FROM BENEATH UNIVERSAL 7 PIN
i
110 VOLT
NEON LAMPS
No I
11
INDICATES
NF
IIOV
1
FILAMENT
CONTINUITY FOR MOST
VALVE TYPES
2 Li1(F
1
i
I
3
IOV
-
001NF
_
1
IIOV
1
1
1
0 50
1 1
N
1 1 1
IBOV
1
aP
60
110V p!F
7
, IIOV
B
110V
240V
FIG. 3
.01
50,
. 3
The older type of shorts tester is shown in Fig. 3.3 and operates on single phase
and requires a total voltage of 880 volts for 8 lamps. This may be used as a hot or cold
cathode shorts tester, but has the disadvantages that it only indicates electrode continu-
ity of those electrodes which draw sufficient current to operate lamps to which they are
connected. It also has the severe disadvantage that the voltages cannot be arranged
conveniently with regard to the electrode configuration for semi -universal testing, as
all
voltages are in phase, and quite high voltages may be developed between adjacent
electrodes in the absence of complicated switching.
When using this type of tester with either hot or cold filament (or cathode), it
is
essential to connect grid No. 1 to the lowest voltage in the chain with respect to earth,
otherwise damage may be done to the valve, or else good valves may indicate as having
a short circuit. The other points in the chain should be connected in the
same sequence
as the grids and plate.
A fairly satisfactory form of single lamp
shorts and continuity tester commonly used
in service testers is shown in Fig. 3.4. The
electrode switch S is rotated with the
cathode heated. If the neon lamp glows F G.3.4 Eh
on one side only the particular electrode is
connected ; if on both sides a short -circuit Fig. 3.4. Simpler type of shorts and
is indicated. If only one electrode indicates continuity tester, employing a single
short- circuit, the short is to heater (fila- lamp.
ment) ; if two or more electrodes indicate a for the
short -circuit, they are shorted together. The continuity test only works well
electrodes close to the cathode, the sensitivity decreasing towards the anode.
The same circuit Fig. 3.4 may also be used for a cold cathode shorts test.
(c) Heater (or filament) current -This is measured with a
d.c. heater source,
using the heater circuit as in Fig. 3.1. The heater voltmeter current passes through
the ammeter A, so that it is necessary for greatest accuracy to subtract the voltmeter
94 (ii) GENERAL CHARACTERISTICS 3.3
current from the current reading. However, with a 1000 ohms- per-volt meter, the
current is less than 1 mA and the error is quite small with heater currents of the order
of 0.3 A. When measuring filament currents of small battery valves (e.g., 50 mA),
the ammeter zero may be set to read zero when the voltmeter is at its nominal reading
(e.g., 1.4 volts). The current is normally measured with no connections to any other
electrodes. See also Ref. S12, Sect. 2.1.
(d) Heater -to- cathode leakage -This may be tested by applying the maximum
rated direct voltage between one heater terminal and cathode and measuring the
current with a microammeter (say 0-500 'sA) with a safety resistance in series (not
greater than 100 000 ohms). The leakage current should be not greater than 20 µA
for ordinary 6.3 volt 0.3 ampere valves, or 50 to 100 µA for power valves, with 100
volts applied.
The heater may be operated with either a.c. or d.c. supply if below 35 volts ; at 35
volts and above it is usual to operate it on a.c. supply. The voltage between heater and
cathode should be applied with both polarities, except in the case of rectifiers where the
heater must be negative with respect to the cathode.
(e) Inter-electrode insulation -This is not normally a regular commercial test,
except to meet service specifications. The test voltage specified is normally 300 volts
for small valves or 500 volts for those which are able to stand the higher voltage.
The insulation resistance may be measured by any ordinary type of insulation tester,
but preferably by an electronic megohm -meter. In this test, the cathode should always
be positive. The testis carried out on a valve which has reached a stable temperature
under normal operating conditions. The minimum permissible insulation resistance
usually specified is 10 megohms (Ref. S2).
This test is not used between heater and cathode.
(f) Emission-The purpose of the emission test is to ensure that the cathode
emission is adequate to provide the peak and average space currents for the particular
application of the type by a margin which has been found to be satisfactory for good
life and performance.
Emission is normally tested by applying a suitable direct positive voltage to all grids
connected together with the plate, and measuring the total cathode current.
Because of the very high value of the emission current normally drawn from the
cathode, the resulting excessive dissipation of the inner grids and the gas produced as a
result of this dissipation and also as a result of ion bombardment of the cathode, the
emission test is a damaging test and should be performed as seldom as possible and then
always very carefully.
In order to avoid damage to the cathode by this test, the applied voltage must be
sufficiently large to approach current saturation without drawing sufficient current to
damage the valve due to excessive dissipation of the inner grids. At the same time the
test must be of sufficient duration to enable the stability of the emission to be indicated
without being so long as to give risk of poisoning the emission. A safe value generally
accepted for the duration of the emission test is 3 seconds. During this period the
emission should not fall below a value recommended by the valve manufacturer.
In general, emission is a minimum reading. As most of the cathode current goes to the
No. 1 grid, which is usually wound with very fine wire, and treated lightly during
manufacture, the applied voltage must be restricted to a value such that the emission
current drawn does not result in damage to the valve. The voltages normally used in
manufacture are listed below.
Emission testing with applied alternating voltage is also used, but in this case the
test becomes one intermediate between a peak emission test and a high current trans -
conductance test. In practice, there is no exact and simple correlation between a.c.
and d.c. emission tests due to the fact that the a.c. emission current depends in part on
the emission capability of the valve as well as on its geometry. Experience has shown
that mains frequency a.c. emission tests tend to be unreliable owing to the fact that the
emission " gets a rest " between cycles and, in a practical case, the rating of peak to
average emission current drawn cannot be made sufficiently high to be effective.
Voltages given below are based on JAN -IA (Ref. S2) ; values of voltages and mini-
mum emission currents are given by Ref. 65.
3,3 (ii) GENERAL CHARACTERISTICS 95
Signal diodes are normally tested for emission with an applied direct voltage of
10 volts, except that 20 volts are normally applied to type 6H6.
Power rectifiers are normally tested with applied voltages as listed below (suffixes
of valve types have been omitted) :
10 volts -LABS, 6AC7, 6AG5, 7E5, 7F8, 7G8, 7V7, 7W7, 26A7.
15 volts -6BE6.
20 volts -1D8, 1J6, 2A3, 3A8, 6AB7, 6AG7, 6SD7, 6SQ7, 6SH7, 7G7, 7H7, 7J7,
7L7, 7Q7, 7R7, 757, 19, 28D7, 35L6, 70L7.
25 volts- 117N7.
50 volts -6A6, 6B5, 6L6, 6N6, 6N7, 12A5, 24A, 53, 59, 79.
All the voltages listed above are subject to variation by different valve manufacturers.
(g) Direct interelectrode capacitances-The following information is based on
recognised engineering practice by the manufacturers of American -type valves, and is
in line with the I.R.E. Standard (Ref. Si, S12), the R.M.A. -NEMA Standard (Ref. S8)
and with JAN-1A (Ref. S2).
The capacitances which are measured are direct interelectrode capacitances, and not
the total (self) capacitances which are the sum of two or more direct capacitances.
The measured value of inter -electrode capacitance of a valve is dependent upon the
valve shield or, in the absence of a shield, upon the geometry of the external environ-
ment. It is therefore necessary for the test to be made under strictly specified conditions
of valve shielding and of electrode and internal valve screen connections.
Published values of interelectrode capacitances are stated either for the condition
where the valve is shielded with a specified standard valve shield (see below) or is
unshielded. Interelectrode capacitances are normally measured with the cathode cold
and with no direct voltage applied to the electrodes. The base pins and leads are
shielded from each other and from other elements of the valve so that they and their
connections do not form part of the capacitance being measured. All external metal
parts integral with the valve should be connected to the cathode ; such parts include
lock -in valve bases, metal base shells, and pins with no connection. When capacitance
measurements are made on indirectly- heated valves the heater should be connected
to the cathcde unless in special cases the measurement is between the cathode only and
other elements, or the heater and other elements. For valves having elements other
than a control grid, a plate, and a cathode, the additional elements of the active section
including internal shields should be connected to the filament or cathode by the shortest
possible connections. For a multiple -unit valve structure, all elements of the other
section(s) should be earthed except when reading inter -section coupling capacitances.
The test sockets should be Standard Capacitance Sockets as specified by R.M.A. -
NEMA (Fig. 3.4A). Valves with top caps should be tested with the R.M.A. -NEMA
Standard Cap Connectors (Fig. 3.4B).
The connections to be made to the electrodes while testing for interelectrode
capacitances are given below :
Input capacitance is tested by measuring the capacitance between grid and cathode,
the latter being connected to heater, screen and shields ; the plate is earthed.
Output capacitance us tested by measuring the capacitance between plate and
cathode, the latter being connected to heater, screen and shields ; the grid is earthed.
Capacitance from grid to plate for all types of valves is measured with all other
electrodes earthed.
96 (ii) GENERAL CHARACTERISTICS 3.3
In multiple unit valves, the capacitance between the grid of one section and the
plate of the other is measured with all other electrodes earthed ; similarly from one
plate to the other.
The input capacitance of a diode is measured between its plate and cathode,
the latter being connected to heater and shields ; other sections are earthed.
The capacitance between diode -plate and grid or plate of other sections is
measured with all other electrodes earthed.
In a converter, the r -f input capacitance is measured between the signal grid
and all other electrodes connected together.
The mixer output capacitance is measured between the mixer plate and all other
electrodes connected together.
The capacitance from oscillator grid to oscillator plate is measured with all
other electrodes earthed.
The oscillator input capacitance is measured between oscillator grid and cathode,
the latter being connected to heater and shields ; the oscillator plate and all electrodes
of the other section being earthed.
The oscillator output capacitance is measured between oscillator plate and
cathode, the latter being connected to heater and shields ; the oscillator grid and all
electrodes of the other section being earthed.
The capacitance between oscillator grid and signal grid, or oscillator plate
and signal grid, is measured with all other electrodes earthed.
In converters in which there is usually a r -f voltage between cathode and
earth (e.g., 6BE6), the oscillator output capacitance is measured between cathode
and heater, the latter being connected to screen and shields ; the oscillator grid being
earthed.
The oscillator input capacitance is measured between oscillator grid and all other
electrodes connected together. The capacitance between oscillator grid and cathode
is measured with all other electrodes earthed.
Standard sockets, cap connectors and shields for use in the measurement of
valve capacitances (R.M.A. -NEMA, Ref. S8).
(1) Sockets
The construction and shielding of capacitance sockets and leads shall be such that
when the holes for the insertion of the base pins are covered with a grounded, flat metal
plate, the capacitance between any one socket terminal and all other socket terminals
tied together does not exceed 0.000 10 µµF for receiving valves.
B
1 1
.
Ali iiiii. i
The socket shall be so constructed that the base of the valve under test will seat
on the face plate.
Table 1.
Base Designation Max. diameter (A)* Max. diameter (B)*
Standard 4, 5, 6 or 7 -pin 0.250 in. 3 in.
Octal 0.175 in. 3 in.
Lock -in 0.093 in. 3 in.
7 -pin miniature 0.075 in. 21 in.
9 -pin miniature 0.075 in.
'see Fig. 3.4A.
21 in.
(2) Cap connectors.
Standard Cap Connectors shall be made as shown in Fig. 3.4B. A thin insulating
film may be placed on surface X in Fig. 3.4B. Dimensions in
inches shall be as
tabulated below.
Designation B C
Medium 0.556 0.850
Small 0.352 0.750
Miniature 0.242 0.750
i'iï/ï'
MIN
2
\_\_\_\_\r`
,\
Fig. 3.4B. RMA -NEMA Standard Cap Connectors for use in measurement of
valve
capacitances (Ref. S8).
(3) Shields.
Standard shields are shown in Fig. 3.4C. It is recommended that these shields be
used as indicated below, or as specified by the valve manufacturer.
Max. seated height
Shield with without
No. top cap top cap
308 GT glass types with T9 bulbs* 3 2/ in.
309 G glass types with T9 bulbs* 31 3 in.
311 G glass types with ST12 bulbs* 4166 3 B in
312 G glass types with ST14 bulbs* 4B. 4 in. -116
N4 316
No.308
3tft i:
o -1
©
No. 315
No.309
AMAX. RADIUS ALLOWABLE ON ALL INTERNAL EDGES
d
6
0.0001 to 2 µµF, between any two electrodes of a valve with all the other electrodes
earthed, at a frequency of 400 Kc/s.
Capacitances above 2 piF are usually measured on a simple capacitance bridge at
1000 c/s. Commercial bridges having a logarithmic scale are available for this purpose,
reading from 0 to 30 µµF. However, audio -frequency bridges are out of the question
for very low capacitances, because the insulation resistance between electrodes may be
lower than the reactance of a 0.0001 µµF capacitance. Even at 1000 Kc /s the re-
actance of this value of capacitance is 1500 megohms.
The capacitance measured with an audio -frequency capacitance bridge is frequently
higher than that measured with a capacitance bridge operating at radio frequency.
This effect is associated with the presence of resistive films on the insulators supporting
the valve electrodes. Such films may be caused by deposition of getter or by the evap-
oration of metal from the cathode either during the manufacture or during the life of
the valve. The value of the resistance may be as high as 1000 megohms or more, but is
capable of modifying the measured value of the capacitance and prevents the prediction
of the performance of the valve at frequencies higher than that of the measurement
(Ref. 23).
The actual capacitance of a valve depends on its operating conditions and the im-
pedances in the electrode circuits. For purposes of specification testing the valve is
tested with the cathode cold and without any voltages applied to the electrodes. In any
particular case under specified operating conditions the interelectrode capacitances
will differ from the static capacitances measured in this way. For further information
see Chapter 2 Sect. 8(iiil and the references there listed.
FIG. 3.5
The plate supply impedance is that of transformer T1 which should be one having
good regulation and low winding resistance.
In a transformer the plate supply impedance per plate is given approximately by
RsR N'R,r,+R,., (1)
where N= voltage ratio of transformer at no load (primary to half secondary in the
case of full-wave rectification)
(half) secondary voltage
primary voltage at no load
R,ri= resistance of primary winding in ohms
and R, =resistance of secondary winding in ohms (or half-secondary in the case of
full-wave rectification).
If the transformer plate supply impedance per plate is less than the required value,
two equal resistances should be added at points XX in Fig. 3.5 to make up the defi-
ciency. If the transformer plate supply impedance per plate is greater than the required
100 (iii) SPECIFIC DIODE CHARACTERISTICS 3.3
value, the test will not be so severe since it will reduce tile peak plate current and also
reduce the output voltage or current.
In view of this limitation, a rectification test which may be used and which should
be satisfactory for all practical purposes is to make use of the maximum published
conditions with a condenser input filter. These conditions include the maximum
r.m.s. applied voltage, the maximum d.c. load current, the minimum plate supply
impedance per plate and the maximum filter capacitance for these conditions, if
published. In cases where the latter information is not available, it is usually safe to
test with the maximum value of filter capacitance used in the published curves or
typical operation data, provided that the plate supply impedance per plate is equal to
the minimum rated value or to the value used for the curves or typical operation data,
whichever is the greater. In all cases it is essential that the maximum peak plate
current rating per plate be not exceeded.
The rectification test may be used for the purpose of checking the output voltage
and so determining end of life. The circuit constants for the rectification test may be
obtained by the following procedure. The value of R L is given by Eel, where E0
and Io are the direct voltage and current respectively across the load resistance under
maximum ratings. The value of E, is obtainable from the usual operational curves
published by valve manufacturers, and is the direct voltage for the maximum r.m.s.
applied voltage at the maximum value of load current (Ia). If curves are published
for more than one value of filter capacitance, the curve corresponding to the highest
capacitance should be selected. The rectification test should be carried out with the
value of CL equal to that used for the derivation of E0, while the total effective plate
supply impedance per plate should be as specified above. Under these conditions
an average new valve should give a voltage Ea and a current Io approximately equal to
the published values, and individual valves will have values either higher or lower than
the average as permitted by the manufacturing tolerances. The end -of-life point is
commonly taken as 80 per cent. of the value obtained with a valve having bogie charac-
teristics (R.M.A., Ref. S7).
The rectification test may be carried out either with full-wave rectification as in
Fig. 3.5 or with half -wave rectification, each unit being tested separately with half the
total average load current. The test is usually performed by inserting a cold valve and
waiting until it has attained normal temperature.
In all cases, rectifier valves are required to operate satisfactorily in the rectification
test without arcing or sputtering.
Where valve failure has been due to arcing or sputtering, the conditions under
which this occurred may be reproduced for the purposes of testing either by
(1) allowing the valve to heat up to normal temperature from cold, the heater and
plate supply voltages being applied simultaneously, or
(2) operating the valve at normal temperatures with heater and plate voltages
applied, and then " keying " the plate voltage.
(b) Sputter and arcing
Sputter and arcing are closely akin. When an indirectly- heated close- spaced
rectifier is warming up with plate voltage applied, there is no space charge and the
electrostatic field gradient at any sharp peaks on the cathode surface may be high
enough to vapourize the coating material and lead to the formation of an arc. Even
when there are no sharp points on either cathode or plate, a discharge of gas leading to
an arc may occur when there is no space charge, and the peak current required is
greater than the available emission. Sputter does not usually occur with directly -
heated rectifiers.
Signal diodes may be tested in a rectification test as an alternative or addition
to the emission test. Diodes in diode- amplifier valves may be tested as half-wave
rectifiers with R L =0.25 megohm, CL =2µF and an applied voltage of 50 volts
r.m.s.
(cl Back emission-Back emission is emission from the plate to the cathode during
the half -cycle when the cathode is positive with respect to the plate. Any appreciable
SPECIFIC DIODE CHARACTERISTICS 101
3.3 (iii)
One convenient form is shown in Fig. 3.7 where E, and E, are adjustable in steps
of 10, while R, is variable from 5000 minimum to 100 000 ohms maximum.
Range Grid tap Plate tap E,/E1 R, min. R, max.
ohms ohms
1 -10 µmhos C B 1/10 10 000 100 000
10 -100 B B 1 10 000 100 000
100 -1000 B C 10 10 000 100 000
1000 -20 000 A C 100 5 000 100 000
Values of the other components may be :
C1 =0.5 µF, C, =1 µF, both 400 V working
C, =one gang condenser section
T, =step up transformer, 1 : 2 turns ratio
7'2 electrostatically shielded output transformer, with primary
resistance not greater than 5J ohms.
The reactances have been selected to give an error in mutual conductance not greater
than 1 per cent. in each case.
"For definitions, see Chapter 2 Sect. 1.
104 (iv) TRIODE, PENTODE, BEAM TETRODE CHARACTERISTICS 3.3
R1 =10 ohms, R2=90 ohms, R, =900 ohms, R,=10 000 ohms max., L1=5 henrys
min., L2 =20 henrys min. at max. plate current. C1 =0.5 µF, C, =0.5 µF, C, may be
two paralleled sections of a gang condenser.
Choke L1 should be of very low resistance, so that the voltage drop caused by any
reverse or positive grid current does not affect the point of operation. Transformer
T, should be electrostatically shielded.
The amplification factor is given by
R,
R
where R= effective total grid resistance in grid arm of voltage divider (i.e., either
R1 or R1+R, or R1 +R,+R,).
With R, adjustable from 1000 to 10 000 ohms, position 1 of S, will give a range of
p from 100 to 1000, position 2 from 10 to 100, and position 3 from 1 to 10. A satis-
factory value of signal voltage is about 10 volts peak. Too high a value will result in
operation over a non -linear portion of the characteristic, while also introducing the
possibility of positive grid current with some types of valves operated at low grid bias.
Capacitance currents may be balanced out by adjusting the value of C, and by moving
switch S, to the position giving capacitance balance.
NULL 4
POINT
INDICATOR
NULL
POINT
INDICATOR
R,
-cc+ - Ebb +
OHIO
Fig. 3.8. Bridge test for measuring Fig. 3.9. Method of testing for plate
amplification factor. resistance.
The amplification factor test of Fig. 3.8 may be used for pentodes with the addition
of a suitable screen supply. For values of amplification factor over 1000 a higher ratio
of the plate and grid arms of the bridge is required, that is R, /R. This will result in
some loss in accuracy at the higher values of µ due to capacitance effects.
References to measurement of mu factor and amplification factor :Refs. 2, 15,
16, 17, 18, 21, 22, 78, 79, 80, 81, 82, 84, S12.
(G) Plate resistance
Plate resistance may be measured conveniently by the dynamic voltage ratio method
developed by Tuttle (Ref. 2) and an adaptation is shown in Fig. 3.9. The value of
plate resistance is given by
R,R,
rn (4)
R,
where R,, R, and R, may have any convenient values of resistance.
One possible combination of resistances is :
3.3 (iv) TRIODE, PENTODE, BEAM TETRODE CHARACTERISTICS 105
SINUSOIDAL
SIGNAL Re SMn T SIGNAL
SOURCE .INF SOURCE POWER
OUTPUT
VOLTAGE OUTPUT
A.C. METER
FIG. 3.11
Fig. 3.10. Method of testing for a.c. Fig. 3.11. Method for testing for power
amplification (after RMA). output.
(H) A.C. amplification
The standard R.M.A. test for a.c amplification (Ref. S7) is shown in Fig. 3.10.
This is applied to audio frequency voltage amplifier pentodes, and to a-f voltage
amplifier high -mu triodes This is a resistance -coupled amplifier test normally per-
formed at zero bias with a 10 megohm grid resistor. The internal impedance of the
signal source should not exceed 2500 ohms. The test is made at rated heater or filament
voltage and normally with a plate load resistor of 0.5 megohm, and an appropriately
by- passed screen resistor in the case of a pentode. In performing this test, it is
important to specify whether the output voltage is to be read across the plate load
resistor alone (as in the R.M.A. test of Fig. 3.10) or across a capacitively- coupled
following grid resistor of specified value. It is also important, if accurate results are
to be obtained, that the valve voltmeter used be true r.m.s. reading-that is, one having
a square -law characteristic.
(I) Power output
Power output from the plate may be measured under the conditions indicated in
Fig. 3.11. The peak value of the sinusoidal voltage E, applied to the grid is normally
106 (iv) TRIODE, PENTODE, BEAM TETRODE CHARACTERISTICS 3.3
equal to the bias voltage E,,, unless otherwise specified. The signal frequency
is
preferably either 400 or 1000 c /s, in order to reduce the size of L and C. The
dance of the choke L at the signal frequency should be at least, say, 7 times
impe-
the im-
pedance of the load, while the reactance of C should be less than, say, one seventh
the load impedance. of
The power output meter for approximate and routine testing may be of the rectifier/
d.c. meter type. In this case the load resistance will be incorporated in
the power
output meter as drawn in Fig. 3.11.
For accurate measurements, however, it is essential to measure either the
voltage across, or the current through, a load resistor of specified value true r.m.s.
and unity
power factor. Generally the latter measurement is the more convenient as the current
through the load may be measured by means of a thermo -couple ammeter.
The power output of a Class B amplifier may be measured either with push-pull
operation or on each section separately. In the latter case the load resistance is one
quarter that of the plate -to-plate value for push -pull operation, and the unit not
under test is tied to the cathode. The power output from one section: owing to shifting
of the loadline, is greater than half the power output with push -pull operation (Ref.
58), but the relationship must be established experimentally in each case. Alternatively,
a resistive load may be connected directly in the plate circuit to give a closer approach
to normal push -pull operation. The value of the resistance is arranged to be one
quarter of the plate -to-plate load resistance with push -pull operation, and the plate
supply voltage equal to the specified plate voltage for normal push -pull operation.
Under these conditions the power output is given by (Ref. 58) :
P02.47 (Ib- 0.25Ib,)2RL (4a)
where P.-power output from one section, in watts
Ib= direct plate current in amperes, as measured by a d.c. meter
Ibe =plate current with no signal, in amperes,
RL =load resistance per section, in ohms.
The value of Ib. may be taken, with sufficient accuracy for many purposes, as the
pubi shed value.
In all cases with Class B amplifiers the specified value of input voltage is applied
to the grid circuit in series with a specified value of resistance. The plate
voltage
source should have good voltage stability, an electronic voltage regulator being satis-
factory. If a plate source impedance is specified, a series resistor of the appropriate
value must be inserted in the plate supply circuit.
The relevance of a particular output measured under specified conditions is signi-
ficant only in relation to the distortion present under those conditions.
(J) Distortion
Distortion in the power output test may be measured by one of several suitable
methods, provided that the signal source is truly sinusoidal. The effect of harmonics
from the signal source may be reduced by connecting a low-pass filter between
signal source and the grid. the
The distortion may be measured by a harmonic analyser, of which several types have
been described in the literature. When merely the value of the total harmonic
dis-
tortion is desired, as in determining the undistorted output, those analysers which
measure the root- mean- square value of all harmonics present are preferable to those
which measure the separate harmonics.
The method of Suits (Ref. 85) is a particularly good example of the type of analyser
which measures the harmonics separately. The Suits method requires only the
simp-
lest apparatus, and where laboratory facilities are limited this advantage may outweigh
the disadvantages involved in the computation of the total harmonic distortion.
The Belfils analyser (Refs. 86, 87) utilizes an alternating- current Wheatstone -bridge
balance for the suppression of the fundamental, and is particularly useful for direct
measurement of the total harmonic distortion. For maximum convenience, the
frequency of the audio -frequency source should be very stable. This instrument can
be operated so that it is direct reading, by maintaining a constant input voltage.
3.3 (iv) TRIODE, PENTODE, BEAM TETRODE CHARACTERISTICS 107
In the McCurdy -Blye analyser (Ref. 88), low- and high -pass filters are used to
separate the harmonics from the fundamental. This instrument is superior to the
Belfils type in that the frequency of the source may vary somewhat without necessi-
tating readjustment.
A differential analyser especially designed for power-output work has been described
by Ballantine and Cobb (Ref. 89).
If an iron-cored choke is employed for shunt feed in the plate circuit (Fig. 3.11),
care should be exercised in its selection or design to avoid the generation of harmonics
in it due to the non-linear and hysteretic behaviour of the iron.
For measurement of distortion see Chapter 14 Sects. 2, 3 ; Chapter 37 Sect. 3.
(K) Microphony
There is no published standard test for microphony outside of Service or manufac-
turers' specifications. However an indication of the Microphony of valves may be
obtained using the same conditions as the R.M.A. standard audio frequency noise
test (see below). It is reasonable to adjust the amplifier gain to be of the same order
as the gain of the subsequent portion of the amplifier in which the valve is to operate.
Nevertheless, valves unless specifically designed for low -level high-gain pre-amplifier
use, should not be expected to be non -microphonic and free from noise at low levels
when tested under such conditions.
In radio receiver factories the valves are usually tested in the chassis in which they
are to operate, the set being tuned to a signal generator with internal modulation, and
the volume control set at maximum. Any microphonic sounds which quickly die
away are not considered as cause for removal of the valve, but a sustained
howl is so considered. Such a valve may not be microphonic in another
position in the chassis or in another model of receiver. Actions to be
taken to reduce microphony have been described in Chapter 35 Sect. 3(iv). Not-
withstanding the considerations above, all valves cannot be expected to be non -
microphonic when placed directly in the acoustic field of, and close to, the loudspeaker
in a receiver, particularly in small cabinets.
Ref. S2.
(L) Audio frequency noise
The R.M.A. standard for audio frequency noise test is with normal voltages applied
to all electrodes, and the plate of the valve under test is coupled to the input of an
amplifier with a frequency range of 100 to 2000 c/s (f 5 db of response at 400 c /s).
A dynamic loudspeaker with a rating of at least 2.5 watts is connected to the output
of the amplifier. The minimum external grid resistance is 100 000 ohms. The gain of
the amplifier and the plate load resistance of the valve under test are to be as specified.
The valve under test may be tapped lightly with a felt or cork mallet weighing not over
ounce. Any objectionable noise heard in the loudspeaker, is cause for rejection
(Ref. S7). See also Ref. S2.
(M) Radio frequency noise
The R.M.A. standard test for radio frequency noise is for normal voltages applied
to all electrodes of the valve under test ; the plate is coupled to the input of a r-f
amplifier (of a radio receiver) at a frequency within the range oí50 to 1600 Kc /s.
The minimum band-width is to be 5 Kc /s at 10 times signal. The audio frequency
response is to be within ±5 db of the response at 400 c/s over the frequency range
from 100 to 2000 c /s. A suitable tuned transformer giving a resonant impedance of
50 000 to 200 000 ohms is to be used to couple the valve under test to the amplifier.
The external grid circuit impedance of the valve under test is to be from 5000 to 25 000
ohms. Other conditions of test are as for the audio frequency noise test (Ref. S7).
See also Ref. S2.
(N) Blocking
Tests on grid blocking are normally applied only to certain types subject to this
characteristic. A valve may be tested for blocking by either of two methods. In the
first method a high positive grid voltage (say 100 volts) is applied instantaneously
with normal electrode voltages and the maximum value of grid resistance. The plate
current should return to its initial value on the removal of the applied voltage.
108 (iv) TRIODE, PENTODE, BEAM TETRODE CHARACTERISTICS 3.3
The second method (" cross- over ") is to apply a gradually increasing positive grid
voltage with substantially zero grid circuit resistance, and to plot the grid current
characteristic for positive values of grid voltage up to the point at which the grid
current changes from positive to negative (" cross- over " point). See Chapter 2
Sect. 2(iii).
(0) Stage gain testing
A circuit such as that shown in Fig. 3.11A can be used to measure the gain of a r-f
pentode operating under typical i -f amplifier conditions. Stage gain testing has the
advantage of evaluating at the one time and under operating conditions at the appropri-
ate frequency the effects of all valve characteristics which influence gain, such as mutual
conductance, output impedance, input resistance, reverse grid current and even the
effect of high screen or plate currents in increasing control grid bias or decreasing
screen grid voltages
VALVE
VOLTMETER
+ EbD
FIG. 3.13
Fig. 3.13. Side -band method of measuring conversion transconductance.
The voltage across the resonant circuit comprising the oscillator frequency, the signal
frequency and the two resulting sidebands are applied to a radio frequency amplifier
followed by a diode detector across whose diode load is connected a valve voltmeter
to indicate the amplitude of the audio frequency components. If an r-f amplifier is not
used, the region of non -linearity of the diode characteristic results in the calibration of
the system being dependent on the voltage developed across the tuned circuit and would
thus be unsatisfactory for valves of low conversion conductance. However, the r -f
amplifier may be omitted provided the signal applied to the diode is sufficiently high
to ensure linear rectification. Where this is not so, the simplified circuit may not be
suitable for measurements of low values of conversion conductance.
If the r -f amplifier is omitted, the signal voltage El may be about 0.5 volt peak, and
R1 may be 10 000 ohms.
The tester must be calibrated against the standard phase reversal method and the
circuit may be adjusted so that the meter is direct reading.
(e) Mixer plate resistance
This may be measured either with a self -excited or driven oscillator, the value of
plate resistance being not affected by the method used. Measurement of plate resistance
is performed as for a pentode, except for the excitation of the oscillator.
(f) Mixer transconductance
This may be measured as for a pentode valve, except that the oscillator should be
either self-excited or driven.
(g) Oscillator grid current
The oscillator performance of a converter valve depends upon several valve and
circuit parameters that are not simply defined.
112 (v) SPECIFIC CONVERTER CHARACTERISTICS 3.3
The oscillator grid current may be used as the criterion of oscillator performance
in a self- excited oscillator, operating under specified conditions. The circuit should
be designed to draw typical peak and average total space currents. The Boonton
Converter Oscillator test circuit is widely used to check oscillator performance, owing
to its simplicity of design and ease of adjustment and control and, while not giving an
exact criterion of oscillator performance, does provide a dynamic test of the usual
parameters under conditions that simulate average receiver operation.
The circuit (Fig. 3.14) is a series -fed Hartley having a fixed feedback ratio and having
the tank circuit impedance variable so that the magnitude of oscillation may be
adjusted. Normal voltages are applied to all electrodes.
It is necessary to specify the oscillator plate voltage, oscillator plate series
resistor (if used), oscillator grid resistor (Re) the inductance of each section of the coil
(L1, L2), the mutual inductance between the two sections (M), the capacitance shunted
across the whole coil (C), the grid coupling capacitance (CO, and the resonant imped-
ance of the tuned circuit (Rd). Normally a coil having a higher reactance than the
specified value of resonant impedance is used and is shunted by a variable resistance
(R1) to give the specified resonant impedance. Bias is obtained by means of a capacitor
and a wire -wound gridleak. The wire-wound grid resistor is used because it has con-
siderably higher impedance to radio frequencies than to direct current. The high
radio -frequency impedance diminishes the shunting effect of the gridleak on the tuned
circuit. The rectified grid current is read on a microammeter (suitably by- passed) in
series with the grid resistor. Before the test oscillator is used, the relationship between
the tuned impedance of the tank circuit and the setting of the variable resistor must
be determined
The R.M.A. standard (Ref. S7) and JAN -1A (Ref. S2, sheet dated 19 July, 1944)
specify the component and coil design values as being : C =100 µµF (mica) ; C,=
200 µµF (mica) ; R1 =50 000 ohms potentiometer ; R, =50 000 ohms wire -wound
resistor ; L1 =83 µH ; L2 =48 µH : M =23.3 µH ; coil diameter -1.25 in. ; winding
length = 59/64 in. ; wire No. 30 a.w.g. enamelled copper ; turns = 83 ; tap at 33
turns from anode end. See also Ref. S12.
The constants quoted above have been selected to simulate average circuit conditions
in broadcast receivers for composite converters having transconductances less than
1500 mhos at zero bias. For valves having higher transconductances or for circuits
above broadcast frequencies, the circuit with the constants given may not be satis-
factory, as spurious oscillations may make it impossible to obtain correlation between
test readings and receiver performance.
The ability of a valve to oscillate when the shunt impedance of a tank circuit is low
is one criterion of the value of a valve as an oscillator. The oscillation test is made by
applying the desired electrode potentials to the valve under test and reading the rectified
grid current at some known setting of tank- circuit impedance. A valve that will not
oscillate with the tank -circuit impedance below about 9000 ohms is considered to be a
weak oscillator. A valve that will oscillate with a tank -circuit impedance as low as
3000 ohms is considered to be a strong oscillator. The minimum tuned impedance at
which oscillation will start or cease as indicated by the rectified grid current will
depend on the oscillator characteristics of the particular type and will vary from type
to type.
3.3 (v) SPECIFIC CONVERTER CHARACTERISTICS 113
SIGNAL
GENERATOR
FIG. 3.1S
Fig. 3.15. Semi-schematic diagram of equipment for measuring admittances
by the
susceptance- variation method (Ref. S12).
T is the 4 terminal transducer to be measured, Y, and Y, are calibrated variable- admittance
elements which may be of various forms, such as coils or capacitors. V, and V, are signal
-
frequency voltage- measuring devices ; these may be simply crystal or diode voltmeters
or
heterodyne receivers. Variable admittances Y,, and Y1, are used for coupling the input
or output circuits to the signal oscillators.
Fig. 3.16. Circuit diagram of equipment used for measurement of short -circuit input
admittance (Ref. 51).
A practical form (Ref. 51) for the frequency range 50 to 150 Mc /s is shown in Figs
3.16 and 3.17. The valve under test is used as a part of a resonant circuit which includes
a continuously-variable inductor and a small concentric -cylinder capacitor built on a
micrometer head. The high-potential end of the inductor is connected to the high -
potential electrode of the micrometer capacitor inside a cylindrical cavity open at the
top. Fig. 3.17 also shows the positions of some of the by -pass capacitors used with the
octal socket. These are button -type, silver -mica capacitors of approximately 500 µµF
capacitance. The socket is of the moulded phenolic type. Terminals 1, 3, 5 and 7 are
connected directly to the mounting plate at a point directly below the terminal in each
instance. Terminals 2 (heater) and 6 (screen grid) are by- passed to ground, and a lead
is brought from each of these terminals through the mounting plate to a small r -f
choke and a second by-pass capacitor. Terminal 8 (plate) is by- passed and fitted
with a spring contacting the terminal for one of the circuits. Terminal 4 (grid) has
only the contact spring. A similar arrangement is used with miniature valves. In
this case, the socket is of the wafer type with mica -filled rubber insulation. Terminals
2, 3 and 7 are grounded and terminals 4, 5 and 6 are by- passed to ground. Terminals
1 (grid) and 5 (plate) have contact springs. The by -pass capacitors are closer to the
mounting plate than in the case of the octal socket ; the capacitors at terminals 4 and
6 overlap the capacitor to terminal 5. A 10 ohm resistor, mounted inside a cylindrical
shield to minimize lead inductance, is connected between socket terminal 5 (plate)
and the by -pass capacitor. This component was added to suppress a parasitic oscilla-
tion observed with certain type 6AK5 valves. The resistor was found to have no
3.3 (vi) TESTS FOR SPECIAL CHARACTERISTICS 115
FIG. 3.17
RF CNOKE
BLOCKING CAPACITO.
AT TERMINAL Of SPLIT
VARIABLE INDUCTOR SPACING
RING
CALIBRATING CAPACITO.
MOUNTING SUSST MUTED
FOR TUBO MOUNTING
SCREEN GRID
CAPACITOR
O IGN.PO TE NT IAI.
ELECTRODE Of MICRO-
METER CAPACITOR
TNOSPNOR SRONEE SPRING
Fig. 3.17. Physical arrangement of circuit elements of Fig. 3.16 (Ref. 51).
measurable effect on input admittance readings obtained with valves, either of the
6AK5 type or of other types, not subject to the parasitic oscillation.
In order to obtain susceptance values, the circuit must first be calibrated for the
capacitance required for resonance at each test frequency. The circuit is calibrated
by determining the inductor settings for resonance with each of a number of small,
disc-shaped, calibrating capacitors substituted for the valve. The insert in the lower
left corner of Fig. 3.17 shows a cutaway view of the cavity with one of the calibrating
capacitors in place. The length of the phosphor -bronze contact spring used with the
calibrating capacitors is approximately the same as that used with the valve. Thus,
the inductance of this lead is accounted for in the calibrating procedure. The reference
terminals for the valve are the socket plate and the grid terminal of the socket or,
possibly, a point on the grid terminal a little inside the body of the socket. The
calculated inductance of the contact spring is 4.5 milli-microhenries per centimeter of
length within about ± 25 per cent., but the difference in effective lengths of the springs
for the socket connection and the calibrating connection is not more than 2 or 3
millimeters.
The range of the micrometer capacitor is sufficient for measurements of the circuit
with the calibrating capacitors, with most cold valves, and with some valves under
operating conditions. For other cases, adjustment of the inductor is required.
If the conductance being measured is small enough to be measured by the available
range_of the micrometer capacitor, the susceptance and conductance variations between
1 /V/2 voltage points are given by
dB= cod
and AG =O
where w is the angular frequency of measurement and d C is the micrometer capaci-
t ance variation between the half-power points.
The calibration curves used at each test frequency are :
(1) Capacitance for resonance
(2) Slope of the capacitance curve
(3) Conductance at resonance of the circuit with the calibrating capacitors.
These three quantities are plotted against the inductor adjustment readings. Since
the conductance values for the calibrating capacitors themselves are too small to affect
the calibration appreciably, the conductance curve corresponds, essentially, to the
equipment.
116 (vi) TESTS FOR SPECIAL CHARACTERISTICS 3.3
is taken. The transducer is removed and one of the standard resistors placed across the
line bridging the two conductors. The position of this resistor along the line is then
adjusted and the system is readjusted for resonance with the calibrated capacitor until
the voltage, as measured at the end of the line, is the same as before. By (15) we have
Gil =G
l[sin 1312 (16)
R sin 131
where R is the resistance of the standard placed x centimeters from the short-circuited
end of the line. It is evident that a resistor must be selected having a resistance value
near to but not larger than the reciprocal of G.
Since the transmission line should have low loss and low characteristic impedance,
a coaxial line is desirable. The line will require a longitudinal opening or slot in order
to permit one of the standard resistors to bridge the line at an adjustable position to
satisfy the required voltage condition. A valve voltmeter is capacitively coupled
to the open end of the line across which the electron-tube transducer may be attached.
Socket and filter arrangements for wire-lead valves can be attached to this line.
Radiation difficulties arising from the longitudinal opening in the line, together with
the increasing difficulty in obtaining resistance standards at frequencies much above
300 Mc/s, appear to make this type of measuring equipment impracticable for measure-
ments at higher frequencies on surface-lead valves. Ref. S12.
(B)Short-circuit feedback admittance (y,,)
Susceptance-variation method of measurement
See (A)1 above for general description.
(a) With the input termination still set at the value for resonance obtained in step (c)
in (A) above, excite the output circuit through Ye,. In the event that oscillation diffi-
culties are encountered, detune the output circuit Y2 or load it until oscillation stops.
(b) Record the voltmeter readings V, and V2.
The magnitude of the feedback admittance is then given by the relation
Yi2i - z12131 [ (1 + 2 772)1 ' + ni (17)
dB, r 1 + n2 (18)
or iyi,i _1
413,1V,
or IYizIjÏ (19)
where AG, and AB, are the values obtained in the preceding measurement of y11.
Ref. S12.
(C) Short-circuit output admittance (y22)
The short-circuit output admittance may be measured by following the procedure
outlined for the short-circuit input admittance, the signal being coupled through Ye,.
If the subscripts 1 and 2 are interchanged, all of the above formulae concerning
may be used to relate y22 to the measured data.
y
Ref. S12.
(D) Short-circuit forward admittance (y21)
The magnitude of the forward admittance may be measured by following the
procedure outlined previously for the measurement of the magnitude of y22. If the
subscripts 1 and 2 are interchanged, all of the formulae concerning Yi, may be used to
relate y22 to the measured data.
Ref S12.
(E) Perveance
(1) Perveance of a diode
The perveance of a diode may be derived by plotting the current against the three-
halves power of the voltage, when the slope of the curve will give the value of G.
Voltages should be chosen sufficiently high so that such effects as those produced by
118 (vi) TESTS FOR SPECIAL CHARACTERISTICS 3.3
contact potential and initial electron velocity are unimportant. If this is not possible,
a correction voltage E should be added to the diode voltage, where
E =(3 /2)Ibrb -eb
PLATE
VOLTAGE
REGULATOR IzoSCREEN VOLTAGE
REGULATOR
Oo SCREEN
CATHODE
o
GRID
o
PULSE
STABILISED
GENERATOR
SUPPLY GRID
HEATER O
VOLTAGE
HEATER
FIG. 3.18 SOURCE
t
Fig. 3.18. Circuit for measuring electrode currents using single pulse generator method
(I= peak-reading current indicator, E =peak-reading voltage indicator, V= moving coil
voltmeter). If high accuracy is required, correction should be made for the voltage drop
across the current indicators. The regulation of the electrode supplies is not important
for point by point testing, provided that the shunt capacitors are sufficiently large.
The plate voltage versus plate current characteristic may be traced by maintaining
the grid (and screen, if any) at constant voltage and by applying the pulse voltage to
the plate. This requires a peak voltage of several hundred volts for receiving valves,
together with high peak currents. The grid voltage must be maintained constant, the
regulation requirements being severe. The regulation requirements of the other
electrodes are not important, if large capacitors are used.
A method of showing the two axes for any single characteristic is described in Ref. 73
which makes use of a standard vibrator.
References to curve tracer methods 37, 39, 69, 70, 71, 72.
:
(a) Visually inspect all valves 100 per cent. for mechanical defects.
(b) Test all valves 100 per cent. for shorts and open circuits.
(c) Select at random from all the valves a representative sample (or samples) and
test for a suitable dynamic characteristic using established statistical acceptance pro-
cedure. The test selected should give a satisfactory indication of the overall per-
formance of the type in its normal application.
References to sampling inspection : Refs. 42, 43, 44.
3.5 (i) PURPOSE AND SCOPE OF SERVICE TESTING 121
pentodes. The method of testing for mutual conductance with a high grid swing is
not the equivalent of a power output test. Neither mutual conductance nor emission
testing is a satisfactory substitute for a.c. amplification testing of high -mu triodes or
a -f pentodes.
It is obvious, therefore, that any ordinary commercial service tester does not give a
true indication of the condition of all types of valves, and its readings should be used
with discretion.
(iv) Methods of testing characteristics in commercial service testers
(a) Shorts testing
Testing for shorts should be carried out at mains frequency or on d.c., and the
voltage applied to the neon lamp circuit should not exceed 110 volts r.m.s. The use
of higher voltages results in some good valves being classed as filament -to -grid short
circuits.
With a.c. supply, it is usual to design the circuit so that resistances over 0.5 megohm
do not indicate as shorts ; this is to avoid capacitance effects. With d.c. supplies there
is no similar limitation, but different models vary considerably in their shorts testing
sensitivity.
The most satisfactory switching arrangement is one which tests for shorts from any
pin to all other pins tied together. The circuit of Fig. 3.4 may be used, except that the
cathode need not be heated.
(b) Continuity testing
Continuity testing may be carried out in a manner closely resembling shorts testing,
but with the cathode hot. Positive or alternating voltage is applied through the neon
lamp to each electrode in turn ; a glow indicates continuity.
The circuit of Fig. 3.4 may be used in a simple service tester, but a more elaborate
tester such as the hexaphase (Fig. 3.2) is much more satisfactory for general use.
For further information on continuity testing see Sect. 3(ii)b.
(c) Heater to cathode leakage
The maximum leakage resistance from heater to cathode in new valves varies from
1 or 2 megohms for power valves to 5 megohms for ordinary 6.3 volt 0.3 ampere
voltage amplifier valves and even higher for rectifier types having separate heaters
and cathodes. A neon lamp will normally glow slightly when used in a shorts tester
and applied between heater and cathode. A neon lamp may be de- sensitized either
by shunting by a resistor (of the order of 0.1 to 0.5 megohm) or by a mica condenser
(a value of 0.01 p.F gives a sensitivity of about 5 megohms-this is only useful with
a.c. supply).
(d) Emission testing
Of the 7 commercial testers which included emission testing of amplifying valves,
5 used a.c. testing, 1 used half -wave rectified a.c. testing, and I used d.c. testing. Owing
to the poor regulation obtainable from most service valve testers with d.c. supplies,
other than batteries, it is doubtful whether this has any practical advantages over a.c.
testing. With a.c. supply, the voltage in most service testers is 30 volts r.m.s. ; this is
only a compromise which is far from the optimum for all cases. For laboratory
testing of emission see Sect. 3(ii)f.
(e) Mutual conductance testing
Of the
17 commercial testers having a mutual conductance test, 7 used some form of
a.c. signal voltage applied tc the control grid, and 6 used some grid shift method, the
methods used by the other 4 being unknown.
All the American mutual conductance testers examined apply an a.c. signal to the
grid, and measure the signal current in the plate circuit ; all but one of these use a
mains frequency signal, the exception being one with a 5 Kc/s signal and a tuned signal
plate current meter.
In most cases this is not an accurate g,,, test for all types of valves because the signal
voltage on the grid is fairly high- between 4 and 5 volts r.m.s. in some cases- sufficient
to run beyond grid cut -off in one direction and well on to the " flat top " in the other,
124 (iv) METHODS OF TESTING IN COMMERCIAL TESTERS 3.5
with short grid-base valves. Even those testers with reduced signal voltages have
quite appreciable errors with some types of valves.
On the other hand, all but one of the examined English mutual conductance testers
use a grid shift method. This introduces some errors due to plate supply regulation
and curvature of the characteristics, but these can be allowed for in the calibration.
(f) Plate conductance testing
This is used by one American tester only as a reasonably close equivalent of the
mutual conductance test, the amplification factor being assumed to be constant.
Pentodes are triode- connected. A moderately high signal voltage is applied, this being
possible with less error than with direct g measurement owing to the slight compensa-
ting change of µ with plate current.
(g) Reverse grid current testing
Of the post -war testers examined, all those of English design incorporate reverse
grid current testing (often referred to as " gas testing ") but this test was incorporated
in only two of American design. The usual methods of testing are either by grid current
meter or by a shorted grid resistor. For further details see Sect. 3(iv)A.
A valve which passes the reverse grid current test on a service tester may not operate
satisfactorily in a receiver. The usable maximum value of negative grid current depends
upon the application, the mutual conductance of the valve, the resistance in the grid
circuit, the type of bias, the bias voltages, the total dissipation of the valve and its
ventilation. The valve may be checked in the actual receiver after running for a
period sufficiently long for the valve to reach its stable operating temperature ; it
may be tested by variation in the plate current of the valve under test when its grid
resistor is shorted.
(h) Power output testing
This is not incorporated into any of the commercial service testers examined, but
is the most satisfactory test for all power valves. Triode testing of pentode and beam
power valves is quite satisfactory. Methods of testing are described in Sect. 3(iv)I.
(i) Conversion conductance
Only one of the testers examined provided a test for conversion conductance, the
method being by phase reversal. For further details see Sect. 3(v)Bd.
A satisfactory alternative is to measure the transconductance of the mixer, this
being the method commonly employed.
(j) Oscillator mutual conductance
None of the testers provided a test for oscillator grid current under self-oscillating
conditions, although this is the most satisfactory form of test. For details see Sect.
3(v)Bg.
The method commonly employed is to measure the oscillator mutual conductance.
The measurement of oscillator grid current in a suitable receiver is a simple way of
checking the oscillator section of a converter.
(k) Noise
The usual form of noise test is to connect the two neon lamp terminals to the aerial
and earth terminals of a receiver, during the shorts test. For other forms of noise test
see Sect. 3(iv)L and M.
(1) A.C. amplification
This test should be carried out with zero bias and a high value of grid resistor, as
described in Sect. 3(iv)H. No test with fixed or self-bias is able to determine how a
valve will operate with grid resistor bias.
(v) A.C. versus d.c. electrode voltages in testers
All but four of the commercial testers examined, which made provision for some form
of g,, test, used d.c. electrode supplies. The general trend seems to be in the direction
of using d.c. supply to all electrodes, and the additional circuit complications appear
to be justified.
For comparison between laboratory testing of emission by a.c. or d.c. methods,
see Sect. 3(ii)f.
3.5 (vi) PRE -HEATING 125
SECTION 6: REFERENCES
1. Benjamin, M., C. W. Cosgrove and G. W. Warren. " Modem receiving valves : design and
manufacture " Jour. I.E.E. 80.484 (April 1937) 401.
2. Tuttle, W. N. " Dynamic measurement of electron tube coefficients " Proc. I.R.F. 21.6 (June
1933) 844.
3. Lehany, F. J., and W. S. McGuire. " A radio- frequency interelectrode- capacitance meter "
A.W.A. Tec. Rev. 7.3 (April 1947) 271. See also Ref. 5.
4. Pockman, L. T. "The dependence of interelectrode capacitance on shielding " Proc. I.R.E.
32.2 (Feb. 1944) 91.
5. Dobbie, L. G., and R. M. Huey. " A.W.A. interelectrode capacity meter Type R1782 for screen
grid valves " A.W.A. Tec. Rev. 2.4 (Oct. 1936) 107.
6. Van der Biil (book). ` The Thermionic Vacuum Tube " (McGraw -Hill Book Co. 1920).
7. Pidduck, F. B. (book). " A Treatise,on Electricity " (Cambridge University Press, 1925).
8. Hague, B. (book). ` Alternating Current Bridge Methods ' (Sir Isaac Pitman & Sons Ltd.,
London, 5th edit., 1943).
9. Hartshorn, L. (book). ' Radio Frequency Measurements by Bridge and Resonance Methods "
(Chapman & Hall Ltd., London, 1942).
10. Thomson, J. J. (book). Elements of Electricity and Magnetism" (Cambridge University Press,
5th edit., 1921).
11. Jeans, J. (book). " The Mathematical Theory of Electricity and Magnetism " (Cambridge Uni -
versity Press, 5th edit., 1943).
12. Smythe, W. R. (book). " Static and Dynamic Electricity " (McGraw -Hill Book Co., and ed. 1950).
13. Maxwell, J. C. (book). Electricity and Magnetism" (published 1892).
14. Valley, G. E., and H. Waltman (editors). " Vacuum Tube Amplifiers " Radiation Laboratory
Series, McGraw -Hill Book Co., 1948.
15. Reich, H. J. (book). Theory and Applications of Electron Tubes " (McGraw -Hill Book Co.,
1939).
16. Chaffee, E. L. (book). " Theory of Thermionic Vacuum Tubes " (McGraw -Hill Book Co., 1933).
17. Scroggie, M. G. (book). " Radio Laboratory Handbook " (Wireless World, London).
18. Terman, F. E. (book). " Measurements in Radio Engineering " (McGraw -Hill Book Co., 1935).
20. Wireless Telegraphy Board Specification K1001 - radio valves.
21. Hund, A. (book). High Frequency Measurements" (McGraw -Hill Book Co., 1933).
22. Miller, J. M. ' A dynamic method for determining the characteristics of three -electrode vacuum
tubes " Proc. I.R.E. 6 (June, 1918) 141.
-
23. James, E. G., and B. L. Humphreys. " Resistive films in valves effect on interelectrode capaci-
tance " W.E. 26.306 (March 1949) 93.
24. Booth, R. H. (letter). " Triode interelectrode capacitances " W.E. 26.309 (June 1949) 211.
25. Walsh, L. ' A direct capacity bridge for vacuum -tube measurements " Proc. I.R.E. 16.4 (April
1928) 482.
26. Wheeler. H. A. " Measurement of vacuum -tube capacities by means of a transformer balance "
Proc. I.R.E. 16.4 (April 1928) 476.
126 REFERENCES 3.6
27. Hoch, E. T. " A bridge method for the measurement of interelectrode admittance in vacuum
tubes " Proc. I.R.E. 16.4 (April 1928) 487.
28. Loughren, A. V., and H. W. Parker. " The measurement of direct interelectrode capacitance of
vacuum tubes " Proc. I.R.E. 17.6 (June 1929) 957.
29. Jones, T. I. " The measurement of the grid -anode capacitance of screen-grid valves " Jour. I.E.E.
74.450 (June 1934) 589.
30. Astbury, N. F., and T. I. Jones. " A capacitance attenuator and its application to the measurement
of very small capacitances " Jour. Sci. Insr. 13.12 (Dec. 1936) 407.
31. Barco, A. A. " An improved interelectrode capacitance meter " R.C.A. Rev. 6.4 (April 1942) 434.
32. Tillman, J. R., and A. C. Lynch. " Apparatus for the measurement of small three -terminal
capacitances " Jour. Sci. Insr. 19.8 (Aug. 1942) 122.
33. Proctor, R. F., and E. G. James. " A radio -frequency and conductance bridge " Jour. I.E.E. 92
Part III. 20 (Dec. 1945) 287.
34. van Suchtelen, H. " Applications of cathode ray tubes : iv-the recording of diagrams " Philips
Tec. Rev. 3.11 (Nov. 1938) 339.
35. Douma, Tj., and P. Ziilsra. " Recording the characteristics of transmitting valves " Philips Tec.
Rev. 4.2 (Feb. 1939) 56.
36. van der Ven, A. J. H. " Testing amplifier output valves by means of the cathode ray tube "
Philips Tec. Rev. 5.3 (March 1940) 61.
37. Millman, J., and S. Moskowitz. " Tracing tube characteristics on a cathode ray oscilloscope "
Elect. 14.3 (March 1941) 36.
38. Easton, E. C., and E. L. Chaffee. " Pulse -type tester for high-power tubes " Elect. 20.2 (Feb.
1947) 97.
39. Webking, H. E. " Producing tube curves on an oscilloscope " Elect. 20.11 (Nov. 1947) 128.
40. Herold, E. W. ` An improved method of testing for residual gas in electron tubes and vacuum
systems " R.C.A. Rev. 10.3 (Sept. 1949) 430.
41. Kauzmann, A. P. " New television amplifier receiving tubes " R.C.A. Rev. 3.3 (Jan. 1939) 271.
42. Dodge, H. F., and H. G. Romig. " Single sampling and double sampling inspection tables "
B.S.T.J. Jan. 1941 20.1 (Jan. 1941) 1.
43. Dodge, H. F., and H. G. Romig (book). " Sampling Inspection Tables " (John Wiley & Sons
Inc., New York, Chapman & Hall. London, 1944).
44. Keeling, D. B., and L. E. Cisne. " Using double sampling inspection in a manufacturing plant "
B.S.T.J. 21.1 (June 1942) 37.
45. Davies, J A. " Quality control in radio tube manufacture " Proc. I.R.E. 37.5 (May 1949) 548.
46. Harris, J A. " On the co-ordination of circuit requirements, valve characteristics and electrode
design " J. Brit. I.R.E. 9.4 (April 1949) 125.
47. Slonczewski, T. " Transconductance as a criterion of electron tube performance " B.S.T.J.
28.2 (April 1949) 315.
48. British Radio Valve Manufacturers' Association (booklet). " Radio Valve Practice " (August 1948).
49. British Standard Code of Practice BS1106 (Feb. 1943). " The use of radio valves in equipment."
51. Application Note. " Input admittance of receiving tubes " (AN -118, April 15, 1947) Radio
Corporation of America.
52. " Tube Handbook " HB3, Radio Corporation of America.
53. " Radio Components Handbook " (1st ed. 1948, Technical Advertising Associates, Cheltenham,
Penn., U.S.A.).
55. Hartkopf, R. E. " Simple valve tester-checking insulation, mutual conductance and emission "
W.W. 42.12 (Dec. 1946) 386.
56. Everett Edgcumbe Service Valve Tester, W.W. 42.12 (Dec. 1946) 416.
57. " Dynamic mutual conductance tube tester " (Hickok design) Service 16.3 (March 1947) 2.
58. Heacock, D. P. " Power measurement of Class B audio amplifier tubes " R.C.A. Rev. 8.1 (March
1947) 147.
59. Walker, H. S., C. G. Mayo and J. Tomlinson " Portable sub-standard of mutual conductance "
B.B.C. Research Dept. Report H.009 (27 Oct. 1938).
60. " A universal tube checker " Radio Craft (Feb. 1944) 292.
61. Dewar, J. A. ' Dynamic tube tester " Radio Craft (Nov. 1944) 94.
62. Planer, F. E. " Valve Testing -rapid determination of amplification factor and anode impedance
under operating conditions" W.W. 51.6 (June 1945) 20.
63. Simpson. ' Dynamic tube tester " Elect. 19.2 (Feb. 1946) 338.
64. " Hickok, Tube Tester Models 532C and 532P " Elect. 19.3 (Mar. 1946) 282.
65. Blackburn, J. F. (editor). " Components Handbook " (M.I.T. Radiation Laboratory Serles,
No. 17, 1st ed. 1949, McGraw -Hill Book Co.).
66. Laws, F. A. (book). " Electrical Measurements " (McGraw -Hill Book Co. 2nd ed. 1938).
67. Appleton, E. V. (book). " Thermionic Vacuum Tubes " (Methuen & Co. Ltd., London).
68. Livingston, O. W. " Oscillographic method of measuring positive-grid characteristics " Proc.
I.R.E. 28.6 (June 1940) 267.
69. Kozanowski, H. N., and I. E. Mouromtseff. " Vacuum tube characteristics in the positive grid
region by an oscillograpphic method " Proc. I.R.E. 21.8 (April 1933) 1082.
70. Chaffee, E. L. " Oscillographic study of electron tube characteristics " Proc. I.R.E. 10.6 (Dec.
1922) 440.
71. Schneider, W. A. " Use of an oscillograph for recording vacuum tube characteristics " Proc.
I.R.E. 16.5 (May 1928) 674.
72. Chaffee, E. L. ' Power tube characteristics " Elect. 11.6 (June 1938) 34.
73. Walker, A. H. B. "Cathode-ray curve tracer" W.W. 50.9 (Sept. 1944) 266.
74. Zepler, E. E. ' Triode interelectrode capacitances " W.E. 26.305 (Feb. 1949) 53.
75. Jones, T. I. " The dependence of the inter -electrode capacitances of valves upon the working
conditions " Jour. I.E.E. 81 (1937) 658.
76. Humphreys, B. L., and B. G. James. " Interelectrode capacitance of valves-change with operating
conditions " W.E. 26.304 (Jan. 1949) 26.
77. Young, C. H. " Measuring interelectrode capacitances " Tele-Tech 6.2 (Feb. 1947) 68.
78. Terman, F. E. " Radio Engineers' Handbook " (McGraw -Hill Book Co., 1943).
79. Dawes, C. L. (book). " A Course in Electrical Engineering " Vol. 2 (McGraw-Hill Book Co.
3rd edit., 1934).
80. Hickman, R., and F. Hunt. " The exact measurement of electron tube coefficients " Rev. Sci.
Instr. 6 (Sept. 1935) 268.
81. Eastman, A. V. (book). " Fundamentals of Vacuum Tubes " (McGraw -Hill Book Co. 1937).
82. Albert, A. L. (book). " Fundamental Electronics and Vacuum Tubes " (The Macmillan Co.,
New York, 1938).
3.6 REFERENCES 127
83. Aiken, C. B., and J. F. Bell. " A mutual conductance meter " Comm. 18.9 (Sept. 1938) 19.
84. Millman, J., and S. Seely (book). " Electronics " (McGraw -Hill Book Co. 1941).
85. Suits, C. G. " A thermionic voltmeter method for the harmonic analysis of electrical waves "Proc.
I.R.E. 18 (Jan. 1930) 178.
86. Belfils. G. " Mesur du ' residu ' des courbes de tension par la methode du pont filtrant " Rev.
Gen. d'Elect. 19 (April 1926) 526.
87. Wolff, Irving. " The alternating current bridge as a harmonic analyzer " Jour. Opt. Soc. Amer.
15 (Sept. 1927) 163.
88. McCurdy, R. G., and P. W. Blye. " Electrical wave analyzers for power and telephone systems "
Jour. A.I.E.E. 48 (June 1929) 461.
89. Ballantine, S., and H. L. Cobb. " Power output characteristics of the pentode " Proc. I.R.E. 18
(March 1930) 450.
90. " Curve generator for tubes " developed by M. L. Kuder of N.B.S., Elect. 23.7 (July 1950) 144.
91. " Heater-cathode leakage as a source of hum " Elect. 13.2 (Feb. 1940) 48.
92. Zepler, E. E. and S. S. Srivastava " Interelectrode impedances in triodes and pentodes " W.E .
28.332 (May 1951) 146.
93. Wagner, H. M. " Tube characteristic tracer using pulse techniques " Elect. 24.4 (Apr. 1951) 110.
Additional references will be found in the Supplement commencing on page 1475.
Standards Specifications.
SI. Standards on electronics (1938) The Institute of Radio Engineers. U.S.A. See also S12.
S2. Joint Army-Navy Specification JAN-1A (U.S.A.).
S3. Joint Army -Navy Specification JAN -S -28 Sockets, electron tube, miniature.
S4. RMA -NEMA' Standards for electron tube bases caps and terminals (Feb. 1949) -RMA
Standard ET -103A.
S5. RMA -NEMA Dimensional characteristics of electron tubes (Jan. 1949) -RMA Standard
ET -105A.
S6. RMA -NEMA-JETEC Standard ET -106A (Feb. 1952) Gauges for electron tube bases.
S7. RMA Standard ET -107 (Dec. 1946) Test methods and procedures for radio receiving tubes.
S8. RMA -NEMA' (Dec. 1947) Standards for measurement of direct interelectrode capacitances
RMA Standard ET-109.
-
S9. RMA -NEMA* (Feb. 1948) Standard for designation system for receiving tubes -RMA Standard
ET -110.
S10. RMA -NEMA' (Feb. 1949) Designation system for electron tube shells -RMA Standard ET -112.
SI 1. "Tube Ratings" -RMA Standard M8-210 (Jan. 8, 1940).
S12. " Standards on electron tubes : methods of testing 1950 " Proc. I.R.E., Part 1, 38.8 (Aug. 1950)
917 ; Part 2, 38.9 (Sept. 1950) 1079.
THEORY OF NETWORKS
by F. LANGFORD- SMITH, B.Sc., B.E.
Section Page
1. Current and voltage ... .. 128
2. Resistance ... ... 130
3. Power ... 133
4. Capacitance ... 134
5. Inductance ... ... ... 140
6. Impedance and admittance ... 144
7. Networks ... ... ... ... 158
8. Filters ... 172
9. Practical resistors, condensers and inductors ... 186
Batteries (or cells) may be connected in series as in Fig. 4.2 and the total voltage
is then equal to the sum of the voltages of the individual batteries (or cells). When
calculating the voltage of any intermediate point with respect to (say) the negative
terminal, count the number of cells passed through from the negative terminal to the
tapping point, and multiply by the voltage per cell. When batteries are connected
in series, each has to supply the full load current.
DRY ZELLS ACCUMUL A FORS
Batteries (or cells) are occasionally
connected in parallel as in Fig. 4.3.
sY 0 i In this case the terminal voltage is
the same as the voltage per battery
CURRENT (or cell). The current does not
Sv
410
FLOW
Q 0 necessarily divide uniformly be-
tween the cells, unless these all have
Sv O
identical voltages and internal re-
s
o sistances.
Direct current may also be ob-
CURA ENT tai ned from a d.c. generator, or
FLOW
from rectified and filtered a.c.
FIG. 4.2 supply. In all such cases there is a
Fig. 4.2. Cells in series. certain degree of ripple or hum
which prevents it from being pure
d.c. ; when the a.c. component is appreciable, the supply may be spoken of as " d.c.
with superimposed ripple (or hum) " and must be treated as having the characteristics
of both d.c. and a.c. When we speak of d.c. in a theoretical treatise, it is intended to
imply pure, steady d.c.
(ii) Alternating current
The ordinary form of d.c. generator actu-
ally generates a.c., which is converted to
d.c. by the commutator. If a loop of wire
is rotated about its axis in a uniform
magnetic field, an alternating voltage is 5v
generated across its terminals. Thus a.c.
is just as fundamental as d.c. The usual FIG 4.3
power -house generates 3 phase a.c., but
in radio receivers we are only concerned Fig. 4.3. Cells in parallel.
with one of these phases. A " sine wave "
alternating current is illustrated in Fig. 4.4, where the vertical scale may represent
voltage or current, and the horizontal scale represents time.* A cycle is the alterna-
tion from A to E, or from B to F, or from C to G.
Most power supplies have frequencies of either 50 or 60 cycles per second i.c /s).
The period is the time taken by one cycle, which is 1/50 or 1/60 second, in these
two cases.
'For mathematical treatment of periodic phenomena see Chapter 6 Sect. 4.
130 (ii) ALTERNATING CURRENT 4.1
The precise shape of the wave is very important, and the sine wave has been adopted
waveform, since this is the only one which always has the current
as the standard a.c.
waveform of the same shape as the voltage, when applied to a resistance, inductance
or capacitance.t In practice we have to deal with various waveforms, some of which
may be considered as imperfect (" distorted ") sine waves, while others are of special
shapes such as square waves, saw-tooth, or pulse types, or rectified sine waves. How-
ever, when we speak of a.c. in any theoretical treatise, it is intended to imply a dis-
tortionless sine wave. Other waveforms may be resolved by Fourier Analysis into a
fundamental sine wave and a number of harmonic frequency sine waves, the latter
having frequencies which are multiples of the fundamental frequency. This subject
is treated mathematically in Chapter 6 Sect. 8.
Wher deriving the characteristics of any circuit, amplifier or network, it is usual
to assume the application of a pure sine -wave voltage to the input terminals, then to
calculate the currents and voltages in the circuit. In the case of a valve amplifier
(or any other non-linear component) the distortion may be either calculated, or
measured at the output terminals. If the device is to operate with a special input
waveform (e.g. square wave), it is usual to resolve this into its fundamental and har-
monic frequencies, and then to calculate the performance with the lowest (funda-
mental) frequency, an approximate middle** frequency, and the highest harmonic
frequency-all these being sine waves.
(iii) Indications of polarity and current flow
In circuit diagrams the polarity of any battery or other d.c. voltage source is usually
indicated by + and -
signs ; alternatively it may be indicated by an arrow, the head
of the arrow indicating positive potential (e.g. Fig. 4.2). A similar convention may
be used for the voltage between any points in the circuit (e.g. Fig. 4.14A). The
direction of d.c. current flow is indicated by an arrow.
In the case of a.c. circuits a similar convention may be used, except that an arbitrary
instantaneous condition is represented (Fig. 4.18A).
SECTION 2 : RESISTANCE
(i) Ohm's Law for d.c. (ii) Ohm's Law for a.c. (iii) Resistances in series (iv)
Resistances in parallel (v) Conductance in resistive circuits.
(i) Ohm's Law for direct current
All substances offer some obstruction to the flow of electric current. Ohm's Law
states that the current which flows is proportional to the applied voltage, in accordance
with the equation
I =E /R (1)
where R is the total resistance of the circuit. For example in Fig. 4.5 an ideal battery,
having zero internal resistance, and giving a constant voltage E under all conditions,
is connected across a resistance R. The current which flows is given by eqn. (1)
above, provided that
I is expressed in amperes,
E is expressed in volts,
and R is expressed in ohms.* *E R
In a circuit containing more than one battery (or other 5'urce of direct voltage),
the effective voltage is determined by adding together those voltages which are in the
same direction as the current, and subtracting any opposing voltages.
Ohm's Law also holds for any single resistor or combination of resistances. The
voltage drop across any resistance R no matter what the external circuit may be,
is given by
Voltage drop = IR, (4)
where I is the current flowing through R,.
If the battery has any appreciable internal resistance the circuit must be modified
to the form of Fig. 4.7 where R, is the equivalent internal resistance. Here we have
four resistances effectively in series and I = E /(R, + R, + R, + R,). The
132 (iv) RESISTANCES IN PARALLEL 4.2
voltages between any of the points A, B, C or D will be less than in the corresponding
case for zero internal resistance, the actual values being
R1 + R, + Ra
x voltage for R; = 0.
Ri +R1 +R2 +R2
If R, is less than 1% of (R1 + R2 + Rs), then its effect on voltages is less than 1%.
(iv) Resistances in parallel
When two resistances are in parallel (Fig. 4.8) the effective total resistance is given by
R,R2
(6)
R R1 + R2
When R1 = R2, R = R1/2 = Rs /2.
When any number of resistances are in parallel (Fig. 4.9) the effective total resistance
is given by
1 1
1
}
1
R2
+R ....1
s
(7)
When two or more resistors are in parallel, the total effective resistance may be de-
termined by the graphical method of Fig. 4.10. This method* only requires a piece
of ordinary graph paper (or alternatively a scale and set square). As an example,
to find the total resistance of two resistors, 50 000 and 30 000 ohms, in parallel take
any convenient base AB with verticals AC and BD at the two ends. Take 50 000
ohms on BD and draw the straight line AD ; take 30 000 ohms on AC and draw
CB ; draw the line XY from their junctjon perpendicular to AB. The height of
XY gives the required result, on the same scale.
Fig. 4.8. Two resistances in parallel. Fig. 4.9. Several resistors in parallel.
If it is required to determine the resistance of three resistors in parallel, the third
being say 20 000 ohms, proceed further to join points E and Y, and the desired result
ven by the height PQ. This may be continued indefinitely.
50,000 50,000
4RO00 40.000
g
effective resistance of two or
eop0o more resistors in parallel (after
Wireless World).
10,000 10,000
OA Y O a
FIG.410
The same method may be used to determine suitable values of two resistors to be
connected in parallel to give a specified total resistance. In this case, select one
value (C) arbitrarily, mark X at the correct height and then find D ; if not a suitable
value, move C to the next available value and repeat the process until satisfactory.
*" Resistances in Parallel-Capacitances in Series," W.W. 48.9 (Sept. 1942) 205.
" Diallist " " Series C and Parallel R," W.W. 51.4 (April 1945) 126.
4.2 (v) CONDUCTANCE IN RESISTIVE CIRCUITS 133
When a network includes a number of resistors, some in series and some in parallel,
firstly convert all groups in parallel to their effective total resistances, then proceed
with the series chain.
(v) Conductance in resistive circuits
obviously the reciprocal of the resistance
G = 1/R
-
The conductance (G) of any resistor is its ability to conduct current, and this is
(8)
Applying Ohm's Law, we derive
I = EG (9)
The unit of conductance is the mho (i.e. the reciprocal ohm).
When resistances are in parallel, their effective total conductance is the sum of their
individual conductances-
G = G, +G2 +Ga +.... (10)
When a number of resistors are in parallel, the current through each is proportional
to its conductance. Also,
I, G,
/total Gtodal
SECTION 3 : POWER
(i) Power in d.c. circuits (ii) Power in resistive a.c. circuits.
IG.4 II
instantaneous current at any point in the cycle. The curve of instantaneous power
(P) may be plotted point by point, and is always positive.
The heating of a resistor is obviously the result of the average or effective power,
which is exactly half the peak power.
Pyrale = Emax Imax (3)
Emax Imax Emax Ima,
Emax / \2
Pay
/2 X
\2 (4)
Triangular
Sine wave Square wave wave
(isosceles)
Form factor ( -= r.m.s. /average) 7r /2 V2 = 1.11 1.00 2/\/3 = 1.15
Peak /r.m.s. \/2 = 1.414 1.00 -/3 = 1.73
R.M.S. /peak 1/ Vi = 0.707 1.00 1 /V3 = 0.58
Peak /average n/2 = 1.57 1.00 2.0
Average /Peak 2 /7r = 0.64 1.00 0.5
SECTION 4 : CAPACITANCE
(i) Introduction to capacitance (ii) Condensers in parallel and series (iii) Calcula-
tion of capacitance (iv) Condensers in d.c. circuits (v) Condensers in a.c. circuits.
rJ ©
cl
T T
c4
FIG. 4.12 FIG. 4.11
1 1 1 1
(4)
C= C1+CE+Ca+...
When only two condensers are connected in series :
C CXC,
C, + Ca
- (5)
When two or more equal condensers (C,) are connected in series, the total capaci-
tance is
C = C, /2 for 2 condensers
and C = C, /n for n condensers.
Note The curved plate of the symbol used for a condenser indicates the earthed
:
(outer) plate of an electrolytic or circular paper condenser ; when this is not applic-
able the curved plate is regarded as the one more nearly at earth potential.
AK
C - µµF when dimensions are in centimetres
AK
or C = when dimensions are in inches
4. 5d µµF
where A = useful area of one plate in square centimetres (or inches). The useful
area is approximately equal to the area of the smaller plate when the square root of
the area is large compared with the gap.
K = dielectric constant (for values of common materials see Chapter 38
Sect. 8. For air, K = 1.
d = gap between plates in centimetres (or inches).
Capacitance with air dielectric, plates 1 mm. apart
C = 0.884 It µF per sq. cm. area of one plate.
Capacitance with air dielectric, plates 0.10 inch apart
C = 2.244 µµF per sq. inch area of one plate.
When there are more than two plates, the " useful area " should be interpreted as
the total useful area.
Cylindrical condenser (concentric cable)
7.354K
C =
logo D/d µµF per foot length
where D inside diameter of outside cylinder (inches)
d= outside diameter of inner cylinder (inches)
and K = dielectric constant of material in gap.
(iv) Condensers in d.c. circuits
An ideal condenser is one which has no resistance, no leakage, and no inductance.
In practice, every condenser has some resistance, leakage and inductance, although
these may be neglected under certain conditions of operation.
- 0632E
TIME
FIG.a l4
Fig. 4.14. Condenser charge and discharge (A) Circuit (B) Discharge current
characteristic (C) Charge voltage characteristic (D) Discharge voltage characteristic.
i = E É -t/rC
(6)
4.4 (iv) CONDENSERS IN D.C. CIRCUITS 137
where e
t
=
=
base of natural logarithms (.
2.718)
time in seconds after closing switch S,
E = battery voltage
r =resistance in ohms
and C = capacitance in farads.
The voltage (e) across the condenser is
e = E -ri = E (1 -
E-t /rC) (7a)
and the curve (diagram C) is of the same shape as the current curve except that it
is upside down. The voltage never quite reaches the value E, although it approaches
it very closely.
The charge on the condenser is given by
q = Q (1 -e/)
where q = instantaneous charge on condenser
(7b)
i = -R /ac (8)
which is of the same form as diagram B, except that the current is in the opposite
direction.
The curve of voltage (and also charge) against time for a discharging condenser is
in diagram D, and is of the same shape as for current, since e = Ri. These charge
and discharge characteristics are called transients.
In order to make a convenient measure of the time taken to discharge a condenser,
we adopt the time constant which is the time taken to discharge a condenser on the
assumption that the current remains constant throughout the process at its initial
value. In practice, as explained above, the discharge current steadily falls with
time, and under these conditions the time constant is the time taken to discharge the
condenser to the point where the voltage or charge drops to 1 / e or 36.8% of its initial
value. The same applies also to the time taken by a condenser in process of being
charged, to reach a voltage or charge of (1 - 1 /e) or 63.2% of its final value.
The time constant (T) is equal to
T =RC (9)
where T is the time constant in seconds,
R is the total resistance in the circuit, either ihr charge or discharge, in ohms,
and C is the capacitance in farads.
This also holds when R is in megohms and C in microfarads.
i.e., i = dq /dt
But q == Ce = CE,,, sin wt
Therefore i == (CE, sin wt)
dt
Therefore i = wCEm cos wt (10)
Eqn. (10) is the equation of the current flowing through the condenser, from which
we may derive the following facts :
1. It has a peak value of wCEm ; the current is therefore proportional to the applied
voltage, also to the capacitance and to the frequency (since f = w /2se).
2. It has the same angular velocity (u,) and hence the same frequency as the applied
voltage.
3. It follows a cosine waveform whereas the applied voltage has a sine waveform.
This is the same as a sinewave advanced 90° in phase -we say that the current
leads the voltage by 90' (Fig. 4.15).
Considering only the magnitude of the condenser charging current,
Im = wCEm (peak values)
Therefore I,m, = tCE,m, (effective values)
Where I and E occur in a.c. theory, they should be understood as being the same as
I,m, and E,m,.
This should be compared with the equivalent expression when the condenser is
replaced by a resistance (R) :
therefore be the applied (line) voltage. It will be seen that the current I leads the
voltage E by an angle 4) where
I 1 1 IX,I (l0a)
=wC RI
tan 4,
wCR R
IfR= 0, then tan 4,= oo, and 4. =90 °.
Since 4, is the phase angle of the current with respect to the voltage, the angle 4,
in the circuit of Fig. 4.16A is positive.
The instantaneous current flowing through the circuit of Fig. 4.16A is therefore
given by
i = I°, sin (wt + 4,) (10b)
where ro = tan-' (1 /wCR).
Fig. 4.16. (A) Resistance and capacitance across a.c. line (B) Vector diagram of
voltage relationships (C) Vector diagram of impedance (Z) with its real component
(R) and reactive component (1 /wC). -j
Using the j Notation
The operator signifies a positive vector rotation of 90 °, while -j signifies a
j'
negative rotation of 90 °. Instead of working out a detailed vector diagram, it is
possible to treat a circuit problem very much more simply by using the j notation.
For example in Fig. 4.16, we may equate the applied voltage E to the sum of the
voltage drops across R and C :
I
E =RI --j wC
the -j indicating 90° vector rotation in a negative direction, which is exactly what
we have in diagram B.
It is sometimes more convenient to put -jI/wC into the alternative form +1 /jo,C
which may be derived by multiplying both numerator and denominator by j (since
j' = - 1). Thus
E = RI + IjjwC (12)
From (11) we can derive :
Z R -j (1 /wC)
= E/I = (13)
For example, in Fig. 4.16A let R = 100 ohms and C = 10µF, both connected in
series, and let the frequency be 1000 c /s.
Then w = 2,r X 1000 -- 6280
1 /wC = 1/(6280 x 10 x 10 -e) = 15.9 ohms.
and Z= R- j(1 /wC)
100 -j 15.9 (13a)
Eqn. (13a) indicates at a glance a resistance of 100 ohms in series with a negative
reactance (i.e. a capacitive reactance) of 15.9 ohms. Values of capacitive reactances
See Chapter 6 Sect. 5(iv).
140 (v) CONDENSERS IN A.C. CIRCUITS 4.4
for selected capacitances and frequencies are given in Chapter 38 Sect. 9 Table 42.
Even if R = 0, we still write the impedance in the same form,
Z =0 -j 15.9 (13b)
Thus Z is a complex* quantity, that is to say it has a " real " part (R) and an " imagin-
ary " part (1 /wC) at 90° to R, as shown in Fig. 4.16C. The absolute magnitude
(modulus) of Z is :
1ZI
and its phase angle ¢ is given by
='R2
+ (1 /wC)2 (14)
tan 0 _
wCR - 1
(15)
which is the same as we derived above from the vector diagram, except for the sign.
The negative sign in eqn. (15) is because ¢ is the phase shift of Z with respect to R.
In a practical condenser, we may regard R in Fig. 4.16A as the equivalent series
loss resistance of the condenser itself.
It is obvious from Fig. 4.16B that
cos 0 = R/Z (16a)
where Z is given by equation (13) above.
Cos ¢ is called the " Power factor."
If d is nearly 90 °, X is nearly equal to Z, apart from sign, and so
cos 4. R /IXl (16b)
where 1X1 indicates that the value of the reactance is taken apart from sigi. With
this approximation we obtain
Power factor = cos 4 wCR (17)
Note that the power factor of any resistive component is always positive. A negative
power factor indicates generation of power.
The total power dissipated in the circuit (Fig. 4.16A) will be
P =EI cos 4 = EI x power factor (17a)
2
E'
R[l + X 2/R 21 (17b)
where P = power in watts
X, = 1 /wC
and cos ¢ is defined by eqn. (16a) or the approximation (16b).
The Q factor of a condenser is the ratio of its reactance to its resistance-
Q = 1 / w CR = tan ¢ 1 /(power factor) (18)
SECTION 5 : INDUCTANCE
(i) Introduction to inductance (ii) Inductances in d.c. circuits (iii) Inductances
in series and parallel (iv) Mutual inductance (v) Inductances in a.c. circuits (vi)
Power in inductive circuits.
di
= -L
-
e (2)
T =L /R (4)
where L = inductance in henrys
and R = resistance in ohms.
The decay of current follows the law
i = (E
Rt,'L) (5)
R
The rise and decay of current are called transients.
(iii) Inductances in series and parallel
Inductances in series
The total inductance is equal to the sum of the individual inductances, provided
that there is no coupling between them
L = Lt -i- L2 La ... :
(6)
Inductances in parallel
The total inductance is given by eqn. (7), provided that there is no coupling between
them :
(7)
1
L =
1 1
-FLz+ La-i-...
1
di
ez = -M dt1 (8)
where e, = voltage induced in the secondary,
di,
= rate of change of current in the primary in amperes per second,
dt-
and M = mutual inductance of the two coils, in henrys.
(Compare Equations 2 and 8.)
M may be either positive or negative, depending on the rotation of, or connections
to, the secondary. M is regarded as positive if the secondary voltage (e,) has the same
polarity as the induced voltage in a single coil.
The maximum possible (theoretical) value of M is when M = L1L2, being the
condition of unity coupling, but in practice this cannot be achieved. The coefficient
of coupling (k) is given by
k = M/ L1LZ (9)
so that k is always less than unity.
If the secondary is loaded by a resistance R2 (Fig. 4.17), current will flow through
the secondary circuit.
FIG. 4.17
FIG. 4. IA
Fig. 4.17. Two inductances coupled by mutual inductance (M) with the secondary
loaded by a resistance.
Fig. 4.18. (A) Equivalent circuit diagram of practical inductance (B) Vector
diagram of voltage relationships.
(Fig. 4.18A). If there is any other resistance in the circuit, it may be added to the
resistance of the inductor to give the total resistance R.
If an alternating voltage is applied across L and R in series (Fig. 4.18A) the vector
diagram may be drawn as in (B). The current vector I is first drawn to any con-
venient scale ; the vector of voltage drop across R is then drawn as RI in phase with
I ; the vector of voltage drop across L is then drawn as wLI so that I will lag behind
it by 90° -hence wLI is drawn as shown ; the parallelogram is then completed to
give the resultant E = ZI with a phase angle such that tan 96 = wL /R.
Using the j notation we may write :
E= RI +jwLl (11
where +j indicates 90° vector rotation in a positive direction.
From (11) we can derive :
Now the average value of (sin ut cos cot) over one cycle is zero *,
Therefore P = Em Im (sin' wt cos 0)
=i Em Im - cos 2(ot) (cos
(1 )
= Er Irma cos 02n (15)
\.
POWER ENERGY TAKEN
BY CIRCUIT
41*
e
Fig. 4.19. Power in an inductive
circuit with applied sine -wave ,.:
voltage (e).
4 ENERGY GIVEN BAC K
BY CIRCUI
360°
FIG. 419
From eqn. (15) the power is equal to the product of the effective voltage and current
multiplied by cos ¢, which is called the Power Factor, its value being given by
cos 0 = R /ß/R2 + (wL)2.
When the load is purely resistive, L = 0 and the power factor = 1 ; the power is
therefore P =
From (15) we may derive for the general case :
Q = wL /R = tan ¢
Power factor = cos 0 = R/Z
tan 0 1 /cos ¢ (error < 1% for 0 > 82 °)
ponent.
It may therefore be written as
Z = R + jX where X = (wL - 1/wC).
For example, let R = 500 ohms, L =20 henrys, and C = µF, all connected in 1
series across an a.c. line with a frequency of 50 c /s.
Then w = 2,r x 50 = 314
1 /wC = 1/(314 x 10 -8) = 3180 ohms
X = wL - 1 /wC = 6280 - 3180 = 3100 ohms
(the positive sign indicates that this is inductively reactive) and
Z= R +jX= 500H- j3100.
If L had been 5 henrys, then X would have been
1570 3180 = -
1610 ohms -
which is capacitively reactive.
The phase angle is given by tan ¢ = X /R. The magnitude of the impedance
is given by IZI _ VR' + (wL 1 /wC)' (la) -
A ß
1JL3
O
RI e I across a.c. line (B) Vector
diagram of voltage re-
-)0.)c lationships.
FIG. 4 20
This may be illustrated by means of a vector diagram (Fig. 4.20 B) where I is drawn
to any convenient scale. The voltage -dçop vectors are then drawn, to the selected
voltage scale, RI in phase with I, jwLI at 90° in advance of I and -jl /wC lagging
90° behind I. The simplest method of combining the three vectors is to take either
cLl or // wC, whichever is the greater, and then to subtract the other from it, since
these two are exactly in phase opposition. In diagram B, (Ll is the greater, so that
-
I/ wC is subtracted from cLl to give (wL 1/01C)/. The two remaining vectors,
RI and j(wL -
1/0,C)/ are then combined by completing the parallelogram, to give
the resultant ZI which, of course, must be equal to the applied voltage E.
A special case arises when wL = 1 /wC in eqn. (1) ; the reactive component
becomes zero and Z = R. This is the phenomenon of series resonance which
is considered in greater detail in Chapter 9.
Special cases of equation (1)
R and L only : Put 1/0,C = 0, Z = R + j wL
R and C only : Put wL = 0, Z = R
Z = j(wL
j/ w C = R + 1 /j wC
1 / wC)
--
L and C only : Put R = 0,
R only: Z = R +j0
L only : Z = 0 + j wL
C only : Z = 0- j /wC = 1 /jwC
Note that Q = !tan ¢ = IX /R. I
Û
3
N >^i
i--.. -.
-"N
c,
óa
.j
a
U+a á+
+
U+
w
VÚ
V .4
.3
a
3+
:,..-1
+â a UU
I ...
I
..
I
+
3
I + a a
^.^
: a
1
D+ a a 3 3 3
U
I
3 3
á á
I
I..
Ì a a
,----..
u
m
'. a U 3
a
k k IN 3Ia kIN a
xyi E O k O
I
+ I + I
+
a 3
II
...-...,
-e- 7a
a
a
u
Uy
p.
;./
+
.
N / a + N
r a
/1
IU
3
`
.á I:4" + Ú 3 U .. V I I
4a 3 3 + a +V + Ú
a I á
+/
á
Ñ
II +
-. 3 >
3
3
\I
^ 1
-
\ :
v. +I U" U
1 3 I 3
i+ ,J 3 á + Û 3
.^ -4 l 3
a
I I
tii çççti
A a 3 + ci '1
o
N
+ r á..13
+ +
a a I
'^
I
+ T.'
+ +
N
a
ao
:ñ
á
"
4,
a a ++ á ä
a' á
(5'
cg
a
+
a
Q
+
a
o
+
a
C)
+
a
+
a
c.)
4.6 (iii) PARALLEL COMBINATIONS OF L, C AND R 147
General case with a number of arms connected in series, each arm being
of the form R + X
Arm (1) Z, = R, + jX, (i.e. R, in series with X,)
:
I, (4)=o)
(4$°-90°) FIG.4.20D
Fig. 4.20.0 Circuit of R, L and C in parallel across a.c. line.
Fig. 4.20.D Vector diagram of voltage relationships.
wL
x
1
R - ( wC- 1
wL
)
R-j(wC- L/ 2)
2
N.
N «
* U H- .. á
N «
a + ...11 ,a I I
N
$ l j+
0 áa U
3
.^11
+
^ +1 3
.6 +
I
>4
-IN -.IN
.sc.) I3 +
'^ .,
1
., IN
.--,
v
a3
'II
,CÇ 1
d i C
k IN a I 3 I fi IN k iN 3
u ca I i -H -H
`NN/
N § A
h
u
/
Itl
.r
:
11
l áÑ
Ú
U
óa á+
1 ± n: a+ 1 U
b
:.L' .'71
-"+ a a a 3
oa Ú 3 á + 4
a
I
B 3 3
.. .. 9?
> a3
II
..}_
Ñ +
/1
^..,a
9
8
9+
Ú áá
.,+ ä á
UÚ
,
áH- Q N
., 3 3
áa
N
á+4
a
~^1
+
+
ä â
^
x+
3 3
,3
1
14
1+
Ñá
I
U
V
3
s
z
i
3 + ...,
+ -.
I
IN â
0
ä ci N C.) U
a
o Û
(5 N a a á 4
l0
a.
4.6 (iii) PARALLEL COMBINATIONS OF L, C AND R 149
Benson, F. A. (letter) W.E. 28.331 (April 1951) 128. Fig. 4.20E. Impedances
in parallel.
Fig. 4.21. (A) Network incorporating 4 elements for impedance calculations (B)
Simplified network with 3 elements.
1 R2
i.e. when CO
(17)
LC _ L2
which can be written in the form
when = CR 2
1
w°
CR' - /
VLC L 1 1 L (18)
where w° = 1 /1/LC
is the value of w when the resistance R is zero.
This condition, namely that the effective reactance is zero, which may also be
expressed as the condition of unity power factor, is one of several possible definitions
of parallel resonance. This is the definition used in Chapter 9. It should be
noted that the expression giving the value of is not independent of the resistance
CO
If LCw2 = 1, which is the condition for resonance if the resistance is zero, the
impedance is given very closely, when R is small, by
Z L /CR (19)
The remaining possible definitions of parallel resonance are conditions to give
maximum impedance. The condition of maximum impedance is sometimes called
anti -resonance. Maximum impedance occurs under slightly different conditions
for the variables C, L and w *.
(b) Condition of maximum total impedance when C is the variable
The maximum value of Z which can be obtained is
IZImnx = R12 - Ri(1 + Q2) (20)
where Q = wL /R1
which occurs when C - R12 w2L2 (21)
1 + V1 + 4w2C'R12
which occurs when L = (23)
or when C - (.02L2
L
(24)
If L is the variable and Q is maintained constant, the maximum value of Z which
can be obtained is
IZlmax =Q2Q/wC (25)
which occurs when L (26)
w2C(1 + Q2)
(Rz )
L
+ (w C2 wiz z ' (Rz ) z + (wCz 12\
3. J and reactive
Determine the combined impedance (Z) by adding the resistive
components of (14) and (15)-
Z = A + jB (37)
1
Rz
where A= R,
(1
z
cull)l 2
Rz/ + ( w Cz
4.6 (iv) SERIES -PARALLEL L, C AND R 153
w C, - wL,
and B = wLt
wC t ( )2 (wC, l
\R,/ Lz/
(.01..2/
z
01 n\-1 Ri L,
% X,
TC' Tc'
FIG.4.21C FIG.4.210 FIG.4.21E
Fig. 4.21. (C) (D) (E) Series-parallel networks for impedance calculations.
Z
1
t
1
Zz
+Z s
1 1
F... (40)
Note : If there are more than two parallel branches it is more convenient to work
in terms of admittance -see Sect. 6(v) below.
Alternatively, if the circuit cannot be divided into parallel branches, treat it as a
series circuit, firstly determining the resistive and reactive components of each section,
and then adding all the resistive and all the reactive values separately, as in (E).
(y) Conductance, susceptance and admittance
In an arm* containing both reactance and resistance in series, the conductance
(G,) of the arm is given by
G, -
R1' R'
+X12 mhos (41)
where X, = (wL, -
1/ w C,).
When there are a number of such arms in parallel, the resultant conductance is
the sum of their separate conductances, that is
G = G, +G, +G, +...
The susceptance (B,) of the arm under similar conditions, is given by
B, - R,' + X12
mhos (42)
where X, = (wL, -
1/ (Xi).
[Inductive susceptance is regarded as positive. Capacitive susceptance is regarded
as negative.]
When there are a number of such arms in parallel, the resultant susceptance is the
sum of their separate susceptances,
B = B, +B, +B3 +. (43)
When any arm includes only resistance, the conductance of the arm is 1 /R,
and the susceptance zero,
i.e. G, = 1 /R, B, = 0.
'An arm is a distinct set of elements in a network, electrically isolated from all other conductors except
at two points.
154 (v) CONDUCTANCE, SUSCEPTANCE, ADMITTANCE 4.6
When any arm has only inductance, the conductance of the arm is zero and the
susceptance is 1/0.1,13
i.e. G, = O B1 = 1 /wL,
When any arm has only capacitance, the conductance of the arm is zero, and
the susceptance is - (DC, = - wC
i.e. Gl = 0 B1
i.e. Y1 =
R1
-1B1 = Rl a + X18 R1 + jXl Z1
G1
- jX1 1 1
(46)
admittance is the reciprocal of the impedance.
indicating that the
Thus the value of the admittance may always be derived from the impedance
1 1 R -jX R -jX
-
Y (tea)
Z R +jX Ra +XE R -jX
G + jB G + jB
Similarly Z=Y= G jB
1 1
- G 2+ e + jB
(46b)
G
The negative sign in front of jB in eqn. (45) deserves special attention
Admittance of inductive arm (R and L in series) :
-
Y = G jB =
R
- wL -j (47)
R, +
Admittance of capacitive arm (R and C in series) :
Y = G +3B=
R +j /wC
+1 /w'Ca-
w'C'R +jwC
w'C'R' +1 (48,
Ra
The admittance of any arm containing resistance, capacitance and in-
ductance in series (Fig. 4.20A) is given by
Y = G jB - (49)
R
where G R2 + X'
X = (wL - 1 /wC)
(wL - /wC)
X 1
and B
R2 - Re + (wL - /wC)$
+ X' 1
- R -+j(wL
R - /wC) 1
(wL - / C)
therefore Y (50)
$ 1 w e
Values of admittance for various series combinations are included in the table in
Sect 6(iii).
(X, = - 1/wC)
0.45 O 0.4 043 03
8025
07
0e
-- 0.9
aP- 1-0
q.- 1.1
0
- 14
4 _ 1.4
o
a- 25 0I
3
ó-
á
1 1
b
1 1 1 1 1 I I
1 1 1 1 I
0 01 09 09 04 03 06 07 04 09 hl 19 13
1
1.4
I
13
1
19 M
1 1
lib 19
1 1
2.0
G FIG. 4.21F
Fig. 4.21F. Chart for conversion between resistance and reactance, and conductance
and susceptance
Z= (R± jX) 10", Y --- (G+jB)10- ".
156 (v) CONDUCTANCE, SUSCEPTANCE, ADMITTANCE 4.6
finity, so the point Y moves from O (at zero frequency) through C to A (at resonance)
and thence through E to O again (at an infinite frequency). The angle ¢ is the angle
by which the current i leads the applied voltage e ; when 0 is negative the current
lags behind the voltage. Thus when OY lies in the upper portion of the circle the
circuit is capacitive, and when it lies in the lower portion of the circle the circuit is
inductive.
R,
Fig. 4.22.A Series form of imped- xs H
ance ; (B) Equivalent parallel form.
FIG.4.22A FIG. 4.22
I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 19 20
SERIES RESISTANCE Rs
Fig. 4.22C. Chart for conversion between series resistance and reactance and
equivalent parallel resistance and reactance.
158 (vi) CONVERSION : SERIES TO PARALLEL IMPEDANCE 4.8
at R, = 2 and X, = 7.5, so the appropriate series elements are R, = 2000 ohms and
X, = 7500 ohms.
SECTION 7 : NETWORKS
(i) Introduction to networks (ii) Kirchhoff's Laws (iii) Potential Dividers (iv)
Thevenin's Thereom (v) Norton's Theorem (vi) Maximum Power Transfer Theorem
(vii) Reciprocity Theorem (viii) Superposition Theorem (ix) Compensation Theorem
(x) Four -terminal networks (xi) Multi -mesh networks (xii) Non -linear components
in networks (xiii) Phase -shift networks (xiv) Transients in networks (xv) Refer-
ences to networks.
over a current ratio 10 000 000 to 1. See K. B. McEachron " Thyrite, a new material for lightning
arresters " General Electric Review (U.S.A.) 33.2 (Feb. 1930) 92.
q.7 (i) INTRODUCTION TO NETWORKS 159
Rectifiers, whether thermionic or otherwise, are non -linear devices ; they are
frequently represented by an equivalent circuit having a fixed series resistance in the
conducting direction and infinitely high resistance in the other. This approximation
is very inaccurate at low levels where they have effective resistance varying as a func-
tion of the applied voltage, while some rectifiers pass appreciable current in the reverse
direction.
Most elements (resistors, capacitors and inductors) transmit energy equally in either
direction and are referred to as " bilateral ", but thermionic valves operate only in
one direction (" unilateral ") ; when the latter form part of a network it is necessary
to exercise care, particularly if they are represented by equivalent circuits.
C1 C,
If in any part of the circuit there are two or more elements of the same kind in
parallel (whether R, L, or C) the resultant should be determined and applied to the
equivalent circuit diagram.
An exception to this rule is where it is merely required to calculate the output
voltage from a passive resistive 4- terminal network. In this case it is sometimes
helpful to arrange the network in the form of a potential divider, or sequence of
dividers, and to use the method of Sect. 7(iii).
It is very important to mark on the equivalent circuit diagram the directions or
polarities of the applied voltages (whether direct or alternating) and the assumed direc-
tions of the currents ; if any one of the latter is incorrect, this will be shown by a
negative sign in the calculated value. A clockwise direction for the flow of current
around any mesh is conventional.
In some cases it may be found simpler to reduce a passive 4- terminal network to
an equivalent T or +r section -see Sect. 7(x)-than to analyse it by means of Kirch-
hoff's Laws.
Definitions
An element is the smallest entity (i.e. a distinct unit) which may be connected
in a network -e.g. L, C or R.
An arm is a distinct set of elements, electrically isolated from all other conductors
except at two points.
A series arm conducts the main current in the direction of propagation.
A shunt arm diverts a part of the main current.
A branch is one of several parallel paths.
A mesh is a combination of elements forming a closed path.
A two -terminal network is one which has only two terminals for the application
of a source of power or connection to another network.
A four -terminal network is one which has four terminals for the application of
a source of power or connections to other networks. The common form of four -
terminal network has two input and two output terminals ; this term is used even
when one input terminal is directly connected to one output terminal, or both earthed.
160 (i) INTRODUCTION TO NETWORKS 4.7
Examples
An amplifier is a four -terminal active network.
An attenuator is a four -terminal passive network.
A conventional tone control is a two -terminal passive network.
This law is illustrated for a.c. in Fig. 4.24 which follows the same general rules as
for d.c. The instantaneous generator voltage e is shown in an arbitrary direction and
the directions of the currents are then determined.
At point b,
Tab - -ibl ibe =
following the general procedure as in the d.c. case.
4.7 (ii) KIRCHHOFF'S LAWS 161
(2) The total rise or fall of potential at any instant in going around any closed
circuit is zero.
This is illustrated for d.c. in Fig. 4.23, taking each closed circuit in turn and pro-
ceeding clockwise in each case. Any voltage source is here regarded as positive if it
assists in sending current clockwise around the closed circuit (i.e. voltage rise). The
voltage across any impedance is regarded as negative if the current arrow is in the same
direction as the direction of travel around the closed circuit (i.e. voltage drop).
Circuit dab c d: -
+ 6 16 iab l0 ib , = 0 - (1)
Circuit d a b f g c d + 6 -
16 iab + 8 i3O 0 -4=
(2)
f 4- -
:
Circuit b c g f: + 10 i b c = 0 (3)
Applying Kirchhoff's first law to point b,
Tab + =Ib = ib, (4)
To find the values of the three unknown currents, it is necessary to apply three
suitable equations.
[Similarly for all other cases -the total number of equations must be equal to the
number of unknowns. The number of equations based on Kirchhoff's first law should
be one less than the number of junction points ; those based on his second law should
equal the number of independent closed paths.]
Equations (1), (3) and (4) would be sufficient, since (2) merely duplicates parts of
(1) and (3).
From (4) : is b = i b b c -it
Applying in (1) : +6- 16(ib, --irb) -10ió, =0
Therefore -
+ 6 16 i b , + 16 it b 10 t b c = 0 -
Therefore -
+ 6 26 i b c + 16 in, = 0 (5)
Adding twice (3): + 8 - -
20 i b , 16 if t, = 0
Therefore + 14 - 46 i b b = 0
Therefore i b c = 14/46 ampere.
The other currents may be found by applying this value in (5) and then in (4).
Kirchhoff's Second Law is illustrated for a.c. in Fig. 4.24. Here again, as in all
cases, it is assumed that we move around each loop of the network in a clockwise
direction.
The voltage across
a resistance is
- -Ri where i is in the clockwise direction around
an inductance is -jwLi the loop.
a capacitance is
Circuit dabcd e
+j(1 /wC)i
-
R,iab + j (1 /wC,)ib, = 0
)
-
:
Therefore
tob - tb, -the -0
io b = t 0, + ib c (8)
Adding (6) and (7)
Rniab + + R,)ib> = e
Applying (8), R,i bf + R,i b c + (j COL' + R,)i b1 = e
Therefore R,i b , + (R1 -f- R2 + wLj)i o, = e j (9)
Multiplying (7) by ( -jR, w C,), remembering that j2 = 1, -
Riib, + (Riw2L1C1 jR,RzwCi)io, = 0
Subtracting (10) from (9),
- (10)
[(R1 + R, -
Riw'L1Ci) + j(wLi + R,R2wC,)] iar = e
which gives the value of :0, when e is known.
(11)
The value of i0, may be found by substituting this value of iv in eqn. (7) ; i b
may then be determined by eqn. (8).
voltages E, and E2, where E = E, + E2. If no current is drawn from the junction
(or tap) B, the voltage across BC is given by
E2 (RR+2R)E, (1)
RI EI
Ij LOAD
Fig. 4.25. Potential divider across
a d.c. line.
l C
R2 E2 ORE'2
FIG. 4 25
On load
When a current Is is drawn from B, the simplest analysis is to consider the effective
load resistance R, which will draw a current Is at a voltage Es', i.e. R, = E_ /I,.
We now have resistances R2 and R, in parallel, and their total effective resistance is
therefore
R' = R2R3 /(R2 -i R,).
In this case the voltage divider is composed of R, and R' in series, and the voltage at
the point B is given by
Eá R,
R'
R'
_( lE
_ R,R3 E
(RIR2 + R1Rs + R,Ra
E 2
_ (R, R2+ R2 )E _ (R,R,R2
+ R2
)Is (3)
The first term on the right hand side is the no -load voltage E2 ; the second term is the
further reduction in voltage due to /3--this being a linear equation.
VOLTS
250 A
a 150 B
o
0
o
Ñ
100
a
o
O
o
50
IMA%.
5 10 15 20 25
FIG.4.26 CURRENT (rj) FIG.4.27
Fig. 4.26. Graphical method for determining the output voltage from a potential
divider.
Fig. 4.27. A potential divider with a load in the form of another potential divider.
Eqn. (3) may be plotted* as in Fig. 4.26 where, as a typical example, E = 250 V,
R, = 10 000 and R2 = 15 000 ohms, so that (R1 -}- R2) = 25 000 ohms. for E
"Curdy, P. F. " Potential Divider Design," W.W. 50.5 (May 1944) 154.
4.7 (iii) POTENTIAL DIVIDERS 163
no load, is obviously 150 volts (point B) while point C is for the condition of maximum
current (1,333) and zero voltage. The latter may be determined by putting E9 = 0
in eqn. (3) which gives
/,33x = E /Rt (4)
In this condition we have, effectively, a resistance R, only, in series with a load of
zero resistance.
If points B and C in Fig. 4.26 are joined by a straight line, we may then determine
the voltage for any value of load current (I3) between zero and maximum. For
example, with a load current of 10 mA, the voltage will be 90 V (point D).
It is sometimes useful to calculate on the basis of a drop in voltage (based on the
no -load voltage Ea) of so many volts per milliamp of load current ; in this case the
drop is 150 volts for 25 tmA, or 6 volts per milliamp. This rate of voltage drop is
actually the negative slope of the line BC.
The equivalent series source resistance (R,) is given by
Ea 150
R' Im3s 25 x 10 -' - 6000 ohms.
This may be put into the alternative form
R1Ra
R, + Ra R' (5)
" Regulation " is defined differently in American and British practice.
American definition The percentage voltage regulation is the difference be-
:
tween the full-load and no -load voltages, divided by the full -load voltage and multi-
plied by 100.
British definition The percentage voltage regulation is the difference between
:
the full-load and no -load voltages, divided by the no -load voltage and multiplied by
100.
In the example above for a current of 10 mA,
+ R,)
orls=1,,,a3-Eá (R,R,Ra (7)
Special Case 1
If it is known that the voltage drops from Ea at no load to E,' for a load current I,,
then the voltage E,, corresponding to a load current I,, is given by
Es = E, (Ea -
Ez)(I = /Ia) - (8)
Special case 2
If the voltages across the load (E,,, E2) for two different values of load current
(I,z, I,,) are known, the voltage at zero load current is given by
E,I2 E91a
Ea = I -- Is
(9)
This is often useful for determining the no load voltage when the voltmeter draws
appreciable current. A method of applying this with a two range voltmeter has been
described *, and the true voltage is given by
Lafferty, R. E. "A correction formula for voltmeter loading" (letter) Proc. I.R.E. 34.6 (June 1946)
358.
164 (iv) THEVENIN'S THEOREM 4.7
b,
(S - Ez
1)
S -scales
(Es /E )
(10)
,
where S = ratio of the two voltmeter used for the two readings E,, and E
Ez = voltmeter reading on the higher voltage scale,
and E
= voltmeter reading on the lower voltage scale.
If S = 2, then
Es
E,
2 - (E:/E)
Complicated divider network
When a voltage divider has another voltage divider as its load (Fig. 4.27) the best
procedure is to work throughout in resistances and voltages, and to leave currents
until after the voltages have been determined. The final equivalent load resistance
(Rs) must be determined before commencing calculations, then proceed
R, in parallel with R, : R = R4R, /(R4 4- Rs)
-
Rs in series with R' R' = Rs + R'
:
Then E2 E (12)
+
and E4 )E. (13)
(R, + R'
A somewhat similar procedure may be adopted in any divider network.
A B
Fig. 4.29. (A) an im-
pedance, carrying a current
I L, connected to two ter-
'` minals of a network ; (B)
Norton's equivalent circuit.
FI G.4.29
breaks down under other conditions. It is therefore only applicable to valves when
these are being operated to give negligible distortion.
Z1 - ZA +ZAZB
ZB + Zc Z.9
Z1Z2 + Z2Z3
Z2
+ Z1Z3
ZAZc
Z3 -
Z_4+ZB+Zc Z. - Z1Z2 + Z2Z3Z1
+ Z1Z3
The bridged T section of Fig. 4.33E may be reduced to the equivalent T section
of Fig. 4.33F, or vice versa, by using the same equivalent values as for equivalent T
and H sections (above).
4.7 (x) FOUR- TERMINAL NETWORKS 167
indicating the meshes to which they are common. The applied alternating
suffixes
are also numbered E1, E2, etc. where the suffix indicates the mesh number.
voltages
If there
is more than one voltage source in any mesh, E1 etc. will indicate the vector
sum of these voltages.
There must be the same number of mesh equations as there are meshes in the net-
work. The mesh equations are written in the general form for n meshes as
Zull 4 Z12I2 -f- Z13I3 +
- Z1 ,II>< = Et
+ Z22I2 + Z22I3 + Z2 ?tin = E2 (16)
I
Z21I1
Z 3I1 + Z 2I2 + Z +
where each Z is of the form R -{- j(wL
,I,
Z = ... - E
1 /wC) and Z11, Z22, etc. is called the self -
impedance of the individual mesh, i.e., the impedance round the mesh if all other
branches of the network other than those included in the mesh in question were open-
circuited.
For example, in Fig. 4.33G-
Z11 = Z10 +Z13 +Z12
= R14) +jX10 + R13 +PXl2 + R12 +jX12
= R10 +J(wLio -
1 /wCio) + R1, +J(wLi3 - /wC13)
+ R12 +j(wLis 1 /wC12) - 1
(17)
Note that the same as Z12 etc. and that the signs of the mutual impedances
Z21 is
may be positive or negative (see below).
An impedance that is common to two branches is considered to be a positive mutual
impedance when the arrows representing the corresponding mesh currents pass
through the impedance in the same direction ; or a negative mutual impedance if
the arrows pass through the impedance in opposite directions.
Thus in Fig. 4.33G the arrows representing the corresponding mesh currents pass
through the impedances Z12, Z133 Z24 and Z34 in opposite directions, so that these
constitute negative mutual impedances, and have negative signs in eqn. (16).
A mutual inductance may be defined as positive or negative according to whether
it acts with a polarity the same as, or opposite to, that of a corresponding common
inductance *.
These linear simultaneous equations may be solved by elimination, but any more
complicated network would have to be solved by the use of Determinants, for which
see any suitable mathematical textbook.
The total current in any common branch may be determined by the difference
between the two mesh currents.
The method of handling mutual inductance is illustrated by the 2 -mesh network
of Fig. 4.33I. Here
Zu = (Rio .- R12) +1(wL10 -F (0L12 1 /wC1o) - (22)
Z22 = (R,, -F- R20) +1(wL12 + wL20 1 /wC20) - (23)
(24)
Z,, = Z,, = R,, + 7(w1.12 ± wM12 )
and the two mesh equations will be as eqn. (21).
The polarity of M12 in eqn. (24) must be determined in accordance with the accepted
convention, as described in connection with eqn. (16)
A solution of eqn. (16) by means of Determinants shows that in the general case
the Determinant D is given by
Z1.1Z12 Z1 n
D= Z21Z22 Z2 n (25)
Zn1Zn2 . Znn
and the current I in the kth mesh that flows as the result of the voltage E, acting in
the jth mesh is
B
Ik = E1
D
(26)
where B is the principal minor of D multiplied by (-
This minor is 1)5 +k.
formed by cancelling the jth row and the kth column
and then moving the remainder together to form a new
determinant with one less row and column than D,
where D is the determinant defined by eqn. (25).
The row cancelled corresponds to the mesh containing
the input voltage, the column cancelled to the mesh
containing the required current.
The input impedance of a passive network with
a single applied voltage (Fig. 4.33J) is given by
FIG. 4.33I Input impedance = E, /I, = D /B1, (27)
where El = voltage applied to input terminals
Fig. 4.33 I. Two-mesh net- I, = input current
pork incorporating mutual D = determinant defined by eqn. (25)
inductance.
and B,1 is the minor of D obtained by cancelling the
first row and column.
The transfer impedance of a 4- terminal network (Fig. 4.33K) is defined as the
ratio of the voltage applied to the input terminals to the resulting current through
the load impedance connected to the output terminals (i.e., the nth mesh).
-o--
1 ir 3
-
FOUR
TERMINAL z,
PASSIVE
o- NETWORK
O
4
FIG. 4.33.1 FIG. 4.33 K
RI
RI
Eo
FIG.4.34
Fig. 4.34 (A) A phase -shift network providing full range phase shifting from 0 to 180°
with 6 db attentuation for all degrees of phase shift (B) vector diagram of voltage
relationships.
It may be shown, when R, = R,' = 1/ u, C, that
E° 2R,Rz
IE _>}
and ¢ = tan
R9_Rt s
which is illustrated by the vector diagram Fig. 4.34B, where ER, and ER,' are equal,
and both equal to E, ; ER, and E, are at right angles, but their vector sum is E.
Point P is therefore on the circumference of a semi -circle.
It is assumed here that the impedance is not a function of time ; this occurs in the
and lamp filaments. case of barrettes
Lafferty, R. E. " Phase -shifter nomograph," Elect. 19.5 (May 1946) 158.
4.7 (xiv) TRANSIENTS IN NETWORKS 171
SECTION 8 : FILTERS
(i) Introduction to filters (ii) Resistance -capacitance filters, high -pass and low -pass
(iii) Special types of resistance-capacitance filters (iv) Iterative impedances of four -
terminal networks (v) Image impedances and image transfer constant of four -terminal
networks (vi) Symmetrical networks (vii) " Constant k" filters (viii) M Derived
filters (ix) Practical filters (x) Frequency dividing networks (xi) References to
filters.
1. The grid coupling condenser and grid resistor of an amplifier (high pass resist-
ance- capacitance filter).
2. The series condenser and variable resistor of a conventional tone control (low
pass resistance -capacitance, with adjustable attenuation).
3. The smoothing filter of a power supply, including one or two inductances and
two or three capacitances (low pass filter).
4. An overcoupled i-f transformer (tuned band pass filter).
5. A tuned aerial coil or r-f transformer (tuned narrow band pass filter).
See Chapter 6 for mathematics.
(1)
A/122 + X' V1 + (X /R) 2
If we select a frequency (f1) at which $X,I = R, then 1E0 /Ed = 0.707 which is
practically equivalent to an attenuation of 3 db. This frequency is the reference
point used for design ; it is called the theoretical cut-off frequency. Its value is
given byt
fl = 1 /(2irRC) c/s (2)
where R and C are in ohms and farads (or in megohms and microfarads). The value
RC is called the time constant and is measured in seconds (see Sect. 4(iv)) so that
i.é. not including a valve or generator.
to nomogram to determine the value of Ji is given by E. Frank " Resistance Capacitance Filter Chart "
Elect. 18.11 (Nov. 1945) 164.
4.8 (ii) RESISTANCE-CAPACITANCE FILTERS 173
.0666
lllllllll
12 500 1000
20 SO 100 200
FREQUENCY N CYCLES PER SECOND
F1G.4.36
-capacitance
Attenuation in decibels versus frequency for a resistancemegohm.
Fig. 4.36. (R) = 1 If
coupling or filter (Fig. 4.35A) in which the total resistance
R, + r,RL/(r6 + RL). If
applied to a resistance -coupled valve amplifier, R = the second term is less than 1%.
r,, is less than 10 000 ohms, the error in neglecting
taken as (R, + RL) with a sufficient accuracy
If the valve is a pentode, R may be most
for purposes.
of C by 2 and similarly in proportion for other
If R = 0.5 megohm, multiply valuesresistances.
174 (ii) RESISTANCE-CAPACITANCE FILTERS 4.8
One of these attenuation characteristics is shown in Fig. 4.36A together with the
straight line AB which is the tangent to the curve and has a constant slope of 6 db/
octave (or 20 db /decade). The point of intersection between AB and the zero db
line is point A which corresponds to the theoretical cut-off frequency fl. For ease
in calculation, the " straight -line " approximate characteristic CAB is sometimes
used in calculations in place of the actual attenuation characteristic, the maximum
error being 3 db.
In Fig. 4.35(B) there are two such filters in cascade and further sections may also
be added. A two- section filter in which R1 = Ra and C1 = CE will have somewhat
more than twice the attenuation in decibels of a single section filter, and the ultimate
slope of the attenuation characteristic will approach 12 db /octave.
A
0
60b /Octava
ACTUAL
ATTENUATION
CHARACTERISTIC
-6
X
1s
IH 16
-21 21
0.1f, r, lof,
FIG. 4.36A
Fig. 4.36A. Actual and approximate attenuation characteristics, using the theo-
retical cut-olf frequency f1 as the reference frequency.
Icoo
FREQUENCY G S
FIG. 4 DE
Fig. 4.38. Shows the attenuation characteristics for the low-pass filter of Fig. 4.37
for the values of R and C shown below :
R (ohms) C(µµF)
Curve A B C D E
400,000 10 20 40 80 160
200,000 20 40 80 160 320
100,000 40 80 160 320 640
50,000 80 160 320 640 1280
20,000 200 400 800 1600 3200
10,000 400 800 1600 3200 6400
Therefore E °
E;
-
R z + X ,2
X `
- /
V1 -F (R/X,)e
1 (4)
21 O4
FIG.4.39 FIG.4.40
E;
ti 40f2 x R,R2C1C2 (6)
+ Z1Z3 + (8a)
Z" = ` 2`
Z12 =
% 41
2
+ Z1Z3 - 2`
(8b)
If we now define a quantity P, called the iterative transfer constant, such that
P = loge ('5 /!2) (9a)
where I, = input current
and I2 = output current
the formulae for the L section may be written in the alternative forms
Z1e'
Zkl = cp Z3(SP
-- 1) - (9b)
ZkY = Zkl E_ P
where cosh P =
Z1 Z1 + 2Z3
1 + 2Z, 2Z 3
3
(See Chapter 38 Sect. 21 Table 73 for hyperbolic sines, cosines and tangents.)
*Scott, H. H., "A new type of selective circuit and some applications." Proc. I.R.E. 26.2 (Feb.
1938) 226.
tWolf, A. " Note on a parallel -T resistance -capacitance network," Proc. I.R.E. 34.9 (Sept. 1946) 659.
See also Hastings, A. E. ' Analysis of a resistance -capacitance parallel -T network and applications,"
Proc. I.R.E. 34.3 (March 1946) 126P ; McGaughan, H. S. " Variation of an RC parallel -T null net-
work," Tele-Tech. 6.8 (Aug. 1947) 48.
:Cowles, L. C. " The parallel -T resistance-capacitance network " Proc. I.R.E. 40.12 (Dec. 1952) 1712.
4.8 (iv) ITERATIVE IMPEDANCES OF FOUR-TERMINAL NETWORKS 177
z
Fig. 4.41 shows a four terminal network
with generator and load.
FIG.4.41
FIG. 442
FIG. 4.43
2
FIG.4.44
to Z1 and at the same time to give an output impedance (looking from terminals
3, 4 with generator connected) equal to Z12. Under these conditions the impedance
on each side of terminals 1, 2 is an " image " of the other (since they are both identical)
178 (v) IMAGE IMPEDANCE 4.8
and similarly with the impedances on each side of terminals 3, 4. The two image
impedances are given by
- /
/(Z4 + Zc)(Z,,ZB + ZAZC + ZBZc)
(Z B + Z r)
(12a)
2 4 2 4
FIG. 4.44A FIG. 4.44B
Fig. 4.44A is a single H section network terminated in its image impedances.
Fig. 4.44B is a single L section network terminated in its image impedances.
The transfer of power is indicated by the image transfer constant B whose value
is given by
E,I, Il /I,
9 = loge .E212
= loge!,
Z ¡t
= loge I= 1, (16)
provided that the network is terminated in its image impedances,
where E, and I, are voltage and current at terminals 1, 2
E, and I, are voltage and current at terminals 3, 4
I,' and 12' are currents at terminals 1, 2 and 3, 4
respectively with transmission in the reversed direction.
FIG. 4.45
When every section of a filter is working between its image impedances, there are
no reflection effects. Fig. 4.45 shows a three section group connected on an image
impedance basis. This is equivalent to a single network having image impedances
Z¡, and ZI, respectively,
and B = (a, + ax + xa) + j(ß, + ßs + ßa) = a + jß (16a)
where 9, = ac, + etc.jß
The real part (a, + a, + a, = a) of the image transfer constant B is called the
image attenuation constant, and the imaginary part (ß, + ß, + ßa = ß) is called
the image phase constant.
4.8 (v) IMAGE IMPEDANCE 179
Where the values of the elements of the T and II networks are known, the value
of the image transfer constant is given by
T Network (Fig. 4.44) :
tank = + ZAZc + ZBZ c) (16b)
V (ZAZB
(ZA + Zc)(ZB + Zc)
R Network (Fig. 4.44A) :
ZB -
Z/a
tanh B
Z (17b)
Zf
Za
tanh B Zc - (17c)
For a II Section (Fig. 4.44A):
Z, = VZI1Zr, sinh B (17d)
Z2 - 1
1 1 (17e)
Z/2 tanh B Z.
Zl - 1
1 1 (17f)
Z tanh Z,
an L Section (Fig. 4.44B)
O
For :
with zero resistance or vice versa. Some popular combinations are given below :
FIG. 4.46
Fig. 4.48 is a half-section terminated in its image impedances, which in this case
are unequal. Two such half -sections, with the second one reversed left -to -right, are
equivalent to a single T section.
ZI = A/Z1Z2 + (Z12 /4) (25f)
Z1Z2
ZÍ = (25g)
At +
(Z1/4Z2)
1
An ideal filter in which the reactances have zero loss has zero attenuation for all
frequencies that make (Z, /4Z2) between 0 and -1 ; this range of frequencies is called
4.8 (vii) " CONSTANT K" FILTERS 181
the pass band. All other frequencies are attenuated and are said to lie in the stop
band (or attenuation band) of the filter.
Low pass filter -constant k type
Fig. 4.49 shows three forms of simple low -pass filters of the constant k type. In
each case
ZI =jwL
and Z2 = 1 /jwC.
Now ZIZ2 == R 2, therefore L/C = R 2, where R may have any convenient value.
The mid- series image impedance is
Z1 = R'/1 -
(wL /2R)2 = R1/1 (f/fo)2 - (26)
where fo = 1 /n1/LC), fo = cut -off frequency.
When f = fo, Z1 = 0 and there is infinite attenuation if L and C have no resistance.
The mid -shunt image impedance is
R R
ZI
1/(1 -becomes
(wL /2R)2 1/l - (f/fo)2 (27)
When f = fo, ZI infinite.
L2 L'2 L/2
s I
Z} Z j Li /2-
20 ` 04 20
T 04
F IG. 4.49
Fig. 4.49. Three varieties of low -pass constant k filters: (A) T section (B) Half
section (C) H section.
With both T and II arrangements, the ideal filter has zero attenuation for frequencies
less than f,, a sharp cut -off at fo, and a very rapid attenuation immediately above f,.
However the rate of attenuation gradually falls as the frequency is increased, and
approaches 12 dh /octave for the single section at frequencies much greater than fo.
Both image impedance characteristics are purely resistive below fo and purely re-
active at higher frequencies.
The phase shift varies from zero at zero frequency to 180° at fo, but is constant
at 180° for all frequencies higher than fo.
2
rIG 4 50
Fig. 4.50. Three varieties of high -pass constant k filters (A) T section (B) Half
section (C) I7 section.
High -pass filter-constant k type
Fig. 4.50 shows three forms of simple high -pass filters of the constant k type
In each case
Z, =1 /jwC
Z2 =
Now ZIZ2 = R 2, therefore L/C = R2, where R may have any convenient value.
The mid -series image impedance is
Z
= R1/1 -
(1/2R 0,C)2 = R1/1 (fo /f)2 (28) -
where fo = 1 /(4,n/LC), f,, = cut-off frequency.
182 (vii) " CONSTANT K " FILTERS 4.8
FIG. 4 51
mL /2 O mL /2 /m
2C¡¡ O. 2C /m mZl/2
mC
i Z Zt
l
Z
I I I
L 41^ C
o
LOW PASS HIGH PASS
Fig. 4.51. (C) Low pass (D) High pass T section m- derived filters (E) m- derived
half-section for matching purposes.
Values of Z, and Z, are as for the prototype (constant k) filter. A low-pass T section
series m- derived filter is shown in Fig. 4.51C, and an equivalent high -pass section
in Fig. 4.51D. In both cases the shunt arm becomes resonant at a frequency foo
given by
Low pass foo - "TrA/(1
1
-m =)LC
(30)
Two forms of shunt m- derived filters are shown in Fig. 4.52. The IT section (B)
may be joined at either end to a constant k section or half-section of mid -shunt image
impedance Z1'. The T section requires the medium of a half-section to match the
impedances before being so connected.
ZI (36)
(A) Z1m 1 + (1 -
m2)(Zi /4Z2)
(B) Z1' is as for Fig. 4.47 (Eqn. 22)
Therefore ZI,Zlm' = Z1ZÍ = 128 (37)
4
FIG. 4.52 C
4
FIG.4.52 D
The filter type has identical circuit connections but the condensers and inductors
are unequal (see Chapter 21 Sect. 3).
18db : available with filter type. This is the maximum rate of attenuation normally
used with loudspeaker dividing networks (Fig. 4.53C).
Ls = Ro /(217f,) Ls = (1 + m)Ro/(2Irf o)
Ls = 2Ro /(2nf ,) LI = Ro/(277f c)
L7 = Ro /(1 m)(2nf,) Ls = R0/(471,)
Cs = 1 /(2nf R0) Cs = l/(nf.Ro)
C6 = l /(4nf ,Ro) C4 = 11(1 + m)(2/rf eRo)
C7 = (1 + m) /(2srf,Ro) Cs = 1/(2arf rR0)
(L in henrys ; C in farads)
See also Chapter 21 Sect. 4.
SERIES PARALLEL
FIG. 4'SS
Fig. 4.53. Frequency dividing networks (A) 6db (B) todb (C) t8db for the octave
beyond the cross -over frequency. L.F. indicates low frequency, H.F. indicates high
frequency speakers.
(a) Tolerances
Every resistor has tolerances in resistance, and the price increases as the percentage
tolerance is made smaller. Composition resistors are usually obtainable with the
following tolerances :
± 5% For critical positions only
t 10% Desirable for semi -critical use in radio receivers and amplifiers (e.g.,
plate, screen and bias resistors)
* 20% For non-critical positions only (e.g. grid resistors).
" High stability " carbon resistors are available with resistance tolerances of ± 5 %,
± 2% and ± 1% (Ref. A26).
Wire -wound resistors are available with almost any desired tolerances in resistance
(± 5 %, 10% are usual values in radio receivers).
Comment on tolerances in components
When a manufacturer of resistors or capacitors selects simultaneously for large
quantities of each of three tolerances, ± 5 %, f 10% and + 20 %, there is a distinct
possibility that the -4- 10% tolerance group may be nearly all outside the + 5%
tolerances, and therefore in two " channels " differing by more than 10 %. Similarly
with f 10% and ± 20% tolerances. It is therefore good engineering practice to
design on the expectation of a large percentage of components lying close to the two
limits.
(b) Stability
The resistance of carbon resistors tends to drift with time. Ordinary composition
resistors may drift as much as ± 2% during storage for 3 months at 70 °C and normal
humidity (Ref. A27). See also Ref. A35.
Some high stability carbon resistors are limited to a maximum change in resistance
of ± 0.5% after 3 months' storage at 70 °C (Ref. A26).
(c) Dissipation
Composition resistors are usually available with nominal dissipation ratings of
I, i, 1, 2, 4 and 5 watts (JAN- R -11). The English RIC /113 standard (Ref. A27)
includes ratings of 1 /10, 5, 5, $, 1 and 2.5 watts. Other manufacturers produce 4,
4, 14 and 3 watt ratings.
English high stability resistors are available with 1 /10, 4, 5, 4, 5, 1, 15 and 2
watt ratings (Ref. A26).
4.9 i) PRACTICAL RESISTORS 187
In accordance with the American R.M.A. Standard and JAN -R-11 characteristics
A, B, C and D [see Chapter 38 Sect. 3(i)], composition resistors may be used at their
maximum ratings with ambient temperatures up to 40 °C.
The JAN -R-11 characteristic G and some English resistors (Refs. A3, A26, A27)
may be used with maximum ratings up to an ambient temperature of 70 °C. In
one case (Ref. A27) higher ratings than the nominal value are permitted below 70 °C
ambient temperature (see below).
At ambient temperatures greater than the maximum rating, resistors may only be
used with reduced dissipation and maximum voltage, in accordance with derating
curves.
The following is a typical power dissipation derating curve for the temperature
limits 40 °C and 110 °C (JAN -R-11 types, A, B, C, D) :
Temperature 40° 50° 60° 70° 80° 90° 100° 110°
Dissipation 100% 83% 66% 50% 33% 17% 5% 0
The following is the English derating curve for high stability Grade 1, RIC /112
(Ref. A26) :
Temperature 70° 80° 90° 100° 110° 120° 130° 140° 150°
Dissipation 100% 87.5% 75% 62.5% 50% 37.5% 25% 12.5% 0
The following is the English rating and derating curve for Grade 2, RIC /113
(Ref. A27) :
Co.) :
Size 1 2 watts
Voltage from zero to 200 350 500 500 volts
Fall in resistance 2.1 2.5 1.3 1.5
The voltage coefficient is defined as (Ref. A27) :
Voltage coefficient - 100 (R, -
R,)
R$(E, -E,)
where R, and E, are the resistance and voltage respectively at the normal maximum
rating, and R, and E, are the values at one -tenth of E,. Limiting values of the voltage
coefficient are (Ref. A27) :
0.025% per volt for values below 1 megohm
0.05% per volt for values above 1 megohm.
This effect is much reduced by the use of " high stability " resistors, a typical value
being 0.4% fall in resistance for 1 megohm, with voltage change from zero to 500 volts
(Dubilier). The limiting value of the voltage coefficient for Specification RIC /112
(Ref. A26) is 0.002% per volt.
Some applications require a resistor whose resistance falls as the applied voltage is
increased ; a wide range of characteristics is available (e.g. Carborundum " Globar"
ceramic resistors).
See Chapter 38 Sect. 3(i) for standard specifications.
(h) Effect of humidity on resistance
The effect of humidity is to increase the resistance by up to about 3% under normal
conditions. Extreme tropical humidity may cause an increase in resistance generally
less than 10 %. Some insulated resistors have less than 1% change in humidity due
4.9 (i) PRACTICAL RESISTORS 189
to humidity tests (e.g., I.R.C. type BTA). See Chapter 38 Sect. 3(i) for humidity
tests and limits.
(i) Capacitance of resistors
Every resistor has a capacitance which, at the lower radio frequencies, may be con-
sidered as a capacitance between the two ends of the resistor (usually between 0.1 and
1.0 µµF for composition resistors). This capacitance may usually be neglected in
normal applications in radio receivers. At higher frequencies it is necessary to con-
sider the capacitance as being distributed along the resistance element. This leads
to a reduction in resistance which, unlike the end-to-end capacitance, is not removable
by tuning [see (k) below]. References A5, A6, A7, A8, A9, A14.
- -
.89
.40
.79
.19
.61
.84
.14
.57
.80
-
.11
.75
Allen Bradley GB -1 0.85 .60 .48 .24 .17 .12
Allen Bradley EB} 0.90 .68 .57 .46 .23 .15
Speer SCT } 0.92 .70 .60 .35 .27 .20
Reference A14.
(m) High stability cracked carbon resistors (Refs. A10, A26, A28)
These not only have high short and long period stability and close tolerances (up
to ± 1 %) but also have low noise, low voltage coefficient and low temperature co-
efficient. They have practically no non -linearity of resistance and the inductance
is very low except for the higher resistance values. They are manufactured in Eng-
land with resistances from 10 ohms to 10 megohms and dissipations from } to 2 watts.
The extremely high resistance values are only obtainable in the higher wattage ratings.
See also (b), (e), (f) and (g) above, and Chapter 38 Sect. 3(i) for standards.
The historical development, constructions and special features of cracked carbon
resistors, with a very extensive bibliography, are given in Ref. A28.
These resistors are particularly suited for use in low -level high -gain a -f amplifiers,
and in r-f applications up to 100 Mc /s. The inductance varies from 0.001 µH for
a small 100 ohm resistor to2µH for a large (spiral element) 1 megohm resistor.
f
The I.R.C. deposited carbon resistors have tolerances of ± 1%, 2% and + 5 %.
The voltage coefficient is approx. parts per million per volt. The temperature
10
coefficient varies linearly from -
0.025 to -0.05 (10 MQ type DCH) or
(10 MQ type DCF). Maximum dissipations for high stability are } and } watt ;
0.065-
when high stability is not essential the values are J and 1 watt.
Ambient temperature 20 °C 40 °C 60 °C 70 °C
8 µF 350 V " micropack " plain foil 148 125 85 32 mA
16 µF 350 V " micropack " plain foil 250 200 110 50 mA
8µF 450 V etched foil 88 67 33 10 mA
16 µF 450 V etched foil 162 122 62 20 mA
16 µF 450 V plain foil 300 260 160 85 mA
32 µF 450 V plain foil 500 405 230 100 mA
With multiple capacitor units, only one of the units is normally intended for use
as the first filter condenser ; see catalogues for identification.
See Chapter 38 Sect. 3(x) for standard ratings.
(d) Paper dielectric condensers
Impregnated paper forms a very useful dielectric, being intermediate between
electrolytic and mica condensers as regards cost, size and leakage for a given capacit-
ance. It is usually manufactured in units from 0.001 to 0.5 µF, larger values being
built up from several smaller units in parallel in one container. The impregnating
material may be resin, wax, oil or a synthetic compound. Some impregnating mater-
ials enable condensers to withstand extremely wide temperature ranges (e.g. 50° -
to + 125 °C -Sprague " Prokar " with plastic impregnant). Waxes may be used for
moderate voltages and temperatures, as in radio receivers (from -
30 °C to + 65 °C
for R.M.A. Class W). Other impregnants are used for higher temperatures (e.g.,
85 °C, as R.M.A. Class M ; 100 °C as T.C.C. " metalpack " and " metalmite ").
The permissible insulation resistance at 25 °C is not less than 5000 megohms for
capacitances up to 0.15 µF, falling to 1000 megohms for 1 µF, but this falls rapidly
at higher temperatures, being 35% of these values at 40 °C.
A typical 1 microfarad wax paper condenser designed for audio frequency applica-
tions has the following characteristics :
in capacitances from 100 to 10 000 µµF in tubular form and from 0.02 to 4µF in
rectangular metal boxes (Ref. B11).
Mineral oil is used as an impregnant for working voltages from 1000 to 25 000
volts and operating temperatures from -30° to + 71°C (T.C.C. " Cathodray ").
The insulation resistance of a mineral oil impregnated capacitor is greater than that
with petroleum jelly impregnation, in the ratio of 2.5 to 1 at 0 °C, rising to 12.5 to 1
at 70 °C (Ref. B13).
Paper dielectric condensers are made in two forms -inductive, and non -inductive.
The former is limited to a-f applications, while the latter may be used at radio fre-
quencies.
Ordinary paper dielectric capacitors should not be subjected to high a.c. potentials.
Special types are produced by some manufacturers for use under these conditions,
for example with vibrator power packs and line filters.
For Standard Specifications see Chapter 38 Sect. 3(iii) and (iv).
Metallized paper dielectric condensers
This type utilizes a metal- sprayed or metal- evaporated paper dielectric instead of
the more conventional metal foil and paper construction. This construction results
in considerable reduction in size, while it also has a partial self-healing property in the
case of breakdown. The insulation resistance of unlacquered condensers is quite
low-of the order of 100 megohm microfarads-but some of those with a lacquered
film have an average insulation resistance as high as 8000 megohm microfarads at
25 °C (Refs. B14, B15).
There is a gradual reduction in the insulation resistance due to the self-healing
property, the degree depending on the number of punctures. It is desirable for the
total circuit resistance to be not less than 500 or 1000 ohms, to reduce the carbonising
effect of the arc. However if the circuit resistance is high, there may be insufficient
current to clear completely any breakdown, and the insulation resistance may fall.
Consequently, this type of condenser should not be used in high impedance circuits
without seeking the advice of the manufacturer.
A metallized paper dielectric unit should be used with discretion as the first filter
condenser following a thermonic rectifier, since the high peak breakdown current
may damage the rectifier unless the circuit resistance is sufficiently high.
The ratio of reactance to resistance (Q) of one lacquered unit with a capacitance of
2 is 200 at 0.5 Kc /s, 140 at 2 Kc /s and 60 at 10 Mc/s ; a 0.1 µF unit has Q = 98
at 10 Mc /s (Ref. B15).
The inductance may be made very low, and these condensers are very useful for
a -f and r-f by- passing. The usual (English) temperature limit is 71°C for d.c. opera-
tion, 60 °C for a.c. In tubular form (wax-coated) these are available from 0.0001
to 2µF with voltage ratings 150, 250, 350 and 500 V d.c. (RIC /136). Larger sizes
are available with capacitances up to 20 µF (400 V d.c. or 250 V a.c.) and voltages up
to 550 V d.c. (4 µF).
Voltage- temperature derating curves for Astron (U.S.A.) are 100% up to 86 °C,
linearly down to 38% at 120 °C for units up to 1µF ; larger units are 100% up to
76 °C, down to 22% at 120 °C (Ref. B16).
The effect of 5000 hours' operating life on capacitance is negligible up to 65 °C and
8% at 100 °C. The effect of the same operation is to increase the power factor from
an initial value of 0.5% to 0.6% at 65 °C, or 0.8% at 100 °C (Ref. B16).
The paper is usually impregnated with wax, although mineral oil has also been
used. Mineral wax impregnated units are generally preferred because of their
higher breakdown voltage, although their capacitance falls about 10% as the tem-
perature is increased from 50 °C to 85 °C. Mineral oil impregnated units have more
constant capacitance with temperature change.
These condensers are damaged by moisture and the unit is therefore well dried
initially and hermetically sealed to prevent the ingress of moisture.
References B10, B12, B14, B15, B16.
Standard Specifications- Chapter 38 Sect. 3(ix).
196 (ii) PRACTICAL CONDENSERS 4.9
(i) Definitions
is a transformer in which the winding reactances
are
An ideal transformer and winding
infinite, and in which winding resistances, core loss, leakage inductancesbetween any two
capacitances are all zero. In such a transformer the voltage ratio of loading,
windings is equal to the turns ratio of the windings, under all conditions windings
in Fig. 5.1. Also, in such a transformer the currents in any two
as illustrated
all load conditions.
are inversely proportional to the ratio of turns in the windings under
E2 NZ
E, N,
12 N,
Il NZ
voltages
Fig. 5.1. Ideal two -winding transformer. E, and E, are alternating of
(r.m.s.). I, and I, are alternating currents (r.m.s.) but the indicated directions at
current flow are at a chosen instant and correspond to the direction of voltage
that instant. Similar remarks apply to Fig. 5.2.
their
Modern iron -cored transformers often approach so closely to perfection for may
particular purposes that their analysis on the basis of ideal transformer theory
give useful practical approximations for design purposes.
5.1, separate
A double -wound transformer is one in which, as illustrated in Fig.
primary and secondary windings are used to permit isolation of the primary and second-
ary circuits except through mutual inductive coupling.
199
200 (i) DEFINITIONS 5.1
Auto -transformers may be used with economy in some cases : a single winding
is tapped to give the required turns ratio, which may be greater or less than unity,
between primary and secondary. An ideal step-up auto -transformer is shown in
Fig. 5.2.
E, N,
E, N,
I, N1
I, N,
Fig. 5.2. Ideal step -up auto- transformer.
An auto -transformer is always more economical to construct than a two winding
transformer as indicated in the following tabulation :
Total Volt-ampere Transformer
Type of Winding Ratios rating of windings Output V.A.
Auto 10 : 9 or 9 10 : 20 100
Auto 2 : 1 or 1 : 2 100 100
Auto 3: 1 or 1 : 3 133 100
Auto 5 : 1 or 1 5 : 160 100
Auto 10 : 1 or 1 : 10 180 100
Double any ratio 200 100
The currents in the primary and secondary sections are exactly 180° out of phase,
and the resultant current flowing through the common portion of the winding is the
difference between the two. When the ratio is 1 : 2 or 2 : 1, the currents in the
two sections of the winding are equal.
In ideal transformer theory there is no distinction between auto- and two-winding
transformers and they need not therefore be considered separately in this section.
FIG.
For an ideal transformer with a centre -tapped primary as shown in Fig. 5.4,
the transformed load measured across the whole primary (between terminals, P, P)
is equal to the transformed value R,. If however only one half of the primary is used
(between either of terminals P and terminal C.T.) the transformed load presented
is f R1.
E1 N,
E2 N2 RI CT
R, = (N1/N2)'.R2
= (El/Ece.R, á
FM. 5.4
Fig. 5.4. Ideal transformer with primary centre-tap and loaded secondary.
As a practical example of the primary centre tap, consider the use of the trans-
former of Fig. 5.4 to feed a 500 ohm line (R2 = 500 ohms). If the transformer has
an impedance ratio (N1 /N2)$ equal to 10 1, the transformed load across the whole
:
of the primary, e.g., when the primary is fed by a push -pull amplifier, is 10 x 500 =
5000 ohms. If, however, only one half of the primary were used for connection to a
single-ended amplifier, the load presented to the amplifier would be 1250 ohms.
For an ideal transformer with a winding tapped for load matching, as
shown in Fig. 5.5, the calculation of the tap to be selected for any particular load
follows from the application of eqn. (1).
Rl = (Nl/NAB)2RAB
E1
N1
or Rz = (N1 /NBD)$1? RD
R1 URNS and so on.
P
FIG.5S
Fig. 5.5. Ideal transformer with secondary tapped for load matching.
and, since the transformer is an ideal one, these relations hold irrespective of the
relative values of the loads.
R1 R2 R3
ER, /
(N,)2.
\/+
2 \N,/ 2 +R, 2
We therefore have
1 1 1
(4)
R, R2(Ni /N2) 2 + R,(N, /N2) 2
so that the total load R, presented by the primary is equal to the parallel combination
of the two transformed loads R2(N, /N2)2 and R,(N, /N,)2.
If an additional winding of N, turns is connected to a load R, we obviously have
in the same way
1 1 1 1
R, RAB(N,
and Ra (such as two loudspeakers) which are required to operate with power inputs
of WI and Wa respectively, so that W1 = W2 + Wa. The required transformer
turns ratios are then given by
R1W1 = E12; R2W2 =
E22; R,W3 = ES Z
(6a)
= N22/N12 (6b)
so that R2W2 /R1W1 = E22 /E12
and R3W8/R1 W1 = Ea 2 /E12 = N32/Ni 2 (6c)
For example if W2 = 3 watts R2 = 500 ohms
Wa = 4 watts R, = 600 ohms
and W1 = 7 watts R1 = 7,000 ohms
we have
N22
N12
500 x 3
7000 x 7
1
32.7
. Therefore -_ -
N2
N1
1
5.7
and similarly Na /Nl = L/4.5.
Expressions such as (6) may be written in the more general form
N /N1 = "V (R /R1)(W /W1) (7)
where N = number of turns on secondary n,
N1 = number of turns on primary,
R = load applied to secondary n,
R1 = transformed total primary load,
W = watts in load R,
and W1 = total watts input to primary,
but it must be noted that these relations hold only when all loads are connected so
that the specified input conditions to the primary do exist. Eqn. (7) is also applicable
to determine the turns ratios of two or more separate transformers feeding two
or
more loads from a common amplifier which delivers W1 watts into a total load of RI
ohms.
FIG. 5.8
Fig. 5.8. Ideal transformer with multiple loads and tapped secondary, which is
effectively identical with Fig. 5.6.
-
A useful equivalent circuit of a practical transformer is shown in Fig. 5.9.
0-- ao !
, I
I lCyr
6ö
L2
Ro a
LpI
g
I
N2
ÓTUR
NSI
1 1 I ;J
iliDEAL TRANFORM
FIG.5.9
Fig. 5.9. Equivalent circuit of a practical transformer.
In this equivalent circuit we have an ideal transformer with a turns ratio N, /N,
(equal to the turns ratio of the actual transformer), with the incidental characteristics
of the actual transformer represented by separate reactances and resistances.
r, = the resistance of the primary winding
L, = the equivalent primary leakage inductance
r, = the resistance of the secondary winding
L, = the equivalent s__ ondary leakage inductance
R, = the equivalent core -loss resistance (including both hysteresis and eddy
current losses)
L, = the inductance of the primary winding
C1, C, = the primary and secondary equivalent lumped capacitances
C = the equivalent lumped capacitance between windings
R, = input resistance of transformer on load
and R, = the load resistance across the secondary.
Such an equivalent circuit is capable of representing a practical design with con-
siderable accuracy, but actual calculations would be tedious and in some cases very
difficult.
The reactances and resistance. shown therein have varying effects on the input -
output voltage ratio according to the frequency of the signal which the transformer
is handling (Ref. 1). In general, the equivalent circuit can be presented in three
distinct simplified forms for use when considering the transformer operating at low,
medium and high frequencies respectively (Figs. 5.10B,C,D).
Audio transformers can be conveniently dealt with in this manner, whereas power
transformers operating over a very limited frequency range can be more simply de-
signed on the basis of Fig. 5.10A.
The whole of the equivalent circuit of Fig. 5.10A can be referred to the primary,
as in Fig. 5.11 where the ideal transformer has been omitted and r, and R2 multiplied
5.2 (i) GENERAL CONSIDERATIONS 205
by the square of the turns ratio. This is often a convenient way of making calcu-
lations.
IDEAL TRANSFORME
F IG. S IO A
Fig. 5.10A. Simplified equivalent circuit for calculating the effect of losses.
FIG. 5.10 B
Fig. 5.10B. Low frequency
equivalent circuit.
Fig. 5.10C. Medium fre-
quency equivalent circuit.
Fig. 5.1OD. High frequency
FIG. 5. IOC equivalent circuit.
R = Total winding resist-
ance referred to prim-
ary
L, = Total leakage induct-
ance referred to prim-
ary
}V('}j{6
FIG. 5.100 C = Total winding capac-
' itance referred to
RM `f
T primary
E¡ R Eo RL = Secondary load re-
sistance referred to
primary
However, the difference between the two definitions is quite small for small per-
centages of voltage regulation.
The regulation of audio transformers when operating over a limited frequency range
(say 200-2000 c /s), and of power transformers working at 50 -60 c /s., i.e. the ratio of
E2 /Ei, is affected mainly by the copper losses.
Thus E2 /Ei 77T and I2 /I1 1 /T, (1)
where i = efficiency
and T = turns ratio.
As a further consequence the impedance ratio is changed, and, making the same
assumptions as above,
Ri ri r2 /T2 + R2 /T2 (2)
At low frequencies, the reflected impedance is altered by the shunt effect of the
primary inductance, while at high frequencies, a similar change is caused by the
leakage inductance and winding capacitances. This is covered in Sect. 3(iii).
The damping of a loudspeaker, connected as a load to the secondary of an output
transformer, is also affected to some degree by the losses. Where an accurate in-
dication of the damping factor is required, these losses should be taken into account.
Refer to Chapter 21 Sect. 3.
Rigid clamping is less important since there are no free laminations to vibrate under
load. For this reason also, combined with the somewhat discontinuous nature of the
material, the acoustic noise emitted by the block is considerably reduced, especially
at higher frequencies.
PHYSICAL PROPERTIES OF CASLAM
Property Caslam I Caslam 2 Caslam 3
Maximum permeability 860 at 4 Kg 1000 at 4 Kg
Effec. permeability 500 at 10 Kg 830 at 10 Kg Similar
Hys. loss for Bmas = 10 Kg. at 50-e 7000 7000 ergs /cc/cycle to
Coercive force B,,,a: = 10 Kg. at 2.4 2.4 oersteds grade 1
Sat. flux density 18 Kg 18 Kg
Total a.c. loss, B,, = 10 Kg. at 50-' 2.5 2.75 watts /lb.
Density 7.0 7.4 gms /cc
Max. Oper. temperature 110 °C 110 °C 150 °C
Resistivity (ohm -centimetre)
(a) Normal to plane of laminae 0.04 0.03
(b) In plane of laminae 0.003 0.002
The a.c. loss is almost entirely hysteresis, the eddy current loss being less than
10% of the total. For this reason, the a.c. permeability at is approximately
equal to the d.c. figure and at higher acoustic frequencies, blocks of CA SLAM
compare favourably in magnetic properties with stacks built up from normal silicon
iron sheet.
Dust cores generally suffer from low permeability and, to reduce eddy currents,
particle size has to be reduced ; this causes further reduction in permeability.
Other new ferromagnetic materials such as the ferromagnetic spinels and FERROX-
CUBE are described in Chapter 11 Sect. 3(v)E, and find their applications principally
at frequencies above the audio range.
(iii) Frequency response and distortion
(Reference 31)
(a) Interstage transformers -Class A and B
At the mid- frequency* the amplification is very nearly equal to the amplification
factor of the valve multiplied by the turns ratio of the transformer, where the secondary
is unloaded.
At low frequencies the gain falls off due to the decrease in primary reactance. The
R (1)
\wLo /
where L = primary inductance
and R = plate resistance plus primary resistance.
The response will fall off 3 db at a frequency such that ß,L0 = R. At a frequency
such that
wLo = 2R (2)
the response will be down approximately 1 db from the mid-frequency level.
At high frequencies the leakage reactance and shunt capacitance, in conjunction
with the plate and winding resistances, form a low Q series resonant circuit. Above
this resonant frequency the gain will fall off rapidly. In the neighbourhood of re-
sonance the change in gain will depend on Q0, the Q of the resonant circuit. This
factor can be varied by adding external resistance or by winding the secondary partly
with resistance wire (Ref. C3). The resonant frequency can be varied by changing
the value of the total leakage inductance, L, or the interwinding capacitances. These
are both functions of the transformer structure. See also page 518.
By careful choice of core material, lamination dimensions and method of sectional-
izing the winding, it is possible to achieve a frequency response extending beyond
the normal audio range (Refs. C2, C4, C5, C6 and Figs. 5.12 and 5.13A).
"The mid-frequency is the frequency at which maximum gain is obtained.
210 (iii) (a) INTERSTAGE TRANSFORMERS 5.3
111
sheet steel or thick cast iron outer cases are used. More often a special alloy, such
as Telcon 2129, is used. This material has magnetic properties similar to RADIO-
METAL, but is suitable for deep drawing. Inner shields are often of copper to give
shielding from electric fields (Refs. C11, C12). Some improvement is possible if a
core type structure is used in place of a shell type, owing to the cancellation of stray
voltages induced in the winding by external fields.
As an example of the use of high permeability alloys, the following design problem
is presented. Calculate the primary turns for a 50 ohm to 50 000 ohm transformer
working at zero level (1 dbm). The frequency response must not fall more than 1db
below mid-frequency response, at 50 c/s Distortion must not exceed 1% at zero
level at 50 c /s. The source impedance is 50 ohms resistive and the secondary is
unloaded. Core material to be used is MUMETAL.
1st Step. Calculate primary inductance.
For 1 db attenuation, wLo = 2R (Eqn. 2).
L, = 2R /w = 2 x 50/2 x it x 50 = 0.32 henry.
2nd Step. Calculate primary turns.
Assume square stack of Magnetic and Elec. Alloys No. 35 lamination. Length
of magnetic path, 1 = 4.5 inches
Cross sectional area = 0.56 square inches.
3.2A µN8 L. x 108 x 1
L, 10 8 x 1
Therefore N' -
3.2 x A x
N2
_0.32x108x4.5 Therefore N = 90 turns.
3.2 x 0.56 x 10'
3rd Step. Calculate working flux density.
Primary voltage E = W.Rv
where W = input power
Therefore E = x 10 -8 x 50 = 0.224 V
0.224 x 108
and B - 4.44 x 50 x 90 x 0.56
= 2000 lines per square inch
= 310 Gauss.
4th Step. Determine percentage distortion.
Referring to Fig. 5.13B and using curve FIG. 5.13 8
for wLo = 2R, it will be noted that the
percentage distortion is approximately
0.75 %. This assumes that the perme-
o
e
6
U
.UOunr =:;, 1O,
:::::
u...
unn
II 1111
ability value, µ, is still 10 000 at the opera- 3 111.n11111=111/11/ 111111
ting flux density. In practice may ex- 2 1/.11111
ceed this figure and thus the distortion as
calculated above may be larger than
be measured in a finished transformer. In
would
ae p__
M "1'
OM11111
.:R:_Ol:C::::a
this problem, for simplification, no account
has been taken of the stacking factor, which
0$
0.4
03
i3i.."...ÌÌ
iarerii:ri
would modify the result slightly.
When the secondary of the transformer 0.2
is loaded, R in eqn. 1 then becomes R4
o,
as in eqn. 3 and R in Pig. 5.13B is read as
RA. The calculation for distortion then ó X
ó$ :m:
follows in a similar manner to the unloaded ama, -9au,f
secondary example worked earlier.
Fig. 5.13B. Total harmonic
(ïü) (c) Output transformers distortion plotted against flux
The factors affecting the frequency res- density for different ratios of
ponse of output transformers (Ref. C13) primary reactance to R where
are similar to those affecting interstage R = plate resistance -}- primary
transformers. Refer to Figs. 5.10B, C resistance (Ref. C9).
and D.
212 (iii) (c) OUTPUT TRANSFORMERS g. 3
The response falls off from the mid- frequency gain by 3 db at a low frequency
such that
wLa = RA (3)
Attenuation is 1 db when
wL0 = 2RA (4)
where R, - (r, + RW)RL
r9 + Rw RL
r9 = plate resistance of valve
RL = load resistance referred to the primary
and Rw = total winding resistance referred to the primary.
The response falls off from the mid- frequency gain by 3 db at a high frequency
such that
wLs = RB (5)
Attenuation is 1 db when
wLs = 0.5 RB (6)
where RB= rD +Rw +RL
and Ls = total leakage inductance referred to primary.
The gain at the mid- frequency = E0 = µR Lt (7)
RB
where E, = voltage input to grid of output valve
Eo = voltage output across RL
and o = amplification factor of output valve.
It is thus possible to specify the primary and leakage inductances permissible when
the frequency response requirements are known. An example will illustrate this.
Determine the minimum primary inductance and maximum leakage inductance
permitted in an output transformer designed to match a pair of Class A 2A3 triode
valves with a 5 000 ohm load. The response is to be within 1 db from 50 to 10 000
c /s. The plate resistance of each valve is 800 ohms. Neglect Rw.
Make all calculations from plate to plate.
For a fall of 1 db at 50 cycles per second,
wLo = 2R A.
5000 x 1600
Now RA - 5000 + 1600 -
1200 ohms approx.
wLo = 1200 x 2 = 2400
Lo = 2400/2 x n x 50
7.6 henrys approx.
This is the value that would be measured on a bridge at low induction.
For a fall of 1 db at 10 000 cycles per second,
wLs = 0.5 RB
Now RB = 5000 + 1600 = 6600 ohms.
wLs = 0.5 x 6600 = 3300
Ls = 3300/2 x 7T x 10 000
= .052 henry approx. = 52 millihenrys.
Note particularly that distortion requirements may necessitate an increase in Le
and a decrease in L.
The following table indicates the relationship between the low frequency attenuation
and the ratio of wLo /RA :
TABLE 1
The inductances required for various values of R 4, for a bass response loss of 1
Where the source impedance is high, the high frequency fall off is determined
largely by Ls and the winding capacitances as in the case of interstage transformers.
Similar devices to that employed in the construction of interstage transformers can
be used to extend the range of output transformers. The use of RADIOMETAL,
specially annealed high silicon content steel (such as SUPERSILCOR), and grain
oriented steels, are all common in better quality output transformers (Refs. C14, 15).
It is feasible to " build -out " a transformer into a half section filter and thus main-
tain the impedance, viewed either by the source or the load, constant over a wide
frequency range. In addition the phase angle variation is reduced towards the ex-
tremes of the range. This reduces distortion and maintains full power output to a
greater degree than otherwise possible.
This idea can be applied to interstage transformers and output transformers quite
successfully (Refs. C19, C20). Even large modulation transformers and class B
driver transformers for broadcast equipment have provided improved performance
when treated in this way (Refs. C16, 17, 18).
In radio receivers and record players advantage can be taken of this " building -
out " procedure, to limit the high frequency response to any given point, say 6000 c /s,
with rapid attenuation thereafter. This usually involves only one extra component ;
a condenser across the secondary winding of the output transformer. This is quite
an effective " top " limit, more so than the normal tone control. The output trans-
former is designed to have the necessary amount of leakage inductance for the net-
work to function as intended.
The winding resistances are not of major importance in interstage transformers,
but assume greater significance in output transformers. An appreciable amount of
power may be lost unless the resistances of both primary and secondary are kept to
reasonable proportions. In the normal good quality transformer the total resistance
reflected into the primary side is approximately double the measured d.c. resistance
of the primary. The total winding resistance (referred to the primary) will vary be-
tween 10 and 20 per cent of the load resistance, which means an insertion copper loss
of 0.5 to 1 db. This extra resistance must be considered when choosing the turns
ratio to reflect the correct load (see Eqn. 2, Sect. 2), otherwise an impedance error
of 10 to 20 percent will occur. Core losses will not materially affect the calculation
as these losses do not reach their maximum except at full power at the lowest audio
frequency of interest.
Distortion in output transformers
When a transformer has its primary connected to an audio frequency source of
zero impedance, the waveform of the voltage on the secondary will be the same as
that of the source -in other words there is no distortion.
When a transformer is connected in the plate circuit of a valve, the latter is equiva-
lent to a resistance r9 in series with the source and the transformer primary. If the
secondary of the transformer is also loaded by a resistance R,, this is equivalent (as
regards its effect on distortion) to a total primary series resistance R,
1 1 1
where
R = r,, + R,(Ni/N2)$
N, = primary turns
and N, = secondary turns.
In the following treatment the symbol R is used to indicate the total effective primary
series resistance, whether it is caused by r, alone or by a combination of this and
secondary loading.
The resistance R in series with the primary causes a voltage drop proportional to
the current flowing through it, which is the magnetizing current. Now the form of
the magnetizing current is far from being sine -wave, since it is distorted by the non -
linear B-H characteristic of the core material.
This distorted current waveform has no bad effect when R is zero, but results in
distortion of the voltage waveform which becomes progressively greater as R is in-
creased, for any one fixed value of Bmaz.
5.3 (iii) (c) OUTPUT TRANSFORMERS 215
The resulting harmonic distortion with silicon steel has been calculated by Dr. N.
Partridge, and his results are embodied in the formula which follows (Refs. C24, C25,
C26, C27).
V,, 10° R4(1 RA 1
V,
-SH 8n' N'A
1
f`
4Z, J
(8)
This formula can be modified to include the core stacking factor, 90 %, and to use
inch units instead of centimetres. It then becomes
V,,
V,
=
5.54SH1RA
(1 4Z,
- RA (9)
where V5 = the harmonic voltage appearing across the primary,
V, = the fundamental voltage across the primary,
SH = the distortion coefficient of the magnetic material,
1 = the length of the magnetic path,
N = number of primary turns,
RA = resistance (or equivalent resistance) in series with the primary, (refer
under eqn. 4, Sect. 3),
A = cross -sectional area of core,
f = frequency of fundamental in cycles perfrequency
second,
2zr fL
Z, = impedance of primary at fundamental
and L = inductance of primary in henrys at chosen flux density.
In most cases the final term (1- R 4/4Z,) can be omitted with a further simpli-
fication.
The right hand side of this equation gives the value of the fractional harmonic dis-
tortion ; the percentage harmonic distortion may be obtained by multiplying this
value by 100. The formula holds only for values of RIZ, between 0 and 1 ; this
limits its application to output circuits having a maximum attenuation of 3 db.
u0
100
2 SO
SM
u
60
.4
0 40
20
s
00 40 60 70
10 20 30 x SO
3
Klo)
e
o
O - 10 20 30 s 40 50 60 70
FIG. 5.13D X10
Peak Flua Density (11M. per
Both of these figures have been adapted from those published by Partridge (Ref.
C25).
An example is quoted to demonstrate the use of these formulae.
Determine the transformer distortion produced when a pair of KT66 valves (very
similar to type 6L6-G) connected as triodes are operated in conjunction with a trans-
former having the following characteristics.
Lowest frequency of operation 50 c /s.
Maximum flux density 40 000 lines /sq. in.
Core stack of M.E.A. 78 Pattern lams. 2 ins.
(Refer to Sect. 5 and also Fig. 5.18C for lamination data).
Operating conditions of KT66 valves
Plate voltage 400 volts
Plate resistance (per valve) 1450 ohms
Load resistance (total) 10 000 ohms
Power output 14 watts
1st step. Calculate primary voltage
E _ '/ WR = A/l4 x 10 000 = 374 V.
2nd step. Calculate primary turns
E x 108
90% core stacking factor)
N 4 x IBA (allowing
374 x 108
4 x 50 x 4 x 104 x 1.25 x 2 1870 turns. -
3rd step. Determine µ from Fig. 5.13A
= 4300.
4th step. Calculate Z,
277f 2.88 N'A
(.
2.7 fL)
Z,
1081
2 x x 50 x 2.88 x 1.87 x 1.87 x 108 x 1.25 x 2 x 4300
7r
108 x 7.5
= 45 300 ohms.
5th step. Calculate RA (parallel resistance of plate and load resistance).
RA
(r, + Rw) RL (see eqn. 4, Sect. 3)
rD } R1 + RL
assuming R w = 400 ohms
5.3 (iii) (c) OUTPUT TRANSFORMERS 217
then
(2900 + 400) 10 000
2480 ohms.
RA
= 13 300 =
6th step. Calculate - (R4/4Z,)1
- (RA /4Z,) = - (2480/4 x 45 300)
1 1 = 0.986.
Thus this factor can be neglected without serious error.
7th step. Determine SH from Fig. 5.13C
SH = 60.
8th step. Calculate fractional distortion
VA . 5.54 SH1R A 5.54 x 60 x 7.5 x 2480
V, NMI 1.87 x 1.87 x 106 x 2.5 x 50
_ .014.
Thus percentage distortion is .014 x 100 or 1.4 %.
Note High fidelity output transformers may be designed with distortion less than
:
From eqn. (11) it will be seen that the insulation between the sections is the limiting
factor. Fig. 5.13E shows that the largest value of N2 for a given number of sections
is achieved when there is a half section at the end of each winding structure. Al-
though series connections are shown, similar results can be obtained by paralleling
the sections. In this case all the turns in each paralleled section must be equal. An
MING ARRANGEMENT
example of the use of Fig. 5.13H to
M10/K'. ARM/CEMENT W N'
determine leakage inductance follows :
n.nw,;°TM!", I ,e1/1 . II . II.
Q
25
gcóm:144
á®®áápflo
total secondary winding height is
.24 . a00312111201:0.:100 0.16 inch.
:aGaa00Q0:,121 a = 0.4 inch.
Referring to Fig. 5.13H, intercepts
N2 = 16.
b r;s'I!ry!i:';
íÿ.. ri!'%%'1'ii
!11.ia.i11,/. .,,.i
E
PRIMARY
c
T
b
i\
M/i/,n,l; 'summon
üïlü.n.
_4:
%,1.,ii/iil!
..n.isi:
iiPs'ssI r..,'///R!r./rsn
Z:::t::D. ::io
iÚ%RÌÌÏ/%iÑÌÌÌIG.0 S
1".
- o ,`o,..
`~
1.`10111\..\\\11
C\\\I11\ ',:r=---
11IUNI I.:31\5;
1`1:`
M---w
0:::
\,.:
1
6cd
SECONDARY
is...iori ssi.Inini!im
WINDING CROSS SECTION
L usasommmin a
ilOG%%iPRiÌ
FIG. 5. 13F
%%V:dl/ 5IIIII11W1
N' FIG. S 13G
LEAKAGE INDUCTANCE
,W -% be.°
MICROHENRIES
. e% 5b%//IIH////////////I lAt %-/NI%
1.° +P al aa"` ,' ///
II/L%%1,2i/ií/.'LÍí'SiR'/.%'L%/Wd%iGiS/.'/.%%i/./.%2%/.%i/%i3?S/./,"/.Í
1P'/.LiL8.ilí/C:i/i/iO4/.%%StenIi//ii/G//Si/.%%G5G%iIí//
db. %
`'
íi/./.%'Li
1'/.'/.'/,%Ì%ríRi/ %%Gr5M/.%!%L%%7Lr5,i%2%.// Ì%%/ÌL7.%%%Cí!í%%W1r7/%%E0 54036I65/
I'./.'I./..1.1íCi5/.A'I./iLC1Cir%'I.4I,/i1i60'/./L.IS'/./I/'/.I./.I/SCiU'/II.rD'/.IVGr
MIIlIHENRIEs
i
/
1 %'I. i/2.I.i:ll LO7lIO / % %I../7.C%i!I LL7...I0/.Ì
200iiI'I%I.IIL.....0%1.II.
i...u.i.....aanui..iu.a...............v.e..í
.GL.L11.%'I1I.I %%%
1%lL76.l.1/1/% 4 dÍ 44 4l1L76641/l% 75e. O%74iILO6 G.]1! 17O56Sf%46
/
15LiL/7iSLí1,21O77í11f/%%0:IG/SSS.71ÌLO75d I,/I'/.L,/%LWiIÌOO..1I./70
%/I.'IiSi11./%%19//i%Viiii..so1,ri9r.viiiiiii.ni.
%I.IS.1..%....L.1%%%1I1...V.11I.I/III%iI Ìd.7./.Í1
L1II ;%./
:...
i1I:!
.1
Á7L%4i/li e
b
b
%/.iI ÍÌO...I.I
ii..iiii
1/%%6e'/.SL?'i//Á7í9.756%%611LIyí/ii/-Lí//.755/%%0%1L:%í/%
,
I'/.'I.%'/.i%IL1!'I.7!%77/L/'/.í'I.'/.7'/.í%IRi!'I.7MM%'/.Ir77771:IrL7'/.nrL%ki/ Q
1%%/.11CÁ5/.IG.r'd%C/./, rILL'/..6i1,o1%!4/SC7././.C5/.%'/..1C.C11%/. dSrL'/..11r
,a.L...II.1.CI..9a.L.G.91.111IR.IS[i
30O1Ì'/ai6IíI4!I./"P.1/Si .L!'/....rSI.!IO/..!.. ÌLOOrSrÁ"LOI..1.E7r
R 1%/ ....q, ÌL.7..,/IL..1
1
./y, VL; Ï
'/./L..1.11:I.l7SSSR..1 %.Í.1.11.I.//IS:I..1'/./'I.1:/ISI,IÍSS SI./Í%
,uGn.i..1
iÌï-ii7:ïlJiiÌi/ïIs%ii iiiÌáSieiiiiÌiiVÌiïii%iiiJ iei.Ìa%ÌI
.a.S/LLV....Si.v..u..i.
Á2i/0551
í9:t8/.%51/,%G:1íi;''/,%Ì.íI.%.Íi0
.
1
_
Z 1%///Li1L16//i/í/L%I%///í./Á6LgIL/I/I/I/.%/1Ll%I7./I/L/í
5001%%05e02i/E%2tiR05.e*/05S01ti/I%i/75e*%J201E2iee $
/ .
\\\ . /Ì
6001//154/L/21%ß07/:J//4/4ii5/irL/0L/:/4/L'/Gi4/I15líG.L/S//Ii
//%6/21/1 IR.IIJ'/%5 i/%.1IEriI:/Ì'// /%1%4IIRIL f. L .1
700/%/L6/2Z!I./16/iit/SII5//II 1'I//%/%I/IALGr2!ICI1Í%/I/
IL9/.Ii..S.I,IIL..r.II.. tl
ÌI%,I7S /SILI.IÌLL'.9.57u0I.%SISS.Ì...SU.Í.L7SÌ'IS
zI7IL.aS.ILO7..ILL.%ILLfvIiLSíO
.i..1 /su..csn..amv..u"Y.uVO.r.U...md9....rr.
III7/Y/19./LY7/%%%IIIW'
'..1.iir:
a.ec.icci5
1ar ccans.ácc
1\`.\`\`b:Ci\\
lì:\SSC:2::0\:Ç:l\\á
... \\.
:.IJi.a.
iü..um..iiu.
\\e
2as..u.aaic....an.oc.w
1
g:
.`U.\P
\`\Ni.\`\ CeCS:IS.J.s..
\\.\\`V.0 .
V\
\\NOSIR\.\`\.
i.,. `\.
ieccvceaa:
711LP.%%/C. %¡O7III%D7r1!ÌL%I1J%%.7
ï.a.I%L...,/..I,..IS
i=.. ..uí=.o......u...... si5i
..1..i.s...v..4..nr .5i. ==:
.
\.\ 1\..`.`
co..ìCcc..`o
ICCS\\`\C6\.\'\ I
v:v.oc.sa.ecss..e1
.v3Cy,1¡O\\1b82I\0 N.\\\:.2::Ci O:\\1 p
aaac2Cg.
0'1iwaocnaaaoc.saearwa.c.aaa..clsas...w1ea.ecve `aac...s.cCQa wale
.
20
i, OS...e.vesa.e.aaam
L'J
° am.acs.e.aowseia.a .me.sv.vacc.n.av°voaac.mcsa g
0J....cu. so.. sa.s. ,!'Cr...7....a.a
oc\.`\r9o7oocVloOOa
aam.óws rauc..3
\\SS`3C5:`\061.O\67 Oa:ilìNbC.\`.acb110Nr\\.V\C3i.V`..`bSCON\I
c:o.ccl0ee\\\7cCN:m\occ.\
o.u.e L.... .. ..`a...
ó
w,.uc.a.cc.acw:a.àcbtcaccaa.c0.ac1ca.cwas1 3
031.\`\\\1C3I3:0`\S`.\`N.\`OV8C3C\OC\\SC\S0C3C3`Oa`\SC.`\
mow 14
i 02a..`
Za a '\\J
1avw.1racw.a \o o.u,a. vü3ocw.
laa\,\'\64:ÌN\\Ì.i\SI\N.\'\\6Ì:Ì\\S\1S\\\I\\\S\64:Ì\\\\\S
ti\4\ aa.aa
\\OÑ.i..Z
..C.JISaccrZarw.
`\f. oácucav.a
e
.
c.ss..c
\N..64C1\\`\535SR\1 \¡\SSC3C3:0'\.i5\\8
\\I
.. ..o.`. 5
CS:V\ `6j1,\\`e11C\S:.\`.DCS:O\`\S`I`ß111
2
1
..---...--.
\C1\S5C3C3:0\\C3\\SI
ii
IyC1\S5C3C3C\\33u\\
ADD
INCHES
01
'c-"
..uuu..orLL..LUG.aIIIwiL. uv
---.....-----.=...-----....----°--r-----
°°was..u.us..uso111110ous..os...mos.ouu111..so
1-.
n.ou.o
m..00...n.I.II .oIii<cl.i.7...v.r4ilr.v.ur....o....0
.1U
..u.1111L/L/0Il!7I/%I%.y75A% 7/../717/1iQ
4
u..V.u..111
6 I
sáÌiíiii:ii4.!ii:
=m.iiióii:iii.á+iii i!iiiiiiiG.iiIiÌiIIÌmolI 08
i:ïiii¡%iiiiiiiiiiióiiiiii:.iiiiiiiii
ii.o....IríL...r.L.I:r11 MIL.1USO......11
I
MIN.....15. AII'L1%.%/.7.roMG.:.515.S.v7.a.'I
....r..r:1....:./ i ii 5,
......:i..i ;..i.'.ii...
.....o.a i...r........
. i=.LG./SU%.5/.iS%'I.I/%L.1191171.5 .eouw...
111Ì i/2/SIrSÍi/i%1i i iili OiCíE'/.ii.2.1L%/lE%./%111111611 04 .°
,.....a..i_..........i
oov/enin..ua
rL1..171.1.11.1...1u .11
u 04 g OS
,
in the same layer. The capacitance effect between adjacent turns in the same layer
is neglected, only that between layers being considered. Capacitance between wind-
ing and core, and electrostatic screens, if used, must also receive attention.
Fig. 5.13J illustrates a cross section of a piece of winding in which layer inter-
leaving is used. It is seen that the dielectric between adjacent conductors in consecu-
tive layers is complex both in shape and material. The turns on the top layer shown
fall so that each turn drops in the space between two turns on the second layer, corru-
gating the interleaving material with a slight resulting increase in capacitance com-
pared with that between the middle and bottom layers shown. Due to the spiral
form of each layer, the position of turns in consecutive layers to one another will
change at different points round the direction of winding, thus the capacitance be-
tween any pair of layers will automatically take up the average value. The composite
dielectric is made up of conductor insulation, most commonly enamel, interleaving
material and the triangular shaped spaces left between adjacent turns and the inter-
leaving material. These spaces will be filled with dry air if the windings are dried
out and hermetically sealed, or with impregnating compound if the windings are
vacuum impregnated. The latter procedure will give rise to a somewhat higher
capacitance.
Coppa Conductors Enamel
In practice the major controlling factor determining the total capacitance between
two adjacent layers of winding is the thickness of the interleaving material, thus dis-
tributed capacitance may conveniently be estimated in terms of the thickness of inter-
leaving material, giving this material a value of dielectric constant empirically ob-
tained, allowing for the average effect of the other dielectrics in the composite arrange-
ment.
ance referred to the high potential end and considered as a pair. Thus the capacit-
ance of n whole layers referred to the whole winding becomes
4 (n - 1)
CI
3 n2
For large values of n the capacitance becomes inversely proportional to the number
of layers. Fig. 5.13Q illustrates a typical winding shape together with the dimensions
as used in the related diagrams. The capacitance per layer, given in the foregoing
formula as Cl, is proportional to the product of the length per mean turn L.. and the
length of layer L
Vertical sectionalizing
It is sometimes of advantage to sectionalise the winding as shown at Fig. 5.13L,
each vertical space being filled completely before proceeding to fill the next one.
FIG. 5.13 L
Although physically the winding will have the same overall cross -sectional dimensions
its self-capacitance will be equivalent to that of a winding having 1/N x layer length
and N x n layers, where N is the number of vertical sections. The distributed
capacitance of the winding due to such sectionalizing thus is reduced by the factor
1 /N'. Note that this rule applies only to referred interlayer capacitance and does not
apply to capacitance between the top and bottom of the winding and adjacent wind-
ings or screens. The various reduction factors for vertical sectionalizing are given
in Table 4.
TABLE 4
1 1 1 0.5 0
2 0.25 0.75 0.25 0.125
3 0.111 0.704 0.185 0.185
4 0.0625 0.6875 0.1875 0.219
5 0.04 0.68 0.168 0.24
6 0.0278 0.676 0.176 0.255
co 0* 0.667 0.167 0.333
The method of winding is here changed so that for the purposes of thi s column,
the dimensions L. and T. will change places.
Effect of mixing windings
In the design of a transformer it is often necessary to mix the primary and secondary
windings in order to reduce leakage inductance. This arrangement will generally
be a disadvantage as regards minimizing winding capacitance, since it exposes greater
surface area of winding in proximity to either the other winding or an earthed screen.
If the ratio of the transformer is fairly high, then from the high impedance winding
the whole of the low impedance winding appears at common audio potential, usually
earthy. But if the ratio of the transformer is not very high, capacitance between
222 (v) WINDING CAPACITANCE 5.3
points at differing audio potentials in the two windings may have serious effects, and
it is generally best to arrange the windings so as to avoid such capacitance.
Fig. 5.13N shows a cross- section suitable for an inter-valve transformer designed
to operate two valves in push-pull from a single valve on the primary side. The H.T.
end of the primary is earthy and is therefore diagrammatically earthed. The high
potential end of the primary is adjacent to the earthy end of one of the half secondaries
so that the capacitance between windings at this point is effectively from anode to
earth. The two high potential ends of the secondary are remote from the primary
and so minimize the possibility of unbalanced capacitance transfer from primary to
one half secondary.
(a)
i Anode
Half Secondary
Grid
Primary
Fig. 5.13M. Arrangements using vertical sectionalizing. See Table 1 (Ref. C29).
Fig. 5.13N. Push-pull secondary intervalve transformer secondary arrangements
(Ref. C29).
Another problem which often arises is in the design of push -pull output trans-
formers, particularly for Class AB or Class B circuits, where it is essential that each
half of the primary be well coupled to the whole secondary. From the viewpoint
of leakage inductance and winding resistance, it is unimportant whether the secondary
sections are connected in series or parallel. Fig. 5.13P illustrates three arrangements
for a transformer of this type, each of which may be best suited under different cir-
cumstances. At (a) is an arrangement which gives minimum primary capacitance,
but suffers from the defect that leakage inductance and winding resistance are unequal
for the two primary halves. For Class A operation using valves requiring an optimum
load of high impedance this arrangement is sometimes the best. At (b) is an arrange-
ment intended to equalize winding resistance and leakage inductance from each half-
primary to the whole secondary as well as primary self-capacitance. This arrange-
ment is particularly suited to circuits employing low loading Class AB or Class B
operation. The alternative arrangement shown at (c) results in a slightly lower re-
ferred capacitance across one half only of the primary. In general this unbalance is
not desirable, but if leakage inductance is adequately low, the coupling between all
the windings may be so good that the reduction in capacitance may be apparent across
the whole primary.
Anode
Anode Mode
Anode
Anode l
TABLE 5.
{
uFKrtE WINDING
CAPACITANCE
WINDING FACTOR
FACTOR
{SCREEN I
SCREEN
One Centre
One Centre ARRANGEMENT
ARRANGEMENT
side skit
CONNECTION saty { CONNECTION
11-1 11-^
T
1-1
li
.055 .611
I- .22
1 I
2.11
-1 Ti1I
-1-1-r,11
-
1111
- 1 I
15
'
2
51i,
.5
I
- ^1111-1
1-11-11-1 I-
J J jl 3
I-1.9. ..
MS
1.5
45
.
DIELECTRIC THICKNESS Ta MILS NUMBER OF LAYERS REF. LINE
.1..MI.M,L"I/0I.=I..8111
4 20 30 20
..1111.1.111..11
2 3 6 8 10
..SSIM..NI.1IS:%/I./:% IIII.
10 0 6 4 2
iiiiiiiiiiiil.'i:iiiïiiiiiiésiiiiiii5:iï i:iii:i
...11111..1 .1G4i:/1/%IMIN.IM1..M11M..1IM..SSSIIIiiPI/%
MIM1= r'/
..11181.%'%4
..81818818!/%%
I'I r ..i
, IIrI.
'Area MinaWAS=
.1%III181%I.81.81ENO
I ,
1111/,!I.'I./.%'/.1411111f1111XICOI.%IIi'/.%'/.'i/i%2I
INMI,I..I..'.I.I..M,I.IIat
r.'III,.
ÌS1
/i/
IMM
I.%%:I%'/.'/.81.818181..8181881./'I%II/i%%i/IIr/5//LI/./
1111IM/'%'/,/.I./.ÍHWAGA.MEO111M111O'/,'/.PI I/%/.I MI./,ÍíM%I.I O 06k.
I . -01
001S
11!%'//GG'/./,IÌ'/.'P.11111%%ÌG/
.O%S%OIO1%IIL155//r// %.1o11...o%' y%
..zi/%ivrinoizizEriiim.m..ri/%C!.%n.i/.«iuin
%ilí/i//%/%I c 04y
I%S' Irr,'
/G6S/L"/./'/./%'/,IE//,/,1111U6G"/Ii'I,I,/O/hÍML4SC//!'LG1
.`
'//.'/.%Í:ÍG/,%%I.I/.%:%I/.VA1111 %%IL/ /i§11 1EZ%%1%%%i1
. IIR/i.195LUÌi%5%.1.IPoli%9V/.%:%.4'.
i"Sili:iiiC..ii'.3S%i:Ciiiïtii::iiir 15%I/.'I.1
's.Iiiiiïïï
.. %iiii . .
LUII%Ì' 5%%%'
.
ri.
I/t'/.ISII%'.%I S%%LI/"/.%:11%iISLIi%%S/ÌÌSI%pS!%S%!/'I./UStIIS1
"i.MI.i... iï ïiï
%t%'/.'/.%.ME%'/.%%iG%I/OÌ51L!'/."/.Í iD."/././íIIG%51 -00Iú
..........W'...r.
/ i
WAS
(N,I
iiiiiiái iiiï:í%1ïiiïiiiïïió".ïiiï ïii:::ii%ï
I1/ÍJ.1%I %%%I,1/%1L/1/S:"I.I.%/165151II1/%51/.i5% ISÍ/%I I41/%.IiS
S OO
1
/I/IRi61.,r/II8181.8111...R1.%I/Ii%5%I18151/I/%.Ì'I ii'.%%iRil/I/.1
.
80
TIRASi18III.181.111..111.ÍII.' I4%5/'III%./' -' /I%L/II/11 6o
T- 0.3
/.i%14I 1111 L\111LÍ Ii/íII//.Íí%Ii%i'./ Ii'/././i I//I.ÍSr11411
0.2 %Í/IS11111111//,Í5
Ir/oRS181..181.1111..111%%%L15UI915%%'I.% MOGSM%i///%i 5!1066EiO11
%% ,.15%%01' m..1 40
/r./11..1...111..111; ;1%4;oD %iï,:%;,145;01%m..11
/.11111111111//II/I.iI//4/Ii/I/ II//iI/d1111 30
I
11111111111/L1,,Í'/.%%1%%%SOi111111
2 4
3 6 8 IO 2004 .06.08.1
LENGTH OFMEAN TURN Lmt INCHES
DIELECTRIC CONSTANT K .
..111M.1M111111..11 %/1/111..MU%!!1/1%E 1/511..111%..1
2
4
.3 3 4
SECTION
IM
MINN MIE
TURNS
6 .8
FACTOR
M1 E
FACTOR
MIME
2
1
=1M
I
3
11
FIG.
4
S.Ì7
20
Fig. 5.13R and 5.13S. For distributed capacitance due to layer winding (Ref. C29).
Refer length of winding L,,, on their respective scales, to intercept at A then along the
horizontal reference lines to intercept with the thickness of dielectric material T4,
at B. From this point refer along the slanting reference lines to intercept with the
empirically determined value of dielectric constant k, at C, then along the horizontal
reference lines to the unity reference vertical at D. From this point refer down the
slanting reference lines to intercept with the number of layers n, at E, whence the
referred capacitance is read off on the scale at the right.
For interwinding, or winding to screen capacitance As above, for the winding :
/'
!!1!P!
.Ii
- iiiiill11u111111rf
V11Á'."11
%% wmu.nlEmomE.a65!Avo
i151Eí1111111f./,i6S:imoIt
slt
41:Mg4COVZ
%i%Í1111I %.1111111111E0.811111%1x%
1'SIiÌI:i/ %iÌ
z other winding is all at low poten-
tial refer down the slanting re-
ference lines to intercept with the
section capacitance factor ob-
% cI a. to: tained from Table 4. For inter-
_ ---- -- -
F_.. TUC; ;WA
z.sweesiaMI1r..Stíl/IAIIE. AioIE 'A' Vim/ winding capacitance where there
ri?z1.i". 1aa :16115UVi i:1 TAM is appreciable potential in the
II %Ì % Ir" jj p ':s.i
11; veo Scream or
adjacent portions of both wind-
MM^'"' ings, individual attention will
:::6 `s#610'4 s
/¡
f,4OWSNISIBM1116.46:1
%./4X11111111 /
i
/L11T¡
1 Ilr RS8 ipV
Fi OA
Lora
Itr=tion
be necessary or each interwinding
space, and the turns factor scale
will assist here.
Ln,t y T4 Section
factor FIG. 5.13S
5.3 (v) WINDING CAPACITANCE 225
random winding the layers should be built up so that at all times during winding the
top surface is level. Failure to do this will not greatly affect the self -capacitance,
but will result in increased danger of breakdown due to electrical or mechanical
stresses. As the number of layers is always large, it is convenient to reduce the cal-
culation to simple terms of the winding dimensions.
The essential variables are length of winding L,,, length of mean turn L.1,
:
winding thickness T. (see Fig. 5.13Q), and number of turns, T. Considering variation
of each of these quantities in turn, the others being taken as constant : variation
of L. will vary the referred capacitance per layer as before, and additionally the
effective number of layers will vary inversely as L.} ; variation of L,, simply varies the
.1..111./.1
2 3
................GI
.............R.I.I.í.
4 6 8 10 4 2010'
..r.1...161..11....r1....s1r1// 800
I.'L.4r/ri/Y1..M..ILi..L/.%.
.i/Í6i...I:.S..Ì..1..:
42 3 6 8 10
Ì......
2 3 6 10
%
....
..ou...n..l.r/ioÍ6rPi.e..ISi
..1M1.111I././iI
.....:I.. ilái.
iiiirr.u.u/.ii.
i.11 .1
,.
Ri
sn:.s.r...I.ded1s.!./
r:'/.Y. Iiiar%.!i.i r:11.a rL./_SI.I rr í.:i m/ 5. / 600
i:0i" wm..t-.o.:..o:ic::r-:..2.:.$
Ì..' n4......l..I..._i_...I".1I5!/
%I./..i.Mí...I.1L.....i.Iíl I..iVí!YíR./ 400
.....o..G/./ArirIBÌ.Iro dtw1..R:.111r..8.míGa6/Q:Ca.rl./
l.ii1/1gIí1:rÍfRÍri/ifMiI81S6oisuli/s.rLi/.í/IlTrílliz/:6/
on1GrPR'//It// //LIsRnef!;QíEilir:ír:eSeí:a!L7,%L!S.li!51/ 300
11II i..
2i1;Gíiiiíi/,/E1I/'.3ell:iJSri!.si:í:rita:aliie:i::íríll
.-./. _. .....:..../
......%r.../-.1i.1.iÍ.:i-.IS1:e./
...%r./111.1././Ir.
...r/d.n...15n
.v1.li.rirn!!:
ul.iil/m .ius...ss.ríl...
ai/s.!:6:..e:.m.r.6G.y%.í1ssirsi/
11z1/ .I:iinilwiSíG.!:6:trilzr..if!.!.r:.'a.rslr
/iwi/1%/45ULieR1r%%afigIiI:Iíi%0:íSÍ:/Si/iMI:1r:
III61.'I._i.í ./.Yil.i/:.i.I...i../il......I./
"/_LiIS%1\/il/ .
v.iz..n....v.nn...asi.iiise...I..a..r
10ïi.I./IIIR.I.....I..
WIMP ...rr/.w1GZm7r111Í.s.RÍa irl....l:G/5:/
IoO
e
o7ar7.%7rOr..
i u =ö/:S°
n5u : s-lri.
ISÏ..n.ÌIr I.ri!..ÌÌ / 1.%./.1t.i:RiRYRIrni
_402...1n.í / so u
s:?25'::6:::':e°ÿ:!"':'::': / 60
9
Y. 6 oi...o....%.i...6..t..I.1a.ws.S.Yía iK
/LroI...9/..68.//Ir Gr4/ sm.S%Cir9.1./%:ISE.ím/:./
.undlL.1plri/%%1.6...i/i/C1..rL4IírIn/%:/:í
Ì..e15!./ 40
I:r2l.ru%51:1it/%:m/a/:%era i/ioSe%í!%P"I.E/5!/ n/im
4 iG/E1//%1G11/7t%%/ / é:$í/2í:íit$t:%Ìisii:il1í1// 20
. . .....,... .iid:$rdr_.:
S 011..i/ i.l"nLYÍ/i_i/.C..S...
%iRSiQiiÌ .LF%.I/o..6.
$i;iaii%a'fo/sfri%raL/rYr
L.S G.rí!z5r:6
%C/ . s5.í1vi!.zz:rzSr.1/.ri2./.:í:!.M/:51
9q.. . /.I.I.I'I._ÍI
/%5D/ %/Ii,l!/6R/%//41N/í5es'%i2 YíI311......I.._ir .i..Ì.I.
/
10
AEI AP,
2
IrL..r.ln 'I.II.I..%/III.'I.._ili.1//..5r:..ÍI..I...riMOri....1
/.r.r/IilydfP...%trri.!%Sormr3/.r:/.:.rs..I/mIireal.z/Í!./
MAU
0 a::ni_//RO1li/Ui %us71r;ii:rÌea!:.E..ri0./iirr:.:iy/:1 6
iizir;i i=- I.'ll! .ÌII.
I./..I '
srlG%i?Se_:..-::i:J-
iI.ÍIíI .Lr/:...i.i..iralir.Ip¡s%
i_
.......R...t,i1......a__.....z.....:........._.r.Íi..Yí
.
/
//'/r 6v/.r...r,nl.m..inr...m.sv
.
./IrrISr
i51I//.rar
Ì%/g1//7.6Rr/%il 3r1I/!i... LS/iri I/ v /
/%:14.r:i/ t%r56.!9 r
oveal¡.%:1r
. /..4/i/./
%/ÁífLLl;/f1:1/%Ir1I.IfIfC V-:/IGiEE'rI/'i r- %/./.
i Rí:i1n22í!lssí;í; %í%:;í íg íí!,%I íll
.1 % i1
3
04/I/.%G6ií61GLG511E 1.....Í...G...íI."
a11.1....I.R %CíiIIL%GI
04
/.r.ii:r-11i ïíMciil: i i iï:ii i:iii?4' I..!.!:Ií1!I.I./.II..!..1i..!./
%i!:i%i: %iiia i=Siiii%ai i i
/r:I/Ifl.fif .ilifflR%Lf.l/r!5!%Idií:í:/_a!%er!í/./!i!ii5/z1r s
7/iÍ./fl'fiIIII./.IRIR/.1r4111ff1 S!y:!5i%:eí:i!7:MPAr:10Rí.e:í!2r.6r
w*4 mum rA ..1M...111..81....1M...111.11
04 0.3 02 01 02 03 -04 06 0801 02 03 0.4 04 04 I
WINDING THICKNESS T, INCHES
FIG. 5.13T
Lmt Turns
.m.c.:wiIIÌ
.v.zi ::.?1nm..
41:viviir..uil. ,,.
,--11:4D:'
-,.cz.5ri:.
Fig. 5.13T and 5.13U (Ref. C29).
II
s.
2'/.GIí1!'IR I..a11111i2,i6a1 For distributed capacitance due to
%iíl! i!i%I
/I/!1P 211-::: s!ZíilíÍl
/iÓ AM Iroldl
random winding. Refer length of wind-
I/ iÌ %%i\Si.b Í .i....
4 %/GI! ill/41A is
% ü CO ing L,,, and length of mean turn
R. D. H.-8
226 (v) WINDING CAPACITANCE 5.3
(b) 0.5 ; (c) 0.375. Thus the total capacitance referred to the whole primary for each
method of connexion is,
(a) 300 µµF -1- 240 µµF = 540 µµF.
(b) 600 µµF + 240 µµF = 840 µµF.
(c) 450 µµF + 240 µµF = 690 µµF.
Example 2
An intervalve transformer, to operate from push -pull to push -pull, uses a simple
arrangement having both windings all in one section : L
= 0.6 in, L,,,, = 2.5 in,
T (each winding) = 0.1 in. ; insulation between windings, T d, k = 3 ; Turns,
4 000 c.t. /12 000 c.t.
Distributed capacitance, using Figs. 5.13T and 13U L
= 0.6 in., L,,,, = 2.5 in.,
T = 0.1 in., T (primary) = 4 000, is C, = 58 µµF. For secondary, T = 12 000,
C, = 34 µµF.
Capacitance coùpling between one -half primary and one -half secondary, using
Figs. 5.13R and 13S, actual capacitance, L= 0.6 in., L,,,, = 2.5 in., Td = 10,
k = 3, is C, = 100 µµF. Both windings wound in same direction, turns factor
across this capacitance referred to whole primary is 5 + 14 = 2, so referred capacit-
ance is 400 µµF. Windings wound opposite directions, turns factor referred to
primary is 14 - 5 = 1, so referred capacitance is reduced to 100 µµF. Referring
-
1/6 =
these two values to secondary, turns factors are 1/2 + 1/6 = 2/3 or 1/2
1/3, giving capacitance values referred to whole secondary of 45 µµF or 11 µµF
respectively.
A complete analysis would need to consider leakage inductance between each prim-
ary and each half secondary, and separate source and load impedances applied to each
half. For this purpose, the primary and secondary shunt capacitances across each
half would be 116µµF and 68 µµF respectively, while the capacitance coupling would
be 1,600 µµF or 400 µµF referred to half primary. The secondary shunt capacit-
ances referred to the half primaries would be 610 µµF.
Example 3
A direct coupled inter -valve transformer is arranged as at Fig. 5.13M : L,, = 1.5
in., L,,,, = 6 in., T d = 20, k = 1 (air spaced) ; T. (each whole winding) = 0.2 in. ;
Turns 4 000/12 000 c.t.
Distributed capacitance, using Figures 5.13T and 5.13U, L= 1.5 in., L,,, = 6 in.,
T = 0.2 in., T (primary) = 4 000, gives C, = 400 µµF. Secondary, T = 12 000
gives C, = 225 µí1F (or 450 µµF per half secondary).
Interwinding capacitance, using Figs. 5.13R and 5.13S, actual capacitance, L _
1.5 in., L,,, = 6 in., Td = 20, k = 1, give RC, just over 90 µµF say 100 µµF.
'Vertical sectionalizing will reduce the distributed component in each case, but
will also vary the interwinding capacitance. Using the information in Table 4 the
results may be presented as in Table 6.
In practice three sections for the primary and four or five sections for each half-
secondary will be best, remembering capacitance reduction is more important in the
5.3 (v) WINDING CAPACITANCE 227
secondary. By making the earthy end of primary and secondary at opposite ends of
the vertical groups, interwinding capacitance coupling effects are minimized.
TABLE 6
1
2
400
100
100
75
500
175
450
112.5
-
12.5
450
125
3 44.4 70.4 115 50 18.5 68.5
4 25 69 94 28 22 50
5
6 -
16
- 68
-
84 18
12.5
24
25.5
42
38
The loss of a speaker matching transformer is the inverse ratio of the power
delivered by the secondary of the transformer to a pure resistance equivalent to the
rated load impedance, to the power delivered by the same source if the transformer is
replaced by an ideal transformer of the same impedance ratio, expressed in db. For
measurement, the transformer shall be connected as for distortion measurement (see
above).
The impedance presented by the ideal transformer to the source shall be taken as
RT1 = VTR' /WTR
The power delivered to the secondary load is (VTS' /RTL) where VTs is the voltage
across the load resistance RTL.
The power delivered to the ideal transformer is (VTl' /R TO where VT! is the voltage
across the load resistance R T1.
The loss is given then by
VTit /RT.
Loss = 10log10VTS'
/RTL
The loss shall be measured at 400 c/s and at a value of source voltage at which rated
power is delivered to the ideal transformer. The loss shall not exceed 2 db (equival-
ent to 63% minimum efficiency).
AIR GAP
FLUX=
IRON
FIG.5.I4
A typical magnetic circuit is shown in Fig. 5.14 where we have the greater part
of
the magnetic circuit through iron, and a short length (the air gap) through air. In
this case the total reluctance is the sum of the reluctances of the iron -circuit and the
air gap. The flux ¢ is caused by N turns of wire carrying a current I amperes and
producing a magnetomotive force given by
F = (4n/10) NI 1.257 NI (2)
where F = magnetomotive force in gilberts,
N = number of turns,
and I = current in amperes.
Thus 1 ampere turn 1.257 gilberts. (3)
Instead of considering the total flux, it is often more convenient to speak of the
per square
flux density, that is the number of lines (maxwells), per square inch orcentimetre).
centimetre (c.g.s. electromagnetic unit = 1 gauss = 1 maxwell per square
The symbol for flux density is B. Thus (4)
¢ = BA
where = total flux, per
B = flux density, either in lines per square inch or in gauss (maxwells
square centimetre),
the cross
and A = cross sectional area of magnetic path (practically equal to or square
sectional area of the core in the iron section), in square inches
centimetres respectively.
or the
The magnetizing force (also known as the magnetic potential gradient of
magnetic field intensity) is defined as the magnetomotive force per unit length
path :
230 (i) FUNDAMENTAL MAGNETÌC RELATIONSHIPS 5 4
H = F/1 (5)
where H = magnetizing force in oersteds (or gilberts /centimetre),
F = magnetomotive force in gilberts,
and 1 = length of path in centimetres.
Alternatively, if F is expressed in ampere -turns, H may be expressed
in ampere -
turns per inch or per centimetre.
The permeability (fi) is defined by the relationship ;
µ = B/H (6)
where B = flux density in gauss (maxwells per square centimetre),
µ = permeability*,
and H = magnetizing force in oersteds.
In air, H is numerically equal to the flux density (B).
Permeability is the equivalent of conductivity in
electrical circuits. Permeability in iron cores is not
constant, but varies when the flux is varied. The --'-,--
u,_ -M
Point D is that at which the applied negative magnetizing force brings the value of the
residual magnetism to zero.
The length OC is called the coercive force (strictly this applies only for symmetrical
cyclically magnetized conditions).
The average permeability is the slope of the straight line EOA (shown dashed
in Fig. 5.16).
The locus of the extremities (A or E) of the normal hysteresis loops of a material
is called its normal magnetization curve ; this is the same as the B-H curve of
Fig. 5.15.
When alternating current is passed through the winding, the iron will pass through
the hysteresis curve ACDEFGA (Fig. 5.16) each cycle. The maximum value of H
is called H,,,,, x, and the corresponding value of B is calledB,,
the curve being norm-
ally symmetrical in the positive and negative directions in the absence of a direct
current component.
When dealing with alternating currents, it is usual to quote ampere -turns per inch
in r.m.s. values ; the corresponding values of flux density may be quoted either in
terms of B or Borax. It is obvious that H,ax = \H, but B, x is normally greater
than \/2B.
1. That the flux confines itself entirely to the iron over the whole length of the iron
path (in practice there is always some leakage flux, which is more serious when there
is an air gap).
2. That the flux is uniformly distributed over the cross -sectional area of the iron.
In Fig. 5.14 we therefore have :
Total magnetomotive force F 1.26 NI from (2)
Total reluctance R = Rr. + R n
where Rr, = reluctance of iron path
and Rr, = reluctance of air path.
Total flux 4' = F/R = F /(Rj, + Roo) from (1).
The reluctance of the air gap is given by
R r, = F/0 = Hl /BA = kl /A (7)
where 1 = length of air gap,
A = equivalent area of air gap, allowing for fringing
(a +1)(b +1)
and a, b = actual dimensions of pole faces.
Values of k are given by the table below :
The reluctance of the iron path is not constant, so that the best approach is graphical
The magnetic potential difference (U) is the equivalent of potential difference
in electrical circuits. The sum of the potential differences around any magnetic
circuit is equal to the applied magnetomotive force.
Applying this to Fig. 5.14,
F = Ur, + Ur, (8)
232 (ii) THE MAGNETIC CIRCUIT 5.4
1.
1,9
DECREASE
AA R
(a) By using a turn of shim copper or brass of full winding width between primary
and secondary, taking care to insulate the ends to avoid a shorted turn. This shield
is then earthed.
(b) By winding the earthed low voltage filament between primary and secondary.
(c) By winding the high voltage secondary in two separate halves, one on top of the
other, over the primary. The innermost and outermost leads are joined and become
the centre tap which is then earthed. The two leads from the middle of the winding
then become the high tension outers, and are connected to the rectifier plates. In
this way the capacitance from primary to secondary is made very small. It is im-
portant in this method of construction to ensure that adequate insulation is used be-
tween the two plate leads within the winding, as the whole of the potential difference
of the high tension winding appears between them.
With the normal method of construction the leakage inductance between either
half of the secondary and the primary is unequal. In large transformers this becomes
M.K.S. unit : See Chapter 38 Sect. 1.
234 (i) POWER TRANSFORMERS-GENERAL 5.5
important and, to maintain balance, the primary is wound between the halves of the
secondary or the latter are wound side by side over the primary.
TRANSFORMER STEELS
Silicon
Content* M. and E.A. Baldwin Sankey Allegheny Armco
4% Silcor 1 Quality 5 Super Stalloy Transf. C Trancor 2
31% Silcor 2 Quality 4 Stalloy Transf. D Trancor 1
21% Silcor 3 Quality 3B 42 Quality Electrical Spec. Elec.
1% Silcor 4 Quality 1 Lohys Armature Armature
*This applies exactly to Silcor (Magnetic and Elec. Alloys Ltd.).
The core losses for the various. Silcor grades measured at 50 c /s. with 0.014 inch
sheet, at two different values of flux density are shown below.
Watts lost per pound
B,,,,,, Silcor 1 Silcor 2 Silcor 3 Silcor 4
10 Kilogauss 0.59 0.63 0.89 1.32
13 Kilogauss 1.04 1.07 1.51 2.24
It will be observed that for small changes in flux density the core loss varies as the
square of the flux density. For radio power transformer work, core materials similar
to Silcor 2 are commonly used. Measurements of losses can be made with a low
power factor wattmeter, or by the three ammeter method. A system suitable for mass
production testing has been described recently (Ref. D9).
Cold rolled, grain oriented 2.7 per cent silicon steels are becoming of increasing
importance. Typical trade names for this material are " Hipersil " and " Crystalloy "
(Ref. 10). Flux densities in excess of 110 000 lines per square inch (17 kilogauss),
can be employed without high core loss. Owing to low core losses at such high flux
densities, the application of this material to small transformers for electronic equip-
ment is increasing. It can be used in strip form, being wound around the winding
in one method of assembly. In another method, the strip core is sawn in half and the
two halves clamped together around the winding. Either method results in a con-
siderable saving in material and labour over the present method of punching lamina-
tions and hand stacking the winding with them.
The most popular laminations in current use are the E and I " scrapless " variety
(Refs. D11, 12, 13, 14). These are so dimensioned that the I is punched from the
window of the E thus avoiding wastage. The usual ratio of dimensions of these
laminations are as follows : Window height 1, tongue 2, window width 3, I- length 6,
magnetic path 12.
Similar laminations are stamped by several different firms :
Standard Lamination Sizes
Pattern Number
Tongue
9/16 in.
¢
M.E.A.
18
145
-
Baldwin
392
--
Sankey Allegheny
EI -56
EI -625
-
Chicago
F000
1
-1
35
147 -
217
-
70 EI -75
EI -11
D000
B000
1
11 -
29
-
430 111
158
EI -12
EI-112
1000
13000
11
11
11
78
152
120
420
-
362 -
133
149
EI-125
EI -138
EI-13
-
14000
5.5 (ü) CORE MATERIAL AND SIZE 235
Laminations for small power transformers are generally 0.014 inch thick ; thicker
laminations up to 0.025 inch are occasionally used, but result in increased losses and
shortened lamination die life.
To determine the core size, it is first necessary to estimate the power requirements.
Then we may apply the empirical relation
=VV.A (1)
5.58
where A = cross -sectional area in square inches
and VA = voltamps output.
For ease of production an approximately square stack is desirable. Thus having
determined A, the tongue size can be estimated from VA. Where high voltage
windings, or windings operating above ground, or at a high potential between other
windings, are used, it becomes necessary to employ a lamination which has a different
ratio of dimensions from those quoted earlier, the window height being increased
to allow room for extra insulation. The input current taken by a transformer on
no -load is commonly called the magnetizing current. In fact, it consists of two
components in phase quadrature. The in -phase component, usually small, is the
iron loss plus a very small copper loss. The quadrature portion is the true magnetizing
current. Their vector sum does not normally exceed about a third of the full load
current.
For 240 volts 50 c/s 230 volts 50 c/s 117 volts 60 c/s
1860 1780 755
N= A N A N A
where A = gross cross -sectional area of core in sq. ins.
The flux density employed depends on the application, the power rating, the core
material and the frequency. For oscilloscopes and pre -amplifiers, densities of
40 000 to 50 000 lines per square inch (about 7 000 gauss) are used. For small
transformers below about 50 watts, densities up to 90 000 lines per square inch (14 000
gauss) are used, gradually decreasing to about 65 000 lines (10 000 gauss) as the
transformer size increases to several hundred watts.
Having chosen a suitable flux density, the turns required for each secondary winding
may be calculated.
It can be shown that, approximately,
E, . N,
N, =
E,./ where n = transformer efficiency.
As a first approximation it may be assumed that i = 0.85 and 'VT = 0.92.
The values thus obtained for the secondary turns may be checked by more detailed
calculations once wire gauges have been chosen and winding resistances calculated.
For this purpose the equivalent circuit of Figure 5.11 may be used.
If a secondary is to feed a rectifier and the d.c. output of the valve is specified,
reference should be made to Chapter 30 or a valve data book to determine the required
236 (iii) PRIMARY AND SECONDARY TURNS 5.5
secondary voltage. Allowance must be made for any voltage drops due to the d.c.
resistance of the rectifier filter.
secondary may be taken approximately as 1.1 times the direct current to the load.
For further details see Chapter 30 Sect. 2.
Choke input filter : The r.m.s. current in each half of the transformer secondary
may be taken approximately as 0.75 times the direct current to the load. For further
information, see Chapter 30, Sect. 4 (also Sect. 3). With half-wave rectification
there will be a d.c. component of the current which will affect the transformer design
if the total d.c. ampere turns are considerable. For half-wave battery chargers using
bulb rectifiers, this must be taken into consideration when selecting core size and wire
gauges (Ref. D15).
In the case of transformers supplying full-wave rectifiers, the full load primary
current can be estimated by calculating the total secondary loading in voltamps,
allowing an efficiency of 85% as a first approximation. This is then the primary
input in voltamps which, when divided by the primary voltage, will give the desired
current. Where secondaries feed a resistive load, the loading is the product of the
voltage and current. Where a secondary feeds a full-wave rectifier the load is the
product of the direct current and the direct voltage output from the rectifier plus the
power lost in the rectifier. This latter can be calculated, for a condenser input filter,
from data presented in Chapter 30, Sect. 2. It should be noted that indirectly heated,
close- spaced rectifiers such as the 6X4, are more efficient than types such as the 5Y3-
GT, with its heavier filament power and lower plate efficiency. This is one reason
why the former are almost exclusively used in the majority of small a.c. radio receivers.
The primary input current as calculated above should be accurate to within 10%
for small transformers.
Wire gauges and copper losses
For the usual type of radio receiver power transformer it will be safe to choose wire
gauges on the basis of 450 -800 circular mils* per ampere (i.e. 2830 -1590 amperes per
square inch), but values up to 1000 circular mils per ampere (i.e. 1270 amperes per
square inch) may be desirable in larger units, or if a high flux density is used with high
loss lamination steel. The latter figure is easy to remember as 1 circular mil per
milliamp.
In practice, wire gauges may be chosen arbitrarily on a basis of (say) 700 circular
mils per ampere (i.e. approximately 1800 amperes per square inch) and check cal-
culations should then be made to see that
(a) the build of the winding (i.e. winding height) is satisfactory for the window
space available,
(b) the copper loss is not so high as to cause excessive temperature rise,
(c) the voltage regulation is satisfactory.
Wire tables in this Handbook will simplify these calculations (Chapter 38 Sect. 19).
silk, varnish and wire enamel (Refs. D16, 17). The temperature rise in the winding,
as measured by the change of resistance method, will be about 10 °C lower than the
maximum (hot spot) temperature. Thus with an ambient temperature of 40 °C
(104 °F) plus a margin of 10 °C for the difference between measured and hot spot
temperature, it will be seen that the maximum permissible rise is 55 °C, as measured
by the change of resistance.
It is common practice to allow 10 °C margin for change in line voltage, frequency,
or operation in situations with restricted ventilation. Thus 45 °C is generally accepted
as the maximum permitted rise above ambient when measured by resistance change.
The temperature difference between winding and core varies between 10° and 20 °C
according to the distribution of losses. This means that even with an ambient tem-
perature of 25 °C (77 °F), the core temperature may be 60° (140 °F). This will feel
quite hot to the touch, although the internal temperature may be will under the per-
mitted maximum. Measurement of the core temperature may be made with a spirit
thermometer if good thermal contact is maintained between the core and the thermo-
meter bulb.
The winding temperature rise can be calculated by measuring the cold resistance,
R at an observed temperature T,. After a heat run at full load the hot resistance R
is measured and the ambient temperature T2 is again measured.
Taking the temperature coefficient of resistivity (a) of copper as 0.003 93, the tem-
perature rise T is found from the formula
R- Ro
T Roa
To correct for any changes in ambient temperature during the heat run it is neces-
sary to subtract the difference T, -
T, from T to find the actual rise. An example
will illustrate this.
The primary resistance of a transformer measured when cold was 30 ohms at an
ambient temperature of 20 °C. After an 8 hour full load run, the resistance was 36
ohms, while the ambient temperature was then 18 °C. It is required to find the
winding temperature rise.
36 -30 6
T
30 x 0.00393 30 x 0.00393 -51 C.
N1 = E1ND /En*/.q
= 300 x 1130/240V0.85
= + 1530.
1530
Turns per volt = N1 /E1 = 1530/300 = 5.1.
Secondary (i) turns = (6.3 x 5.1) = 32.
Wire gauges :
Assume a current density of 600 -700 circular mils per ampere. Referring to
wire tables, a suitable primary gauge is 29 A.W.G. enam.
Turns per layer = 95. Refer Figs. 5.18C and wire tables in Chapter 38 Sect. 19.
No. of layers = 1130/95 = 12.
R.M.S. current in secondary (ii) is (0.06 x 1.1) = 0.066 A.
mmom
I
100
90
80
11111= mw1IIII
1111111111= L.
11II11rIIIIIII111
1111 111111111111
IP4
111 IIIII11111 4
loo
90
60
70
-_-- ...
-- ....
IIIIIIIIIIIiIIlllllIlIIIIIIII!;IIIII
,
--- - ...........------
=_-- .......------
.MMMMMM.ii
4-....
0--:
====
80
70
60
- --iii-.111IIIIIm---w
.1.
WM11II00 MIPA MEN I1IIIIIIII N
---IIN111w---IA
rrÌr N
IIIIIMMI-111
iIMII1INI/ IIIIMMIIM-NI
IIMIII11IM
MIi°-
IIIIII%IIIIIIII111III I=
40
IIIII/.AI
BIIIMIIIIIM111RLMIIIIIIII111
111Vd1= II111111111M
NNIIIIIIIIIIIM IIII1111111111 11 40
30
-- -
¡° ----
/IIIIIII1 MIN-MI MIMI
-IIIMMIII1I1 1Iïiiiiiiiliiiii
1111111ii1111111111N11111I1I1I111111
M -.iii
......
IIII1=--1
1MON NM
IIIM1III MIME
MIIIM
30
20
II=1111II11 I1I1II1111
ANINIIIIIIMMIIMIIIIII11111E 20
10
01 02 03 04 05 06 07 08 09 1010
W- Watts per Sq.inch Surface
FG. 5.1ßR
Fig. 5.18B. Temperature rise versus loss per square inch of cooling surface
for
ambient temperature 25 °C (Ref. D.1.)
5.5 (vi) TYPICAL DESIGN 239
Primary build = No. of layers x (enam wire diam. --F interlayer insulation)
-'
= 12 x (12.2 + 2) x 10 inches
= 0.170 inch.
Secondary (ii) build = 19 x (6.9 + 1) x 10 inch-'
= 0.150 inch.
Secondary (i) build = 2 x (32.4 + 5) x 10 inch -'
= 0.075 inch.
Allowing 50 mil former thickness and 10 mil insulation between windings and over
outer winding, total build is (0.170 + 0.150 + 0.075 + 0.050 + 0.040) inch
= 0.485 inch.
Winding height = 0.562 inch.
Build expressed as a percentage of window height
= 0.485/0.562 x 100 = 86.5 %.
In this particular design, the heater winding is wound between primary and high
tension windings to serve as a static shield.
Mean length of turn calculations-see Fig. 5.18D.
A = Former build plus insulation = 0.050 + 0.010 = 0.060 inch
B = Primary build plus insulation = 0.170 + 0.010 = 0.180 inch
C = Secondary (i) build plus insulation = 0.075 + 0.010 = 0.085 inch
D = Secondary (ii) build plus insulation = 0.150 + 0.010 = 0.160 inch
4S = 2 x (stack + tongue) = 4.5 inches
Primary mean length of turn = 5.44 inches
Secondary (i) mean length of turn = 6.27 inches
Secondary (ii) mean length of turn = 7.04 inches.
Winding resistance :
If each value for the mean length of turn is multiplied by the number of turns
in its own winding and then divided by twelve, the resulting quantity will be the
number of feet of wire in each winding. By referring to the wire tables, the re-
sistance in ohms per thousand feet for any particular gauge can be found. Dividing
the wire length by a thousand and multiplying by the resistance per thousand feet,
will determine the winding resistance.
Primary resistance = 5.44 x 1130 x 81.8/12 x 1000 = 42 ohms
Secondary (i) resistance = 6.27 x 32 x 12/2 x 12 x 1000 = 0.09 ohms
Secondary (ii) resistance = 7.04 x 3060 x 261/12 x 1000 = 470 ohms
Copper loss :
Primary copper loss = (0.19)2 x 42 = 1.52 watts
Secondary (i) copper loss = (2.6)2 x 0.09 = 0.61 watts
Secondary (ii) copper loss = (0.066)2 x 470 = 2.05 watts
Total copper loss = 4.18 watts
Iron loss :
Efficiency :
Efficiency = Power output /(power output plus losses)
= 39/(39 + 6.5) x 100 = 86%.
Regulation, calculated from British definition :
FIG. 5.18C
LAMINATION AND COIL DATA
P
D--r
EXAMPLE:
24E WIRE ON SQUARE 1.25 CORE
LAMINATION
SIZE 1.25
OVERALL 3.75 5 3.12 [aIiOMw_i_i7
4 HOLE MTG.- -3.12 X 2.5
CORE
WEIGHT 2.95 LB5.
AREA 1.56 SW
MAGNETIC PATH- -7.5
COIL -.24E WIRE
WIRE - -- 0.576AMPS
MAX
WIRE-
WIRE
--
61 TURNS/LAYER
--
-.0213 DIAMETER AP
INSULATION-.00_4 - KRAFT
TUBE
BUILD
--
---.050
-.538 MA
PAPER
X.
LAYER INSULATION
Lamirntioa Width
si:e center Leg, in. A 0.500 0.625 0.750 0.875 1.000 1.125 1.250 1.375 1.500
Area, Sq. In. B x L) 0.250 0.292 0.422 0.573 0.750 0.950 1.17 1.42 1.69
W1 °dO
Dimensions
Length, in. D 0.812 0.937 1.125 1.31 1.50 1.69 1.87 2.06 2.25
Width, In. B 0.312 0.312 0.375 0.437 0.500 0.562 0.625 ' 0.687 0.750
Build, Max. In. S 0.256 0.256 0.315 0.367 0.430 0.483 0.538 0.590 0.646
Coil Length
Dimension, 'Dyer."' P 0.750 0.875 1.06 1.25 1.44 1.56 1.75 1.94 2.12
Length
Winding, In. R 0.500 0.625 0.750 1.00 1.12 1.25 1.37 1.56 1.75
Core Tube Thickness, In. T 0.035 0.035 0.050 0.050 0.050 0.050 0.050 0.050 0.060
"E" & "I" Piece Length, In. F 1.62 1.87 2.25 2.62 3.00 3.37 3.75 4.12 4.50
Overall
Dimension, Width, In. M 1.31 1.62 1.87 2.19 2.50 2.81 3.12 3.44 3.75
Length, In. H 2.50 2.81 3.12 3.44 3.75
4 Hole
Mourning Width, In. J 2.00 2.25 2.50 2.75 3.00
Dimensions Diameter
Hole, In. .. .... .... .... .... 0.219 0.219 0.219 0.219 0.281
2 Slot Width, In. G .... .... 1.62 1.95 2.16 2.44 2.78 2.94 3.12
Mounting
Dimension. From End, In. K .... .... 1.12 1.31 1.50 1.69 1.87 2.06 2.25
Width,
"E" Piece Overall, In. E 1.06 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00
Dimensions Width, Outer
Leg, In. C 0.250 0.312 0.375 0.437 0.500 0.562 0.625 0.687 0.750
"I" Piece Width, In. L 0.250 0.312 0.375 0.437 0.500 0.562 0.625 0.687 0.750
Volume,
Cu. Inches Area o 0.810 1.46 2.54 4.03 6.00 8.55 11.70
F
15.60 20.20
For Vol.
Square Stack Weight, a
Center Leg Pounds Density 021 0.37 0.64 1.05 1.50 2.14 2.95 3.90 5.06
Area,
Sq. Inches A , A 0.250 0.390 0.562 0.766 1.000 127 1.56 1.90 225
Magnetic
Path Length. In. I 3.25 3.75 A50 5.26 6.00 6.75 7.50 8.24 9.00
N<
MEAN TURN
KAN
N\`frTURN.
TURN. TUBE .
BULD
PRI
OF
(i) General
Iron cored inductors fall into several different categories depending upon the cir-
cuit requirements. In some applications these inductors may have to carry a.c.
only, in others, both a.c. and d.c. They may have to work over a wide range of
frequencies, or at any single frequency up to the ultrasonic range. Iron -cored
inductors are employed as smoothing and swinging chokes in power supplies, as
equalizer elements in audio frequency equipment, as modulation chokes and as filter
elements in carrier equipment, to mention a few varied applications.
following specifications -
To design such an inductor it is therefore necessary to know some or all of the
flux densities in the core, and on whether the core laminations are interleaved or
whether there is an air gap in the magnetic circuit.
The length of the magnetic circuit is the length measured around the core at the
centre of cross section of each magnetic path. Referring to Fig. 5.19, the path taken
will be along the centre of each leg except where there are two windows, when each
path through the centre leg will be along a line one -quarter of the way across the leg.
In this latter case only a single path round the window is considered in calculating
the magnetic circuit length.
Eqn. (1) is properly applicable only when the cross -sectional area of the core is
uniform throughout the magnetic circuit. Where the cross section is non -uniform,
a conservative value of L will usually be obtained by using for a the minimum value
of the cross -sectional area. For more accurate calculations in such cases reference
should be made to a suitable text -book (e.g. Ref. El).
(1) Choose an available core and calculate from eqn. (1) the number of turns re-
quired to give the necessary inductance assuming a value of effective per-
meability of between 1 000 and 5 000 depending on a very rough estimate of
the a.c. flux density in the core.
(2) Calculate the a.c. flux density for this number of turns in relation to the known
frequency and voltage across the coil.
(3) Correct the estimate of the number of turns using a revised value of perme-
ability for the calculated flux. If the first estimate was far out, repeat steps
(2) and (3).
244 (iv) DESIGN WITH NO D.C. FLUX 5.8
'I I
llu11111 11111n11n11I11111111111111111111111111111l
ÌIÌIÌIÌÌÌIÌtÌIIIIa11111111111111111111111111111111111.11.1
°iriimimüliIliiiuunun.lin.l
öiimii8liüñl'Il° llllllllllllu
.millmmuuuuno,::
.wmuuumu:.um.nüÑluuumnumuumw
III
..................
II:L'IIIiL'CIRIIIIIIIIIIIII:I8':
ÌÌÌÌÌÌIÌÌÌÌÌIIÌÌI:Ì ñiñiiiÌÌIÌi.IIÌÌÌÌII:Ì:::ÌÌ:lÓUl.luiiii.ii\1iÌÌÌIÌÌIIIÌIIIIIiulllnurlmumu1u111u1e liiiiñi
1111111111111111111 IIiI10I11I11L11111111111111111111;
1111111111111111111 IIIIII'Alllllll.
111II11111111111111.IIIIlldllllllll.1111111111111
IllllIIIIIIIIIIIIII:IIIIII:IIIIIIIIII 1111111111111111111.1
l
1....\U111111111 7111 11ulnnini1mnnmm1111mmuull.l 1
....'II1111111111IIIIIIIIIIII11I1.IIIIIIIIIIIIIIIIIIIIIII
I....'IIIIIIdII11111111111Illllll1111111111111111111111.
IL..IM'IIIIIIIIIIIIIIIIIIIIIIIIIII'llllllllllllllllllllll
IIIllllnll ...IM
lllllulllll....IM
IWIIIIIIIIl111111 1111111111111111111'111111111IL'111111111111I1I111111111111 III111111111111111111'1111111111/111
IIIIIId1111II11111
IIIIIIIIIIIIIIIIIII IIIIIIIIIIIIIIIIIIIId1I11111III11I111111I111111111111111111111111111111111IIIIIII 11111111111111111 1111111111111111110111 8O
l: ll - =_=
llll
.-..... 1i155tt - -___
va
lEE m
IM
1111111111IIII1111111111111111111111111111111111111111111:11111111 11111111'1111111111111G1111111111111111111111111111111111111111111.
uuu 1 '..I I
iiúiiiiñiiiiiiiiiiiiilll:i1iiiiiiiiii::::Ii
..nu uuu ,.:::
iiiiMáiiiiiiiiúiiiiiii:.iliiiüiiiiiiñiiiiiiiiiiiiiilmuon
... 1 .
..i
u..
iiiiiiiiiiii
IIII11111111
IIIIIIIIIIIII
IIIIIIIIIIII'
f
u.u.mununuumuiu.nluwllu.
.Mllllllllhlllllllllf. nullllllllllullllllllllllluullll
.MJIIIIIIIIIIIIIIIIIII'i i.InIII1IIIIlI1111111111111111u1u1
...
MINIM
M..
IIIIIIIIIIIIIIm MIM111111111111111111I1
1111111111111111111II111ÌÌÌIIIIIIIIIIIII4ÌÌ11111111111111111111MMMIIU1111111111111111111111i1dIIIIIIIIII:IIIIIg111111'
IIIIIIIIIIIIII1111111111111I
\M
1111111111111111111:1111111111111111111111111111111
ÌÌÌÌÌÌÌÌÌ\ ÌÌÌÌÌÌÌÌÌhÌÌÌÌÌÌÌÌ
111::.111111111111
I ÌÌÌÌÌÌÌÌÌÌIÌÌÌÌÌÌÌÌÌ O
8
(ri) A1I119V3 48]d
Fig. 5.20. Effective permeability (µ) of silicon steel versus a.c. maximum ux
density, for various values of d.c. magnetising force Ha (by courtesy of Alleghany
Ludlum Steel Corporation).
5.6 (iv) DESIGN WITH NO D.C. FLUX 245
:ÌÌÌÌ
if the air-gap is large, so that 1/ is <
a /1, eqn. (3) approximates to
L = 3.2N8 . a /1082 (4)
FIG. 5.21
IMu11I11IIIIMIIIIIIIIIIINIIIIIMMU11111111
uuu1111IllluilniIIIIIIIIIIIIIIIIIIMMIIIIIIIIII
III11111N1111MMIIIIIIIIIIIIII =MIIIIIIIIII
MIu11111N111INNIIINUUU11111111 1IMINIIIIIIIIIII
iiimuu1111 IIINUmiinuu1111111111111111=IMIII1111111
MNUIIIIIl1111mmUltllllllllin11I111110
uul
1111uuu1111111.II1111111lllllllll
111111u1111111111111lllllllll
=:11111ÌÌÌÌ ÌÌÌÌÌÌ Ì ÌÌ
=MEMO..
M11I11I1I11lllll!,í
I=M1NU1111111IUUUNUNNUIU
I
1/1118..-^e
11I11111111
7 111111
II11111111u\oe%11I111111IIN==RA!Sd
,/
lllIIIII M/5í/!S111111
]l
M
11I11 1
l11111i1I1Nlu1U..°ed /SíIIIIIIII
NN1111I111111uwhrr
M1II11NllluMlullll
111IIINI NllluM
u111111111
MU1111111UI
luau
MI11111IIlItI
=Nulllll H`o,
111111uu/MI_í1!:/I1111111
uu111111'111IN\e.onn,
11111I1I=NIIIN111111/1111
i
1í
ul111111111
l_//1u111111111
uu\e.
o
c+` ed
A dlglad
¿co
G
1101Mulh::
oe dIÍÌII=M 1u111111I
e
or
d11I1111INNNNN111111I11
dÍ11111111M==1IIIIIIII
oo°
./_ 11111111
HIM 1
11 e
i
I
M
S
Nu 0 r/IÍ111u11111111=M11111111
./ "1IIIIIIII..111111111
v `° 1
M!íIIIN tcooi.IIIIIIIIUN=MIIIIIII
003 o I
iNNNI
/%N/ .MM111I111111Mllllllllll
/ ÌÜÌÌÌÌÌÌitÌÌÌÌÌIiiÚÌÌÌtÌÌ
o
AIMMENNNUU111111111u=u=l11111111
-NN .,,c
..e1
.dllll
a°-d1I111111
Core wee of the Material .kP(üratts per pound
1111111111
111lI11111
u1111111111
Igo
Frequency - Cycles per Second
4000 7000
Fig. 5.21. Experimentally determined curve showing values of " a" as a function
of frequency for several magnetic materials (Ref. E2).
(y) Design of high Q inductors
This sub -section follows S. L. Javna (Ref. E2) and is set out in a step -by -step form
for ease of working. The following table defines the symbols used by Javna in the
formulae presented here.
A = Effective cross -sectional area of magnetic-flux path, in sq. in.
A, = Gross cross -sectional area of magnetic -flux path, in sq. in.
a = Empirical constant, see Fig. 5.21.
B = Maximum flux density in the core, in lines per sq. in.
F = Fraction of core-window area occupied by copper wire of coil
f= Frequency, in cycles per second.
g = Actual gap length, in inches.
k = Empirical constant, see Fig. 5.22.
L = Inductance of iron -cored coil, in henrys.
1 = Mean length of magnetic path, in inches.
m = Mean length of a turn of the coil, in inches.
246 (v) DESIGN OF HIGH Q INDUCTORS 5.6
= 1 6.0 ins
w = 1.5 lbs.
A, = 1.0 sq. in. from Fig. 5.18C
A = 0.9 sq. in.
s = 0.75 sq. in. J
m = 5.5 ins. from Fig. 5.18D
F ti 0.3, a typical value for this lamination
k = 1.3 x 10 -8 from Fig. 5.22.
a = 1.987 from Fig. 5.21.
µp is found from Fig. 5.20.
Calculation :
.....1
__..0
N....M
NE....
INI1111
DESIGN OF HIGH Q INDUCTORS
i..:
1111:44::::
1::: ...r
CI1si0iiiCT i==CcSiCia.CX
......
=/....1.11I11
MM11.M....
:.
.__.=_-;:::
=Csss ... Sd::Ii
un1 u111e11.1111.111=w.O.mu.
=IIMME...1. 11111111MMI.I
..uuouMMUM=INInnommuIMMIIIMIluPAuanuu
:
UI11111111MMMIMMIM1111111111111111111MW/ IdIMII11111111
/' ,Illl,vm.IIIIII
1111111111 IIIMINI1111I111111=1111111111W/11111IIIIIII
II1ME11111111111'/.11111111l;11
iliiiiliinuM
.. FIG. 5.22
CEsG37: :1»::t:
.ú
IIMMIIMIIMMINIW..
"
247
1
1111111111111111
I
s'
11a,II
...
'M...
nunn1 I..
.:
....-¡.::.
...... . ..1
.....
d1=11.EHM..u
:
M...
M1..
: COO:::C w
.11
,..
....111111
INt. d1/1d dl/ °MnE1111111111
AM df
P
q 1
111111I1111 o.-
111M1111111 Ill
4111111 dllMEM111UIIIIIII
N111111 c .g`4.I111 P
P
.
o`°.ÌI1 .`°.111ó
IMIIIIIIIIIIIII
:=M
:L:.Cs.
iúuuut ...iiiiiiiii:i:
_.. .. 11.¿ìLI
tI/MIMI
Kr.. 1...
MIMEO
Ammo. 416111.11
ANIIIPI-MIN
-MINIMUM
u.i.. ..... II..E.uou...
L..1 N=MN...0..111.
111M1 .....
51.F..1I1i 1111INIIMMI
1
...m.immlm..u11111111
Cor Lou of the Mo4erlal . Kla watts per pound
1111111.1....UO
MIIIIII1111111111
W1111111111
M1 __ 111111111111111111
44
10 20 d0 70 100
Frequency-
200 400
Cycler pe
700 1p00
Second
2pOO 4p00 7,000 ICPCO
Fig. 5.22. Experimentally determined curve showing values of " k " as a function
of frequency for several magnetic materials (Ref. E2).
Such data may be obtained using the methods given in ` Magnetic Circuits and
Transformers " (Ref. A3) Ch. 7, which sets out the general methods suitable for
calculation of inductance with d.c. flux in the core.
The method developed by Crowhurst (Ref. E18) makes a considerable saving in
time by the use of charts, particularly when a number of inductors are to be designed.
However, when the a.c. flux density is very small, as for example in many filter
chokes, the method developed by Hanna may be used.
M...O.MI..Ei...
.. ..m...E.
......
M
.
.. ./...
......!'
.... .....
....r ...
3
230
II
... 0.0
>i
W0.003 .....
.....Id ......
160
.... .......
............EM
....F.
......
.
........
100
..I,..........
.r!.........
( %10 ó , o
.
mgommlimmommomm
20 40 60
N
!0 100 120 140
T FIG. 5.23
Fig. 5.23. LIE /V plotted against NI/1 with various gap ratio intercepts
(Ref. E13).
(vii) Design by Hanna's Method
(Ref. E12)
Strictly, the method is applicable only when the core to be usediis of constant cross
section throughout the length of the magnetic path, but it gives a useful first estimate
in cases where the cross section is not uniform.
Let N = number of turns in coil
I = direct current through coil
L = Inductance in henrys (at low a.c. flux density)
= length of magnetic gap (inches)
a = length of air gap (inches)
a/1 = air-gap ratio
a = cross -sectional area of core (square inches)
and V = l.a = Volume of core (cubic inches).
5.6 (vii) DESIGN BY HANNA'S METHOD 249
per inch of core path length. The peak a.c. voltage, of double mains frequency,
across the choke, may be taken to be about two thirds of the d.c. voltage at the input
to the filter.
(3) Using Fig. 5.20 note the change in permeability µ when maximum d.c. load
current flows through the choke and check that the value of inductance under these
conditions is sufficient. If it is not sufficient an increase in turns may be satisfactory
and this can be checked by repeating the procedure above. If this is not satisfactory
a small air gap may be necessary, but calculations then become complex.
A choke so designed is commonly referred to as a swinging choke because its
inductance varies with the direct current through it.
(b) The second choke
The design of the second choke of a choke input filter is straightforward, and
Hanna's method may be used after determining the required value of inductance by
the methods set out in Chapter 31 Sect. 1.
(ix) Measurements
Measurements of inductances of iron-cored coils must be made under conditions
similar to those under which the coils are to be used, because the value of inductance
depends to a marked degree on the a.c. flux density in the core and also on any d.c.
flux set up in the core by direct currents in the windings.
Bridge methods are desirable for accurate measurements of inductance but must
be arranged to simulate the operating conditions of the coil being measured. Owen
and Hay bridges are widely used when there is a large flux due to direct current
(Refs. E15, E16).
For many purposes it is satisfactory to determine the effective inductance (or more
accurately, the impedance) of a coil by measuring the current through the coil when
the rated a.c. voltage is applied to it ; but this method is not usually feasible when a
d.c. component is also present.
VALVE
VOLT-
METER
When a d.c. flux must be produced in the core to simulate operating conditions,
the circuit arrangement of Fig. 5.24 may be used to determine the a.c. impedance
of the coil. A valve voltmeter is used to adjust the a.c. voltage drop across the known
variable resistance R to equality with the a.c. voltage drop across the inductance L.
The required value of direct current through the coil is obtained by adjusting the
tapping on battery B and the rheostat R1 (or by adjusting any other variable direct
current source), the current being measured by a d.c. ammeter or milliammeter.
The required alternating voltage across the choke coil is then adjusted by varying
the tapping on transformer T, the voltage across the coil being read on the valve volt-
meter ; this adjustment is not usually critical. The resistance R is then varied until
across R
the same reading is obtained on the valve voltmeter when it is connected
5.6 (ix) MEASUREMENTS 251
SECTION 7: REFERENCES
(A) GENERAL
Al. Still, A. (book) " Principles of Transformer Design " (John Wiley and Sons, New York, 1926).
A2. Still, A. (book) " Elements of Electrical Design " (2nd ed. McGraw -Hill Book Co. 1932).
A3. Massachusetts Institute of Technology (book) " Magnetic Circuits and Transformers " (John
Wiley and Sons, New York ; Chapman and Hall, London, 1943).
A4. Blume, L. F. and others (book) " Transformer Engineering " (John Wiley and Sons, New York,
1938).
A5. Terman, F. E. (book) " Radio Engineers' Handbook " (McGraw-Hill Book Co. 1943).
A6. Lee, R. (book) " Electronic Transformers and Circuits " (John Wiley and Sons 1947).
A7. Beatty, R. T. " Radio Data Charts " (Iliffe and Sons Ltd.).
A8. Roers, H. C. (book) " Electromagnetic Devices " (John Wiley and Sons 1941).
A9. Miner, D. F. (book) " Insulation of Electrical Apparatus " (McGraw -Hill Book Co. 1941).
A10. Connelly, F. C. (book) " Transformera " (Sir Isaac Pitman and Sons Ltd. London, 1950).
All. Crowhurst, N. H. "Winding space determination" Electronic Eng. 23.282 (Aug. 1951) 302.
Additional references will be found in the Supplement commencing on page 1475.
(B) REFERENCES TO PRACTICAL TRANSFORMERS
B I. Honnell, P. M. " Note on the measurement of transformer turns ratio " Proc. I.R.E. 33.11 (Nov.
1945) 808. See also Refs. Al to A10 inclusive.
(C) REFERENCES TO AUDIO FREQUENCY TRANSFORMERS
Cl. Harrison, E. B. " Notes on transformer design " Elect. 17.2 (Feb. 1944) 106.
C2. Koehler, G. " Audio -frequency amplifiers " included in Henney's " Radio Engineering Handbook "
Chapter 8, p. 355, 4th ed., 1950 (McGraw -Hill Book Company).
C3. Webb, E. K. " Response of a.f. transformers at high frequencies " Proc. I.R.E. Aust. 5.3 (Jan.
1944) 3.
C4. Klipsch, P. W. " Design of a.f. amplifier circuits using transformers " Proc. I.R.E. 24.2 (Feb.
1936) 219.
C5. Rohner, A. J. (Chapter) " Ferrous -cored inductors " (" Elec. Eng. Handbook " Edit. by Pender
and McIlwain 4th ed. 1950) Communication and Electronics, Sect. 3-42.
C6. Wrathall, E. T. " Audio -frequency transformers " W.E. 14.6, 7, 8 (June, July, August 1937).
C7. Sowter, G. A. V. " The properties of nickel iron alloys and their application to electronic devices "
J. Brit. I.R.E. 2.4 (Dec. 1941-Feb. 1942) 100.
C8. Peterson, E. " Harmonic production in ferromagnetic materials 3t low frequencies and low flux
densities." B.S.T.J. 7.4 (Oct. 1928) 762.
C9. Story, J. G. " Design of audio-frequency input and inter -valve transformers " W.E. 15.173 (Feb.
1938) 69.
C10. Roddam, T. " Distortion in audio -frequency transformers " W.W. 51.7 (July 1945) 202.
Cl1. Terman, F. E. " Radio Engineers' Handbook " 1st edit. (1943) p. 131.
C12. " Telcon Metals " Booklet Pub. by Telegraph Construction and Maintenance Co. Ltd., Green-
wich, London 1948.
C13. Terman, F. E. (book) " Radio Engineering " (3rd edit. 1947) p. 291.
C14. Baxandall, P. J. " High quality amplifier design " W.W. 54.1 (Jan. 1948) 2.
C15. Lee, R. " Recent transformer developments' Proc. I.R.E. 33.4 (April 1945) 240.
C16.' Strong, C. E. " Final stage class B modulation " Elect. Comm. 16.4 (April 1938) 321.
C17. McLean, T. " An analysis of distortion in class B amplifiers " Proc. I.R.E. 24.3 (March 1936) 487.
C18. Lee, R. (book) " Electronic transformers and circuits 1st edit. (1947) p. 1'63.
C19. Terman, F. E. " Radio Engineers' Handbook " (1st edit. 1943) p. 392.
C20. Mortley, W. S. " The design of low frequency transformer coupled amplifiers " Marconi Review
No. 45 (Nov. Dec. 1933) 25.
C21. Preisman, A. (book) " Graphical Constructions for Vacuum Tube Circuits " (1st edit. 1943
McGraw-Hill) p. 195.
C22. McLean, T. " An analysis of distortion in class B audio amplifiers " Proc. I.R.E. 24.3 (March
1936) 487.
C23. Sah, A. P-T. " Quasi -transients in Class B audio frequency push -pull amplifiers " Proc. I.R.E.
24.11 (Nov. 1936) 1522.
C24. Partridge, N. G. R. " Distortion in transformer cores " W.W. 44.25 (June 22, 1939) 572 ; 44.26
(June 29, 1939) 597 ; 45.1 (July 6, 1939) 8 ; 45.2 (July 13, 1939) 30.
C25. Partridge, N. G. R. " An introduction to the study of harmonic distortion in audio frequency
transformers " J. Brit. I.R.E. 3.1 (June -July, 1942) 6.
C26. Partridge, N. G. R. " Harmonic distortion in audio frequency transformers " W.E. 19.228 (Sept.
1942) 394; 19.229 (Oct. 1942) 451 ; 19.230 (Nov. 1942) 503.
C27. Partridge, N. G. R. " Transformer distortion " W.W. 48.8 (Aug. 1942) 178.
C28. Crowhurst, N. H. " Leakage inductance " Electronic Eng. 21.254 (April 1949) 129.
C29. Crowhurst, N. H. " Winding capacitance " Electronic Eng. 21.261 (Nov. 1949) 417.
C30. Harrison, E. B. " Test methods for high quality audio transformers " Tele.Tech. 9.3 (March
1950) 40.
C31. " Reference Data for Radio Engineers " Chap. 11 (Federal Telephone and Radio Corp. N.Y.
3rd edit. 1949).
C32. Lord, H. W. " Design of broad -band transformers for linear electronic circuits " E.E. 69.11
(Nov. 1950) 1020.
C33. Williams, T., and R. H. Eastop " Harmonic distortion in iron -core transformers " Audio Eng.
35.4 (April 1951) 18.
C34. Crowhurst, N. H. "Transformer iron losses" Electronic Eng. 23.284 (Oct. 1951) 396.
Additional references will be found in the Supplement commencing on page 1475.
(D) REFERENCES TO POWER TRANSFORMER DESIGN
DI. Blume, L. F. (book) " Transformer Engineering " (John Wiley and Sons, Inc. New York, 1938).
D2. Stigant, S. A. and H. M. Lacey " The J. and P. Transformer Book " (Johnson and Phillips Ltd.,
London).
D. M.I.T. Staff (book) " Magnetic Circuits and Transformers " (John Wiley and Sons, Inc. New
York, 1943).
5.7 REFERENCES 253
D4. Reed, E. G. (book) " Essentials of Transformer Practice " (McGraw -Hill Book Co. Inc. New
York 2nd edit. 1927).
D5. Ledward, T. A. " Transformer screening " W.W. 50.1. (Jan. 1944) 15.
136. " British Standard Specification for Steel Sheets for Transformers for Power and Lighting "
B.S. 601-1935 (British Standards Institution, London).
D7. ' Standard Methods of Test for Magnetic Properties of Iron and Steel " A-34 (1940). American
Society of Testing Materials, N.Y.
D8. Partridge, N. G. R. " Harmonic distortion in audio frequency transformers " W.E. 19.8 (Aug.
1942) 180.
139.Medina, L. " A device for the measurement of no-load losses in small power transformers " Proc.
I.R.E. Aust. 7.9 (Sept. 1946) 13.
D10. Horstman, C. C. " Rolled steel cores for radio transformers " Elect. 16.6 (June 1943) 110.
D11. Partridge, N. G. R. Lamination design-influence of shape on the cost and weight of trans-
formers and chokes" W.W. 48.12 (Dec. 1942) 286.
D12. Mawson, R. " The design of stampings for L.F. transformers " Electronic Eng. 16.195 (May
1944) 514.
D13. ` Dry Type Power Transformers for Radio Transmitters " R.M.A. Standard TR-102 Nov. 1947.
D14. "Power Transformers for Radio Broadcast Receivers-Core Laminations, Vertical and Hori-
zontal Channel Frames." R.M.A. Standard REC-120 Oct. 1948.
D15. Prince, D. C., and P. B. Vogdes (book) " Mercury Arc Rectifiers " (McGraw -Hill Book Co. Inc.,
New York, 1927).
D16. British Standard Specification for Electrical Performance of Transformers for Power and
Lighting " B.S. 171 -1936 ; Amendment No. 3, Jan. 1948 (British Standards Institution, London).
D17. American Standards
C57.10 -1948 Terminology for transformers, regulators and reactors.
C57.11 -1948 General requirements for transformers, regulators and reactors.
C57.12 -1949 Distribution, power and regulating transformers other than current limiting.
1318. ' Power and Audio Transformers and Reactors -Home Receiver Replacement Type " C.16.9
(1943) (American Standards Association, New York).
D19. " Reference Data for Radio Engineers " Chapter 11. (Federal Telephone and Radio Corp. N.Y.
3rd edit. 1949).
D20. Engineering a transformer." Booklet (Standard Transformer Corp. Chicago).
D21. Roesche, C. " Practical transformer design and construction " Radio News 37.6 (June 1947) 60 ;
38.1 (July 1947) 58 ; 38.2 (Aug. 1947) 60.
1322. Radio Industry Council, London, Specification " Transformers, Power, up to 2 KVA Rating "
No. RIC /214 (Issue No. 1 -Feb. 1951).
D23. " Components Handbook" Vol. 17 of M.I.T. Radiation Lab. Series. Chapter 4 " Iron -Core
Inductors " (McGraw -Hill Book Co. Inc. New York, 1949).
D24. Hamaker, H. C., and Th. Hehenkamp " Minimum cost transformers and chokes " Philips Res.
Rep. 5.5 (Oct. 1950) 357 ; 6.2 (April 1951) 105.
1324. R.C.S.C. Standard RC.L /191 Cores, magnetic.
D25. R.C.S.C. Standard RC.L /193 Cores magnetic, strip wound.
D26. R.C.S.C. Standard RC.S /214 Specification for transformers, power.
also A10.
MATHEMATICS
By F. Langford -Smith, B.Sc., B.E.
Section
1. Arithmetic and the slide rule 255
2. Algebra 259
3. Geometry and trigonometry ... 272
4. Periodic phenomena ... ... ... 278
5. Graphical representation and j notation 279
6. Complex algebra and De Moivre's Theorem 285
7. Differential and integral calculus ... 289
8. Fourier Series and harmonics ... 299
9. References 304
Mathematics, to the radio engineer, is merely a tool to be used in his design work.
For this reason it is often used in a slovenly manner or with insufficient precision or
understanding.
There are normally three stages in the solution of a problem -
1. Transferring the mechanical or electrical conditions into a mathematical form.
2. Solving the mathematics.
3. Interpreting and applying the mathematical solution.
The first stage is dealt with in Chapter 4 ; the second stage is the subject of this
chapter, while the third stage requires careful consideration of all the relevant con-
ditions. A solution only applies under the conditions assumed in stage one, which
may involve some approximations and limits. In all cases the solution should be
checked either experimentally or theoretically to prove that it is a true solution.
This chapter is not a textbook on mathematics, although it is in such an easy form
that anyone with the minimum of mathematical knowledge should be able to follow it.
It has been written primarily for those who require assistance in " brushing up "
their knowledge, and for the clarification of points which may be imperfectly under-
stood. It is " basic " rather than elementary in its introduction, and could therefore
be read with advantage by all.
Sufficient ground is covered for all normal usage in radio receiver design, except
for that required by specialists in network and filter design.
Reference data have been included for use by all grades
254
6.1 (i) FIGURES 255
Note that 10' means 10 x 10 = 100 ; 10' means 10 x 10 x 10 = 1000 and so on.
The " 2 " and the " 3 " are called exponents, or indices (plural of index).
10' is called " 10 squared "
10' is called " 10 cubed "
10° is called " 10 to the fourth " etc.
10 -1 is called " 10 to the minus 1 " etc.
Examples 2 750 000 is written 2.75 x 108
:
31 = 3
3' =3x3 =9 3- 2= 1/3' =1/9
=1/33 =1/27
3' =3 x3 x3 =27 3 -3
Roots The expression 91 may be written 1/9, where the sign A/ is called the
:
square root. Similarly 27dj3 may be written /27 where the sign called the "is
cube root.
The whole question is dealt with more fully, and more generally under " Algebra "
in Sect. 2.
(iii) Logarithms
We may write 100 = 10'
or we may express this in different language as
2 = log" 100
which is spoken of as " log 100 to the base 10." Here 2 is the logarithm of 100 to
the base 10.
Tables of logarithms to the base 10 are given in Chapter 38 Sect. 20, Table 71.
They are useful in multiplying and dividing numbers which are too large to handle
conveniently by the ordinary procedure.
256 (iii) LOGARITHMS 6.1
A typical logarithm is 2.4785. Here the figure 2 to the left of the decimal point
is called the index, and the figures 4785 to the right of the decimal point are called
the mantissa.
(1) The index of the logarithm of a number greater than unity is the number which
is less by one than the number of digits (figures) in the integral* part of the given
number ; for example the index of the logarithm of
57 640
is 4
5 764
is 3
576.4 is 2
57.64 is 1
5.764 is 0
If the number is less than unity, the index is negative, and is a higher number by
one than the number of zeros that follow the decimal point of the given number ; for
example the index of the logarithm of
.576 4 is -1
.005 764 is -3.
To denote that the index only is negative, the minus sign is usually written above it ;
e.g. 1, 3.
(2) The mantissa of the logarithm is found from the tables. Proceed to find the
first two figures in the left hand column of the table, then pass along the horizontal
line to the vertical column headed by the third figure. To this number add the
number in the difference column under the fourth figure of the given number. The
mantissa is the result obtained by this process with a decimal point before it.
For example, to find log, 0 5764 .
The index is 3
The mantissa is .7604 + .0003 = .7607
Therefore log10 5764 = 3.7607.
Similarly, log10 0.5764 = 1.7607 (note that the index is negative, but the mantissa
positive).
To find the number whose logarithm is given, it is possible to use either
antilog. tables (if these are available) or to use the log. tables in the reverse manner.
In either case, only the mantissa (to the right of the decimal point) should be applied
to the tables.
The procedure with log. tables is firstly to find the logarithm, on the principal
part of the table, which is next lower than the given logarithm, then to calculate the
difference, then to refer to the difference columns to find the number -exactly the
reverse of the previous procedure.
For example, to find the number whose logarithm is 2.5712. The mantissa is
.5712 and the nearest lower logarithm on the tables is 5705, the difference being 7
(in the fourth figure). The number whose log = 5705 is 3720, and to this must be
added the figure 6 which corresponds to the difference of 7 in the fourth column ;
the number is therefore 3720 + 6 = 3726. The decimal point must be placed so
as to give 2 + 1 = 3 digits to the left of the point, i.e. 372.6.
sum = 4.3353
The number whose log = .3353 is 2164.
The decimal point should be placed so as to give 4 + 1 = 5 digits to the left of
the point ; i.e. 21 640.
To the left of the decimal point.
6.1 (iii) LOGARITHMS 257
In a complicated calculation, especially with very large and very small numbers,
it is highly desirable to arrange the numerator and denominator in powers of 10.
For example
75 000 x 0.0036 x 5900 7.5 x 104 x 3.6 x 10-3 x 5.9 x 103
160 000 x 0.000 001 7 1.6 x 105 x 1.7 x 10-6
7.5 x 3.6 x 5.9 x 105
1.6 x 1.7
The slide rule does not indicate the position of the decimal point, and it is necessary
to determine the latter by some method such as inspection ; this is much easier when the
individual numbers are all between 1 and 10 as in the example above. It is also possible
to keep track of the decimal point by noting how often the manipulation passes from
end to end of the rule.
To find the logarithm of a given number, move the I on C scale to the number on
the D scale, then turn the rule over and read the logarithm on the L scale against
the mark (this will be a number between 0 and 1).
To find the decibels corresponding to a ratio, proceed as for the logarithm, but
multiply by 10 for a power ratio or 20 for a voltage ratio.
To find the sine or tangent of an angle, first set the angle on the S or T scale to the
mark, then read the value on the B scale, below 1 on the A scale, and divide by 100.
There are countless special types of slide rules, and in all such cases the detailed
instructions provided by the manufacturers should be studied.
Hints on special calculations on the slide rule
(1) Z = VR2 + X2 = RA/1 + (X' /R2)
Procedure For example if X = 3 and R = 2 set cursor to 3 on D scale, move
:
slide to give 2 on C scale. The value of (X/R)2 is given by the value on A scale op-
posite 1 on B scale -in this case 2.25. Move the slide up to 3.25 (= 2.25 + 1) and
then move the cursor to 2 on C scale, reading 3.61 on D scale as the answer.
(2) If a large number of figures is to be divided by one figure, divide unity (C scale)
by the divisor (D scale) and then, with fixed slide, move the cursor to each dividend
in turn on the C scale, reading the answer on the D scale
(y) Short cuts in arithmetic
(a) Approximations involving IT
n2 may be taken as 10 with an error less than 1.4%
it may be taken as 25/8 with an error less than 0.6%
1 /7r may be taken as 8/25 with an error less than 0.6%
27r may be taken as 25/4 with an error less than 0.6%
1/27r may be taken as 4/25 with an error less than 0.6%
(27r)2 = 39.5 with an error less than 0.06%
50 =v 49 1. 49 {
0.5 x
1/49
1
N 7.0714 (error = 0.004%)
Cubes (n = 3)
= (10 + 0.2)3 ^s 103 + 3 x 102 x 0.2
(10.2)3 1060 (error = 0.1 %)
Cube roots [n = 0.333 ; (n 1) = -0.667J -
fx ± S ti/x f 3(f)2
2
66 = -/64 + 2 ti/64 + 4.042
3 42
For more accurate approximations see Sect. 2 eqns. (82) and (83).
(e) To multiply by 11
To multiply a number by 11, write down the last figure, add the last and last but
one and write down the result, carrying over any tens to the next operation, add the
last but one and the last but two and so on, finishing by writing down the first
e.g. 11 x 42 736 = 470096 (no error)
(f) For approximations based on the Binomial Theorem see Sect. 2(xviii).
(g) For general approximations see Sect. 2(xx).
SECTION 2 : ALGEBRA
(i) Addition (ii) Subtraction (iii) Multiplication (iv) Division (v) Powers
(vi) Roots (vii) Brackets and simple manipulations (viii) Factoring (ix) Proportion
(x) Variation (xi) Inequalities (xii) Functions (xiii) Equations (xiv) Formulae
or laws (xv) Continuity and limits (xvi) Progressions, sequences and series (xvii)
Logarithmic and exponential functions (xviii) Infinite series (xix) Hyperbolic
functions (xx) General approximations.
See Section 6 for Complex Algebra.
Algebra is really only arithmetic, except that we use alphabetical symbols to stand
for figures. It is frequently more convenient to put an expression into an algebraic
form for general use, and then to apply it to a particular case by writing figures in place
of the letters. All algebraic expressions are capable of being converted into arith-
metical ones, and the fundamental mathematical processes of algebra may be used
in arithmetic.
260 ALGEBRA 6.2
(i) Addition
If a, b, and c are all values of the one unit (e.g. all resistances in ohms) we can add
them together to find the sum d, where d will also be in the same unit,
d = a + b + c
For example, if a = 5, b = 10, c= ohms,
15
then d = 5 + 10 + 15 = 30 ohms.
(ii) Subtraction
Subtraction is the opposite of addition, or negative addition, and can only be applied
when the quantity to be subtracted is in the same unit as the quantity from which it
is to be taken. For example let
where the
ñ 1
a
in the form a-"
-n is not a true index (or exponential) but merely a way of writing 1 /a ".
aTM
Therefore
a
= aTM x a"
=aTM " (2)
which indicates that the -n can be treated as though it were a true index.
6.2 (v) POWERS 261
a""
= a(m-m) = a" (3)
am
am
but
am
= 1
Therefore a° =1 (4)
(am)n = am x n = am" (5)
(ab)" = a"b" (7)
(
\a/ - r\b/ = b" (8)
(vi) Roots
1/axa =aor a' =a (10)
va x a x a= aor3a' =a (11)
We may adopt as a convenience the form
\= at (12)
(13)
"A/a= at," (14)
This may be extended to include
= (am)'l" = am/" (15)
so that am'" is the nth root of am._
Note that 1 / '"a = «1 /a; 1 / "/a = "A/1 /a (16)
a
+d
bad±bc a ba±b a a a(d +c)
c cd ' c +c c c +d cd (31)
The sign ± means either plus or minus. When ± and /or + signs are used on
both sides of an equation, the upper signs in both cases are to be taken in conjunction
as one case, while the lower signs are to be taken as the other case.
a c ac a ac
Tx d bd' b bc (32)
a c _a d_ ad
b d b X c bc (33)
x a+b x(c+d) y(a+b
y c +d y (c + d) y\c + d) (34a)
x(c -{- d) -
y(a + b)
y(c + d) (34b)
(ix) Proportion
(1) If a
b
= c
d
then
a
c
-d b
(39)
also
a
b
c
= ad - bc
0 therefore
bd -0 (40)
from which ad - bc = 0 and thus ad = bc (41)
a c e
(2) If = and also = -h
ae cg
then (42)
bf dh
(x) Variation
If y = kx, then y cc x
i.e. y is directly proportional to x.
k
z
If y = then y cc 1 x
i.e. y is inversely proportional to x.
If y = kxz, then y varies jointly as x and z.
x
If y = kz, then y varies directly as x and inversely as z
(xi) Inequalities
The letter symbols below are positive and finite.
If a > b then a + c > b + c, b < a (43a)
c,c a- c>b- -a<c -b (43b)
ac > bc, bc < ac (43c)
a b c
> < (43d)
a
6.2 (xi) INEQUALITIES 263
(xii) Functions
We may describe 3x + 4 as " a function of x" because its value depends upon the
value of x. This is usually written as
F(x) = 3x + 4.
Other typical functions of x are
2x2 +3x +5; x(x2 +3x);
cos x ; log x ; 11x.
In such functions, x is called the " independent variable." It is usual to write
F(a) as meaning " F(x) where x = a."
(xiii) Equations
An equation is a statement of conditional equality between two expressions con -
taining one or more symbols representing unknown quantities. The process of
determining values of the unknowns which will satisfy the equation is called solving
the equation.
An Identical Equation is one which holds for all values of its letter symbols.
A Linear Equation is one in which, after getting rid of fractions, the independent
variable only occurs in the first degree (e.g. x).
Example y = 5x + 3.
:
A Quadratic Equation is one in which, after getting rid of fractions, the inde-
pendent variable occurs in the second degree (e.g. x2) but not in higher degree.
Example : y = 4x2 + 5x + 3.
A quadratic equation in one unknown has two roots, although both may be complex
(i.e. with an imaginary term).
A Cubic Equation is one in which, after getting rid of fractions, the independent
variable occurs in the third degree (e.g. x3) but not in higher degree.
Example : y = 3x3 + 4x2 + 2x + 5.
Note : x and y are usually taken as unknowns ; a, b, c and d as known constants.
2. A term can be moved from one side to the other provided that its sign is changed.
Example If a = b ; then a
:
b = 0. -
3. All signs in the equation may be changed together.
Example: If a x- =y
then a x-b; y. - =b -
-
This is equivalent to multiplication throughout by 1.
Warning
If both sides of an equation are squared, or if both sides are multiplied by a term
containing the unknown, a new root may be introduced.
Solution of equations
(1) Linear equations with one unknown
Example ax + b = O.
:
Solution x = : b /a. -
(2) Linear equations with two unknowns
(46)
Any linear relation between two variables, x and y, can be written in the general form
ax +by +c =0 (47)
or (provided that b is not zero) in the alternative form
y = mx + n (48a)
This type of equation is not limited to one or two solutions, but has a corresponding
value of y for every possible value of x. It is very helpful to plot the value of y against
the value of x on squared paper-see Sect. 5(i). With any equation of this type, the
graph will be a straight line, and it is only necessary to determine
(1) one point on the line
(2) the slope of the line at any point.
The most convenient point is usually x = 0, and in eqn. (48a)
this will give y = n
or in eqn. (47) this will give y = c/b -
The slope of the line is given by the difference of they values of two points, divided
by the difference of their x values.
In eqn. (47) the slope is
while in eqn. (48a) the slope is m.
a/b -
A particular form of eqn. (47) is
y
+ = 1 (48b)
and in this case the line cuts the x axis at x = a and cuts the y axis at y = b. The
slope is equal to b /a. -
An equation of the form
a
x y
-+ -=c
b
(49)
may be solved by regarding 1/x and 1/y as the unknowns, then following a similar
procedure as for an equation in x and y, and solving for 1/x and 1 /y.
(3) Simultaneous linear equations (two unknowns)
aix + bty = c1 x
clbz czb1
- a1b, -- azbl (50a)
aax + bzy = alcz - azc1
y - alb, - a5b1
cz
(50b)
provided that (a ib2 -
a2b1) is not zero.
Alternatively the solution may be derived by determining x in terms of y from the
first equation, and then substituting in the second.
Checking solutions :
After any solution has been found, particularly with more than one solution, it is
highly desirable to check these in the original equation.
(4) Three simultaneous equations (three unknowns)
Given +by +cz +d =0.
a1x+ó1y +c1z +d1 =0
a2x+ bay +c,z-}-d,=0
(xiii) EQUATIONS 265
6.2
Then
d(b,c - bic,) + d 1(bc, - b,c) + d2(b - be 1c 1)
This type of equation is not limited to one or two solutions, but has a corresponding
value of y for every possible value of x. It is helpful to plot part of the curve on
squared paper -see Sect. 5(i). The curve may cut the x axis at two points, or it may
touch at one point, or it may not touch it at all. Let y = 0, then
ax2 -}-bx +c =0
and the points at which the curve cuts the x axis will be
- b ±\/b2 - 4ac (54)
x 2a-
If b2 > 4ac, the curve will cut at two points.
If b2 = 4ac, the curve will touch at one point.
If b2 < 4ac, the curve will not cut the x axis.
(xiv) Formulae or laws
A formula is a law, or rule, generally in connection with some scientific relation-
ship, expressed as an equation by means of letter symbols (variables) and constants.
For example, Ohm's Law states that E = RI where each of the letter symbols
has a precise meaning. If we know any two of the variables, we can determine the
third,
E
R=E and I=R
Another example is
1
2nfG
X,
which gives X, for any desired values off and C. Note that 2n is a constant.
All formulae or laws may be rearranged in accordance with the rules for the solution
of equations, so as to give the value of any variable in terms of the others.
+Rr, L (55)
n->
S
Go
-1 a (62)
Harmonic Progression The terms a, b,
: c, etc. form a harmonic sequence if their
reciprocals
1 1 1
a, b, c, etc.
form an arithmetical sequence.
Example : 1, 4, 1/3, 4, 1/5 is a Harmonic Progression
because 1, 2, 3 , 4, 5 is an Arithmetical Progression
2ab
The Harmonic Mean between a and b is
+ b
6.2 (xvi) PROGRESSIONS, SEQUENCES AND SERIES 267
Note that the Arithmetical Mean between two numbers is greater than the Geo-
metrical Mean, which in turn is greater than the Harmonic Mean.
With any form of progression or sequence, we are often concerned with the sum of
a number of terms of which the general term is given. It is possible to write this
sum in a shortened form, for example
k = n
1 -I- 2 +3 +4 +... +k +... +n= 2k
k = 1
where the Greek letter capital sigma is used to indicate the sum of a number of terms ;
k is merely the general term ; and the values of k are to be taken from k = 1 (beneath
sigma) to k = n (written above sigma).
(xvii) Logarithmic and exponential functions
Ifa" =N
then x is the logarithm to the base a of the number N. This may be put in the form
x = loge N
where the base a may be any positive number except 1 or 0.
The two principal systems of logarithms are
(1) The Naperian (or natural) system, using the base e = 2.718 28 ...
(pre-
ferably written with the Greek E), and
(2) The Briggs (or common) system, using the base 10.
Only one set of tables is required, for it is possible to convert a logarithm to one
base (b) into a logarithm to any other base (a) :
Ifr =10and x = 3
Then y = 109 = 1000.
This may be handled by the logarithmic form of the equation,
x = log, y = log," 1000 = 3 as before.
Logarithmic decrement : If the equation is of the form
y =ae-a`
where (-
b) is negative, the value of y decreases as x is increased, and (- b) is called
the Logarithmic Decrement.
1 + mx +
m(m - 1)
x
m(m - 1)(m - 2) x9 + etc. (78)
1.2 1.2.3
The nth term is
a"
- 1)(m - 2) ... (m - n + 2) x" _
m(m ,
1.2.3 ... (n - 1) (
79
)
The denominator is usually written in the form
(n - 1)
which is called " factorial (n - 1)."
!
(1 ±x)"' =1±mx }
m(m 1) x2 m(m -13(m -2)x9
2I
m(m - ... (m -
1) n + 2)x" -' +...
(n -1)! (80)
which holds for all values of x if m is a positive integer, and for all values of m provided
that x is numerically less than 1.
The Binomial Theorem is useful in certain approximate calculations. If x is small
compared with 1, and m is reasonably small,
(1 + x)m 1 + mx
(1 -x)" 1 -mx l
(1 +
-
x) -"'
x)-"
N 1 - mx r (81)
(1 1 + mx. J
To a closer approximation (taking three terms),
(
(1
1 +
-
x) " '
x)m
1 +
-
mx
mx
+
+
74 m(m -- 1) x'
1 4 m(m 1) x$ (82)
(1 +x) -m 1 -mx+,} m(m + 1) x'
(1 -x)-m 1 +mx +} m(m + 1)x'
Numerical example : To find the cube root of 220.
6.2 (xviii) INFINITE SERIES 269
1
1 1/S
'220 = (216 + 4)1/3 = {216(1 -I-
2:6)}h13 =
6(1 -+- 54
Applying the approximation from the Binomial Theorem,
. 1+ 1+ 1.006 17
( 1+
Therefore 1/220
Case 2 :
54)
6.037. ,.
54
6 x 1.006 17 . 162
We can also expand (a + x)m, which is convergent when x is numerically less than
a:
(a ± x) m -a
= m
± ma ml-1 x +
m(m - 1)
a m-s x s t m(m - 1)(m - 2) a m-e x a
+ . . .(83)
(x"-1
! !
Approximation
a +1 = (a
:
The right hand side of this equation is the value e, which is equal to 2.71828 (to five
l +-1
:
places of decimals). Taking the xth power of each side of this equation,
1
=Em
"s
e= 1
Therefore +
+
1
n
xa xa x* x "-1
and = 1 +x +2 +3 +4! +(n -1) +... (87)
loge (1 + x) = x - xa
+
x3
- 4 + ... x*
(88)
which is convergent if x is numerically less than 1.
For derivation of eqns. (89) and (90) see Sect. 6, eqns. (17) and (18).
(xix) Hyperbolic functions
These are combinations of the sum and difference of two exponential functions.
Em
is called the hyperbolic sine of x, designated by sinh x
Em +
2
f m
Ex - E-'
is called the hyperbolic tangent of x, designated by tanh x
Ei + E
Note E= 1 +1 +2 1 1
+Si +4i +
1 1
. . 2.718 28 (92)
The following may be derived
- sinh2x =
:
cosh2x 1 (93)
sech2x + tanh2x = 1 (94)
coth2x - cosech2x = (95)
- x) = - sinh x cosh - = cosh x
1
(100)
tanh x tanh y=
sinh (x f y)
(104)
cosh x cosh y
(sinh x + cosh x)" = cosh nx + sinh nx (105)
sinh-1x = loge (x + 1/x' + 1) = cosh'1-Vx2 + 1 (106)
cosh-1x = log (x + Vx' - 1) = sinh-1Vx'
1 + x
- 1 (107)
tanh-lx = loge -x (108)
---
1
1
= cosh-1 = sinh'1 (109)
A/1-x2
Exfr -E--(.± o
\/1-x'
sinh (x f - y)
2
= sinh x cosh y ± cosh x sinh y (110a)
Eatv +e--Ixfvl
cosh (x f y) = cosh x cosh y -
sinh x sinh y f (110b)
2
Y- xf
tanh (x f
Exf E-1 Y) tanh x tanh y
y) =
Exf Y+ E-) xf Y) -- 1
txxf
(111)
tanh x tanh y
sinh 2x = 2 sinh x cosh x- 2
2 (112)
cosh x + sinh x = ex
cosh x - sinh x = -$
x' x'
sinhx =x+3-{--5!+ (115c)
8.2 (xix) HYPERBOLIC FUNCTIONS 271
xa x4 x°
cosh x =
4!
+ 6- +
1 + 2 + ...
(115d)
(these are convergent for all real values of x).
Also in complex form (see Section 6)
sinh jx = j sin x ; cosh jx = cos x ; tanh jx = j tan x (116)
sin jx = j Binh x ; cos jx = cosh x (117)
sinh (x + jy) = sinh x cosy ± j cosh x sin y (118)
Note that x and y in sin x, cos x, tan x etc. in eqns. 116 to 119 must be expressed
in radians.
cosh (x ± jy) = cosh x cosy + j sinh x sin y (119)
sinh (2) =
\ /
/cosh x - 1
cosh
(X)
1 /cosh x +
2
1
(120
(120)
2 2
x) cosh x - 1 sinh x (121)
(tanh 2) sinh .x cosh x + 1
1
+S1 ^..1 +S1 -Sa (123)
1 +Ss
(1 +S)" 1 ±n8
1 where n may be integral, (124)
(1 ± S)"
ti 1 f nS fractional or negative 1 (125)
V1 + S N 1 + IS -V1 - S N 1 - IS (126)
-V11 +5
ti 1 -i8 1.-
1
ti t + }S (127)
-
f - (135)
(136)
sinh S S cosh 1 tanh S S. S (137)
sinh 1S S tanh -1S S (138)
sinh (x + 8) N sinh x + 6 cosh x ; sinh (x 6) - sinh x - a cosh x (139)
- -
cosh (x + 6)
tanh (x + S) .
cosh x + S sinh x ; cosh (x
tanh x + S sech2 x ; tanh (x
S)
S) -
cosh x
tanh x -
S sinh x (140)
8 secha x (141)
When L is a large quantity
sinh L ¡EL .cosh L lEL tanh L . 1 (142)
Trigonometrical relationships
When 8 is an extremely small quantity, so that an angle of S radians is a very small
angle, and x is an angle very large compared with S,
272 (xx) GENERAL APPROXIMATIONS 6.2
S.
sin 6
sin- 'S
^'
where K is any integer.
8
6
cos
cos 1 6 . 1
}tr(4K -
1) + 8
tan
M p
p
=n P =A/mn (16)
m a b
also = = (17)
Any triangle inscribed in a semicircle, with the diameter forming one side, is a right
angle triangle.
Equilateral Triangle (Fig. 6.2a)
= a
Each side
Each angle = 60°
m = } a
area
h
=
=
') ah = (//4)a'
\
1f m = ( /2)a ti 0.866a
0.433a'
(18)
(19)
rW.66
ric 6 6
Trapezoid rule
Area = d(y, + y2 + 373 + + -1 + +y) y (53)
where d = width of each strip
and y y2, y2 .... y are measured lengths of each of the equidistant parallel chords.
)
Note that the first (y,) and the last (y do not cut the area, and may be zero if the
surface is sharply curved.
Simpson's Rule
d
Area = (y, + 4y2 + 2y2 + 4y, + 2y,... + 2y -2 + 4y -1 + y) (54)
3
where n must be odd
d = width of each strip
and y, ... y
are measured lengths of equidistant parallel chords.
o
FIG. 6-II
4
Examples of angles in all four sectors are shown in Fig. 6.12. It will be seen that,
for any angle A, the position of OP is the same for a movement of angle A in a positive
direction, or for a negative movement of (360° -
A) ; for example,
-
+ 330° _ (360° 330 °) _ - 30 - °.
In the case of angles greater than 360° we are generally only concerned with the final
position of OP, so that for these cases we may subtract 360 °, or any multiple of 360 °,
from the angle so as to give a value less than 360 °. For example 390° = 360° + 30° ;
800° = 720° + 80 °; 1125° = 1080° + 45 °.
In trigonometry it is also necessary
to define the polarity of the three av
4
sides of the triangles from which
we derive the sine, cosine and
tangent. The hypotenuse (0P1)
Zi'
tat '
1,
is always positive (see Fig. 6.13). .
fIG.6.13
r
positive when P is above X (e.g. P, X,) and negative when P is below X (e.g. = P4).
1st Quadrant All sides positive (OX ,P1).
:
10. sin 421 = fß4(1 - cos A) = f ¡V1 + sin A - }V1 - sin A( ) (99)
cos }A = +V1(1 + cos A) = f }\/1 + sin A + }V1 - sin A)
( (100)
tan }A =
1- cos A sin A (101)
sin A 1 + cos A
278 (iii) TRIGONOMETRICAL RELATIONSHIPS 6.3
The motion of the point X, as it oscillates about O between the extremes A and A'
Simple Harmonic Motion.
is called
Angular velocity (usually represented by the small Greek letter omega- co) is the
number of radians per second through which the point P travels. In each revolution
1.e. uniform rate of rotation.
6.4 PERIODIC PHENOMENA 279
(360 °) it will pass through 27r radians, and if it makes f revolutions per second, then
the angular velocity will be
w = 271 radians per second
where f = frequency in cycles per second.
In most mathematical work it is more convenient to write w than to write 271f.
" cause " and " effect," the " cause " (known as the independent variable) is plotted
horizontally on the x axis, and the " effect " (known as the dependent variable) vertic-
ally on the y axis. In other cases the choice of axes is optional.
Any convenient scales may be used, and the x and y scales may differ.
The procedure to be adopted to plot a typical equation
y = 2x2 + 4x 5 -
is as follows. Select suitable values of x (which will be regarded as the independent
variable) and calculate the value of y for each :
x =3 y= 18 +12 -5 = +25
=2 y= 8+ -5 =
4- 5 =++11
x 8
x= y= 2+
0 -5 =- 5
1 1
x =0 y= 0+
x = - y= 2- 4- =- 5 7
x = - 23
1
y = 8- 128-- =- 5 5
x = - y = 18 - = -- 5 1
Then plot these points, as in Fig. 6.19A and draw a smooth curve through them.
280 (i) GRAPHS 8.5
The tangent at any point (e.g. P in Fig. 6.19A) is a straight line which is drawn so
as to touch the curve at the point. The slope of the tangent, which is the same as
that of the curve at the point, is defined as the tangent of the angle O which it makes
with the X axis. Between points B and C on the curve, O is positive, therefore tan O
is positive and the slope is called positive. Between points B and A the angle O is
negative, and the slope negative. At point B, O = 0 and the slope is zero. It is
important to remember that the curve normally extends in both directions indefinitely
unless it has limits, or turns back on itself. It is therefore advisable, when plotting
an unknown function, to take very large positive and negative values of x and cal-
culate the corresponding values of y, even though the points cannot be plotted on
the graph paper. This will indicate the general trend of the curve beyond the limits
of the graph paper,
+á
x
+X
FIG. 6.I9A -
The equations of some common curves are:
Straight line through origin y = mx (1)
Straight line not through origin y= mx
Circle with centre at origin
Circle with centre at (h, k) (x
= r2
h)2
--I -
-+ ++ dx(y -+ ey = 0
y2
n
x2
k)2 r2
(2)
(3)
(4)
General equation of circle x2 y2 +f= (5)
Ellipse -+ y2-=
x2
a2 b2
1 (6)
Hyperbola
Parabola (origin at vertex)
x2
a2
- y2
-1;2--
1 (7)'
3,2= 2px (8)
Focus is at x = p /2, y = 0 (9)
Areas and average heights
The average height of a curve (i.e. the average length of the ordinates) may be de-
termined by dividing the area beneath the curve into strips of equal width, and then
using either the Trapezoid Rule or Simpson's Rule [see Sect. 3 eqn. (54)] to deter-
mine the area, and dividing the area by the length (abscissa).
Logarithmic paper
Logarithmic ruled paper is frequently used in the plotting of curves, particularly
when the x or y coordinates cover a range of 100: 1 or mare. Single cycle* paper
accommodates a range of 10 1, and may be drawn by hand, using the whole of the
:
C scale, or half the B scale on a slide rule. Two " cycle " log paper accommodates
a range of 100: 1, and may be drawn with the whole B scale on a slide rule. Each
Strictly this should be called single decade.
6.5 (i) GRAPHS 281
of the " cycles " has the same linear length. Additional " cycles " may be added as
desired (for examples see a.v.c. characteristics in Chapter 27).
FIG. 6.19 B
and = 20. 15
These points have been plotted and a smooth curve drawn through them. Similar
curves could be drawn for any other plate voltage, thus forming a " family " of curves.
This is actually a three- dimensional graphical diagram reduced to a form suitable
for a flat surface.
(iv) Vectors and j notation
Any physical quantity which possesses both magnitude and direction is called a
vector. Vectors may be represented on paper by means of straight lines with arrow-
heads. The length of the line indicates (to some arbitrary scale) the magnitude of
the quantity, and the direction of the line and arrow -head indicates the direction in
which the vector is operating. The position of the line on the paper is of no con-
sequence.
Addition of vectors
Vectors may be added by drawing them in tandem, and taking the resultant from the
a beginning of the first one to the end of the
FIG.6.21
last one. Fig. 6.21A shows two vectors,
a and b, which are added together to give
the resultant c. Exactly the same result
is obtained by placing a and b together, as
in Fig. 6.21B, completing the parallelo-
gram, and taking c as the diagonal. Vec-
tors are generally printed in bold face type,
to distinguish them from scalar values,
which have no direction, although they
FIG.6.22 have magnitude and sign (i.e. positive or
negative).
+6
Vector negative sign
A vector (- a) has the same magnitude
as a vector (a) but its direction is reversed.
6 Subtraction of vectors
The vector to be subtracted is reversed in
direction, and then the vectors are added.
In Fig. 6.22, to find a -
b, the direction of b is reversed to give (-
b) and then a
and (-b0 are added to give the resultant (a -b).
Multiplication of a vector by a number (n)
The resultant vector has the same direction, but its length is increased n times
e.g. a x n = na
v FIG. 6.24
where a . b indicates the multiplication of two vectors. From Fig. 6.23 it will be
seen that the scalar product is the product of the magnitude of one vector and the
" projection " of the other on it.
8.5 (iv) VECTORS AND j NOTATION 283
Components of a vector
Any vector can be resolved into two component vectors in any two desired directions.
For example, in Fig. 6.21 the vector c can be resolved into the component vectors
a and b. If the component vectors are at right angles to one another they are called
rectangular components ; in such a case they are usually taken horizontally (along
the X axis) and vertically (along the Y axis). The vector OA in Fig. 6.24 can be
resolved into two rectangular components OB and OC where
OB _ OA cos 0, IOC! _ 10A sin B.
I I I I
I
Polar coordinates
An alternative form of defining a vector OP is
OP = r /O
where r is the magnitude of OP, and L O indicates that there is an angle O between it
and OX (Fig. 6.25).
A graphical device has been described* for the con-
version from complex to polar forms.
Rotating vectors ** (j notation)
If a vector X (OA in Fig. 6.26) is rotated 90° in a
positive direction to a position OB, the new vector is
called jX, and j is described as an " operator "
which rotates a vector by 90° without changing
its magnitude.
If the vector j X (OB in Fig. 6.26) is operated upon
by j, it will be rotated 90° to the position OC,
where it is called i2X, the j2 indicating that it has
been rotated 2 x 90° = 180° from its original position OA.
If the vector j'X (OC in Fig. 6.26) is operated upon by j, it will be rotated 90° to the
position OD, where it is called j3X, the j3 indicating that it has been rotated 3 x 90°
= 270° from its original position OA.
If the vector j3X (OD in Fig. 6.26) is operated upon by j, it will be rotated 90° to
the position OA where it would be called j'X, the j' indicating that it has been rotated
4 x 90° = 360° from its original position OA.
There is an important deduction which is immediately obvious. j2 indicates a
e reversal of direction which is the same as
a change of sign. The operator j2 is
therefore equivalent to multiplication
FIG 6.26
by -1.
tx If the operatorJ2 is applied twice in succes-
sion, the result should be equivalent to multi-
plication by
(-
1 twice in- succession, i.e.
c x 1) x (-
1) = + 1. This is so, because
i.x the operator j'
brings the vector back to its
original direction.
Since the operator j2 is equivalent to multi-
plication by -
1, we may deduce that the
operator j is equivalent to multiplication by
-
1, even though this in itself does not
o mean anything.
Operator Equivalent to multiplication by
J -- 1
i2 1
is -1/- 1
ja + 1
XI
RI ' E
and c = a sin O.
and 2xy = b.
Modulus = r = A/ a2 + b2
Square of modulus of (x + jy) = x2 + y2
Therefore x2 + y2 = r
But x2 -
y2 = a (see above)
Therefore x2 = ¡(r + a) and y2 = j}(r a) -
Therefore x = ± -/,)(r {i
a) and y = + V%(r
The signs should be checked to see which are applicable
a) - (5)
¡
,
E'19 i -I- -r
2,
03 B'
2 4 +j 5
02 05
= 1 +io ..
Grouping the j terms,
E'B = (1
02
-2,x-4, -...)
B'
I j(B - 03
-5 - 05
. .)
But cos 8 = I - 02
+
B'
- .. from eqn. (17) below
.
Thus, in the general case, the trigonometrical operator (cos B + j sin 9)" gives a
rotation of an angle n O which, as explained above, is equivalent to a trigonometrical
operator
(cos n B + sin n B). j
We have, in this way, proved De Moivre's Theorem for the case when n is a positive
integer, and indicated the significance of the Theorem.
Application of De Moivre's Theorem :
1. To express cos n O and sin n O in terms of cos B and sin B, where n is a positive
integer.
Cos nB +j sin nO =(cos B +jsin 8)"
= cos" B + nj cosn -' B. sin B + n(n 1)
j', cos"-2 B. sin' O + .. .
the sine and cosine as in eqns. (15) and (16). When n becomes large, cos (0/n)
may be taken as unity and sin (Bin) as (0/n) itself. In the limit as n tends to
infinity it can then be shown that
cos B = 1 - 92
+ 4j
94
- 08
!
+ .. (17)
sing = B- 0'-j+ 5- 05 97
,Tj +.. (18)
where O is expressed in radians.
3. To express sin and cos B in terms of E'B .
O
sin O =
EiB -E iB
(19)
2j
and cos O =
EB
- --
- E- ,B
-- (20)
2
6.7 (i) SLOPE AND RATE OF CHANGE 289
The lengths of the projections must be measured in terms of the scales to which
the line is drawn. Thus horizontal projections such as CB must be measured, not
in inches, but in the equivalent number of units corresponding to the length CB on
the X axis. Similarly with the vertical projections on the Y axis, which usually has
quite a different scale from that for the X axis.
Definition
The rate of change is the amount of change in the function, per unit change in the
value of the independent variable. The rate of change of the function is therefore
the ratio of change in the function to the change in the variable which produces it.
Consider the equation for a straight line
y =ax +b
where a and b are constants. Let us take two values of x, one equal to x, and the other
(x1 + dx), where dx is a small increment of x.
Point 1 : x = x1 .'. y1 = ax1 + b (1)
Point 2 : x = (xl + dx) .. y2 = a(x, + dx) + b (2)
Subtracting (1) from (2), (y, -y
l) = a. d x
Putting (y, -y 1) = dy, dy
dy
x
Dividing both sides by dx, = a (3)
Here dy is the amount of change in the function for a change dx in the independent
variable. The " rate of change " is defined as the amount of change in the function
per unit change in the independent variable, that is
rate of change = dy /dx (4)
Referring to Fig. 6.34, we have a graph of y = ax + b which is, of course, a straight
line cutting the Y axis at a height b above the origin. The first point (x1, y,) is at P,
and the second point (x1 + dx, y, + dy) is at Q. In the preceding argument we
found that dy /dx = a. In Fig. 6.34 dx = PR and dy = RQ, so that dy /dx =
RQ /PR, which is the slope of the line y = ax + b. Thus
RQ dy
slope =
PR dx a -
Equations (4) and (5) prove that the rate of change is the same as the slope
(5)
As with the simpler case of the straight line, take a second point (Q) with its x
coordinate increased by dx. Also, as before, call the increment in the y coordinate
dy. It will therefore be seen that, commencing at point P, an increment (dx = PC)
in the value of x results in an increment (dy = CQ) in the value of y.
The average rate of change over this increment is, as before, dy /dx which is
the slope of the chord PQ and the tangent of the angle B, which PQ makes with the
horizontal. If now, leaving point P unchanged, we gradually move Q along the curve
8.7 (ü) DIFFERENTIATION 291
(ii) Differentiation
We may express the foregoing argument in the mathematical form of limits
Lt (dy /dx) = tan 0, (6)
-
dx 0
which says that the limit (as dx is made smaller and approaches zero) of dy /dx
is tan 0, or the slope of the tangent PT, which is the instantaneous rate of increase
at point P.
This is given the symbol dy /dx which is " the differential coefficient (or de-
rivative) of y with regard to x," and is spoken of as " dee y by dee x."
It should be noted that dy /dx is a single symbol, not a fraction, and is merely a
short way of writing
Lt (dy /dx).
dx -+ 0
Differentiation is the process of finding the differential coefficient (or derivative).
Some examples are given below and in each case the result may be obtained by con-
sidering the increase in the function which results from an increase dx in the inde-
pendent variable.
Note : u and y are functions of x ; a, b and c are constants.
1. Derivative of a constant [y = c] dy /dx = 0 (7)
2. Derivative of a variable with respect to itself
[y = x] : dy /dx = dx /dx = 1 (8)
3. Derivative of a variable multiplied by a constant
[y = cx] : dy /dx = c (9)
4. Derivative of powers of a variable
[y = x'] : dy /dx = 2x (10)
[y = x'] : dy /dx = 3x' (11)
[y = x`] : dy /dx = 4x' (12)
[y = x "] : dy /dx = nx " -' (13)
This applies for n negative as well as positive.
5. Derivative of a constant times a function of a variable
[y = cx'] : dy /dx = 2cx (14)
[y = c.u] : dy /dx = .,.du /dx (15)
6. Derivative of fractional powers of a variable
[y = xi] : dy /dx = ¡x-1 (16)
[y = x11 "] : dy /dx = (1 /n)xu/"" -1 (17)
7. Derivative of a sum or difference
[y = u f v] :
-
dy /dx = du /dx
dy /dx = 3ax' + 2bx
dv /dx f -
(18)
[y = ax' + bx' cx] : c (19)
8. Derivative of a product of two functions
dv du
[y = u.v] : dy /dx = (20)
udx + vax
[y = (x + 1) x'] : dy /dx = (x + 1).2x + x1.1
= 3x' +2x (21)
9. Derivative of a quotient of two functions
du dv
d'y
dx
= n(n - 1)(n - 2)x -' .... third derivative F "'(x)
"
Application of differentiation
The plate current versus grid voltage characteristic of a triode valve (for constant
plate voltage) is a function of the grid voltage, and follows approximately the law
I, K(1 tE, + E,)3/2
where K, c and E, are constants.= The derivative with regard to E, is dI, /dE
which is the mutual conductance. The second derivative is the rate of change of the
mutual conductance with regard to E
and is useful when we want to find the con-
ditions for maximum or minimum mutual conductance.
In Fig. 6.36 there is a curve with a maximum at point M and a minimum at
point N. It will be seen that the instantaneous slope of the curve at both points M
and N is zero, that is dy /dx = 0.
Part of curve: PtoM M MtoN N NtoQ
Slope (dy /dx) : + ve O -ve O + ve
d'y /dx' : - ve + ve
A maximum is indicated by : dy /dx = 0
d'y /dx' negative
A minimum is indicated by : dy /dx = 0
d$y /dx' positive
x measured in radians.
Positive sign if sin-i x lies in first or fourth quadrant, negative sign if sin- 1 x lies in second or third
quadrant.
'i
tNegative sign if cos -1 x lies in first or second quadrant, positive sign if cos x lies in thlkd or fourth
quadrant.
6.7 (ii) DIFFERENTIATION 293
x = + 1 : y = 1 -6- 1 + 16 = + 10.
x = - 1 : y = 1 -6+ 1 + 16 = + 12.
Some
It is always wise to make a rough plot of the curve to see its general shape.
curves have more than one value of maximum and minimum.
Partial differentiation
Partial differential coefficients, designated in the form ay /ax (the symbol a may be
pronounced " der " to distinguish from " d " in dy /dx) are used in considering the
relationship between two of the variables in systems of more than two variables such
as the volume of an enclosure having rectangular faces, the sides being of length
x, y and z respectively :
v = xy z (44)
Thus, the rate of change of volume with the change in length of the side x, while
the sides y and z remain constant, is
av /ax = y z (45)
$imilarly av /ay = z x, where z and x are constant (46)
And av /az = xy where x and y are constant (47)
In three-dimensional differential geometry, the equation representing a surface may
be represented generally in the form
y = F (x, z) (48)
In this case, the partial differential coefficient ay /ax represents the slope at the point
(x, y, z) of the tangent to the curve of intersection of the surface with a plane parallel
to the plane passing through the x and y axes and separated oy a fixed distance z
from the latter.
Thus ay /ax represents the slope of a tangent to a cross section of a three- dimensional
solid, the partial derivative reducing the three -dimensional body to a form suitable
for two- dimensional consideration. " ay/Ox" is equivalent to " dy /dx (z constant) "
when there are three variables, x, y and z.
*A point of inflection is one at which the curvature changes from one direction to the other (e.g. S
in Fig. 6.36). It is necessarily a point of maximum or minimum slope.
294 (ii) DIFFERENTIATION 8.7
f dx -2b(ax + +
13.
ax + b -f axx+ b = a'C(ax
b)
=- scot-'x +C
.1
x
dx
-f 1 1 1
14. fx' +a' x9 +asdx=
atan-'Q +C (66)
r dx 1
a,dx =2a
1 x -a +C 1 a - x +C
15.
x'- a'- fx'- log x +a 2alogEa +x (67)
dx
17.
f.x' ±a' -f x' 1
±aa'
dx = log, (x + 1x' ± a') + C (69)
Areas by integration
If the area under a given portion of a curve is A (Fig. 6.37), then a small increase
dx on the horizontal axis causes an increase dA in area, where
JA = dx(y + ¡dy).
As dx and dy are made smaller, in the limiting case as dx approaches zero, the value
of (y + dy) approaches y,
r
FIG. 6.37 i.e.
Example
To find the area under the curve y = 3x2 from x = 1 to x = 4.
A= f3x2dx =x3 +C
when x = 1, A = 0,
when x = 4, A = x3
therefore x3 + C = 0, therefore C =
- 1 =43 -1 =63.
- 1
Definite integrals
When it is desired to indicate the limits in the value of x between which the integral
is desired, the integral is written, as for the example above,
x = 5 5
3x' dx or 3x' dx.
= J x r r
These limits are called the hmits of integration, and the integral is called the de-
finite integral. For distinction, the unlimited integral is called the indefinite
integral.
-
The definite integral is the difference between the values of the integral for x = b
and x = a
f a
f(x) dx = [F(x)1 = F(x = b) - F(x = a) (84)
Owing to the subtraction, the constant of integration does not appear in definite
integrals.
Special properties of definite integrals
(a
-
b
Examples :
f ir/2
o
sin B dB = -- cos B17r/2
o
= [ - cos 7T/2 + cos 0] = 0 + 1 = 1 (87)
(iii) INTEGRATION 297
6,7
('o 0
Jn
cos O dO =[ sin B = [sin IT - sin 0] = 0 - 0 = 0 (90)
fo0 1o
2tr
27
sin B dB =[- ces B = [- cos 2n + cos 0] _ - 1 + 1 = 0 (91)
0 o
2a
Jo
2n
cos O dO sin O
o
- [sin 2n - sin 0] = 0 - 0 = 0 (92)
n
J sin' nO dO = }r O - (1 /2n) sin 2B1 n
o
= n/2 (93)
J COO n B dO = J n
0
(1 - sin' n B dO) dO = L B
0
- 7r/2 = 17/2 (94)
0
(' 2ar
sin' n B dO = n
2
o
cos' n B de - ,r (96)
o
27r
sin m0 cos n0dû =O sin m 0 -'n nO dO = 0 (97)
o f o
2ar
sin n B cos n B de =0 (99)
o
yav =4 -1 1 J 41
3x'dx =3 x'
1L 4
_a1
3 [64 - 1
1
J
63
=21.
3. (4,,)2 =2
air
(Im sin 9)2 dB
} - _ Em or Im
FIG. 6.38
Imp B sin 2 B
tar
2n 2 4 (T)
Jo n 21T
/,2
=2ñ [7r- 0 -0 +0] timt
ongle (rodions)
Imp Imp
if
2! 31
+.. + (n
1 n1
s "(x ") t f 1
(100) (a)
(x a)"
h ^ -1 h"
+ f " -'(a) + -n! f"(a) + ... (101)
The sum of the first few terms of Taylor's Series gives a good approximation to f (x)
for values of x near x = a.
Examples of the use of Taylor's Series :
f(x) = cos x
then cos x = f(0) + (0)x + f f (0)x8 +
1! 2! 3! 4!
where f(0) = cos 0 = 1
PO) = -
sin 0 = 0
f"(0) = -
cos 0 = - 1
f "'(0) = sin 0 = 0
f " "(0) = cos 0 = 1
The series may then be written down as
coax= l - xe
+4T-- -67
x* x° x°
-... (105)
Similarly
sinx= x-3-+5¡-+
x'
9-x° x6 xe
(106)
x°
j x=j(x-3I+51-"')
sin
_xb
=
Also it may be proved that
1 +jx +
j2x3
21
j8x3
+ 3! + 4!
j*x*
+ (107)
logo (1 +x)=x-2
x°
+
x°
-4 +... x*
(108)
0
B d Bwhere n = 1,2,3, etc.
n o
nO dO where n = 0,1,2,3, etc.
etc.
Special cases
(1) F(0) is an even function
If the waveform is symmetrical about the y axis (e.g. Fig. 6.39), F(0) is called an
even function and A = 0, giving the simplified form
F(0) = B, /2 + B cos B + B2 cos 20 + B, cos 3 B + ...
1 (2)
This is the equation which applies to all types of distortion introduced by valves.
(2) F( 0) is an odd function
If the waveform is such that the value of y is equal in magnitude but opposite in
sign for plus and minus values of x (e.g. Fig. 6.40), F( 0) is called an odd function and
B = 0, giving the simplified form :
F(e)-odd function
This is the equation for the condition when the fundamental and all the harmonics
commence together at zero.
(3) F(0) = F(0 + rr)-
If the waveform (Fig. 6.41) is such that the value of y is equal in magnitude but
opposite in sign for x = x1 and x = (x1 f
rr), the expansion contains only odd
harmonics
+... +...
:
8E (cos
B+
cos 3 8+ cos 5 B+ l (10)
y = - ` 9 25 J
y
2E(
=T `sing- --+ sin33 0- sin4 4 0 +...
sin 2 0
FIG. 6.43
r
A
E
x
o
302 (i) PERIODIC WAVES AND THE FOURIER SERIES 6.8
Short rectangular pulse (D)
2 sin 2kor cos 2 B
y =E k + ,r
(sin kor cos B -}
y =
2E 2 cos 2 9
1 + -3-
2 cos 4 b
15 +
2 cos 6 0
35
...
(- 1) "i2+i 2 cos n
n2 -1
B
...) (n even) (14)
n
> 2Tr-
0 4
x
Fig. 6.43
methods for carrying out the calculations have been described. Some are based on
equal divisions of time (or angle) while others are on equal divisions of voltage.
In the harmonic analysis of the distortion introduced by valves on resistive
loads, it is possible to make use of certain properties which simplify the calculations :
FIG. 6.44
(1) All such distortion gives a waveform which is symmetrical on either side of the
vertical lines (ordinates) at the positive and negative peaks.
(2) It is therefore only necessary to analyse over half the cycle, from one positive
peak to the following negative peak, or vice versa.
(3) Even harmonic distortion results in positive and negative half cycles of different
shape and area, thus causing a steady (" rectified ") component.
(4) Odd harmonic distortion results in distorted waveform, but with the positive
and negative half cycles similar in shape.
(5) Even harmonics are in phase with the positive fundamental peak, and out of
phase with the negative peak, or vice versa ; they are always maxima when the funda-
mental is zero.
(6) Odd harmonics are always exactly in phase or 180° out of phase with both
positive and negative fundamental peaks, and are zero when the fundamental is zero.
The relative phases of the fundamental (H1) and the harmonics (up to H5) are shown
in Fig. 6.45. The fundamental and third, fifth and higher order odd harmonics
have zero amplitude at 0 , 180' and 360° on the fundamental scale. The second,
FIG. 6.45
l i 1 l 1 1 1 1 1 1 1 1 1 I -
1 1 1 1 l 1 1 l 1
1 1 i 1
ASAngle
L
iiiìíì 91d'
1 6° I50°
135°I
áá
oó ó ó
112°
fourth, and higher order even harmonics reach their maximum values (either positive
180°
O
I
I
H6-
210° 234°
199°
li
225
I
Ó
i-
m
27CP
An 9 It
Dyn°mic
41.
Curve
or negative) at 0', 90 ", 180 °, 270 and 3600 on the fundamental scale.
304 (iii) GRAPHICAL HARMONIC ANALYSIS 6.8
The amplitudes of the harmonics as drawn in Fig. 6.45 have been exaggerated for
convenience in drawing, while their relative magnitudes are quite arbitrary. Their
relative phases are, however, quite definite.
In proceeding with Graphical Harmonic Analysis it may be shown that it is possible
to select thirteen points on the X axis which will enable the exact values of the first,
second, third and fourth harmonics to be calculated (Ref. C9) on the assumption that
there are no harmonics of higher order than the fifth, or that these are negligibly small.
These points are limited to the range from 90' to 270' on the fundamental scale, as in
Fig. 6.45. They may be expressed in terms of the grid voltage E,, the static operating
point being E, and the operating point swinging from E. = 0 on the one side
to 2E, on the other side.
It is only necessary to determine the plate currents at the specified grid voltages,
to insert these into the formulae given in the article and to calculate the. values of the
harmonics.
The preceding exact method has been approximated by R.C.A. (Ref. C10) to give
greater ease in handling. In the approximation there are eleven specified points
in place of 13 in the exact form, the values of the grid voltages being (see Fig. 6.45) :
0 ; -0.191E, ; -0.293E, ; -0.5E, ; -0.691E, ; -E, ; -
1.309E, ; -1.5E, ;
-1.707E, ; - 1.809E, ; -2.0E,.
These have been approximated by R.C.A. to the nearest decimal point, and the
approximate values have been used in the " eleven selected ordinate method " of
Chapter 13 Sect. 3(iv)D and Fig. 13.24.
The equation giving the second harmonic distortion eqn. (28) in Chapter 13 Sect.
3(iv) -only requires the values of plate current at three points. This is an exact
form and is used for triodes in Chapter 13 Sect. 2(i) eqns. (6) to (7b) inclusive and
Fig. 13.2.
The " five selected ordinate method," described in. Chapter 13 Sect. 3(iv)A
and used for calculating second and third harmonic distortion in pentodes, is exact
provided that there is no harmonic higher than the third. It is, however, a very close
approximation under all normal conditions. The same remarks also apply to the
simple method for calculating third harmonic distortion in balanced push -
pull amplifiers, described in Chapter 13 Sect. 5(iii) eqn. (23) and Fig. 13.37.
An alternative method, based on equal grid voltage divisions, has been devised
by Espley (Ref C6). This gives harmonics up to one less than the number of voltage
points. Two applications are described in Chapter 13 Sect. 3(iv) -five ordinates
giving second, third and fourth harmonic distortion, and seven ordinates giving
up to sixth harmonic distortion.
When the loadline is a closed loop, as occurs with a partially reactive load, these
conditions and equations do not apply, or are only approximated.
SECTION 9 : REFERENCES
(A) HELPFUL TEXTBOOKS ON MATHEMATICS FOR RADIO
Elementary
Sawyer, W. W. "Mathematician's Delight" (a Pelican Book, A121, published by Penguin Books,
England and U.S.A. 1943). Possibly the best introduction to mathematics.
Cooke, N. M. and J. B. Orleans, " Mathematics essential to electricity and radio " (McGraw -Hill,
New York and London, 19431. Highly recommended for general use. 418 pages.
Everitt, W. L. (Editor) " Fundamentals of radio " (Prentice-Hall Inc. New York. 1943). Chapter 1 only.
" Radio Handbook Supplement " (The Incorporated Radio Society of Great Britain, London, 2nd ed.,
1942). Chapters 2, 9.
" Radio Handbook " (10th edit., Editors and Engineers, Los Angeles, California). Chapter 28. (Not
11th edition.)
Basic :
Colebrook, F. M. " Basic Mathematics for Radio Students " (Ilife and Sons Ltd., London, 1946).
Complete textbooks, commencing from elementary level :
Dull, R. W. " Mathematics for Engineers " (McGraw -Hill Book Co., New York and London, 2nd edit.
1941). Covers whole ground. 780 pages.
Cooke, N. M. " Mathematics for Electricians and Radiomen " (McGraw -Hill Book Coy., New York
and London, 1942). Generally at lower level than R. W. Dull and less comprehensive. Useful
for those with limited mathematical background. 604 pages.
6.9 REFERENCES 305
Smith, Carl E. " Applied Mathematics for Radio and Communication Engineers " (McGraw-Hill Book
Coy., New York and London, 1945). Excellent treatment in limited space. 336 pages.
Wang, T. J. " Mathematics of Radio Communications " (D. van Nostrand Coy., New York, 1943).
371 pages.
Rose, W. N. " Mathematics for Engineers " Parts 1 and 2 (Chapman and Hall Ltd., London, 2nd
edition 1920).
Textbooks commencing at higher level:
Warren, A. G. " Mathematics Applied to Electrical Engineering " (Chapman and Hall, London, 1939).
384 pages.
Sokolnikoff, I. S. and E. S. " Higher Mathematics for Engineers and Physicists " (McGraw-Hill Book
Coy., New York and London, 2nd edit., 1941). 587 pages.
Toft, L., and A. D. D. McKay " Practical Mathematics " (Sir Isaac Pitman and Sons Ltd., London,
2nd edit., 1942). 612 pages.
Jaeger, J. C. " An Introduction to Applied Mathematics " (Oxford University Press).
Limited application :
Sturley, K. R. " Radio Receiver Design " Part 1 (Chapman and Hall, London, 1943). Appendix IA--
j notation ; 2A- Fourier Series.
Golding, E. W., " Electrical Measurements and Measuring Instruments " (Sir Isaac Pitman and Sons
Ltd., London, 3rd edit., 1944). Chapter 15-Wave forms and their determination.
Lawrence, R. W. " Principles of Alternating Currents " (McGraw -Hill Book Coy., New York and
London, 2nd ed., 1935). Chapter 4-Non -sinusoidal waves.
Eshbach, O. W. " Handbook of Engineering Fundamentals " (John Wiley and Sons, New York ; Chap-
man and Hall, London, 1944). Section 1- Mathematical and Physical Tables : Section 2-
Mathematics.
Also other handbooks.
NEGATIVE FEEDBACK
By F. LANGFORD- SMITH, B.Sc., B.E.
Section Page
1. Fundamental types of feedback 306
2. Practical feedback circuits ... 316
3. Stability and maximal flatness 356
4. Special applications of feedback 389
5. Valve characteristics and feedback 390
6. References ... 399
7. Overloading of feedback amplifiers ... ... 1475
EÉ AMPLIFIER
AMPLIFICATION+A
Ée LOAD
RL
Fig. 7.1. Block diagram of an
11- amplifier with feedback.
/iE
FIG.7.1
an amplifying valve into an oscillator (see Chapter 24). Negative feedback is used
mainly in amplifiers, both at radio and audio frequencies, although this Handbook
covers only low frequency applications.
The voltage ßE0 may be proportional to E0, in which case it is described as voltage
feedback ; or it may be proportional to the current through the load, when it is
called current feedback. These two are, of course, identical if the load is a constant
resistance, since the voltage and current are then proportional.
Bridge feedback is a combination of voltage and current feedback.
A network may be inserted in the feedback circuit to change the phase of the fed-
back voltage, or to change its magnitude and /or phase so as to discriminate between
different frequencies.
Use of symbols
Black (Ref. Al) and others use the symbols µß to indicate the same quantity
which is here indicated by A. Some authorities use the symbols Aß, but in certain
equations (e.g those involving effective plate resistance) it is necessary to introduce
ßµ or µß where p. is the amplification factor of the final valve. The latter form might
be confused with Black's .ß which has an entirely different significance. Hence the
use of ßA in this chapter. A further, but minor, advantage from the use of ßA is
that the beginner, in considering operation at .the mid- frequency, is able to regard the
magnitude for ß as being simply a fraction.
Eol
-4
FIG. 7.2A FIG. 7. 2B
LOAD
AMPLIFICATION .+ A
1}
FIG. 7.2C FIG. 7.20
The magnitude of ß is equal to R, /(R1 + R2). In all cases, for negative feedback,
ßA must be negative and the feedback network must be arranged to give this result.
An alternative method of applying the feedback voltage to produce degeneration
(Fig. 7.2C) connects the feedback voltage in shunt with the input voltage. In
this case the amplification must be negative, that is to say there must be an odd number
of stages unless a transformer is incorporated (Ref. A17), and ß must therefore be
positive.
A still further variation is when the feedback voltage is applied to the cathode
of one of the stages in the amplifier (Fig. 7.2D). In this case the amplification
must be positive, for example an even number of r.c.c. stages with the feedback
returned to the cathode of the first stage, which must have some impedance between
cathode and earth. Alternatively the feedback may be returned to the cathode of the
penultimate stage of a multi -stage r.c.c. amplifier. The value of ß should be con-
sidered as negative on account of the method of connection of the feedback loop.
The following treatment is based on the circuit of Fig. 7.2B but the results may
also be applied to any amplifier with the feedback applied externally to the amplifier
and in series with the input voltage. It is understood that in all cases the polarity
of the feedback voltage is arranged to give negative feedback.
E, - ßA 1 (12)
Thus the feedback reduces noise or hum voltages introduced by the amplifier in
the same proportion that the gain is reduced, provided the following conditions are
fulfilled :
(1) That the signal input voltage is increased with feedback, to maintain the
signal output voltage constant.
(2) That ß and A have the same values for the hum (or noise) frequency as for
the signal frequency.
(3)The voltage fed back does not include any voltages other than the fraction
of the amplifier output voltage. In many circuit arrangements this con-
dition is not satisfied ; the most important of these are covered in Sect. 2.
When all noise is considered as originating at the input (e.g. thermal noise), the
signal-to -noise ratio is unchanged by the feedback.
(E) Effect of voltage feedback on the output resistance of the amplifier
Every linear network which has a pair of output terminals may, as regards its ex-
ternal effects, be replaced by its equivalent open-circuit voltage in series with its
equivalent resistance.
The following treatment is based on a single stage, but may be applied to any ampli-
fier provided that the values of it and r,, are as defined below.
Fig. 7.3A shows the equivalent circuit of the amplifier without feedback :
E, = FLET
E,
\r RL
RL)
(13)
A = E; -
rD µRL
+ RL (14a)
where µ = amplification factor (i.e. amplification with an infinite load resistance)
and r, = plate resistance of final stage.
If the amplifier has more than one stage kt must be defined by
(14b)
where At = amplification between input and grid of final stage (with due regard to
phase reversal)
and µz = amplification factor of final stage.
Fig. 7.3B shows the equivalent circuit of the amplifier with voltage feedback : **
RL \
E, = (15)
Rd.
*'Based on B. D. H. Tellegen Inverse feed -back Philips Tec. Rev. 2.10 (Oct. 1937) 289.
7.1 (ii) (E) EFFECT OF FEEDBACK ON OUTPUT RESISTANCE 311
Transformer
Gain IPA ißid r, /r,' impedance
ratio
1/2 1 12.8 1/13.8 1/4
1 /10 9 115 1/116 1 /100
1 /100 99 1270 1/1271 1 /10000
With feedback, the same current will flow, but the input voltage will be
E,' = E,(1 ßA) -
Therefore r,' = E,' /11 = (E, /11)(1
Therefore r,' = r,(1 ßA). - ßA) - (21)
Thus in this case the input resistance is increased by feedback in the same proportion
that the gain is decreased. This applies to all cases in which the feedback voltage
is applied in series with the input voltage, including the current feedback case of
Fig. 7.4, as will be shown later.
On the other hand, if the feedback voltage is applied in shunt with the input voltage
the input resistance will be decreased as proved below for Fig. 7.2C-
The input resistance of the amplifier without feedback is
r; = E: /l1
With feedback, the current I, will be unchanged, but there will be an additional
current Is through R, and of direction such as to increase the current from the source,
which will therefore become (I, + Is). The input voltage applied to the input
terminals of the amplifier will be the same as without feedback
Il
Therefore r,' = r,
1, + le
which is less than r,.
Other special cases are covered in Sect. 2.
F C. 7.4 FIG. 73
Fig. 7.4. Block diagram of an amplifier with negative current feedback. The
resistance Rs is very much smaller than the load impedance.
Fig. 7.5. Block diagram of an amplifier with bridge feedback.
Without feedback but with Rs in position (Fig. 7.4)
E, µRL
A E, r, +Rs +RL (22)
With current feedback,
E, E0 E0 1
(A) the amplification factor with current feedback is the same (µ) as without
feedback
(B) the effective plate resistance with current feedback is
r,' = r, + µRs (25a)
and r,' /r, = 1 + µRs /r, (25b)
The increase in plate resistance (eqn. 25b) is proportionally greater than the decrease
in gain (eqn. 24a).
When the voltage is fed back over more than one stage, the effective plate resistance
with feedback is given by
r,
= r, + A,µ2R,, (25c)
where A, = amplification between input and grid of final stage (with due regard to
phase reversal)
and µ2 = amplification factor of final stage.
The " looking- backwards " output terminal impedance of the amplifier is
equal to r á in parallel with R L,
r'R (r, + µR,)RL
R,, r,
+ RL r, + /AR, + RL
for single stage
or R °' _
+ (r
A, µsRs)R L
for more than one stage.
r, + Aiµ2R2 + R1.
(C) The ratio of amplification with and without feedback is given by
A (µ + l)Rs + r, + RL
(26)
A' Rs + r, + RL
which may be put into the form
A /A' = 1 + A(Rs /RL) = 1 yA. - (27)
(D) Approximations when p. is very large
If µ is very large,
A' ti RL /Rs 1/(- y) (28)
which is independent of the amplifier characteristics.
Also, if µ is very large,
E;' /Rs (29)
which gives an output current which is constant, irrespective ,.f RL.
(E) Input resistance with negative current feedback
In the circuit of Fig. 7.4-
Without feedback
r, = E; / I,
With feedback
E;' E; + (Rs /RL)E° +
r,' - I,
- - E; (123/R 1;44E,
Therefore A' - AE
E
AE,
E; + AE,R1 /(R1 + R2) + R,AE; /RL
A
Therefore A' = 1 (32)
+ AR1 /(R, + R2) + AR, /RL
Comparing this with (2b) and (24a), it will be seen that the denominator in (32)
includes the second terms of the denominators of both (2b) and (24a). Equation (32)
may be put into the alternative forms
µR
r,, + (µ + 1)R, + RL(1L (33)
`g + µR1) /(R1 + R2)
and A' - µ
1 + µR,/(R1 + R2)
RL
r,, + (µ + 1)R, (34)
RL
+ µR1/(R1 + +
Equation (34) indicates that - 1
1 1
which is the same as for voltage feedback
1 + µR1 /(R1
R2) + 1 - ßµ
alone (equation 18).
(2) the effective plate resistance is given by
r9 + (It +1)R, r,, -
(1.t +1)(VRL)
where y =
= 1
- R3/RL.
-A1µ2ß 1 - A1µ2ß (35c)
This arrangement is shown in the simplified block diagram of Fig. 7.6. The effect
of positive feedback on At is to increase its gain from A1 without feedback to A' with
positive feedback. The effect on the whole amplifier, so far as effective plate resist-
ance, distortion in the final stage and gain are concerned, is the same as though A1
with positive feedback were replaced by another amplifier without feedback but with
gain A'. The positive feedback in At will increase the distortion of this stage, but
_
I
EFFECTIVE_
GAIN A'
Fig. 7.6. Simplified block
I
INPUT OUTPUT
diagram illustrating one possible
arrangement of combined posi-
tive and negative feedback.
FIG. 7.6
will decrease the distortion of the whole amplifier provided that the initial distortion
in A1 is small in comparison with that in A2. A general purpose triode may be used
in the first stage, and yet have an effective stage gain greater than that of a pentode
with a plate load resistor of 0.25 megohm.
Practical amplifiers using this principle are described in Sect. 2(xi).
The effect on the input resistance depends on the method of applying the input voltage.
316 (vi) COMPARISON BETWEEN TYPES OF FEEDBACK 7.1
fore frequently referred to as " cathode loading " in distinction to the conventional
" plate loading." As a result of 100 per cent. negative voltage feedback inherent in
a cathode follower, both the distortion and the output impedance may be very low.
The basic circuit of a triode cathode follower is shown in Fig. 7.7. It is understood
that the input circuit must be conductive to provide a grid return path, and that a
suitable grid bias value is achieved by inserting an appropriate battery at, say, point D.
The input voltage is applied between grid and earth while the load resistance (R k)
is connected between cathode and earth. The voltage developed across R k is the out-
put voltage, and the maximum output voltage is slightly greater than for the con-
ventional case with R k as the plate load resistor. The stage gain, however, is neces-
sarily less than 1 since the input voltage is equal to the grid -to- cathode voltage plus
the output voltage (with a medium -mu valve the voltage gain normally approximates
0.9). Since the stage gain is always less than unity, the input voltage is always
greater than the output voltage. Thus the input voltage may be very much greater
than the grid bias, but grid current will not flow until the instantaneous difference
between the input and the output voltage exceeds the bias.* In some cases the input
voltage will be very high as, for example, when a high power output is required from
a cathode- loaded low -mu triode operated with a high plate voltage. In extreme
cases there may be difficulty in obtaining sufficient voltage in the preceding stage to
excite fully the cathode follower.
FIC 7 7
Since a cathode follower does not amplify the input voltage it might, at first sight,
be thought that the arrangement fulfilled no useful purpose. Its principal usefulness,
however, lies in its impedance characteristics. The input impedance is high while
the output impedance (R a') is low, and the whole device may therefore be regarded
as a kind of Impedance Transformer -see (F1) below.
The low output impedance of the cathode follower makes it particularly useful as a
driver stage for Class AB, or Class B amplifiers, or for pulse techniques where a low
impedance source is required.
With the cathode follower, the grid and cathode are in phase with one another, so
that there is no reversal of polarity as with plate-loaded amplifiers -this characteristic
is sometimes useful in amplifiers, and particularly with pulse techniques.
A cathode follower will work into practically any value of load resistance, although
there is a broad optimum value for any particular application.
The effect of interelectrode and wiring capacitances on the gain is generally negli-
gible up to a frequency of 1 Mc /s.
It is important to remember that all electrode voltages must be stated with respect
to the cathode ; these will normally have the same values as for plate loading.
The graphical treatment is given on pages 390 -394.
More precisely, the difference between the bias and the voltage of the grid current cross -over point.
318 (i) THE CATHODE FOLLOWER 7.2
Ifµ is
9
For 250 volts between plate and cathode, µ = 20, = 1900 micromhos and
r, = 10 500 ohms. The value of R k for greatest power output may be assumed
to be approximately equal to 2r
so that a value of 20 000 ohms may be adopted
(see Sect. 5 for graphical treatment). The effective plate resistance is therefore
r,' = r, /(µ + 1) = 10 500/21 = 500 ohms,
and the gain will be
See Sect. 5(i) for graphical treatment.
7.2 (i) THE CATHODE FOLLOWER 319
20 000
A' (21/20)(20 000) + 525 - 0.93.
Example 2 : type 6V6 (triode connection)
With 250 volts between plate and cathode, µ = 9.6, g,, _ 4000 micromhos and
r = 2400 ohms. If R k is made equal to 5000 ohms we have
D
R, = -
and the optimum value of bias resistor (Fig. 7.10) is given by
gm(Ebb
Ebb
2Ek,.) -
If Ex, is small compared with Ebb, R, is thus approximately equal to 1 /gm.
( 4h )
.
1 /gm = r,, /,u, the maximum peak output voltage will be given approximately by
Ekm
Special case (5)
11E,,bR /r,, (41)
If R k is small compared with /gm and R, alone used to provide bias, the bias will
1
not be optimum and the maximum peak output voltage will be given by
µR b un2Ebb
Ekm (4m)
r,,[r,, + (µ + 1)R k7 (µ + 1)n + 1
where Rk is neglected in comparison with and n = Rk/ro. r
The ratio of the output voltage with correct bias (eqn. 41) and that with self bias
from cathode resistor (eqn. 4m) is given by
Ratio - 1 +
2
1 /µn (4n)
(Fl) The " looking -backwards " output terminal impedance (R o')
The impedance, seen when looking backwards from the output terminals of Fig. 7.7
into the amplifier, is equal to r,, in parallel with R k. This corresponds to the
plate -loaded case where we have to consider r,, and R L in parallel.
r,'Rk Rk 1
Ro'- + Rk (rn + Rk)/r + gmRk g, +- 1/Rk + 1/rv (5a)
similar case with plate -loading, it is usually desirable to have R, considerably greater
than R k. R, may take the form of a volume control.
Alternatively, R k may be selected to suit the grid bias, while R, is then calculated
to provide the desired total a.c. load.
When the circuit of Fig. 7.8 is used as a power amplifier stage, it may be "shown
(Ref. C32) that maximum power is dissipated in R, when R, = r, and Rk = y 2r
and that when these conditions are fulfilled the power dissipated in R, is given by
P,,,,,, = E,2/93r,
where E1 = plate supply voltage E bb less the voltage where the tangent to the
E,, = 0 characteristic cuts the axis (point B in Fig. 7.8A)
and r, = plate resistance of valve at junction of d.c. loadline with E,, = 0
characteristic (point D in Fig. 7.8A).
If in Fig. 7.8 the load impedance is re-
moved to a distance and the leads between
the cathode follower and the output ter -
minais are lengthened, the " link " be- A.c. IOADUNE
tween the amplifier and the output may t
b
o
be used as a low impedance loaded a -f line, W`
although it is loaded unequally at the two
ends. `q oc LawuNE
FIG. 7-9 7
The maximum power output with the same load resistance as for plate loading will
be approximately the same. The distortion and the output resistance will, however,
be very much lower than for plate loading, and the damping on the loudspeaker will
be greater.
A special case of some importance is when V, is the driver valve of a Class B power
amplifier (see Chapter 13 Sect. 7).
The optimum load resistance may best be calculated graphically (see Sect. 5).
The resistance of the transformer primary will provide at least part of the voltage
for grid bias ; the balance may be supplied by a suitable resistance (shunted by the
322 (i) THE CATHODE FOLLOWER 7.2
usual by-pass capacitance) between the cathode and the upper transformer primary
terminal.
Alternatively, the primary of the transformer may be connected in place of R,
in Fig. 7.8, and C, may be adjusted to tune the primary to a suitable low frequency
to extend the frequency band to lower frequencies or to give bass boost (Ref. C14).
(H) Grid circuit arrangements and their effect on input resistance and R0'
If the voltage drop in R k is greater than the desired grid bias, then a convenient
arrangement is to return the grid resistor (R0) to a tapping point (X) on the cathode
load resistor, the position of X being adjusted for optimum operating conditions
(Fig. 7.10) ; the bias will be
E,, = R,10 (7)
where E,, is the bias in volts, R2 is expressed in ohms and Ib the plate current in
amperes. The total load resistance is (R, + R2) which corresponds to R k in Fig.
7.7. The correct value of C should always be selected on the basis of the input re-
sistance (R,') and not on the basis of R,.
Fig. 7.10. Cathode follower with grid bias supplied by a tapping point (X) on the
load resistance.
Fig. 7.11. Cathode follower with separate resistances for bias (R2 by- passed by
C2) and load (R,).
Fig. 7.12. Cathode follower with alternative form of grid bias circuit.
The circuit of Fig. 7.10 does not apply the full degree of feedback (as in Fig. 7.8)
unless the output resistance of the preceding stage is very much smaller than R,.
In other cases the feedback is reduced by a factor
r R t /(r, + R,,) for the preceding stage.
R
/(R, + R,) where Ro =
The input resistance (R;') due to the grid resistor (R, in Fig. 7.10) is greater than
R, because the voltage across R, is less than
Thus R,' = E; /i
where E; = the input voltage
and i = the current through R, (Fig. 7.10).
Now i = (E; -
E,) /R,
and E = A'E;R, /(R, R2).
E; E,R, E;R,
Therefore R,'
E, -E
R,
E,[1 -
A'R1/(R, + R2)]
Therefore R,' _
Special cases
1 - A'R,/(R1 + R2) (8)
:
If R2 =
If R, =
0, i.e. R, returned to cathode
R, returned to earth
: R,' = R, /(1
A') - (9a)
0, i.e. R;' = R,
: (9b)
As a practical example take type 617 (triode connection) with the same conditions
as previously (Example 1). The cathode load resistance (R, + R2) is therefore
20 000 ohms. The cathode current (from the published data) is 6.5 mA while the
grid bias is -8 volts. The resistance of R2 is therefore 1230 ohms and R, is 18 770
ohms. If R, is taken as 1 megohm the input resistance will be
R,' - -
1
1
1
Ró R, R, l (10)
g n r - R, + R, 1
R, + R2 + R, + R2 + rD
1
Fig. 7.13. Cathode follower with positive grid bias from voltage divider across
B +, in order to counteract the too -great negative bias from Rk.
Fig. 7.14. Direct coupling to the grid of a cathode follower.
Fig. 7.15. Triode cathode follower showing capacitances.
(I) Input capacitance (purely resistive load)
The capacitance from grid to plate (C,,,) is effectively across the input (Fig. 7.15).
The effect of the capacitance from grid to cathode is reduced by the ratio of the voltage
from grid to cathode to the voltage from grid to earth. Thus
Ci' = Con + (1 -
A')C,k (11)
or C,' = Co9 + C,k/(1 + A) (12)
where A has the same meaning as with plate loading, being the stage gain without
feedback, and A' is the actual stage gain, being always less than 1 (eqns. la and lb).
Miller whereby the C, is reduced by the factor 1 /(1 + A).
Note the reversed Effect k
to cathode (CA k) are across the output (Fig. 7.15). The voltage across the capacitance
from grid to cathode (C95) is (1 -
A')e; and the current through Ck is, therefore,
(1 - A') e;. j (C, k. The current which would flow through C, k if connected directly
across R k is A'e1. j oi C, k. Therefore, the effect of C, k connected between grid and
cathode is the same as that of a capacitance of [(1 A') /A'] Cg = C,k/A con-
nected across R k. It is assumed that the source of input voltage has a resistance and
a reactance each negligibly small in comparison with the reactance of C, k. With
this assumption the effect of the capacitance from grid to plate (C,,) is zero, since
it merely shunts the input voltage.
Thus, the output capacitance (Co') under the assumed conditions is given by
Co = C, k + Co k -
C, k/A
This capacitance is effectively shunted across R,' (see eqn. 5).
(13)
In the case of type 6J7 -GT (triode connection), C, k = 17.0 µµF, Ck k = 10.5 µµF
(approx.), C,k = 2.6 µµF, and A' = 0.93 (see Example 1).
Therefore Co' = 17.0 + 10.5 + 2.6 (0.07/0.93) = 27.7 µµF.
This is effectively in shunt with the output resistance (R,') of 487 ohms. In the
plate -loaded case there would be a capacitance of 17.0 µµF shunted across 6900 ohms.
In all practical cases, both for input and output capacitances, it is necessary to make
allowances for stray capacitances in addition to the valve capacitances.
(K) Special considerations with pentodes
The discussion so far has been principally confined to triodes, or pentodes con-
nected as triodes. It is readily seen that if a pentode is operated with the screen volt-
age equal to the plate voltage, without special precautions
to isolate the screen and plate, the result is equivalent to
triode operation. The reason is that there is, in the
cathode circuit, an impedance which is common to both
ó
5 plate and screen currents, so that the signal voltage drop
PIC. 7-16
across it affects the voltages of both plate and screen.
Under these conditions the valve ceases to behave like a
Fig. 7.16. Pentode pentode.
cathode follower with
screen by-passed to If it is desired to obtain pentode operation, the voltage
cathode. between screen and cathode must be maintained steady
and free from signal frequency fluctuations. This may be done by connecting an
impedance (either a voltage dropping resistor or a choke) between the plate and the
screen, and by- passing the screen to the cathode by means of a sufficiently large
capacitance (Fig. 7.16).
The screen dropping resistor R,, is effectively in parallel with R k so far as signal
voltages are concerned. Unless Rp, is more than 20 times greater than R k, allowance
should be made for its effect.
If the by -pass condenser (C) is returned to earth instead of to the cathode the result
is, again, effectively triode operation except that the screen may be operated at a
PLATE voltage lower by a fixed amount than
Ioov- that of the plate. For example, if the
static voltage from cathode to plate is
200V 200 volts, and from cathode to screen
SCREEN
100 volts, then the voltage from screen
120V
to plate is 100 volts (Fig. 7.17). But,
CATHODE J___ _
the cathode voltage has a signal -fre-
Qk ISOV 12(1)V
12 =
'SOV-
-T
quency component
y com 1 which may reach a
peak value of, say, 120 volts (see dia-
gram). This makes the cathode 20
EARTH
volts more positive than the screen, or
F 16..7. 17
this may be expressed the other way
Fig. 7.17. Voltage relationships in pentode around as the screen being 20 volts ne-
cathode follower. gative to the cathode. The result is
7.2 (i) (K) SPECIAL CONSIDERATIONS WITH PENTODES 325
plate -current cut-off and severe distortion, which can only be avoided by a reduced
output voltage. In the case of pentode operation, with the screen by -pass condenser
returned to the cathode (Fig. 7.16) this type of overloading cannot occur.
When valves having high mutual conductance are connected as triodes, it may be
found that parasitic oscillation occurs at a very high frequency. This may be pre-
vented by connecting a 100 to 500 ohm non -inductive resistor in the link between
plate and screen, using short leads.
(L) Voltage gain with pentodes
Owing to the very high values of /.4 with most pentodes, eqn. (2) may be taken as
being almost exact :
gmRk (2)
A' 1 +gmRk
where g,,, = pentode mutual conductance under operating conditions in mhos.
In Example 1, the pentode g m may be taken as being (I ,/1 k) x triode g,,, ; i.e.
0.8 x 1900 = 1520 micrcmhos.
With Rk = 20 000 ohms,
1520 x 20 000 x 10-6
A'
1 + 1520 x 20 000 x 10 -8 -
0.97.
Values of A' up to about 0.99 are practicable with pentodes, using higher values of
load resistance.
(M) Input capacitance with pentodes, screen by- passed to cathode (Fig. 7.16)
The input capacitance with pentode operation is different from that with triode
operation of the same valve. The capacitance from grid to plate may be neglected,
and the total input capacitance is approximately given by
= (1 -
A') X pentode input capacitance (14)
In the case of type 6J7 -GT the pentode input capacitance is 4.6 µµF and with a stage
gain (A') of 0.97 the input capacitance will be 0.03 x 4.6 = 0.14 µµF. This is
less than one fourteenth the input capacitance of the same valve with triode con-
nection under similar conditions (previously shown to be 1.98 µµF).
If the screen is by- passed to earth the input capacitance will be the same as
for triode operation.
(N) Output capacitance with pentodes (screen by- passed to cathode)
With pentode operation the screen and suppressor are effectively at cathode potential
at signal frequency (Fig. 7.16). The " pentode output capacitance " is the capacit-
ance from the plate to all other electrodes except the signal grid and will therefore
take the place of the plate -to- cathode capacitance in a triode (eqn. 13). The output
capacitance is therefore given by
Co' = pentode output capacitance + Ckk
+ (1 /A) x pentode input capacitance (15)
In the case of type 6J7 -GT with a stage gain of 0.97, the pentode output capacitance
is 12 µµF, and the pentode input capacitance 4.6 µµF, so that Co' is 12 + 10.5 + 0.14
= 22.6 µµF. This may be compared with 27.7 µµF for triode operation under
similar conditions.
(0) Circuit to avoid screen current through the
load resistance Z
A ti g,aRk
+ gfrymR
D
+ RL) /r,, (20b)
Ratio of gain with and without feedback
From equation (24a) Section 1,
A' 1
A' 1 1
-
A - 1 - yA µyRL (24)
where -
y = Rk/RL.
r, + RL(1 - y)
Example (1) 6J5 with µ = 20, r,, = 20 000 ohms (under resistance -coupled
:
Input resistance As the grid resistor R, (Fig. 7.20) is returned to earth, the input
:
resistance is R,. If it had been returned to cathode, eqn. (30) Sect. 1 could then be
applied, putting R, in place of r;
Input resistance with R, returned to cathode
:
FIG. 7 22
Fig. 7.22. Cathode degenerative amplifier with a small degree of current feedback.
Fig. 7.23. Cathode degenerative amplifier with a large degree of current feedback.
Fig. 7.24. Cathode degenerative amplifier with an alternative method of obtaining
grid bias.
If a higher degree of feedback is desired there are several alternatives. In Fig. 7.23
(R, + R2) = R for signal frequencies, but only R2 is effective in producing grid
bias. C, should have a reactance much lower than (Rs + R,). As a slight modi-
fication, C, may be returned to any tapping point along R, and 1222 to give any desired
degree of feedback between zero and maximum. An alternative modification is to
by -pass R2, and to return C, to any tapping point along R,. Finally, the arrangement
of Fig. 7.24 provides a positive voltage from the voltage divider R
R, which may be
adjusted to give the correct bias voltage from grid to cathode.
When RL and R k are not purely resistive, they may be replaced by ZL and Zk in
the expressions for gain which then become vectors.
Current feedback is undesirable in transformer loaded amplifiers for it tends to
stabilize the output transformer's magnetizing current (i.e. make it sinusoidal) and
thus produce a distorted output voltage.
B+
(B) Phase splitter
If Rh is made equal to RL, we have the well known
" phase splitter " (Fig. 7.25), in which the output
from P to earth is equal and opposite in phase to that
from K to earth. Obviously (R, + R,) = R h = RL,
r
and the same signal current (I,) passes through R L,
R, and R,. The grid resistor R, is returned to a
point on Rh to give the desired grid bias (E, _ -
R21,).
FIG. 7 2S
The gain to either output channel is then
Fig. 7.25. Conventional tiRt
phase splitter.
A
- (u + 2)RL + r,
(28)
The capacitive component of Zp is the sum of the input capacitance of the following
stage, wiring and stray capacitances, and the output capacitance of P channel (eqn. 33),
and similarly with ZK (eqn. 34). The output capacitances of the phase splitter norm-
ally differ by only 1 or 2 µµF, having a negligible effect on the balance at 10 000 c
see also pages 522 -523.
/s-
Input capacitance (Fig. 7.25)
-
C¡ = C,k /(1 A') + Cov(1 + A') (32)
Output capacitance
P channel : Co' = 2C9k + C (1 + 1 /A') (33)
K channel : Co' = 2C,k + C,k(1/A' 1) + CT, -
The equivalent source impedance that determines the high frequency attenuation
(34)
in that the voltage across the voice coil may not be great enough to provide sufficient
degeneration. It is important to avoid accidental reversal of the secondary terminals.
This, and all its modifications, may be treated as pure voltage feedback following the
formulae of Sect. 1, at least in the middle frequency range where ß is equal to T, /T
the transformer turns or voltage ratio. It has the properties of maintaining constant
the voltage across the secondary, thereby avoiding the effects of transformer leakage
inductance at high frequencies, and of reducing the transformer distortion. Un-
fortunately, at very low and very high audio frequencies, the phase angle introduced
by the output transformer tends to cause instability, the tendency being more pro-
nounced as the feedback is applied over 2 or 3 stages. This subject is treated in detail
in Section 3. a,
FIG. 7-26
The push -pull form is merely a mirror image of the single -ended variety, and
involves two separate secondary windings on the input transformer, and a centre -tap
on the secondary of the output transformer, which is earthed.
In either case, if the feedback voltage is greater than desired, the connection may be
taken from a voltage divider across the voice coil (or each half of the transformer
secondary).
FIG.7 27 FIG 7 2$
avoids d.c. through the secondary winding, but the impedance of the preceding stage
(R 0) affects the value of ß, A being unchanged,
R, T,l Ro
(37a)
P
R, + R2/ T,/ Ro
In this case the input resistance is decreased because the feedback voltage is applied
r
in shunt from grid to cathode,
= R, /(1 ßA).-
This principle is also incorporated in Fig. 28.3 where feedback is taken from the
(37b)
INPUT
This arrangement reduces the amplitude distortion resulting from saturation of the
output transformer core at low frequencies, and improves the low- frequency response.
It does not, however, counteract the effect of leakage reactance at high frequencies.
A = voltage gain of valve from grid to plate
ß = R, /(R, + R,)
and r{ is extremely high, being unchanged by the feedback (valve input impedance
alone).
The same method may be applied to a push-pull stage, but a transformer is required
having two separate secondaries, and each valve must have its own feedback network.
This circuit is not suitable for use with resistance coupling.
(v) Voltage feedback from plate-r.c.c. input
If itis desired to employ negative feedback from the plate to a resistance -capacitance
input coupling, it is impossible to apply the feedback voltage in series opposition to
the signal voltage. The only manner of applying it is, so to speak, in shunt with the
input signal.
The simplest circuit is Fig. 7.30 in which V, is r.c. coupled to V2, and a feedback
path R,C, is provided from plate to grid of V2. For normal applications C, is merely
a blocking condenser to isolate grid and plate for d.c. The resistance R, acts as a
shunt path for signal frequencies in much the same way as the Miller Effect ; the
circuit behaves as though an additional shunt resistance equal to R, /(A2 + 1) were
connected from grid to earth, A, being the numerical voltage gain of V2. This de-
creased input resistance causes a reduction in overall amplification. The amplifica-
tion of V, is not affected, but V, now has a heavy a.c. shunt load, leading to increased
distortion and possible overloading. As an example, for the gain to be reduced to
7.2 (v) VOLTAGE FEEDBACK FROM PLATE 333
half, the additional a.c. shunt resistance across R, will be equal to the total resistance
of r,,, RL and R, in parallel. A larger than normal value of coupling condenser (C)
will obviously be required.
As a modification of Fig. 7.30, R, may be made zero and C, may have a capacitance
of about 10 to 100 µµF to provide degeneration principally at high audio frequencies.
This has been used very successfully in small receivers with very little margin of gain,
but is only fully effective when no plate by -pass capacitor is used.
An improved circuit is Fig. 7.31 (Refs. El, E3) in which R, and R2 form a voltage
divider across the primary of the output transformer, RL being connected to the
junction (X). If (R, + R_) is at least 10 times the load resistance of V,, the divider
will have little effect on V,. It is obvious that the degree of feedback can be adjusted
from zero to maximum by moving the tapping point X on the voltage divider from
the B+ end to the plate end. If R, + RL >Rwhere R, = r,,R, /(r,, + R,),
(39)
R, R,
then ß - R, + RL R, + /23 (38)
(
38
An electrically equivalent circuit is Fig. 7.32 (Ref. E2) in which the two resistances
Ra and R, take the place of the three resistances RL, R, and Ra of Fig. 7.31. For
identical operating conditions :
Fig. 7.31. Voltage feedback applied in series with the load resistor. The reduction
in effective resistance occurs in the load resistor itself.
Fig. 7.32. Electrically equivalent circuit to Fig. 7.31, using one less resistor.
Both these circuits (Fig. 7.31 and 7.32) have the effect of reducing the effective
load into which V, works. This has the merit of extending the response to higher
audio frequencies. In Fig. 7.31 the effective value of RL becomes
RL' = RL /(IPIA, + 1) where IßI = R, /(R, + R,)
and Aa is the numerical voltage gain of V,. For example if If! = 0.1 and A, = 17,
the effective load resistance changes from RL without feedback to 0.37 RL with feed-
back. The load Rí into which V, works is, however, not constant because A, varies
due to distortion. This circuit is usually limited to values of IfI not greater than say
0.05 to 0.1 for typical applications.
A similar effect occurs with the equivalent circuit Fig. 7.32 in which the feedback
causes the effective value of R, to change to R, /(AZ + 1). In both circuits V, is
preferably a pentode, although a high -mu triode may be used with less effectiveness.
334 (vi) VOLTAGE FEEDBACK OVER TWO STAGES 7.2
Mezger (Ref. E4) has shown that it is equally valid to regard the plate resistance of
V, (and consequently its amplification factor) as being reduced by feedback, leaving
the external resistors unchanged.
FIG. 7.33
FIG. 7.34
Fig. 7.33. Amplifier with voltage feedback from the plate of V, to the cathode of
V also incidentally incorporating negative current feedback due to the unbypassed
resistor Rk.
Fig. 7.34. Simplified a.c. equivalent of Fig. 7.33 for calculations.
Also E, = A1A,E,1 = R +
+ RR rv1
A2E"
See Fig. 35.3 for one possible arrangement.
tThe Editor is indebted to Mr. E. Watkinson for this method.
7.2 (vi) VOLTAGE FEEDBACK OVER TWO STAGES 335
µ1R ll (( Et 11J
Aa
R + Rk + r,l/ \ 1 + ßAlAa + µ1R k/(R + Rk + r,l)/
-/ R + Rk + µ1R
r91 / \ R+Rk+r,1+µlRk+ßA1A,(R+Rk+r1)j
`gaEí
AR -R k+r,1) 1
l
- E, r(R + (µ1 +
AaFhR
1)Rk + r,l + ßA1A,(R + Rk + r,1)/
Aaµ1R
Therefore E° = E¡
(1 + µ1 + MIAs) + R
+ d [1 + ßA1,42]
(41 1)
` 1 + ßA1A2 / r
44214?
Therefore `q (R +r1
9
A j[Rk(1 +µ1 +ßAlAal +R +r,l] +ß441A,] \Aaµ1R/
1 +ßA1A, / [1
Therefore
A' R + r91
A
[R
k(1 + µ1 + ßAlAa)
+ R + r,1] [1 +. ßA1As]
(41m)
``
L 1 + 13A1Aa f
Output impedance
The effect of the unbypassed cathode resistor depends on the resistance of the
source (R,).
Let = R, /(R, + R,)
8 (42a)
Let voltage E. be applied across RL, from an infinite impedance source with no
input voltage applied to the amplifier.
Eo
Then l° = Eo RLa
Eo E0ß8A1 o,
(42b)
+ rza + Rr + Rk
The final term in eqn. (42b) is the effect of the voltage E a fraction of which is °,
fed back across R k, and amplified by V1 and applied to the grid of Va. The effect
in the plate circuit of Va is the same as though a voltage E 0f8A1µ,;were applied to a
series connection of r,, and. RL2.
Thus the " looking backwards " output terminal impedance is given by
E° 1
R, I° 1 1 1 ßS441µ2 (42c)
RLa +ra +R, +Rk +r,a +RL,
i.e. the effect of the feedback is to add another impedance,
r R
ß8A1µ,
in parallel with the output, as shown in. Fig. 7.41.
Input impedance
AIR k
E( E°1 [l + R + R k+ r1 + ßA1A:] (42d)
Ri
Ir' Eal/Rol
F+1R k (42e)
Therefore Ri' = R,[1 -F
Rk r91
ßA1A2
336 (vi) VOLTAGE FEEDBACK OVER TWO STAGES 7.2
Conclusions
Eqn. (41m) shows that the effect of feedback in reducing gain in this circuit (Figs.
7.33 and 7.34) differs from that calculated from considering the effects of voltage and
current feedback separately.
Eqn. (4M) indicates that the two feedback voltages are effectively added so far as
their effect on gain is concerned. This is in line with Duerdoth's multiple feedback
theory in Sect. 3(v)E.
Eqn. (42c) shows that the effective plate resistance of V2 differs from that of an
equivalent amplifier without current feedback only on account of 8. When the
source impedance is zero, I has the value unity, and the effective plate resistance of
V2 is identical with that of an equivalent amplifier with voltage feedback only.
It is obvious that current feedback in V, cannot directly affect the plate resistance
of V2. The only indirect effect (through I) is the result of partial application of the
feedback voltage to the grid of V which must also affect the gain. In other words,
the impedance of the input voltage source affects the gain, the effective plate resistance
and the output resistance Ro'.
If the source is reactive, some non -linear frequency characteristic will occur
for example, if the source is a crystal pickup there will be some measure of bass
-
boosting.
The circuit of Fig. 7.33 tends to produce peaks at low and high frequencies in the
response characteristic unless the feedback factor is low. An analysis of the peaks
has been made by Everest and Johnston (Ref. H5) based on the ideal condition of
zero -impedance screen circuits and unbypassed cathode resistors ; it also omits any
complications arising from an output transformer. Methods for reducing or elimina-
ting these peaks are described in Sect. 3(vi) and (vii).
The condenser C, in the feedback network of Fig. 7.33 is normally only a blocking
condenser ; it may, however, be designed to produce a fixed degree of bass boost.
Alternatively, C, may remain as a blocking condenser and a second condenser inserted
at point X to provide bass boosting ; the latter may be shunted by a variable resistance
to form a continuously variable tone control (see Chapter 15 Sect. 2 and Fig. 15.11).
It will be demonstrated in Sect. 3 that feedback over two stages of r.c. coupling is
normally stable ; this does not necessarily hold if cathode or screen by -pass condensers
are used or if an iron -cored transformer forms any part of the feedback loop.*
Similarly the reactance of the output transformer primary at 50 c/s is equal to the
plate resistance of the 6AQ5 in parallel with the reflected load and winding resistances
(coupling 3).
COUPLING COUPLING 2 COUPLING 3
6AV6 6AQ5
O-0051+F 0-0INF 5,000/1: 1.711
Fig. 7.35. Simple feed-
back amplifier in which
setting of volume con-
trol alters feedback.
Movement of the input volume control is assumed to have no effect on the fre-
quency (124 c /s) at which the reactance of the input condenser is equal to the resistance
in series with it, say 2.55 megohms (coupling 1).
The amplification factor, µ, of the 6AV6 is 100 and its plate resistance, r,, 80 000
ohms, so the mid -frequency 6AV6 gain, At [see Chapter 12 Sect. 2(vii)] is
100 x 73 000
A, 48
80 000 + 73 000
where 73 000 ohms is the value of R8 and R, in parallel.
The 6AQ5 µ is 210 and its rD 52 000 ohms so that the 6AQ5 gain
210 x 5000
242 18.4.
52 000 + 5000
The output transformer impedance ratio is 5000/1.7 so that its gain
As = A/1.7/5000 = 1/54.
Total gain from 6AV6 grid to voice coil :
1 -ßA 1+ 16 x 4 9
As the volume control is turned up, a smaller amount of the negative feedback
voltage is applied to the 6AV6 control grid, and when the control is 50 000 ohms
from its maximum setting only one quarter of the previously used feedback voltage is
effective (ignoring the shunting effect of R,). In this case
1 1
1 - ßA 3
The low frequency response curves will be calculated for each of these volume
control settings, but it is first necessary to obtain the response of the amplifier without
feedback. This is done in Table 1. Lines 1 and 2 of Table 1 are obtained from Figs.
12.9A and B with 124 c/s taken as the frequency for 70.7 per cent. voltage gain (when
the reactance is numerically equal to the resistance) in line 1, and 50 c/s in line 2.
Thus for the 25 c/s response in line 1
Actual frequency
Frequency for 70.7 per cent. frequency response 2 -
so that attenuation = 1 db and phase shift = 27 °, and so on for each of the other
frequencies in the two lines.
Line 3 is obtained by adding twice the attenuation and phase shift of line 2 (because
there are two 50 c/s couplings) to line 1.
Line 4 is the result of adding the mid -frequency gain of the amplifier (24.1 db)
to the attenuation and phase shift at each of the tabulated frequencies. This gives
the frequency response as plotted at A in Fig. 7.36.
In line 5 the gain A is expressed numerically, giving the actual gain at each fre-
quency.
Gain expressed in decibels of voltage gain -see Chapter 19 Sect. 1(vi)A.
338 (vi) (B) FEEDBACK FROM OUTPUT TRANSFORMER SECONDARY 7.2
Table 2 sets out the calculations necessary to obtain the modification of the fre-
quency response by the application of feedback when ß = 4. Line 1 results -
from converting the attenuation expressed in db in line 3 of Table 1 to fractions, and
line 2 is obtained by multiplying the mid-frequency gain, the attenuation and phase
(for the frequency concerned) by P. The result is ßA which is plotted in Fig. 7.37.
25
=:eeeE9iiiii
fi::!ls:::I
..11
.......
6:::: ..........
r_-
-51
: :
20
:.......;
..... :..--:::a:::s:: . .
GAIN(dD)
IS
=Mr ï
E u .°
.
:i':.:.- -
::::.
10
::C::....iMail
ai:::-C : u
:::
:::::i:: i
: °. ïouunu.::Gi:I:.
..;:! u
:in.o
..
r.:
uuoumYu..--__
...... '
S
iluil :. ..... =:
: .u::
u. :: nuuu
::--::::::::::°:::
E::?:::cáii:ii:::e::.i::::::.: w:
.:.
D
i::::::::::ii!::::'
'@EEiE.i..iE'--
3::-
-5 ::=iiiiiEljli!!:ï
::..::..
....::::::i::8I:::l:
to
FREQUENCY (CA) FIG. 7.76
Fig. 7.36. Frequency response of amplifier of Fig. 7.35 with zero feedback and
two selected values of feedback.
For the method of plotting this curve and for its significance see Sect. 3 of this Chapter.
It should be noted that once the ßA curve has been obtained the effects of feedback
on the gain and phase shift of the amplifier can readily be obtained graphically as
discussed below. Lines 3 to 9 of Table 2 are the mathematical equivalents of the
measurements mentioned below (1 and 5) which give the degree of degeneration or
regeneration and the phase shift with feedback between output and input voltages
i
at a given frequency.
MID-
p
y : M FREQUENCIES
e._iY6:7.L7._'W1
\arÌ
mac Yt
6I-711\'11
Il
owin/
..
.
' 2
wv' ili ._
10:= Ir
ï
A
200e/.
10,
nas
O
n
O O
Lines 3 and 4 are the values
ti
__
ú in
,:r
o
vo
r- .
rn as
.. for 'the sine and cosine of the
.a J
en
cJn o,
4J ái
M
angle of phase shift [see
Chapter 38 Table 72 for magni-
..
Ñ O tude ; Chapter 6, Sect. 3(iii)
for sign]. Line 5 expresses
\"
o
CO
NJ
o
J
N
J
N
W
J J
N
O 1 - ßA in the form a + jb.
Multiplying the magnitude of
m ..
ci
en
O
..
n
fV
N
d;
-. :..
N. PA in line 2 by the sine of the
associated angle gives the j term
o and multiplying by the cosine
ñ and adding 1 (because of the 1 in
,á J'r, J Jr Jd; J 1 -ßA) gives the real term. For
S example, at 100 c /s, 1 + 6.3 x
-, .., o ..
ch er
.. 0.4848 = 4.05 and 6.3 x 0.8746
o °
O
1"
O O = 5.5 so that 1
j 5.5.
- ßA =4.05+
e .. ..
-
"
©
ce)
NJ
e1;
NJ
u1
NJ
v.
Jr J Q+
Lines 6 and 7 convert 1
to polar co- ordinates again (the
ßA
... pp .-i
CO
., 1(1
change to rectangular co- ordin-
a in in in
ates was necessary to add 1 to
PA), line 6 being expressed in
..
J ..J
G.-
Ñ .-.
" JO JO JO .-+ O.
the form of the square root of the
sum of the squares of the two
.-i r- in
.. T fV
terms in line 5 at the angle whose
°
4 d tangent is the imaginary term
divided by the real term, e.g. at
J ..JM
Gol
in N ...
"
-°1
J NJ J 100 c/s the angle is tan-1 5.5/
' (1 V1 U1 sO
4.05 = tan-' 1.36 = 54° to the
n
en 0 er, o; ..
.0; in
nearest degree. Where negative
terms occur the appropriate
m
.. Ú1
v g.. g.. g.. quadrant is determined from the
ú J Ni J NJ Jr knowledge of the sign of the sine,
O M O cn ot)
cosine and tangent, all of which
in ó ai sa r- ti are known at this stage.
N O O O o Line 8 is the reciprocal of line
j J Ñ b O ,O 7, so that magnitudes are divided
into unity, and angles are re-
é °O ó NJ
.
ob
in
N
±
G
li;
á N
-
result is plotted in curve B of
Fig. 7.36, while the smaller ßA
:á ó N S N Fig. 7.37.
0
do de
p
Úcri C7
i
x
Several interesting aspects
which are common to all feed-
back amplifiers are brought out
v.4 v.4 v+v v.4 v by these calculations and curves.
340 (vi) (B) FEEDBACK FROM OUTPUT TRANSFORMER SECONDARY 7.2
°:
NJ 1p
V
Ñ O o
tr,
`MO Ts N
NI
Ç Ç so
Ó Ó Ó .+ eri Ó
ú Jm sr
ä
sa
ó I
in in
no
p
d
I
d
I
d
I
á
ó
v
0
ái e-:
óo:
o
°.M.
In
°
W et 0,
.+
N +
O N
NJ
N
NJ Ni
Jm
I
é ö
ó I ó I d ó a'
in
n
NJ
u b O
Ñ + ñ ñ °
Ñ J a n + g
M
Ni NJ NJ
1f
NJ
O
C
N
P
O C Ó N
b
I b M
d I
d I d ó v .Mi .-
¡n
7-1
NI
N
u In
..
J ó
° O
+
,o
M
. i g
N
á g13
Ó
A
Ó
ó
Ó
Jr J NJ
l
co
O I
;T.
O I i ì O .i + ..
N
n
g n uno
o J
co
uD
M
O
w+
a.
N
m
ó
Ni
....
4
p
M
I
OP
O
O
I
a0
O
Ó
ñM
Ni
Ñ
O I
Ni
,p
a'
ol
V
J d,
us
°
C
M
a
-
a
J
M
x
Tr
vi
^
+
a' vi
J J NJ
vo n
J
n m v
á I d d a; ? a7
,c; d I u;
to
V N N
Ui ::0
N VJj O. O ;
T
M N J a'
J
a'
á
ó
` I
íNn
d ó r
K+,
>
M
aó d
°
I v;
6
3st NJ 6
J
:
É J q
°+ á+ ó «
I
? áv
o
o . rn > rn d I
. :« 9 á
N
Ao m..' ..
M ° II G
ó i
..
I I I
..
C2
I
..
N 1
..
(5+ in U ^.
q c m a o
7.2 (vi) (B) FEEDBACK FROM OUTPUT TRANSFORMER SECONDARY 341
1.
s6
i
o
u
.M
nT
^ J
N V t^
o
I
p,
8 NI
N Ñ
a0
Q Ñ
¿ I 1 ° p .., d 1
o co
10 N i
ú CO
J o e ° Ó
o
eV
Oy0
- ñN
Ó d d °
`qQ'.
00
d
Óp
00
d . PJ1
..
d 1 1 I
pú
J m
v á $
N N O Ñ
O
° ñ
O
ñ
O
M J,
N
O
O I 1 oá
R,
Ni
:°
s
°t
,i^,
JAJ
CO
..
r. ° ó
N
OD
O ñ
Vi
ó
O ptp
Ç4 Js NJ
Ñ
Jg1
d I d I d - d ai .
N.
7r
u ÿel y
o
. v
t J WS
v
J$
vie r
+ ap
J
I
g + g
,
d
Ç a0
d
O
I
b
d
O
ip
NJ
g
d
O.i
-.
b
NJ
N
N
J
$ a d
J ó + v°+
J$ + 9,
J J Jg
1
á° ó Ñ
° °
v°'i ñ
á m
d I d I
d d
Co
T
\
a
u
o
NI
O
f1
OD
g + pp pp I
N
J Jr Ja
I
J °,° 4 + =
Ki
NJ
m
d I O d .rn»' > vi d 1
Jd
N
.,
ú
J
^
a
d
N
1.,
i
v1
o d
co
O:
°+
ñ
N
^
ó+
ib
O
o
Ñ
NJ
ñ
eV
o
pJ
tñ
d
I
o
á
°D
I
1
o
J
.
; t..
J
O
Ó
6
yF
O
Ni
JN Q^
} \!
+
1+,
O fol
M1
NI
o
o: Ñ
I O .. M, F, O I
.4i
1 41. ÿ 4
...
en li
.+ieÿ.r ...."I.."I..á
N
+
......
u
iñ w iñ
I
Ti
CO.
ñ re
CO. ai C
.°.
...
I
342 (vi) (B) FEEDBACK FROM OUTPUT TRANSFORMER SECONDARY 7.2
(1) The degree of degeneration (or regeneration) at any frequency marked on the
polar diagram Fig. 7.37 can be obtained by measuring the distance at the appropriate
frequency from the curve to the point 1,0. The same scale is used for the measure-
ment as for plotting ßA and the distance gives the amount of degeneration, e.g. the
50 c/s point on the ßA = -
8 curve is 3.66ßA units from 1,0, so gain reduction
= 3.66 times = 11.3 db (compare 11.4 db Table 2 line 9). Again at 20 c /s, distance
= 0.2 units so gain reduction = 0.2 times, i.e. gain increase = 5 times = 14 db (com-
pare 14.1 db Table 2 line 9).
It follows that at whatever frequency the ßA curve crosses the circle with centre
1,0 and radius 1, the gain of the amplifier is unchanged by feedback. For example,
with ßA = - 8, the locus cuts the circle with radius 1 unit at a frequency of 28 c /s,
which is therefore the frequency at which the gain of the amplifier is unchanged by
feedback.
The same result may be obtained from Fig. 7.36 where the intersection of curves
PA = 0 and ßA =
(2) When ßA = --
8 is at 28 c /s. At lower frequencies the feedback is positive.
8 the low frequency peak is relatively much larger than when
ßA = - 2 (14 db when ßA = -
8 and 2f db when ßA = -
2) and it occurs at
a lower frequency. The reason can be seen from the polar diagram. When ßA is
large the curve is increased in size so that lower frequencies on the curve come closer
to the 1,0 point.
(3) When ßA = -
8, the angle of ßA varies from 0° to 197° but the angle of
-
1 - ßA varies from 0° to 113° to 8° (Table 2). The reason can be seen from the
-
polar diagram. Note the rapid 1 ßA phase shift near the regenerative peak and
the reason from the polar diagram.
(4) From the calculations it will be seen that, close to the mid-frequency, phase
shift correction is approximately equal to 1/(1 -
ßA) e.g. when ßA = 8 at 200 -
c/s in Table 2, a phase shift of 32° is reduced to 4 °.
(5) Although the feedback is regenerative as soon as the ßA curve cuts the circle
with centre 1,0 and radius 1, there is phase correction until the ßA curve cuts the
horizontal axis (between 181 and 16 c/s when ßA = 8). -
Mis ::: iGiii
6 111111I®11111111111111<Jia1111 ïàiai .._
111111iuPl111111111111111
%.í111
111SEíl7;ZI
-4
11.
11111
1111.
ou o 11
vu
'\U\'\
111
4111111111
11 C1PI01111
_ \
111111A11111111111141111111111111111111\\111111111111:1111111111\'1111,1111
11/
[ß:
I 1111
E,4
mm,
.111111111111110011111 111111111111111/í
FEEDBACK LOOPS
FIG. 7.40
of V,. In the case of a 3 stage amplifier, the unbypassed cathode resistor could be
in the first or second stage. It is highly desirable, in general, for some such form of
subsidiary feedback to be used. It is normally desirable that the degeneration due
to the subsidiary feedback should not exceed say 10 per cent. of that due to the prin-
cipal feedback loop, since the subsidiary feedback increases the distortion in the
final stage. For further details and more complicated circuits see Sect. 3(v)E.
344 (vi) (D) MULTIPLE FEEDBACK 7.2
There are two general cases of interconnected feedback loops, the first in which
one loop is completely contained within the other (Fig. 7.39) and the second in which
the loops overlap as in Fig. 7.40. In each of the circuits the gain of the valves VV,
and V, is A1, A, and A, respectively, and the size of the resistors R is such as to intro-
duce negligible current feedback.
For the circuit of 7.39 the gain reduction factor is
1
1 - aA,A,A, - ßA,A,
(with a subject to the same convention as ß) and the effective plate resistance of V,
becomes
ra,
r99
- A,A,FL, - ßflaµs
For the circuit of 7.40 the gain reduction factor is
1
«A,A,
1 - - ßA,A,
and the effective plate resistance of V, becomes
r,,
ßAzp-s
1 -
(vii) Voltage feedback over three stages
When the feedback is taken over three stages, there is a strong likelihood of experi-
encing instability at extremely low and high audio (or ultrasonic) frequencies (see
Sect. 3). There are several popular circuits :
FIG. 7.42
Fig. 7.42. 30 watt push -pull amplifier with voltage feedback from one output
plate to the screen of the first stage. Stability is ensured by adjustment of condenser C,.
(A) Feedback from plate of V, to screen of V, (Fig. 7.42)
This circuit is very satisfactory for push-pull amplifiers. As the feedback is taken
from one plate only, the output transformer should have very tight coupling between
the two halves of the primary (i.e. very low leakage reactance). The amplifier and
feedback network have been designed to give a flat response over a wide band of fre-
quencies, with the intention of introducing additional attenuation at one point only.
No instability has been experienced, and practically no rise in response, at low audio
frequencies. On a resistive load, the high frequency response has a slight peak at
about 40 000 c /s, while on a loudspeaker load the peak is much larger and occurs at
a lower frequency. Attenuation is provided by the condenser C, from the plate of
7.2 (vii) VOLTAGE FEEDBACK OVER THREE STAGES 345
the first valve to earth, with the results on frequency response shown in Fig. 7.43,
for resistive loading.
FIG. 7.43
OuTNI
.OmRKr-c.c[[a RO =cOy6
Conditions of test -input frequencies 60 c/s and 2000 c/s (voltage ratio 4:1)
Power output (r.m.s.)* 4 10 14 20 24 watts
Equivalent power ** 5.9 14.7 20.6 29.4 35.4 watts
Intermodulation 2.9 7.7 10 17.5 42%
*r.m.s. sum of two output frequencies
* *r.m.s. sum x 25/17 to give the single frequency power having the same peak
voltage swing (see Chapter 14 Sect. 3).
Any further increase in the degree of feedback is likely to prove difficult (see Sec-
tion 3).
When a circuit is used which applies negative feedback to the screen of a
pentode, the input capacitance is increased by
µtCvlv2 1ß 1A
where µt is the " triode amplification factor "
C, 1v2 is the capacitance from control grid to screen
1ßI is numerical value of ß
and A is voltage gain from grid to the point from which the feedback is returned.
(B) Feedback from secondary of output transformer
This introduces additional phase shift in the amplifier and requires very careful
design of the output transformer for all but small degrees of feedback. The feedback
voltage is usually taken to the cathode of the first valve, but if there is transformer
input to the first stage then it may be returned to the transformer secondary (as Fig.
7.26). This method may be applied equally to both single -ended and push -pull
amplifiers.
346 (vii) (B) FEEDBACK FROM OUTPUT TRANSFORMER SECONDARY 7.2
Fig. 7.44 is the circuit of such an amplifier using push -pull triodes, with extra-
ordinarily low distortion. It is based on the design of Williamson (Refs. F4, F6).
The first triode (V 1) is direct- coupled to the grid of a phase splitter (V2) which in
turn is r.c. coupled to push -pull triodes (V3 and V4) and thence to the output stage.
No by -pass condensers are used and the only reactances to cause phase shift at low
frequencies are the two grid coupling condensers. The circuit has a number of
refinements which are described in the references.
Fig. 7.44. Amplifier employing voltage feedback from the secondary of the output
transformer, with push -pull triodes. The amplifier is virtually distortionless up
to an output of 11 watts, and has a smooth overload up to 16 watts.
The original version used type L63 valves in place of each half of the 6SN7 -GT,
and type KT66 as triodes in the output stage. The circuit shown (A515) gives almost
identical results, but type 807 valves are used in place of type KT66.
The specifications laid down by Williamson for the output transformer are as
follows :
of 16 watts has been reduced ten times and the important leakage inductance existing
between the halves of the primary has been decreased thirty times.
These improvements mean that this transformer, when used in the original William-
son circuit, will give a greater stability margin or alternatively will permit the use of
a greater amount of feedback.
Distortion will be appreciably reduced in any amplifier by the use of a well designed
transformer of such a type.
Distortion with amplifier A515
Power Output H2 H, H, H, Total r.m.s.
11 watts 0.01 0.04 0.01 0.015 0.045%
14 watts 0.075 0.35 0.01 0.26 0.44%
16 watts 0.04 0.92 0.11 0.45 (overload)
The smooth overload is a particularly valuable feature.
Intermodulation distortion
Conditions of test -input frequencies 60 c/s and 2000 c/s ; higher frequency
12 db lower than 60 c/s level.
Power output (r.m.s.)* 4 6 8 10 12 watts
Equivalent power ** 5.9 8.8 11.8 14.7 17.6 watts
Intermodulation 0.17 0.27 0.72 3.7 8.8%
*r.m.s. sum of two output frequencies
* *r.m.s. sum x 25/17 to give the single frequency power having the same peak
voltage swing (see Chapter 14 Sect. 3).
Note :
In the circuit of Fig. 7.44, a capacitance may be inserted in series with the 5000
ohm resistor in the feedback circuit to provide bass boosting for equalizing purposes
in record reproduction ; see Chapter 15 Sect. 9(ii)B.
A modification of this circuit, which provides for bass boosting, is given in Fig.
15.58A.
A new version of the Williamson amplifier, together with pre -amplifier and tone
control (Ref. F9) is given in Figs. 17.35B,C,D,E,F,G. See also Chapter 13 Refs.
F4, F5, H15.
Modified forms of the Williamson amplifier are given in Refs. F6, F8 and Chapter
13 Refs. H5, H6, H8, H10, H12.
(C) Bridge circuits
Amplifiers can be designed with bridge circuits in both input and output circuits,
which prevent any modification of the input and output impedances and which also
eliminate any phase shift round the feedback loop caused by reactances introduced
through the input or output circuits (Refs. Al, A2, Al 1).
EA
=
i( RLµEeR n
+r,I / \l + RL +r,RL ++2(r,,µ +1)Rk / (44)
r,
and ER = i( µE,RL
+ r, )( -
1
RL + 1 )R
) (45)
(45
348 (viii) (A) CATHODE COUPLED PHASE INVERTERS 7.2
The two voltages thus differ by a small amount, which can be reduced by using a
large value of R k and /or by using high mu valves. If the remaining out -of-balance
is still serious, it may be eliminated entirely by making RLl slightly smaller than RLs
so as to give exact balance.
PR
Gain AZ' - 2r, + RL2 + r (,
_ rrL+ RLa) /Rk(µ + 1)
(47)
If µRk >r,, then A,' - µR L2/ (2r + RL,) (48)
In reality, the values of µ and r,
are not constant and not equal for the two units
in the twin triode. A rigorous analysis (Ref. G17) provides a straight forward method
for the accurate derivation of the amplification when the output is taken from B.
Outputs from A and B (special case of phase inverter).
The gain is somewhat less than with Fig. 7.45, but the circuit is often more con-
venient.
There are many applications of this circuit which are described in References
(G) also C7, D6 and J2 ; see also Chapter 12 Sect. 9(iii).
(ix) Hum
The hum in a practical amplifier is not always reduced by the factor 1/(1 ßA)
as in eqn. (12), Sect. 1 but sometimes is decreased according to a different law, and
-
sometimes is even increased by the application of feedback. All hum arising from
sources within the amplifier and which is independent of the signal level, may be
represented by an equivalent hum voltage in series with the input voltage ; this is
reduced in all cases by the factor 1 /(1 - ßA).
(A) Hum originating in the grid circuit of a single stage amplifier will be reduced
by 1 /(1 - ßA).
(B) Hum caused by heater - cathode leakage is reduced by 1 /(1 - ßA) with
7.2 (ix) (B) HUM CAUSED BY HEATER -CATHODE LEAKAGE 349
E o'= 0.54E]. With a pentode the total hum from plate and screen is always de-
creased by current feedback
E no -E L
RL
RL
+ r + Rk(µ + 1) + RL +
µvx>RL
r + (µthe+screen
µvx)Rk J
( 59 )
where the first term is for the plate, and the second for circuit.
[Example Eno' _ (0.04 + 0.97)E = 1.01En for type
: 6176 -G T.]
Fig. 7.49. Transformer- coupled output with feedback from an unbypassed cathode
resistor.
Fig. 7.50. Parallelfed ,transformer- coupled output with feedback from an unby-
passed cathode resistor.
Case (6) Parallel feed, with feedback from an unbypassed cathode resistor
(Fig. 7.50)
With a pentode, the plate " hum is increased by feedback from an unbypassed
cathode resistor, while the " screen " hum is the same as for transformer -coupling
(Eqn. 59) except for the sign.
E no ' r rr
+ Rk(µ + 1)
+ Rk(µ + 1) RL + + (µ +
µv,DRL
r 11 (60)
n LRL +
The first term also applies to a triode. The approximation is from regarding R k
as the load, instead of R k in parallel with a,L.
[Example of triode : Eno' = 0.46E,, for type 2A3_
Example of pentode : E 0' = (0.95 -
0.97)E = -
This gives almost exact hum neutralization.]
References to hum : B3, B4, B5, B7.
Summary -hum originating in the plate and screen supply voltages
With triodes, the output hum voltage is always less than the hum voltage from
the plate supply. In the examples given above, it varies from 46% to 85% of the plate
supply hum voltage with one exception. The exception is transformer-coupled output
with parallel -feed, feedback being taken from the plate, for which the output hum is
only 1.5% of the plate supply hum voltage (Case 3).
With pentodes or beam power amplifiers the output hum may exceed the plate
supply hum voltage, owing to the effect of the screen. This occurs with Case (2)
which is the conventional transformer -coupled output both without feedback and
with voltage feedback from the plate. The only circuit giving very low output hum is
Case (6) which applies to transformer -coupled output with parallel feed and with
feedback from an unbypassed cathode resistor. The second best is Case (1) with
feedback from the secondary, while the third best is Case (3) with parallel feed, having
feedback from the plate.
The examples given above have been tabulated for ease in comparison, including
some cases also without feedback, and some with the screen circuit perfectly filtered.
352 (x) SPECIAL FEATURES OF FEEDBACK AMPLIFIERS 7.2
+310V
e+ +250V
O
TzONF
neNF
F1G.7.51
Fig. 7.51. Two stage amplifier using combination of positive and negative feedback
(Ref. F10).
and C_ since the plate resistance of V, is relatively small, and the input resistance of
the grid of V, is high. The increase in gain due to positive feedback is about 26 db
at 400 c /s. About one tenth of the voltage on the grid of V, is fed back to the grid of
V2. The resistance of R, is therefore made about nine times that of R2, and Cy has
about nine times the capacitance of C,. Because of the highly degenerative nature
of the phase inverter, the balance is not appreciably affected by the additional load
of the positive feedback network.
Some phase shift in the positive feedback is obtained at extreme frequencies in
the stages V, and Vs due to electrode and stray capacitances, and due to the blocking
capacitors. The input capacitance of the grid of V, causes a further phase shift, so
that the polarity of the product /3, A, reverses from positive to negative at extremely
high frequencies, where ß, and A, apply to stage V,. The input capacitance of V,
is primarily Miller Effect due to feedback through its grid-plate capacitance at very
high frequencies where the overall feedback is positive or small.
In some designs it may be necessary to connect a small capacitor from the grid of
V, to earth, or to use a more elaborate phase shift network to obtain a sufficiently
rapid phase turnover in the local feedback.
At extremely low frequencies most of the local feedback current flows through R,
instead of through C,, so that a phase shift is obtained, which together with the phase-
shifting action of the 0.03S.F blocking capacitors in stages V, and V is sufficient to
cause the desired phase reversal. In practice, the phase reversal frequencies are
placed as far outside the desired pass band as good stability permits.
The output transformer is quite small, the core area being only / in. by } in.
The maximum third harmonic distortion at 400 c/s is 0.24 %, higher harmonics
being relatively small, at 8 watts output, measured at the secondary. At 100 c/s
the highest harmonic is the second, with 0.12% for 8 watts output. At 50 c /s, with
5 watts output, third harmonic distortion is 0.88 %, while at 2000 c/s with 4 watts
output the third harmonic is 0.23 %. The intermodulation distortion is 40% with
no feedback, 8% with negative feedback alone and 1.9% with combined positive and
negative feedback under the following conditions -output 8 watts, 4:1 ratio with
frequencies 60 and 7000 c/s. With frequencies of 100 and 7000 c/s the intermodula-
tion distortion is only 0.84%. The output circuit regulation is quoted as 0.1 db at
400 c/s as compared with 2.7 db for positive feedback disconnected, or 19 db with
no feedback (Ref. F10).
354 (xi) COMBINED POSITIVE AND NEGATIVE FEEDBACK 7.2
e+
1TO
PUSH-PULL
Fig. 7.51A. Direct -coupled triode
amplifier V, and phase splitter V2,
¡GRIDS with positive feedback from the
cathode circuit of V2 to the cathode
of V, (Ref. F11).
B-
RIG. 7.SIA
It is possible to omit the by -pass condensers from the cathode bias resistors in the
final and penultimate stages of an amplifier or receiver (or any other two successive
stages) and to bring the gain back to normal by means of positive feedback, using only
one resistor coupling the two cathodes as in Fig. 7.51B.
This circuit (Fig. 7.51B) may also be used with any desired degree of feedback.
The effective plate resistance of V, may be increased or decreased by the positive feed-
back as shown by the following analysis and Fig. 7.51C. Here R, replaces R, and R8
in parallel, no input signal is applied but a signal generator E is inserted in the plate
circuit. It is assumed that R4 is very much greater than R2. We thus obtain
E ¡R3µ2 IR3IßIAiµ2
(61)
RB
where RA = Rs + r5,, + R3
RB =R8 +r + R3(12 + 1)
A, - Rt + ry, + R2
R2
and ITI -R2 +R,
From (61) we may derive
E
I RB - R3IßIA1µ2R A/RE B
R, r 2,2 + R3[µ2 + 1 -
(Iß A, s2R9) /Re]
and hence r32' = rv2 + R3[112 -(IßlAi/s2Ra) /RB]
where IßJ and A have the values defined below eqn. (61). The second term on the
(62)
right hand side of eqn. (62) is the increase in effective plate resistance due to the un-
bypassed cathode resistor of V2, while the third term is the decrease in effective
plate resistance due to positive feedback. It is obvious that, by a suitable choice of
7.2 (xi) COMBINED POSITIVE AND NEGATIVE FEEDBACK 355
13A1, it is possible to obtain any desired value of effective plate resistance from very
high values to very low, even zero or negative values. If the resistor R. is made
variable, the effective plate resistance may be varied over a wide range. The addition
of overall negative feedback will make the gain nearly constant for all settings of R
and will also decrease the distortion and reduce the effective plate resistance to any
values desired (Ref. F17). See also Ref. J19.
The addition of positive feedback will increase the phase shift and may increase
the tendency towards instability at very low and very high frequencies ; care should
be taken to reduce to a minimum the phase shift within the positive feedback loop.
The general principles of designing for stability are the same as for negative feedback
-see Sect. 3.
It is normally inadvisable to return the positive and negative feedback loops to the
same point in the input circuit. If both are taken to the same electrode, a bridge
network may be used for isolating their effects. Each 6 db increase in gain due to
positive feedback will double the distortion in this stage ; for example 24 db increase
in gain will increase the distortion in this stage by 16 times. This will then be reduced
by the negative overall feedback in accordance with the usual relationship. Refer-
ences to positive and negative feedback -Refs. F10, F11, F12, F13, F15, F16, F17 ;
Sect. 3(v)E.
Fig. 7.51C. Analysis of circuit of Fig. 7.51B to determine the effective plate
resistance of V2.
Fig. 7.51D. Choke coupled phase inverter (Ref. F14).
the input circuit, and thus causes s low effective plate resistance. Design must aim
at a high total inductance to keep phase errors small, and a low leakage inductance
if it is desired to keep the effect of grid current small. A fixed bias voltage is applied
between terminals X, having polarity and magnitude to give, in conjunction with the
voltage drop in the cathode choke, the correct grid bias. The complete analysis
is given in Ref. F14.
The voltage gain is given approximately by
A' µR2
(µ + 2)R2 + r,,
where X > R2
and the phase shift between input and output is given exactly by
r, R
tan ¢ _ (µ
+ 2)R2 + r, X
where X = 2w(L t
M).
X may be taken to include any self-capacitance or other stray capacitance.
As a result of the small phase shift, this circuit may readily be used in feedback
amplifiers.
356 STABILITY, PHASE SHIFT, FREQUENCY RESPONSE 7.3
feedback, thus giving the vector ßA0 as drawn. At any lower frequency f
phase angle displacement will be zero. It will, of course, be negative for negative
of A may be taken from the curve for this amplifier ; let us call it A,. The value
the value
of ß may be regarded as a fixed negative fraction less than unity. We thus have the
value of ßA and the phase angle displacement 0, may also be determined from the
phase angle curve. This may then be plotted as a radial vector from 0 to ßA with
a leading phase angle displacement ¢ which implies that the phase of the feedback
voltage leads the input voltage by the angle 0,.
Similarly for any higher frequency f2 ; we can plot ßA2 with a length proportional
to the amplification at this frequency, and a lagging phase angle displacement W2.
As the frequency increases still further, we have a smaller ßA, with a larger phase
displacement, while in the extreme limit ßA = 0 with 0 = 90° lagging. At the
other frequency extreme (f = 0) we have ßA = 0 with ¢ = 90° leading. In this
case the shape of the locus is approximately a circle which is all in the negative region ;
this amplifier is therefore always stable.
Any amplifier may have its ßA locus plotted by this method ; a typical case is
Fig. 7.53B for a 2 stage r.c.c. amplifier. Here ßA, still has appreciable length with
¢, greater than 90 °, but ßA = 0 when 0 = 180° either leading or lagging.
11, MERV,
I000 b
500
50
I
I
I
1I I
I
I
I
-I I
I I
j
I
I
I I
¡
¡ I I
I i
) f
f / fo
FIG. 7.52
Fig. 7.52. Voltage gain and phase angle shift of an imaginary amplifier. The
f
midfrequency is o and the maximum voltage gain A 0.
The distance from any point on the locus to point K is equal to (1 ßA) where ß -
is negative for negative feedback (see Fig. 7.53B).
If it is desired to find the magnitude and phase angle of ßA at any frequency, it is
only necessary to measure the gain without and with feedback, to calculate the ratio
Ao/A' between them, to draw a circle with centre K and radius Ao /A' units, then to
draw a second circle with centre 0 and radius equal to the measured fed -back voltage
I'A'. The line from the origin to the point of intersection between the two circles
is the complex number ¡3A.
For any amplifier to be stable, the ßA locus must not include the point K
(1,0). It is also desirable for the locus not to cut the X axis beyond the point K,
©+
+2
-4 -4
FIG. 7.53
Fig. 7.53. (A) Locus of ßA vectors (" Nyquist diagram") for a single stage
resistance capacitance coupled triode amplifier with fixed bias (B) Similar locus for a
2 stage amplifier.
358 (ii) CONDITIONS FOR STABILITY 7.3
otherwise oscillation may occur during the warming -up period ; such an amplifier
is said to be conditionally stable. In the case of more complicated circuits,
particu-
larly those with peculiarly shaped ßA loci, it is advisable to plot loci for successively
increasing values of ß up to and somewhat beyond the desired value of ß. The
criterion of stability should be applied to each -that the locus should not enclose
the point K (1,0).
The following rule is also helpful as it covers most cases and avoids having to draw
a ßA locus. If ßA is less than unity at
amplifier is stable.
0 = 180° and at ¢ _ 180° the -
Another useful rule is that the maximum permissible value of ßA is equal to
the maximum value of A (i.e. A,) divided by the value of A at 0 = ± 180°
(whichever is the larger).
Also, if the phase shift is not more than 180° at zero and at infinite fre-
quencies, the amplifier is always stable with any value of P.
The " attenuation " is the attenuation with respect to the maximum value of A
(i.e. A0). Both the attenuation and the loop amplification ßA may be expressed in
decibels.
90°
+3 i
cY
Stabi
Margin
db gain reduction
Gain increased 12 db
-95db
FIG. 7. 54
Fig. 7.54. Additional information furnished by a " Nyquist diagram" (see also
Fig. 7.41 A and B.)
Certain aspects of the polar diagram for amplifiers with feedback are shown in
Fig. 7.54. Circles are shown with centres at point K (1,0) with radii of 0.25, 0.5,
1.0, 2 and 3 units. The circumference of the circle with a radius of 1 unit indicates
the locus of points at which the gain is unchanged by feedback. At all points outside
this circle, the feedback is negative and degenerative. For example, the circum-
ference of the circle with a radius of 2 units is the locus of points having a gain reduc-
tion of 6 db. At all points inside the circumference of the circle with a radius of
1 unit, the feedback is positive and regenerative
although not necessarily unstable.
For example, the circumference of the circle with a radius of 0.5 unit is the locus of
points having a gain increase of 6 db.
Instability occurs when the point K is included by the ßA locus. Straight lines
are drawn radiating from the origin with stability margins of 15 °, 30 °, 60 °, and 90 °.
The 30° line is tangential to the circle with a radius of 0.5 unit, so that the point of
7.3 (ii) CONDITIONS FOR STABILITY 359
tangency has 30° stability margin and 6 db gain increase. At other points along this
line the gain increase will be less than 6 db.
An example of the use of the polar diagram in connection with the design of nega-
tive feedback amplifiers is given in Sect. 2(vi)B and Figs. 7.37 and 7.38.
It would be permissible for an amplifier to have a ßA locus such as the line
OABCDEFG in Fig. 7.54 (see Ref. A29). Here the maximum increase of gain due
to positive feedback is 6 db, which occurs between C and D. The feedback is nega-
tive over the path EFG. The distance OA should not exceed say 0.3 unit, so as to
allow for a possible increase in amplifier gain of about 3 db without further increasing
the peak of 6 db.
Feedback amplifiers which are designed with a small stability margin should have
narrow tolerances on components that have a direct effect on gain or frequency charac-
teristics. It is desirable for ßA to be measured and its locus plotted for the pilot model
of each design for three conditions, with normal (bogie) valves and with valves at the
upper and lower limits for mutual conductance. This is only required in the region
of the low and high frequency peaks of response, which are normally beyond the
working frequency range. Alternatively and more simply, the height of the low and
high frequency peaks may be measured in relation to the response without feedback ;
in accordance with good practice, the rise with feedback should not exceed 6 db, and
lower values are desirable.
A special stability problem occurs when an amplifier may not always be connected
to its correct load. One way of overcoming this problem is to design a high -and -low
pass filter pair for connection between the amplifier and the load, the high -pass filter
being terminated by a resistance (Ref. A29).
If an amplifier is stable with the output terminals open -circuited, short-circuited,
or operated into its rated resistive load, then it is stable under all load conditions.
The measurements of stability on open-circuit and under rated load resistance can
be made with an oscilloscope across the output terminals to check for the presence
of ultrasonic oscillations. A check on the short -circuit conditions can best be made
using a high -frequency ammeter across the output terminals. An oscilloscope of
very high input impedance might also be used by connecting it across the input of
one of the stages in the feedback loop (Ref. A28).
(iii) Relationship between phase angle displacement and attenuation
The phase angle displacement of an amplifier is normally a function of the attenua-
tion characteristic. Because of this fact, it is possible to design a feedback amplifier
on the basis of either the phase angle displacement or the attenuation characteristic.
If the attenuation characteristic* is a straight line wiìh a slope of 6 decibels per octave,
then the ultimate phase angle is 90° and so on in proportion ; 12 db /octave gives
180° phase angle and 18 db /octave gives 270 °. The condition which must be satis-
fied is that the amplifier is a " minimum phase shift network." This condition is
satisfied by most amplifiers, the exceptions being (1) when it contains a transmission
line or equivalent circuit with distributed constants, and (2) when it includes an
allpass section, either as an individual structure or in a combination which can be
replaced by an all -pass filter section plus some other physical structure (Refs. H6,
H10, H11).
Fig. 7.54A shows ultimate slopes of 6 and 12 db /octave and higher slopes. The
12 db /octave attenuation characteristic is the limiting value for stability, being on the
verge of instability. A practical amplifier requires a safety margin between the slope
of its design characteristic and the limiting value of 12 db /octave. A typical design
slope is 10 db /octave, giving an angular safety margin of 30 °.
Any single reactive element such as a grid coupling condenser, a shunt capacitance
or a shunt inductance in conjunction with a resistance provides an attenuation charac-
teristic with an ultimate slope of 6 db /octave. Such a combination is known as a
single time constant circuit. Attenuation characteristics are given in Fig. 4.36 (grid
coupling condenser or shunt inductance) and Fig. 4.38 (shunt capacitance), for a
single time constant in each case.
*Plotted with logarithmic frequency scale.
360 (iii) PHASE ANGLE AND ATTENUATION 7.3
If a parallel -fed transformer is used, the coupling condenser resonates with the
primary inductance at some low frequency and the ultimate slope of the stage is ap-
proximately 12 db /octave below the frequency of resonance. If the secondary of a
transformer is unloaded, the leakage inductance resonates with the distributed capacit-
ance, giving a slope for the stage of approximately 12 db /octave above the frequency
of resonance. Each of these combinations is equivalent to two time constants.
,_
3
\\,
\
6
v 9
tI
",V
.
-M__
/IN,
éhr,,t,
Illft:
?
18 0
C
!III'P
C C1
21
24
UNSTABIE
W
2f, 4f,
FREQUENCY
8f,
FIG. 7.54A
16 ,
+R
B = 2 3 5 10
Max. loss = 6 9.5 14 20 db
Max. slope = 2.25 2.6 3.6 4.8 db /octave
Max. angle = 20° 30° 42° 55°
where B = 1 + Rk(µ + 1) and R = RLRu2
For a pentode B . 1 + g,,,R k.
RL ry+Rv2
9 Ob
12
15
90* 18
80° 21
t; 70°
4 60'
0 SÓ
` 400
30
20°
10°
o
.3fo 3fo 10 fo 30fo 100f0
'Ifo
FIG. 7.546
Fig. 7.54B. Attenuation and phase characteristics of " step circuit " Fig. 7.59A
for particular case with step attenuation 20db (Ref. H18).
70"
60
50
Fig. 7.54C. Maximum
40 phase shift as a function
30 of attenuation in the
circuit of Fig. 7.59A
20 (Ref. H18).
lo
0
o 2 4 6 8 10 12 14 16 18 20 22
Stil) Attenuation (db) FIG. 7. 54C
362 (iii) PHASE ANGLE AND ATTENUATION 7.3
a 80
a7Ó
Sk
x6Ó
v 1'AA
r o 713
x50
4 10k
r4Ó
r3Ó
k .20 Sk
X10°
Ó y/
fo 2!0 4f0 df0 Ife
Fig. 7.55A. Phase angle for attenuation characteristic as shown ;
4f0 6f0
FIG. 7.55A
k=
Of.Ok
1 for
6db /octave, k = 2 for 12db /octave etc. (Based on Bode, Refs. H6, Hl 1).
On account of this fairly close agreement between the phase angle characteristics
¡or the two cases, it is possible to replace the actual attenuation characteristic by an
approximate " straight -line " equivalent. Where the attenuation characteristic
does not have constant slope throughout, it may be replaced by a succession of inter-
secting straight lines having slopes of zero or any multiple of 6 db /octave, either
positive or negative (e.g. Fig. 7.55B). These, in turn, may be resolved into " semi -
infinite "
lines of constant slope, as in Fig. 7.55C. The phase characteristics may
be derived by the use of Fig. 7.55A for each junction. The resultant phase angle
characteristic may be determined by adding these individual characteristics.
A " semi- infinite " straight line is one commencing from a definite point and proceeding to infinity
in one direction only.
7.3 (iii) PHASE ANGLE AND ATTENUATION 363
More accurate results may be obtained in some cases by the use of phase angle
curves for attenuation characteristics having a sloping portion and a horizontal portion,
as given by Bode (Ref. Hll, Chapter 15).
ó
db
s
f1 f,
1, f, 93
Fr°aodncy FIG. 7.556 FIG. 7.55 C
9 db
12
1-15
90° I8
90° 21
o
'1 /o 2f0 o 2f0 51 10 f0
FIG. 7.550
Fig. 7.55D " Straight line" approximate characteristics with an attenuation slope
of 6db/octave (actual characteristics shown with broken lines) and corresponding
phase angle characteristics (Ref. H18).
364 (iii) PHASE ANGLE AND ATTENUATION 7.3
12
6
3)
3
Ob
0
6
(2)
9
12
IS
(I) le
21
24
27
30
33
36
39
42
45
4e
51
.3f0
f. 3f0 IOfa 30f0 100 f.
CIG. 735 E
Fig. 7.55E. Typical attentuation characteristic of amplifier with feedback (broken
curve) and component " semi-infinite" slopes (1, 2 and 3) from which Fig. 7.55A
is constructed (Ref. H18).
The broken line in Fig. 7.55F is the actual phase characteristic, indicating reasonable
accuracy for the method. In all cases the actual characteristic tends to " round the
corners " of the approximate characteristic.
The same curves (Figs. 7.55D, E and F) may be used for the low frequency attenua-
tion characteristics by inverting the frequency scales.
be the permissible feedback factor. One or both cathode by -pass condensers may be
omitted, or the by- passed section of R k may be reduced in resistance. One or both
screen dropping resistors may be reduced in resistance, or replaced by a voltage divider
with other consequential adjustments. The values of the cathode and screen by -pass
capacitors may be adjusted so that their frequencies of maximum phase angle are
" staggered," preferably an the ratio of at least 20 to 1.
The high frequency peak may be reduced by staggering the high frequency response
of the two stages, for example by shunting a capacitor from one plate to earth and
increasing the capacitance experimentally until the peak is sufficiently reduced.
For an exact design method see Sect. 3(vii) below.
The phase angle displacement is increased by the inclusion of an iron -cored trans-
former within the feedback loop, also by any resonance effects, and such cases should
be treated by the method described for multi-stage amplifiers.
140°
RESULT NG _
PHASE CHARACTERISTIC
I
i:iiii
,..,.'--,
,E11I111MIIMA''
-
120°
1o0°
B0°<
60°
á
IIIJJIII-'
ANII MINIM TO BE
SU BTRACTED
40°
20°
0
-If. 3f° f° 3f° Of0 30f° 1001°
FIG. 7.SSF
Fig. 7.55F. Phase angle characteristic derived from Fig. 7.55E (Ref. H18).
Design tests
It is highly desirable to measure and plot both the frequency response and phase
angle over a sufficiently wide frequency range ; both curves may be plotted on the
same sheet of logarithmic graph paper. The required frequency range is the useful
frequency range of the amplifier plus (at each end) one octave for each 10 db of feed-
back plus at least one octave.
When it is desired to achieve the maximum degree of feedback it is essential to
plot the polar (Nyquist) diagram as in Figs. 7.53B and 7.54. For a method of measur-
ing the phase angle see Refs. H7, H19, H20.
An amplifier may be tested for transient response by applying a rectangular wave-
form to the input and observing the waveform of the output with a C.R.O. having
good frequency and phase characteristics up to the maximum test frequency. If
there is any oscillatory response or " overshoot " the feedback may be reduced or
some modification made to the feedback loop to provide the desired degree of damp-
ing.
Another useful design test is to overload the amplifier with input voltages of various
frequencies, and to determine the level to which the input must be reduced to return
to normal linear operation (Ref. A29 discussion).
A U) D R
\ \O
N
WORKING
FREQUENCY RANGE
a \\
N.
40
IRO
C
É
0
120
` 6e
x
o
á
o
ts
10
s
S
sz fo 2f0 5/0 loro
FIG.7.56A
Fig. 7.56A. Bode's method -Curve (1) attentuation characteristic and (2) phase
angle displacement characteristic for constant 150° phase angle displacement above
f
upper working frequency o ; curve (3) constant slope 10 db /octave (Ref. Hl 1).
Alternative methods of design
(a) One method sometines used is the " cut and try " method of constructing an
amplifier and then applying feedback with the feedback factor increasing in steps
until instability occurs, and finally decreasing the feedback to provide a safe margin.
If instability occurs with only a small degree of feedback, there are devices which
may be experimented with, such as
1. Shunting the primary of the output transformer by a condenser.
2. Connecting a very small condenser from the plate of one of the earlier stages to
earth.
7.3 (v) DESIGN OF MULTI -STAGE AMPLIFIERS 367
Fig. 7.56B. Bode's method-Curve (1) as in Fig. 7.56A ; (2) normal attenuation
characteristic of typical amplifier. Shaded portion indicates additional attenuation
required in the feedback loop.
For convenience in feedback design, Curve has been redrawn in Fig. 7.56B with
0 db corresponding to ßA I I
- 1
and 12 = frequency at which the actual attenuation is equal to (iß0AoI + 10) db,
then f, = 12 (10 /n)
number of octaves between fo and f, _ 113oAoi /10
f, and 3.32 log(n /10).
12 =:
The frequency ratio may be derived from the number of octaves by the relation- -
frequency ratio = antilog (0.301 x number of octaves)
or by the use of the table below :
10
20
\
9
30 IF IQROeo)
Y$
D
40
/ ¡
/./
SO
60
0425 1.25 2S S IO 20
Frequency c/s FIG. 7.57
Fig. 7.57. Low frequency attenuation characteristics of multi -stage amplifier with
feedback (simplified treatment).
Low frequency attenuation characteristics
As an example for illustrating the general principle, take an amplifier with nominally
flat frequency response down to 20 c/s (actually it will be 2 or - -
3 db at this fre-
quency). Let the loop amplification PA be 30 db at useful frequencies.
Draw the attenuation characteristic with a slope of 10 db /octave from 20 c/s down-
wards (line ABC in Fig. 7.57). Point B is where the value of the loop amplification
becomes unity. Extend the line AB to C which is at a level 10 db below IßAI = 1.
At C, insert a horizontal " step " CD, the length of which will be given later. Be-
yond D the attenuation will fall at a rate of 12, 18, 24, 30 db or more depending on
the circuit ; each single time constant or non-resonant transformer contributes 6 db,
while each resonant circuit contributes 12 db. The frequency ratio between points
370 (v) DESIGN OF MULTI -STAGE AMPLIFIERS 7.3
C and D is made equal to the ratio of the slopes of DE and AC, which is equal to
the slope of DE divided by 10. The line ABODE, thus determined, is to be used
as a guide to the low frequency attenuation characteristic of the amplifier. In reality,
of course, the sharp corners will be rounded -the important features are that the
slope of the AC region should not exceed 10 db /octave, and that the step should be
sufficiently long.
1
SNF 2NF 5NF
ISMII 6SJ7 ISMII 6V6 IOs000.ff
39MII _
3fM11 c ...
FIG. 7.58
oY
ó
s
10
20
25
30
Miff
_/_
,w
65J7 Grid
_-, -6V6
6SJ7
Screen
/
Fig. 7.58. Three stage amplifier with negative feedback illustrating design method (Ref. H9).
Immg_mi....--_,
ißAI -I
+e300V-
Low unp<donca
35
40 /
/ Screen -Grid Response
45
Totol Response
50
01 1 10 100
Frequency (c/s) FIG. 7.59
Fig. 7.59. Low frequency attenuation characteristics of Fig. 7.428 (Ref. H9).
Example
An example of an amplifier designed in accordance with this method is Fig. 7.58
(Ref. H9). The low frequency attenuation characteristics are given in Fig. 7.59.
The 6V6 cathode, 6V6 screen and 6SJ7 screen characteristics are designed to provide
a very close approach to a total slope of 10 db /octave with a limiting attenuation of
40 db, so providing the basis of a step. The two grid condensers are designed to give
effective attenuation beyond the limit of the step. It is evident that it is possible to
design an amplifier which provides the necessary low frequency attenuation charac-
7.3 (v) (B) TREATMENT BY LEARNED 371
teristics, including the step, without using any resonant circuits. The tolerances on
the frequency- dependent components (R and C) must be small.
The feedback loop does not include the plate circuit of the third stage, so that the
slope of the high-frequency attenuation characteristic is nominally 12 db /octave.
The resonant circuit (incorporating C,) in the feedback path is employed to stabilize
the frequencies in this region by providing a step in the high- frequency attenuation
characteristic.
Corrective networks
The design of simple corrective networks is well summarized in Ref. H9. More
complicated designs incorporate LCR 2-terminal networks as the plate load imped-
ances and (some) cathode bias impedances (Ref. H11). See also Ref. J16.
One of the simplest and most effective methods of providing the required step in the
high frequency attenuation characteristic is the RC network shunting the plate load
resistor of the pentode V, in Fig. 7.59A (based on Ref. H9). The ultimate attenuation
of the step in decibels is given approximately by
attenuation 20 log (R1 -}- R) /R.
The cut-off frequency is given by
fl = 1 /27R ,C
and the " flattening -out " frequency by
f2 1/242C
where R, = R R, /(R t + R1)
and the " flattening -out " frequency is defined as the frequency at which the attenua-
tion is 3 db less than the ultimate attenuation of the step.
C1 .
The value of C1 is selected to give negligible attenuation at f2 ; say
5 /[7t.W (R + RL -}- R,)].
With these step circuits it is desirable to have as many separate circuits as possible,
each with a small step.
Fig. 7.59C. Feedback path including.
CR network to give a step in the
attenuation characteristic, and thereby
a
R, I 1
c LOAD
db /octave, the slope difference is 8 db /octave, and the phase angle between the volt-
ages is therefore 120 °.
Take the intersection of the two characteristics as the origin (as in Fig. 7.60) and
assume that 0 db = (say) 1 volt -both assumptions being for convenience.
Let x be the voltage contributed by the lower slope characteristic at a certain fre-
quency. It may readily be shown that the voltage contributed by the high slope
characteristic at this frequency is xa, where a is the ratio of slopes in db /octave.
These two voltages, x and x' must then be added vectorially, the angle between them
being as given by the table above.
The magnitude of the combined characteristic may then be plotted as in Fig. 7.60,
and the values of the phase angle may be marked along it.
The method of deriving the magnitude and phase angle of the resultant is illustrated
by Fig. 7.61 for the condition where the higher slope characteristic has a slope of 12
db /octave, corresponding to 180 °. Vector OA, having magnitude x and phase angle
+30
INDIV DUAL
CHARACTERISTIC
II
171. ° COMBINED
CHARACTERISTIC
+20 1669°
it
Approaching 12db /octave
,
167.7
156S
"'
+10
db 142.4°
1338° Approach ng 4db /octave
o 120°
95.7*
B3.2°70.3°
10 1,,., 66.2°
63. 7
g.} 62 4°
}
- 20
2x°" + 1
-
i.e. 180° - 0.
1.73 2x5
*between the two voltages,
A similar procedure can, of course, be applied to any other value of maximum slope.
Procedure when lower slope is zero
The foregoing procedure cannot be used when one of the two attenuation charac-
teristics has a slope of zero, that is when one is a fixed voltage. In this case the pro-
cedure is as follows.
"/x2 - 1.93x +
165°
0.26x
1 tan-' - 1 0.97x
An example of the summation of two attenuation characteristics, one having zero
slope and the other having a slope of 10 db /octave, is given in Fig. 7.63. The re-
sultant is asymptotic to 10 db /octave and to the horizontal characteristic, but reaches
a minimum value at the frequency of the point of intersection O. The phase angle
is marked at points along the resultant, and at all frequencies the phase angle is less
than that of the 10 db /octave characteristic, the effect becoming more pronounced
+2 10
147°
+I
I\` s 150° ..
5.6
144 x
\\ P
+1 1
\,,:e., O ;° 3.2
adi
13 no
40 4
a.
+ \n 18
db Volts
\1122° 0 Zero Slope <
o `
\\r 413°
- 41-60 3°/-e
-5 \
iiJ5°
' "29°
56
-10 0 72
150°
-15 O 18
20 9
0.2510 5f0 /° 2fo 4f°
I
FIG. 7.63
Fig. 7.63. Summation of two attenuation characteristics having slopes of zero and
10 db /octave.
at higher frequencies, particularly beyond point D. The fact that the slope of the
resultant in Fig. 7.63 exceeds 12 db /octave to the left of point D, has no effect what-
ever on the stability as indicated by the angles. It is therefore obvious that the re-
lationship between phase angle and slope of the attenuation characteristic which
applies to minimum phase shift networks, does not apply here.
A similar calculation has been made for a slope of li db /octave, Fig. 7.64.
The position of point D on the resultant characteristic is a function of the slope of
the attenuation characteristic.
slope in db /octave 2 4 6 8 10 11
height of point D + 5.7 + 4.7 + 3 0 5.6 11.6 db. - -
(E) Duerdoth's method employing multiple feedback paths (Ref. A29)
In accordance with this method, one or more additional subsidiary feedback volt-
ages are introduced into the feedback loop so as to decrease the phase angle over the
7.3 (y) (E) DUERDOTH'S METHOD OF MULTIPLE FEEDBACK 375
Io
163°
+1 54
16)'lls
+I
`\ 165°
N
32
15e
+5 Ie
^o
de e
Volts
0
149
}O
+-' 1
/
2°
....
1 .. 7ó
-s 1 0 56
1
9
-10
\r 0 e4°
6
165°
092
-Is
1 0.18
-20
4o
I
0.25fo 0.5 fo fo 2!o
FIG. 7.64
Fig. 7.64. Summation of two attenuation characteristics having slopes of zero
and 11 db /octave.
attenuation portion of the amplitude characteristic without appreciably affecting the
characteristics within the working frequency range.
With subsidiary feedback applied over the first stage of a 2 -stage amplifier and loop
feedback over the whole amplifier, the ratio of gain with and without feedback is
given by
A' 1
where A
A 1- ß,A1 - ß2A
= overall amplification
(1)
without feedback
A' = overall amplification with feedback
A, = amplification of first stage without feedback
ß, = value of ß for subsidiary feedback over first stage only
and ßz = value of ß for loop feedback over whole amplifier.
It is evident that when subsidiary feedback is used, the several feedback voltages
must be added at their common point. This holds even when the subsidiary voltage
is entirely inside the feedback loop.
Under the same conditions, the ratio of distortion in the final stage with and without
feedback is given by
D' 1 -
ßiA. A'
-
D 1 - ßiAi -
ßzA A (1 $,A)
Thus the application of negative subsidiary feedback to the first stage increases
(2)
the distortion in the final stage in the same proportion that the gain of the first stage
is reduced, provided that the overall feedback loop is unchanged.
If, however, the subsidiary feedback is positive, the gain of the first stage will be
increased and the distortion in the final stage will be reduced further. The use of
positive subsidiary feedback need not present any stability problems, since the stability
criterion is dependent on (ß,A, + ßzA). See Sect. 2(xi) and Refs. F11, F12.
With multiple loop amplifiers, the usual Nyquist stability criterion applies to any
loop, provided that the amplifier remains stable when the particular loop is broken.
Even with amplifiers which are always stable, the Nyquist diagrams obtained by
measurements of the several loops in turn may have different shapes and the definition
of stability margin becomes meaningless. However, if a mesh of the amplifier can
be found which, if broken, simultaneously breaks all the loops, then there is no possi-
376 (v) DESIGN OF MULTI -STAGE AMPLIFIERS 7.3
bility of oscillation due to the disconnection and the definition of stability margin
remains applicable (Ref. A29).
When employing multiple feedback paths with summation of attenuation charac-
teristics and subsidiary voltages, the crossing point of the characteristics should be
controlled with considerable accuracy. The crossing point will be modified when the
gains of the various stages change owing to changes in the valve characteristics with
age, or to overloading caused by an excessive input voltage. The latter may be avoided
or reduced in more elaborate amplifiers by the addition of a cathode follower stage,
operating as a limiter, as part of the first stage. Precautions to be taken in design
are described in Ref. A29.
Some applications employing multiple feedback paths are described below. See
also Sect. 2(vi)A.
FIG. 7.65
j- LOOP
FEEDBACK
(1) One simple application is the circuit of Fig. 7.65 where the signal voltage across
the unbypassed cathode resistor R k is added to the loop feedback voltage. This is
an application of the principle of summation of attenuation characteristics as shown
in Figs. 7.63 and 7.64. The voltage across R k is practically constant over the critical
region of the high frequency attenuation characteristic, so that its characteristic has
zero slope. This method may only be used when the phase angle of the ßA locus is
always less than 180 °, i.e. the slope of the ßA attenuation characteristic is less than
12 db /octave. This method does not reduce the distortion in the final stage to the
same level as without subsidiary feedback, or to that using one of the alternative
methods below. For this reason the subsidiary voltage should be considerably smaller
than the loop feedback voltage. However, this method will improve the stability
margin of an amplifier and reduce the amplitude of the high frequency peak.
(2) One modification of this method, which has only small degeneration in the
working range, is shown in Fig. 7.66. This appears to have been first described by
Farren (Ref. A7) and applied to one of the stages inside the feedback loop. The
phase shift reaches maxima at low and high frequencies outside the working range,
beyond which it approaches zero in both directions. The degeneration increases
gradually in both directions to the limiting design values provided by Rl and R2.
This arrangement does not modify the fundamental phase shift between the input
and output voltages of the stage due to the phase angle of the plate load impedance.
7.3 (y) (E) DUERDOTH'S METHOD OF MULTIPLE FEEDBACK 377
(3) A further modification by Farren (Ref. A7) is shown in Fig. 7.67 which has the
advantage at very high and very low frequencies that it reduces the phase shift between
the input and output voltages of the stage due to the phase angle of the plate load
impedance which would be characteristic of the stage if there were no subsidiary
feedback.
+30
LOOP FEEDBACK
+20
RESULTANT
+10
db
10
20
0116 f, of,
FIG. 7.69
Fig. 7.69. Summation of loop and subsidiary feedback voltages to provide improved
stability (Ref. A29).
(4) Two terminal networks of the form of Fig. 7.68 are described by Duerdoth as
cathode impedances for improving the stability of feedback amplifiers with an ulti-
mate attenuation of 12 db /octave. A value of L = 10CR2 may be used as a first
trial and maximum feedback and zero phase angle will then occur at a frequency of
approximately
f = 0.048/CR
where C and R are in farads and ohms respectively. An example of the application
of subsidiary feedback on an amplifier with an ultimate attenuation slope of 12 db/
octave is given in Fig. 7.69. In this example the peak of subsidiary feedback occurs
at twice the maximum frequency of the working band. The stability margin in the
critical region has been increased from 10° to 36° with subsidiary feedback, as shown
by the Nyquist diagram (Ref. A29).
It often happens that the addition of the subsidiary voltage results in an increase
in phase angle just above the working frequency band where the magnitude of the
Fig. 7.70. Three stage amplifier incorporating subsidiary feedback in the form of a
4 terminal network. m = 0.87. W = " Wheeler " network (based on Ref. A29).
378 (v) DESIGN OF MULTI -STAGE AMPLIFIERS 7.3
-IIIaMI
I I
4
Low FrtauuI ry Attanus[bn HO Frequency Attdnl0tbll
1' "Iv
r-
TirerrAMIIIIIIIIIMIIIME_
----- : áe°, -----
`
,
,go
/iIII sI
o. 'I, -
ol 02 05 01 2
1111
5
F Fi
1.0 2 S 40 1 04 2 .5
F.FY
u0
L-4
2 5 10 20 50
F/ F, Frequency F/F, y
FIG. 7.72
/ 'I
4
I
-
I
ß 2
: Ao l0
FF
--
l
I
F F,
QAó-10 gji n
.5 ----- !Aax-2 Fteoock -- - - ----------- -- No --- A-2------
No I I
os ß 5 10 2 5 0 'i 2 d 10 2 5 10 20
F/ F, Frequency F/ F,
FIG. 7.73
Fig. 7.72. Relative voltage gain of a single stage r.c.c. amplifier without feedback
and with three selected values of feedback (Ref. l4).
Fig. 7.73. Relative voltage gain of two identical r.c.c. amplifier stages, without
feedback and with four selected values of feedback (Ref. H4).
Graphical method for two -stage r.c.c. amplifier
An ingenious graphical method has been developed by Barter (Ref. H17), and may
be used for determining the R and C values to give any desired height of peaks at low
and high frequencies.
(vii) Design of amplifiers with flat frequency response
(A) Method of H. Mays (Ref. H13). See also J13.
A resistance -capacitance-coupled amplifier Fig. 7.74 may be regarded, so far as
frequency response is concerned, as a tuned circuit with a Q less than 0.5. The
value of this Q is given by
1
Q (3)
b(d +
+ (1 /b)(p + k)
k)
where b = ./R,/12t; d = C, /C; k = Cr /C;
p =C, /C and C= VC,CC +C,C, 4-C,C,.
This method may be applied to multi -stage amplifiers with feedback, it being
assumed that all stages have the same resonant frequency (w, /2n). The conditions
for maximum flat response are then given by :
1 stage I(A0 /A)I = V1 + x$ (4)
= - tan-ix (5)
380 (vii) FLAT FREQUENCY RESPONSE AMPLIFIERS 7.3
(9)
and Qt /Q2 = (n - 1) + 1n(n - (10)
FIG. 7.74
3 stages I(Ao/A)I = VI + a2x4 + x4
= - tan -1(bix
3
- x3)/(1 - bax2)
where x = Q2 Uf2j ` wo - WO)
/
a2 naL_ 2. /
QaeJ
bi
1/n'r,/ Qa8 + 23,,/Q3]
3 Qi
b2
- +4,Vr-F 2Qa]
Q1/Q2 = 2(n - 1) + V/2n(2n - 3) (17)
and Q2 = Q2. (18)
N.B. n must be greater than 1.5.
This has also been extended to four stages (Ref. H13).
Symbols used :
w = 1 /(C1/RLR,).
These equations have been plotted in Fig. 7.75 from which it will be seen that the
feedback factor has no effect on the shape of the curves for 1 or 2 stages, and only a
very slight effect for 3 or 4 stages (curves for limiting values of n are given). Although
the curves are shown only for the high frequency limit of the flat top, they may be
applied to the low frequency side by changing the sign of the abscissae and, in the
case of phase angle, also changing the sign of the ordinates.
This method of design provides the maximum flat frequency response, but the
maximum degree of feedback only occurs at the mid- frequency and it falls off rapidly
outside the frequency band of the selective " stage. The feedback becomes zero
near the knees of the flat top, and positive at still higher and lower frequencies.
This design ensures stability at all frequencies from zero to infinity.
This method also makes it possible to design for any desired response curve, within
the limits of the amplifier. The procedure is to commence with the desired response
curve and to determine the response (either modulus or phase angle) corresponding
7.3
. i-
IMI:::::::
FREQUENCY
7::::::
(vii) (A) METHOD OF H. MAYR
RESPONSE
m:: .011-M.IIII 11111u111
.H11111IMIHI11
PHASE
IM=.1IU11-1M.111
ANGLE
..nll
381
i
_':I::: EIMMCIIM:::c¡;=C.01:::::t ....._
=1u11 ..111.111111 11U/111111IM1I111
.HIIIff0H111f111H1
.: -...::::::
Ì-..:::::: 1111 1111.u1111 111=MI. III
- ....1-M.
O =./H111 I=111111011.11
-.1.I .....-..1
M11111110\ _
MINIMUM 1M111II. 1=./111 11.u.1
ee 56O IIIHIII 1111=.111111>011u111
-I
5
-M......M.
-::Ö:: ...a-......1
.i:nmi-ZMI°1
::i-..::::
C:':::.7g7C:::=C:::::::
ow
ow a.emn1
ma
Mm1u 11111Mu \\.1fiHIII11 III 1I I11
.HII .IIIIII.uI11I-.IIII
=MUM
=1111
STAGES
111111\11111fH11
7111
`II
.u1.- 1
=.1111I -IIM.11111111116Nu111111111111111
NI1111 rIN1111111111=.'_11ME:iii
1111_.
:i:i:i IMM1.111111M111111
MEMOIR
+- 5 -.-...:::Ì ÌÁ-..:::1
-...1
:::::-M.Siii
--IS..131-1=1..YI +
M.1111 1..1111I=11111.1II111IMMEM../III
=ENNUI
.-_.:..:1-. o
---.:::::i.
`o ::ii
1:Z
i
_-... ::.. .....
11:::::i-::::::
I}T.-..I11-......1
.11:Ì-:::::i
WW'.-..tllff=..11
1i::ul-.....
:.nll......-....1
üi. MM.:::::i-::i.: ==. CÇ
:.1-..:....
.:::::I- ,
--..
.r1 THREE
-..:::i- STAGES
2 :
1aiiiiÌÌÌ
ii.1.1 C:
=:i= E:::ñ: iiiiÌiÌiii 1m.iiiÌÌi
+- 5
....11-....u1
i::::: :::::: HIIM./11111IMMIHIII
iiüìlú"uw1uuu
.H111111111111f111111EM111111
iiii\iiiiiiili
ouu.-..LaRf?f;_-m1u-Muu1
uI11e
1 i 1
l!
I-.uI11-..11
MM11111f11111
_= 9E:67+==oEEB:u__-'::.;E
°::
1Ilnll.u11
=MINIM MIMINCCIIII=11.UIIUIMM.uI111
MII1.I11 MINIM
1111111 11111Mum1 'II MINIMUM =BIM
.1111 u..111111111.111111111M=.u1111
1H1
=MIMI 11111.1IU611M.IHIIII=III
11.HII 11111111111QH111111M11111
:=.:' IMHU 111
IM.u. IIIMMnu1111uu111111Mnuu
-1IIIIIMIL I.11
:l .uu-.uu..ouu
f1111 11111\Ju1111
iOIII } um
=.u111 fff/HIII\ì7111111111
. mI
..
°
1 ...
...Y:Ìi.:.1 .1I11 .Hw.uuu_-11111Oi
=MIMI IIMN.uu11=11I11111111
10 CO 041 o-1 10 100
FIG. 7.75
Fig. 7.75. Frequency response and phase angle of frequency response for 1,2,3,
and 4 r.c.c. stages designed for maximum flat response by Mayr's method (Ref. H13).
to two arbitrary frequencies (W, /2n) and (w1 /270, one near the lower and the other
near the upper frequency limit. Then we read from the curves the values of x,
and x2 corresponding to the two chosen response values ; the value of x, corresponding
to the lower frequency is, of course, negative.
The mid-frequency may be computed from
xYWl -x12 (19)
Wpb
- x,W,
= W1WI
Xa(us
stages or
and the ratio /Q, from equation (10), or (17) for or
Q1 2 3
(Q,/ Q2) = 3(n - +'ß/3(3n - 4) for 4 stages. 1)
The Q's of the various stages are then given by :
nxl
1 stage : Q =
Wl Wo
Wo W1
/nos x1 Qt
2 stages : Q = w// w1
i Q1 = Qs
Q2
Qi coo
Wo WI
3 nQs
3 stages : Qs=Qs= 01 Wl
x1
w0
; QI = Q.
QS
Q1
Wo WI
382 (vii) FLAT FREQUENCY RESPONSE AMPLIFIERS 7.3
X1 QI
4 stages : Q2 = Q, = Q4 = v// nl X10 ; Q1 = Qa Qa
(23)
co,.
These amplifiers are made up of one rather selective stage, corresponding to Q13
and a number of equal broadly-tuned stages, corresponding to Q,, Q, With
increasing feedback the selectivity of Q1 must be increased, while the selectivity of
...
Q, etc. approaches a limiting value which is of the order of the Q of a single stage
amplifier without feedback, having the same bandwidth as the complete amplifier
with feedback.
The amplifier is first designed, neglecting all capacitances, to have a gain approxi-
mately n times the final value. The feedback is designed with a purely resistive feed-
back network to decrease the gain n times. The value of C, is then determined at
the minimum practicable value, that is the input capacitance of the following stage
plus an allowance for strays. Then the values of the other capacitances are given by
p
Cn =
woVRLR,
; C,
k
- (24)
b = factor by which the normal value of load resistance is multiplied for one
stage,
7.3 (vii) (B) METHOD OF C. F. BROCKELSBY 383
1/b = factor by which the normal value of load resistance is multiplied for
the other stage,
and ß = fraction of the output voltage fed back to the input.
Equation (26) may be expressed in the alternative form
F = 2S' (26a)
where F = 1 -
ßAo = feedback factor
and S = -(b + 1/b) = " staggering coefficient."
In a two stage amplifier with a high frequency peak, the required staggering to
produce a peak of known value is given by
S2 = (F /2)(1 - 1 -a2) - (27)
where a = ratio of voltage gain at middle frequencies to that at the peak (a is less
than 1).
The results are summarized in the following table :
= b 1 1.5 2 3 4 5
S = 1 1.08 1.25 1.7 2.1 2.6
CD o
o
dD e0
to db
20
20
30
05 a 02 04 07 2 4
0-2 s 2 3
FIG. 7.76
i FIG. 7.77
If the condition of maximal flatness is to hold at the low as well as at the high fre-
quency end of the band, the centre-frequencies of both stages should be identical
(Ref. H13).
Three-stage amplifiers and maximal flatness
It has been shown (Ref. H12) that if a substantially flat response over the maximum
possible frequency band is required, the optimum arrangement for a three -stage
amplifier is to have two stages with wide frequency response and one with narrow.
384 (vii) FLAT FREQUENCY RESPONSE AMPLIFIERS
7.3
This finding is based on the assumption that the amplifier has maximal flatness ;
is not necessarily the most desirable arrangement under all conditions.'
it
The condition for maximal flatness is that
(B2 -f- 2)'
F - 2B + 1 (28)
where F = feedback factor
B = staggering factor = 1 /b
and b = factor by which the normal value of load resistance is multiplied for the
one (narrow) stage.
[B is always equal to or greater than unity.]
Equation (28) may be approximated within 1.4% when B z2 by
F ti
+ (B' /4) 1 (29)
For example when F = 10 (i.e. 20 db feedback), B ti 3.3 and the ratio of band-
widths is approximately 11.
As with the two -stage amplifier, an increased amount of feedback beyond that to
give maximal flatness gives high frequency peaks (Fig. 7.77).
The critical amount of feedback to provide self-oscillation is
F, = 5 + 2B' + 2 /B' = 8Ff -3 (30)
where F, = feedback factor to give critical feedback
and Ff = feedback factor to give maximal flatness.
The stability margin of a three -stage maximal -flatness amplifier is (F, F,) and
in a typical example
-
Ff = 3.16 (i.e. 10 db reduction in gain) for maximal flatness
then F, = 8 X 3.16 -
3 = 22.2 (i.e. 27 db approx.) to give self-oscillation.
The stability margin is therefore 27 -
10 = 17 db approximately.
If the feedback is made very large, and the staggering is adjusted so as always to
give maximal flatness, F, 8F, which for this example becomes 8 x 3.16 = 25.2
(i.e. 28 db approx.) giving a stability margin of 18 db approximately.
io
1I
nl lI
OUTPUT
CIRCUIT
J
11111 Ra
.
(32)
- FIG. 7.79 - FIG. 7.90
Fig. 7.79. Two -stage uncompensated amplifier with feedback (Ref. H15).
Fig. 7.80. Three-stage uncompensated amplifier with two time constants within
the feedback loop (Ref. H15).
Two stage amplifier with constant value of fi (Fig. 7.79)
The following may also be applied to the three stage amplifier of Fig. 7.80 which
has only two time constants within the feedback loop.
It may be shown that the condition for critical damping is that
- ßA, -
2
1
a4 1sa
where A 0 = amplification at the mid- frequency without feedback
ß is negative and independent of frequency
a1 = 5/R1C, - inverse time constant of V1
and a2 = 1 /R2C2 = inverse time constant of V2.
When a higher feedback factor than that indicated by eqn. (32) is used, the transient
response will be oscillatory ; when a lower feedback factor is used, the transient
response will be over -damped.
Equation (32) may be put into the form
1 - ßAo = (1 + K)2 /4K = K/4 + 1/2 + 1/4K (33)
where K = a2 /a1 = ratio of inverse time constants.
This is plotted in the lower curve of Fig. 7.81.
If a fairly large amount of feedback is to be used, the ratio of time constants must be
large.
The resultant inverse time constant of the amplifier is given by
a = J(a1 + a2) = ía1(1 + K) (34)
This relationship is plotted in the upper curve of Fig. 7.81.
386 (vii) FLAT FREQUENCY RESPONSE AMPLIFIERS 7.3
100
10 100 ipoo
K FIG. 7. 61
Fig. 7.81. Conditions for critical damping of two -stage uncompensated a nplifier
(Ref. H15).
Note : The " staggering coefficient " S as used by Brockelsby (Ref. H12) is related
to a, and a, by the equation
S
ai
- as
-a2- (Vk .VK/l 1
(35)
The condition for maximal flatness is therefore given by
- ßAo - (a1 + a2)°
1
(36)
Comparing equations (32) and (36), it will be seen that for any particular ratio of
sa /a1, the condition of critical damping permits a feedback factor of only one half
that required for maximal flatness of the frequency characteristic. The transient
response of the maximally -flat amplifier is therefore oscillatory.
A 2-stage critically damped amplifier will normally have a small amount of positive
feedback at high frequencies, although the rise in response in a typical case will be
less than 1 db above the zero-feedback curve.
feedback amplifier is made identical with that of an amplifier without feedback having
two stages, each having an inverse time constant equal to the geometrical mean of
the inverse time constants of the two stages multiplied by the square root of the feed-
back factor (eqn. 38).
A particularly simple special case is obtained when
as = ai(1 - ßAo)
then a = a3 = «2 = «1(1 -
of «3 for critical damping may be derived from the curves of Fig. 7.82 and
values of « from Fig. 7.83.
a= (al +as+
as) /3 (40)
_ [(asa2as(1 ßA0)] - - 3«,a2a3ßAo 3 (41)
a4
a12 + a22 + - (a2a2 + a2a3 + anal)
a32 (42)
where «3 = inverse time constant of third stage
as = inverse time constant of feedback path
and a = inverse time constant of amplifier.
10
So
,o
10 100
a(2/.,(1 FIG. 7.82
Fig. 7.82. Curves for determining a3 for critical damping of two -stage compensated
amplifier (Ref. H15).
388 (vii) FLAT FREQUENCY RESPONSE AMPLIFIERS 7.3
ioo
lo
loo
FIG 7.53
Fig. 7.83. Curves for determining a for critical damping of two -stage compensated
amplifier (Ref. H15).
the output voltage of the amplifier relative to the input voltage, without feedback.
It is obvious that B = tan- 1(b /a) and IA =- 1/a' + P. I
point is at E,, = 150 volts, and point A is E, b = 100, so that the peak input voltage
7.5 (i) (A) RESISTANCE LOADED TRIODE CATHODE FOLLOWER 391
00 150 200
PLATE TO CATHODE VOLTAGE
FIG. 7.85
Fig. 7.85. Cathode follower characteristics of small general purpose triode. The
solid loadline is for Rk = 15 000 ohms, the dashed loadline is for the addition of an
a.c. shunt load.
amplitude is 50 volts. It is obvious that the opposite swing will extend to E = 200
volts (point B) thus giving a peak-to -peak swing of 2 x 50 = 100 volts. It will be
seen that OA corresponds to E, = - 5 volts, while OB corresponds to - 5.7 volts,
but this is due to the improved linearity with the cathode follower. With plate loading
the loadline would be AD, so that cathode loading causes a longer loadline by the
amount DB, thus resulting in a slightly greater power output.
The voltage gain is given by
A
EB - EA
( 3)
Ecb(B) - Eeb(A)
where subscripts A and B indicate the voltages at points A and B on the loadline.
If the cathode is coupled through a coupling condenser Co to a following grid re-
sistor R, (Fig. 7.8), the a.c. loadline will be drawn through the operating point O with
a slope of -1/R L where R L= R kR,/(R k+ Ru).
The mutual characteristics of cathode followers may be drawn and interpreted by
the method of Ref. C29.
(B) Transformer -coupled triode cathode follower
The sarne form of cathode- follower characteristic may be used when the load is
coupled through a transformer so as to reflect a resistance R k across the primary.
In most cases the load resistance may be taken (at least as a first approximation) to
have the same value as for plate loading. In Fig. 7.86 the loadline AOB extends from
A (the junction of the E, = 0 and the E, b = 100 V curves) to B, on the E, b = 500 V
curve. The operating point O is the intersection of the loadline with the E,5 = 300 V
curve, thus giving a peak input voltage of 200 volts in both directions. The power
output and distortion may be calculated as for plate loading.
If the load resistance R k is increased, the loadline will rotate about O to the position
(say) CD with constant input voltage. This will usually result in a decrease in both
392 (i) (B) TRANSFORMER -COUPLED CATHODE FOLLOWER
7.5
power output and distortion -there is the same advantage (as with plate
keeping the loadline out of the region of bottom curvature. It would loading) in
be possible,
with the higher resistance load, to increase the input voltage so as to extend
line from OC to meet the E1 = 0 curve. the load-
If the load resistance R x is decreased, the loadline will rotate about O to the position
(say) EF. If the input voltage is unaltered, the plate current will cut off
at F and the
valve will run into grid current at E, both resulting in distortion. In
this, the cathode
follower differs from a plate loaded triode, and it is important to ensure that
resistance does not fall appreciably below the designed value. Fortunately,
the load
speaker has an impedance characteristic which, although it rises considerablya loud-
its nominal (400 c /s) value, does not drop appreciably below it. A cathode above
follower
with a loudspeaker load is therefore a good combination from the loading point
view. of
GRID
CURRENT
REGION
1
1
1
/ /
V.
MN IN 1 IiIII
1.1
111III,.11P i 111
ISO
Il111 1
I1.I1 yYMN
11I1 'I1. I1I1
IllllI,1
I
MI ' I
et
á
î<
1.\
II\/1.0II\h.O
I'I
I.IlI. F
p
7
..
.NIi.11III 41111111t
1
11r1
UI
IN
I 1
.
.
loo
01ì101Irl.h
.
IN 1.R
ryl112\r
.11 I..I
.\xLìI.11.I i
I I
I.
M......
I.IIiJIKr} .
1I
.
010.1r11Il
iZ01OI11I I1
ANEW II
á
1114
/,IIIJ/II'I.GO.OIEll
r1.IIIJna.I' O II 111 `
.
.III/Ii.i/"
11
/JG
I.'I.
IIRENE
,
IU\\i.\I, .,
r.
iCUOV
.
AMIN
4
i'G_ /IA
W.
O
..!1I!1Pm IOO
.
1.1'.1I
N.U711UAÌ11IPC.Grl
200
,'
!.r.A, .il!íllí
rn'y 11
current cut -off. A somewhat lower load resistance may be used for fairly high output
voltages.
The transformer primary inductance should have the same value as for plate loading
under similar conditions, if full power output is desired at low frequencies. A trans-
former with low primary inductance will give uniform gain at low level output, but
at high output it will cause plate -current cut -off and grid current as for loadline EF
in Fig. 7.86.
References C14, C16, C17, C29, D1, D2, D3, D8, D9, D10, D11.
(ü) Pentode cathode follower
A pentode (or tetrode) may be connected in several ways arising from the screen
supply and by-pass-
FIG. 7.87
e+ (A)Triode connection (screen tied to plate)
The published " triode " characteristics may generally
be used. If no triode characteristics are available, ascer-
tain the characteristics for the screen voltage corresponding
to the desired conditions (the plate voltage may be equal
INPUT S 0E:5 ouTFUT or higher).
Then I I b -}- I cQ (4)
o g, (pentode g,) x (1k/10 (5)
Fig. 7.87. Pentode ana µ Ee, /E..l (cut -off). (6)
cathode follower, trans-
former- coupled, screen (B) Screen by- passed to earth
by-passed to cathode. There is no exact method using published characteristics.
An approximate method has been described by Shapiro (Ref. D2).
(C) Screen by- passed to cathode-Transformer -coupled (Fig. 737)
The cathode follower characteristics may be drawn by the same method as for a
triode (Fig. 7.88). The procedure for calculating the power output and distortion is :
FIG. 7.88
Fig. 7.88. Cathode follower characteristics of beam power tetrode and typical of all
pentodes. Loadline AOB is normal ; CD high resistance ; EF low resistance illus-
trating grid current and plate current cut -off.
394 (ii) (C) SCREEN BY- PASSED TO CATHODE 7.5
1. To determine the maximum plate voltage and current for the operating point.
Here Eb = 285 V and happens to coincide with the E,b = 300 V curve. If desired,
mark the value of Er corresponding to point O.
2. Try several values of load resistance (i.e. slope of loadline) until an optimum
position for point A is determined, so as to make OA as long as possible, and at the
same time give OB = OA with B slightly above plate -current cut-off.
3. Determine Eb for point A (in this case 50 V) -A is then automatically on the
E,5 = 50 V curve.
4. Draw the other limiting Ebb curve (in this case E,5 = 550 V, giving a peak
amplitude of 250 V each way).
5. Calculate power output and distortion as for a power triode, using the loadline
AOB and the E, b curves.
When a higher resistance load is used with the same input voltage, the loadline
rotates about O to (say) COD, where the power output is lower, the distortion slightly
less, and point C is well negative with respect to E , = 0. When a lower resistance load
is used with the same input voltage, the new loadline may be (say) EOF where E is well
into the grid- current region and F is beyond plate-current cut-off. In this respect the
pentode is similar to the triode, but the pentode is rather more critical regarding low
load resistances ; it is, however, quite satisfactory with a speaker load.
If cathode bias is used, the procedure is as for a plate -loaded power pentode.
(D) Screen by- passed to cathode-Resistance- loaded (Fig. 7.89)
The curves are generally the same as for transformer coupling, and the treatment is
similar to that for resistance -loaded triodes, except that it is necessary to allow for the
d.c. screen current flowing through R k. The screen current may be taken from the
published data for typical operating conditions, or estimated from the ratio of plate
to screen currents. The voltage drop through R k additional to that due to the plate
current will be R k7,2. The procedure is to take as the effective plate supply voltage
the value (Ebb -R 51,2) and then to proceed normally, as for a triode.
(E) Screen voltage from separate supply (Fig. 7.90)
If the screen voltage is obtained from a separate supply, such as a battery, returned
to the cathode, then the voltage from cathode to screen may be maintained constant
without smoothing out the screen current variations. In this case the screen current
does not flow through the cathode resistor R k. The pentode characteristics for a
constant screen voltage may be used without any adjustment for the effects of screen
current.
References D1, D2, D3, D11.
FIG. 7.89 FIG. 7.90
5+
Now consider the general condition when the grid is negative with respect to the
cathode, If the input voltage EÌ is kept constant, then the grid -to- cathode voltage
E, will change when Eb is changed. The relationship is
dE, = IßI( dEb) when E,' is constant.
Therefore dEb = dE, /IßI
Therefore Eb = Ebb -
dEb = Ebb -
AEC /I131 (9)
Eqn. (9) may be used for the calculation of points on the constant E,' characteristics,
as for example :
Ebb = 100 100 100 100 100 volts
!ßi = 0.2 0.2 0.2 0.2 0.2
EÌ = 20 20 20 20 20 volts
- LE, = 0 -1 -2 -3 -4 volts
- .IEc/ißl = 0 -5 -10 -15 -20 volts
Eb = 100 95 90 85 80 volts
These have been plotted in Fig. 7.92. It should be noted that Ebb is the plate
voltage corresponding to zero grid voltage.
(iv) Pentode with voltage feedback -Transformer coupled
Pentodes with a substantial degree of negative voltage feedback (say Ißß z 0.2)
FIG. 7.91 + behave more or less like a cathode follower. Pentodes with a
limited degree of feedback, or none, exhibit some special
peculiarities. Fig. 7.93 shows input voltage curves for three
different values of ß ; the curvature on the ß = - 0.1 curve
E
continues along its whole length while its slope is gradual and
+ bottom curvature extensive.
IPIEb The full effect of a low degree of feedback is illustrated in
Fig. 7.94 where ß = -
0.1. The curves are drawn as in the
_ triode case except for the change in ß. A suitable loadline
Basic AOB has been selected to give
about the same power output
Fig. 7.91. Provided that the load resistance re-
circuit of triode with as without feedback.
negative voltage mains constant, the performance is satisfactory, but the
feedback. value is critical. With constant input voltage (40 volts peak)
Fig. 7.92. Feedback characteristics of small general purpose triode with 20%
negative voltage feedback (ß = -0.2).
396 (iv) PENTODE WITH VOLTAGE FEEDBACK 7.5
FiG. 17.94
Fig. 7.94. Feedback characteristics of beam power tetrode (6L6 Jr 807
Eo.$ = 300 V) and 10% voltage feedback (ß= -0.1).
7.5 (iv) PENTODE WITH VOLTAGE FEEDBACK 397
the valve will run into grid current as soon as the load resistance
either increases or decreases. It is therefore not suitable for
use on a loudspeaker load unless the input voltage is reduced.
A 10% reduction in input voltage (19% in power output) would
be some improvement, but greater reduction is desirable, say
20% to 30% of the input voltage, depending on the operating
conditions. Even with an input reduction giving a power out-
put of only 50% maximum, the pentode is still less flexible
with regard to load resistance than an ordinary triode. This
position improves as the amount of feedback is increased.
See References D4, D12. Fig. 7.95. Basic
circuit of cathode
degenerative triode.
(v) Cathode degenerative triode
The circuit diagram of a cathode degenerative triode is shown in Fig. 7.95 from which
it is evident that
E, = Eí Ek - (10)
The curves of a cathode degenerative triode may be drawn by the procedure out-
lined below, although a special set of characteristics is required for each value of R k.
Fig. 7.96 shows the curves for a typical general purpose triode with RI, = 1000 ohms.
The input voltage curves are straighter than those without feedback, and the two
curves coincide only at Ik = 0. The E;' curves have a lower slope, indicating a
higher plate resistance, than the E, curves.
Take the E, _ - 4 V curve as an example of the calculations. We know that
R k = 1000 ohms and E;' = -4 V. If I = 2 mA, then E k = 2 V and E, _
-Ek=- 4 -2 = -6 V.
b
Ef
Refer to the plate characteristics to find the plate voltage which will give a plate
current of 2 mA at a bias of -6V -the value is Enk = 144 volts.
16
o 400
o 100 200 300 áIX1.
PLATE VOLTS FIG. 7.96
Fig. 7.96. Current feedback characteristics of general purpose triode (6SN7-GT
single unit) with RI, = 1000 ohms.
398 (v) CATHODE DEGENERATIVE TRIODE 7.5
The values of Ei' may be plotted against Ek and therefore also against E, to give the
dynamic characteristic, but this is usually unnecessary. Once the maximum input
voltage has been selected it is only necessary to mark the extremities on the loadline.
For example, if it is desired to swing to E, = 0, then the peak E,' will be 34.7 volts
in each direction. The peak in one direction will be A where E, = 0, and the
will be B, determined by interpolation :
other
E, = -16 Ek =14.7 E,'= -30.6
E, = - 18 Ek =10.1 E,'_ -37.2
Diff. = 6.6
Also the difference between the desired value
is therefore 4.1/6.6.
(- 34.7) and - 30.6 is 4.1. The ratio
Similarly with E k : 14.7 -
10.1 = 4.6.
-
The value of Ek for point B is therefore 14.7 4.6 (4.1/6.6) = 14.7 2.8 = 11.9. -
The voltage gain and power output may be calculated from E, = 5E,,, using the
values of Ek at E, = 34.7 (point A), E,' = 0 (point 0) and E, _ 34.7 (point
The second harmonic distortion may be found from the ratio AO /OB measuredB).
-
volts
H2
(Ek) ; i.e. (64 -
37.3)/(37.3
1%-see Chapter 13 Sect. 2(i).
-
11.9) = 26.7/25.4 = 1.05.
in
Therefore
References D1, D3, D5, D8, D9, DI0, D11.
7.5 (vi) CATHODE DEGENERATIVE PENTODE 399
C14. Schlesinger, K. " Cathode -follower circuits " Proc. I.R.E. 33.12 (Dec. 1945) 843.
C15. Cocking, W. T. " Cathode follower dangers-output circuit capacitance " W.W. 52.3 (March
1946) 79.
C16. Mcllroy, M. S. "The cathode follower driven by a rectangular voltage wave" Proc. I.R.E. 34.11
(Nov. 1946) 848. Letter H. L. Krauss, Proc. I.R.E. 35.7 (July 1947) 694.
C17. Rifkin, M. S. " A graphical analysis of the cathode -coupled amplifier " Comm. 26.12 (Dec.1946)
16.
C18. Smith, F. W., and M. C. Thienpont " Electronic attenuators " Comm. 27.5 (May 1947) 20.
C19. Kline, M. B. " Cathode follower nomograph " Elect. 20.5 (May 1947) 136.
C20. Reich, H. J. " Input admittance of cathode -follower amplifiers " Proc. I.R.E. 35.6 (June 1947) 573.
C21. Kline, M. B. " Cathode follower nomograph for pentodes " Elect. 20.6 (June 1947) 136. Letter
P. M. Hackett 20.8 (Aug. 1947) 250.
C22. Audio design notes, " The cathode follower " (output impedance graph and power output) Audio
Eng. 31.5 (June 1947) 39.
C23. Kline, M. B. " Cathode follower impedance nomograph " Elect. 20.7 (July 1947) 130.
C24. Houck, G. " Gain chart for cathode followers " Tele -tech. 6.8 (Aug. 1947) 54.
C25. E.M.I. Laboratories " Cathode -follower circuit using screen-grid valves," Electronic Eng. 19.229
(March 1947) 97.
C26. Yu, Y. P. " Cathode -follower couplings in d -c amplifiers " Elect. 19.8 (Aug. 1946) 99.
C27. Diamond, J. M. " Circle diagrams for cathode followers " Proc. I.R.E. 36.3 (March 1948) 416.
C28. Sowerby, J. McG. "Notes on the ,cathode follower" W.W. 54.9 (Sept. 1948) 321.
C29. Parker, E. " The cathode- follower " Electronic Eng. (1) 20.239 (Jan. 1948) 12 ; (2) 20.240 (Feb.
1948) 55 ; (3) 20.241 (Mar. 1948) 92 ; (4) 20.242 (April 1948) 126.
C30. Diamond, J. M. " Circle diagrams for cathode followers " Proc. I.R.E. 36.3 (March 1948) 416.
C31. Amos, S. W. " Valves with resistive loads " W.E. 26.307 (April 1949) 119.
C32. Scroggie, M. G. " RC coupled power stage " W.E. 27.318 (March 1950) 81.
Books
C33. Reich, H. J. " Theory and Application of Electron Tubes " pp. 164 -174.
C34. Sturley, K. R. " Radio Receiver Design " Vol. 2 pp. 118 -120.
(D) REFERENCES TO VALVE CHARACTERISTICS WITH FEEDBACK
Dl. Pratt, J. H. " The equivalent characteristics of vacuum tubes operating in feedback circuits"
R.C.A. Rev. 6.1 (July 1941) 102.
D2. Shapiro, D. L. " The graphical design of cathode -output amplifiers " Proc. I.R.E. 32.5 (May
1944) 263. Correction 32.8 (Aug. 1944) 482.
D3. Schlesinger, K. " Cathode -follower circuits " Proc. I.R.E. 33.12 (Dec. 1945) 843.
D4. " Cathode Ray," " Negative feedback -(2) its effect on optimum load and distortion," W.W. 52.3
(March 1946) 76.
D5. Middleton, R. G. " Graphical analysis of degenerative amplifiers " Radio, 30.3 (March 1946) 23.
D6. Radio Design Worksheet No. 51 " Graphical analysis of the cathode -coupled amplifier " Radio
30.8 (Aug. 1946) 20.
D7. Radio Design Worksheet No. 53 " Graphics of negative feedback in cascade," Radio, 30.10 (Oct.
1946) 17.
D8. Mcllroy, M. S. " The cathode follower driven by a rectangular voltage wave " Proc. I.R.E. 34.11
(Nov. 1946) 848. Letter H. L. Krauss, 35.7 (July 1947) 694.
D9. Krauss, H. L. " Graphical solutions for cathode followers " Elect. 20.1 (Jan. 1947) 116 (with
bibliography).
D10. Huber, W. A. " Graphical analysis of cathode -biased degenerative amplifiers " Proc. I.R.E. 35.3
(March 1947) 265.'
D11. Lonsdale, E. M., and W. F. Main " A method of graphically analyzing cathode- degenerated
amplifier stages " Proc. I.R.E. 35.9 (Sept. 1947) 981.
D12. Langford- Smith, F. " The Design of a high -fidelity amplifier, (2) Negative feedback beam power
amplifiers and the loudspeaker" Radiotronica No. 125 (May /June 1947) 53.
D13. Schade, O. H. " Beam power tubes " Proc. I.R.E. 26.2 (Feb. 1938) 137.
See also C31.
Books
Sturley " Radio Receiver Design " Vol. 2 pp. 126-128.
(E) REFERENCES TO VOLTAGE FEEDBACK FROM PLATE WITH R.C.C. INPUT
El. Laboratory staff of Amalgamated Wireless Valve Company, " Inverse feedback," Radio Review
of Australia, 5.3 (March 1937) 64.
E2. R.C.A. " Application Note on an inverse feedback circuit for resistance -coupled amplifiers " No. 93
(June 8, 1938).
E3. Laboratory staff of Amalgamated Wireless Valve Company, " Negative feedback in RC amplifiers "
W.W. 43.20 (Nov. 17, 1938) 437.
E4. Mezger, G. R. " Feedback amplifier for C-R oscilloscopes " Elect. 17.4 (April 1944) 126.
E5. Winternitz, T. W. " A variation on the gain formula for feedback amplifiers for a certain driving -
impedance configuration" Proc. I.R.E. 34.9 (Sept. 1946) 639.
(F) REFERENCES TO PRACTICAL FEEDBACK AMPLIFIERS
(additional to those listed in A)
Fl. Terman, F. E. " Feedback amplifier design " Elect. 10.1 (Jan. 1937) 12.
F2. Day, J. R., and J. B. Russell, " Practical feedback amplifiers " Elect. 10.4 (April 1937) 16.
F3. Reference Chart " Methods of applying negative feedback " Electronic Eng. 17.213 (Nov. 1945) 770
F4. Williamson, D. T. N. " Design of a high quality amplifier " W.W. 53.4 (April 1947) 118 ; 53.5.
(May 1947) 161.
F5. Aston, R. H. " Radiotron 30 watt amplifier A513 " Radiotronics No. 124 (March April 1947) 19.
F6. Langford- Smith, F., and R. H. Aston " The design of a high fidelity amplifier-(3) A design using
push -pull triodes with negative feedback," Radiotronica No. 128 (Nov. /Dec. 1947) 99.
F7. Booth, H. (letter) " Amplifiers with negative feedback " W.W. 54.6 (June 1948) 233.
F8. Sarser, D., and M. C. Sprinkle "Musician's Amplifier" (Williamson circuit) Audio Eng. 33.11
(Nov. 1949) 11.
402 REFERENCES 7.6
F9. Williamson, D. T. N. " High -quality amplifier -new version " W.W. 55.8 (Aug. 1949) 282 ; 55.10
(Oct. 1949) 365 ; 55.11 (Nov. 1949) 423.
F10. Miller, J. M. " Combining positive and negative feedback " Elect. 23.3 (March 1950) 106.
F11. Roddam, T. " More about positive feedback- applied locally it can improve negative-feedback
amplifier performance " W.W. 46.7 (July 1950) 242.
F12. Llewellyn, F. B. U.S. Patent 2 245 598.
F13. Shepard, F. H. U.S. Patents 2 313 096, 2 313 097, 2 313 098.
F14. Seymour, R. A., and D. G. Tucker " A choke-coupled phase- invertor of high accuracy " Elec-
tronic Eng. 23.276 (Feb. 1951) 64.
F15. Banthorpe, C. H. " Positive feedback in a.f. amplifiers " Electronic Eng. 22.273 (Nov. 1950) 473 ;
correspondence 23.276 (Feb. 1951) 70.
F16. Cross, R. M. " Positive feedback in a.f. amplifiers " Electronic Eng. 23.276 (Feb. 1951) 70.
F17. Roddam, T. " Output impedance control " W.W. 56.2 (Feb. 1950) 48.
Fla. Griffiths, E. " Negative feedback -its effect on input impedance and distortion " W.W. 56.3
(March 1950) 111.
F19. Mitchell, R. M. " Audio amplifier damping " Elect. 24.9 (Sept. 1951) 128.
(G) REFERENCES TO CATHODE COUPLED AMPLIFIERS
Gl. Schmitt, O. H. " Cathode phase inversion " Jour. Sci. Instr. 15 (March 1938) 100.
G2. Williams, E. " The cathode coupled double -triode stage " Electronic Eng. 16.195 (May 1944) 509.
G3. Noltingk, B. E. (letter) " The cathode coupled double-triode stage " Electronic Eng. 16.196 (June
1944) 34.
G4. Butler, F. " Cathode coupled oscillators " W.E. 21.254 (Nov. 1944) 521.
G5. Sziklai, G. E., and A. C. Schroeder, " Cathode- coupled wide -band amplifiers " Proc. I.R.E.
33.10 (Oct. 1945) 701.
G6. Amos, S. W. " Push -pull circuit analysis : Cathode coupled output stage " W.E. 23.269 (Feb.
1946) 43.
G7. Pullen, K. A. "The cathode- coupled amplifier " Proc. I.R.E. 34.6 (June, 1946) 402. Discussion
35.12 (Dec. 1947) 1510.
G8. Le Bel, C. J. " Graphical characteristics of cathode - coupled triode amplifiers " Audio Eng. 31.3
(May 1947) 40.
G9. Wheeler, M. S. " An analysis of three self-balancing phase inverters " Proc. I.R.E. 34.2 (Feb. 1946)
67P. -
G10. Bird, E. K. M. " A note on the phase -splitting amplifier circuit " Electronic Eng. 17.198 (Aug.
1944) 103 (Schmitt circuit).
GI1. Schmitt, O. H. (letter) "Re-an original phase inverter " Radio Craft, 13.2 (Aug. 1941) 68.
G12. Cocking, W. T. " Phase splitting in push -pull amplifiers " W.W. 44.15 (April 13th, 1939) 340.
G13. Korman, N. I. (letter) "Cathode- coupled triode amplifiers" Proc. I.R.E. 35.1 (Jan. 1947) 48.
G14. Clare, J. D. "The twin triode phase -splitting amplifier " Electronic Eng. 19.228 (Feb. 1947) 62.
G15. Campbell, N. R., V. J. Frauds and E. G. James " Common-cathode amplifiers " W.E. 25.297
(June 1948) 180.
G16. Rifkin, M. S. " A graphical analysis of the cathode-coupled amplifier " Comm. 26.12 (Dec. 1946)
16.
G17. Ross, S. G. F. " Design of cathode- coupled amplifiers " W.E. 27.322 (July 1950).
G18. Lyddiard, J. A. "Cathode coupled amplifier-analysis and design" W.E. 29.342 (March 1952) 63.
(H) STABILITY AND MAXIMAL FLATNESS
Hl. Nyquist, H. " Regeneration theory," B.S.T.J. 11.1 (Jan. 1932) 126.
H2. Peterson, E., J. G. Kreer and L. A. Ware, " Regeneration theory and experiment " Proc. I.R.E.
22.10 (Oct. 1934) 1191.
H3. Reid, D. G. " The necessary conditions for instability (or self -oscillation) of electrical circuits "
W.E. 14.170 (Nov. 1937) 588.
H4. Terman, F. E. and Wen -Yuan Pan " Frequency response characteristic of amplifiers employing
negative feedback " Comm. 19.3 (March 1939) 5.
H5. Everest, F. A., and H. R. Johnston " The application of feedback to wide -band output amplifiers "
Proc. I.R.E. 28.2 (Feb. 1940) 71.
H6. Bode, H. W. "Relations between attenuation and phase in feedback amplifier design" B.S.T.J.
19.3 (July 1940) 421. Also U.S.A. Patent, 2 123 178.
H7. Watton, A. " Modulated beam cathode -ray phase meter " Proc. I.R.E. 32.5 (May 1944) 268.
H8. Becker, S. " The stability factor of negative feedback in amplifiers " Proc. I.R.E. 32.6 (June 1944)
351.
H9. Learned, V. " Corrective networks for feedback circuits " Proc. I.R.E. 32.7 (July 1944) 403.
H10. Terman, F. E. " Radio Engineers Handbook " (McGraw -Hill, 1943) pp. 218 -226 ; 396 -402.
H11. Bode, H. W. (book) "Network Analysis and Feedback Amplifier Design" (D. Van Nostrand
Co. Inc. New York, 1945).
H12. Brockelsby, C. F. " Negative feedback amplifiers- conditions for maximal flatness " W.E. 26.305
(Feb. 1949) 43.
Letters T. S. McLeod, W.E. 26.308 (May 1949) 176 ; 26.312 (Sept. 1949) 312 ; C. F. Brockelsby
26.310 (July 1949) 247 ; 26.314 (Nov. 1949) 380.
H13. Mayr, H. " Feedback amplifier design -conditions for flat response " W.E. 26.312 (Sept. 1949)
297.
H14. West, J. C. " The Nyquist criterion of stability " Electronic Eng. 22.267 (May 1950) 169.
H15. Flood, J. E. " Negative feedback amplifiers-conditions for critical damping " W.E. 27.322 (July
1950) 201.
H16. Bothwell, F. E. " Nyquist diagrams and the Routh -Hurwitz stability criterion " Proc. I.R.E.
38.11 (Nov. 1950) 1345.
H17. Barter, L. D. " Graphical solution for feedback amplifiers " Elect. 23.11 (Nov. 1950) 204.
H18. Roddam, T. " Stabilizing feedback amplifiers " W.W. 57.3 (March 1951) 112.
HI9. Ragazzini, J. R., and L. A. Zadeh " A wide band audio phase meter " Rev. Sci. Inst. 21.2 (Feb.
1950) 145.
H20. Baldwin, T., and J. H. Littlewood " A null method measuring the gain and phase shift of com-
paratively low frequency amplifiers " Electronic Eng. 23.276 (Feb. 1951) 65.
Also Al, A6, A7, All.
H21. Lynch, W. A. "The stability problem in feedback amplifiers" Proc. I.R.E. 39.9 (Sept. 1951) 1000.
Additional references will be found in the Supplement commencing on page 1475.
CHAPTER 8
-{
but move to and fro through a limited path in the direction of motion of the wave.
403
404 (i) WAVE MOTION 8.1
Electromagnetic waves are also transverse. However, radiated electromagnetic
energy consists of two component fields, respectively magnetic and electric, both in
phase, existing at right angles to each other and to the direction of motion of the
wave. In contrast to sound waves in air, electromagnetic waves in free space have
a velocity of 3 x 108 metres per second.
WAVE MOTION
(A)
o*N
M
(e) O
o M
(c)
N
CHARGED
PARTICLE
MOTION
FIG. 8.2
(i) Introduction
There are several methods of ccnveying information using the radiated waves.
By switching on and off the transmitter in accordance with a prearranged code, audible
sounds will be heard in a suitable receiver. This is known as radio telegraphy.
Alternatively the intelligence in the form of speech or music can be superimposed
on the transmitted radio frequency by a process of " modulation." This is referred
to as radio telephony. There are four major types of this latter system of transmis-
sion, known as amplitude, frequency, phase and pulse modulation (Refs. 17, 18,
33).
(iii) Radiotelephony
(A) Amplitude Modulation (A -M)
This is commonly used in local medium frequency and long distance shortwave
broadcast transmitters. The amplitude of the r -f radiated wave (the carrier)
is
varied at an audio frequency rate according to similar variations of the intelligence
which it is required to transmit. This is usually accomplished by varying
the plate
voltage of a r-f amplifier by the audio signal. The r-f amplifier may be in the
final
transmitter stage, in which case a large audio power is required for complete modula
tion-50% of the modulated r-f stage plate input power. This is called " high --
level " modulation. As an alternative, modulation can be employed at an early
in the transmitter, and linear r-f amplifiers used to raise the power level stage
to that re-
quired. Much smaller powers are necessary for this " low- level " modulation
system, but the Class B linear r -f amplifiers are less efficient than the Class C counter-
parts used in the " high -level " method of modulation. Both systems are used
ex-
tensively, but most high power transmitters employ " high -level " modulation. With
amplitude modulation, the power output of the transmitter varies, but the phase and
frequency of the carrier remain unaffected (Refs. 12, 13).
P -M is not yet in general use and will therefore not be considered further (Refs.
19, 20, 21, 22, 23, 24, 31, 33).
SECTION 3 : REFERENCES
1. Dow, W. G. (Book) " Fundamentals of Engineering Electronics " Chapter 17. John Wiley and
Sons Inc. N.Y. 1937.
2. Everitt, W. L. (Editor) " Fundamentals of Radio " Chapter 9. Prentice -Hall, Inc. N.Y. 1943.
3. Skilling, H. H. (Book) " Fundamentals of Electric Waves " 2nd edit. John Wiley and Sons Inc.
N.Y. 1948.
4. Spectrum Chart " Electronics Buyers Guide " McGraw -Hill Book Co. N.Y. 19B.6B (June 1946) 80.
5. Bennington, T. W. (Book) " Short-Wave Radio and the Ionosphere ". Iliffe and Sons Ltd. London
1950.
6. "CRPL-D Basic Radio Propagation Predictions " Sup. of Docs. U.S. Govt. Printing Office Wash.
D.C.
7. " I.R.P.L. Radio Propagation Handbook " Part 1, 1943.
8. " Radio Propagation Bulletins " I.P.S. Commonwealth Observatory, Canberra, Australia
9. " Handbook for use with Radio Propagation Bulletins " Radio Research Board C.S.I.R., Australia,
1945.
10. Foley, W. R. " Forecasting Long Distance Transmission " Q.S.T. 30.2 (Feb. 1946) 36.
11. Nelson and Hornung (Book) "Practical Radio Communication" 2nd Edit. McGraw -Hill Book
Co. 1943.
12. Sandeman, E. K. (Book) " Radio Engineering Vol. 1, lit edit. Chapman and Hall London 1947.
13. Ladner and Stoner (Book) "Short Wave Wireless Communication.'
14. Tibbs, C. E. (Book) "Frequency Modulation Engineering Chapman and Hall, London, 1947.
15. Rider, J. F. (Book) " Frequency Modulation " Rider Publisher Inc. N.Y. 1940.
16. Hund, A. (Book) " Frequency Modulation " McGraw-Hill Book Co. N.Y. 1942.
17. Hund, A. " Amplitude frequency and phase modulation relations " Elect. (Sept. 1942) 48.
18. Smith, E. J. "Theoretical signal -to noise ratios "(comparison between various modulation systems)
Elect. 19.6 (June 1946) 150.
19. Tibbs, C. " hase and frequency modulation" W.W. 48.9 (Sept. 1942) 210.
20. Farren, L. I. " Phase detectors-some theoretical and practical aspects " W.E. 23.279 (Dec. 1946)
330.
21. Hund, A. " Frequency and phase modulation " Proc. I.R.S. 32.9 (Sept. 1944) 572.
22. Stockman, H. and G. Hok, " Frequency and phase modulation " Proc. I.R.H. 33.1 (Jan. 1945) 66.
23. Jaffe, D. L., ID Pollack, A. Hund and B. E. Montgomery (letter) "Frequency and phase modula-
tion " Proc. I.R.S. 33.3 (March 1945) 200 ; 33.7 (July 1945) 487. See Refs. 21, 22, 31.
24. Crosby, M. G. "Exalted-carrier amplitude and phase -modulation reception" Proc. I.R.E. 33.9
(Sept. 1945) 581.
25. I)eloraine, E. M., and E. Labin, "Pulse-time modulation" Elect. 18.1 (Jan. 1945) 100.
26. Post war prospects for modulation pulse transmission' Elect. 18.7 (July 1945) 244.
27. "Pulse-width modulation-its basic principles described" W.W. 51.12 (Dec. 1945) 361.
28. "Pulse -time modulation -en explanation of the principle" W.W. 52.2 (Feb. 1946) 45.
29. Roberts, F. F , and J. C. Simmonds " Multichannel communication systems " preliminary investi-
gation of systems based upon modulated pulses -W.E. 22.266 (Nov. 1945) 538 ; 22.267 (Nov.
1945) 576. See (30) below.
30. Fitch, E., and E. R. Kraemer, "Pulse modulation" (Ref. 29 above) letters W.E. 23.275 (Aug.
1946) 231 23.277 (Oct. 1946) 288.
31. Stockman, H., and G. Hok " A note on frequency modulation terminology " letter to Proc. I.R.E.
32.3 (March 1944) 181. See Refs. 21 and 22 above.
32. Armstrong, E. H. " A method of reducing disturbances in radio signalling by a system of frequency
modulation" Proc. I.R.E. 24.5 (May 1936) 689.
33. "Reference data for radio engineers " (Federal Telephone and Radio Corporation, 3rd ed. 1949)
CHAPTER 9
TUNED CIRCUITS
based on original chapter in previous edition written by L. G. Dobbie, M.E.
Revised by G. Builder, B.Sc., Ph.D., F. Inst. P.
Section Page
1. Introduction ... 407
2. Damped oscillations 408
3. Series resonance ... 409
4. Parallel resonance ... ... 410
5. General case of series resonance 412
6. Selectivity and gain ... ... 412
7. Selectivity- graphical methods ... 416
8. Coupling of circuits 418
9. Response of stages in cascade ... ... 421
10. Universal selectivity curves ... 421
11. Summary of formulae ... ... 423
12. References 427
SECTION 1 : INTRODUCTION
When a violin is tuned, the tensions of its strings are adjusted to permit vibration
at particular frequencies. In radio, when an arrangement of L, C, and R responds
to particular frequencies, it is called a " tuned " circuit.
In principle, the tuned circuit is similar to a pendulum or violin string, tuning fork,
etc.-it has the property of storing energy in an oscillating (vibrating) state, regularly
changing from kinetic form (magnetic field when current flows through the coil)
to potential form (electric field, when the condenser is charged) and back again at a
frequency called the natural resonant frequency.
FIG. 9.1 In Fig. 9.1 let the condenser C be charged. It will discharge its
energy through the inductance L, causing the current to increase all
the while, until it reaches the maximum when there is no potential
across C. At that instant the energy is all magnetic, and the current
continues, fed by the magnetic field, to build a voltage of reversed
polarity across C. When all the energy has been transferred from L to C,
the voltage across C has its original value, but is reversed in sign, and the current is
diminished to zero. The process then reverses, and repeats itself indefinitely, cycle
after cycle.
407
408 DAMPED OSCILLATIONS 9.2
LC 4LQ
The quantity f = w /27r is called the natural resonant frequency. In the above
formulae L is in henrys, C in farads, r in ohms, w in radians per second and
cycles per second.
in f
The resonant frequency f o (corresponding to r = 0) of the circuit is defined by
2irf, = wo = 1/1 /LC (2)
The ratio of the natural resonant frequency to the resonant frequency is
f/f,, = V71
co oI
-
2nf oI
1 /(4QE) (3)
where Q =
r r
reactance of the coil at resonant frequency
coil resistance
The voltages across the several parts of the circuit under this condition of
series resonance are (Fig. 9.2a) :
Q - r r'v C wo Cr
(12)
It is shown in Chapter 4 Sect. 5(vi) that the power factor is approximately equal
to 1/Q ; when Q is greater than 7 the error is less than 1 %.
See also Sect. 11 Summary of Formulae.
The general treatment of series circuits with L, C and R is given in Chapter 4
Sect. 6(ii).
410 PARALLEL RESONANCE 9.4
ti wol. Q~
/1 //w-C
Y
where Q = w oL /r as before. The current in the inductance and resistance
is very closely equal in value to that in the capacitance,
IL ti -I, E /woL ti woCE - (19)
and is Q times larger than the current in the external circuit
The formulae given so far hold only when the condenser loss is negligible. In
order to generalize our expressions in a simple manner let us first consider the two
circuits shown in Fig. 9.3a and Fig. 9.3b. At (parallel) resonance for Fig. 9.3a we
have found that the current I in the external circuit is
I ECr /L E /Qwoh etc. ;
*Parallel resonance may be defined either as
.
(a) the frequency at which the parallel impedance of the circuit is a maximum, or
(b) the frequency at which the equivalent reactance of the complete parallel circuit becomes zero
(i.e. when the impedance has unity power factor and acts as though it were a pure resistance at
the resonant frequency). This can also be expressed by saying that the parallel circuit has
zero susceptance at the resonant frequency.
For further details see Chapter 4 Sect. 6(iii) and (iv).
Definition (b) is used in this chapter.
9.4 PARALLEL RESONANCE 411
We can now consider the important practical case of the circuit shown in Fig. 9.4,
in which a resistance R appears in shunt with the condenser C. R represents the
effect of all insulation losses in condenser, coil, wiring, switches and valves, together
with the plate or input resistance of the valves.
FIG 9.3 FIG. 9.4
(a ) (b)
In the present case we have R in parallel with our equivalent parallel coil resistance
R , of Fig. 9.3b. The resultant parallel resistance at resonance, which we will call
the resonant impedance is, of course,
1
(21)
RD 1/R -f- Cr /L
Therefore, the resultant value of Q is
C 1 1
Q L . RD (woL /R) (22)
r \/C + (r /w,L)
Note that at the resonant frequency the expression AL /C is equal to the reactance
of the inductance and also that of the condenser, i.e.
w,L = 1/L /C = 1/ (,),,C. (23)
See also Sect. 11 Summary of Formulae.
412 GENERAL CASE OF SERIES RESONANCE 9.5
A°
A
z -RD
r' Z
_ ,1 + Q2(f -f°)2
f° f
(24)
where A, is the voltage gain at the resonant frequency f ° and A the gain at frequency f.
Note that A °/A is the ratio of current at resonance to that at frequency f in the
series case, and the ratio of total current at frequency f to that at the resonant fre-
quency in the parallel case (see below).
The phase angle between the applied voltage and the total current is such that
tan ¢ = + Q l
f° f ) (25)
the positive sign pertaining to the series case and the negative sign to the parallel case.
The equation (24) leads to a simple method for determining Q from the response
curve. We see that when Q[(f/f °) - (f ,If)] = ± 1 the total current will be de-
creased by the factor \/2 in the series case (i.e. decreased to 70.7% of the resonance
value), and increased by the same factor in the parallel case. It will be observed that
at either of the frequencies satisfying this condition the phase angle ¢ is numerically
equal to 45°, as tan 0 = f 1, and that the resistance r' or RD is equal to the reactance.
9.8 (i) SINGLE TUNED CIRCUIT 413
The condition
-
Q(f/fo f./f) _ +
may be written
1
(26)
gmkRD1/LI
/L2
- + k'(RD/r,) (Li /La)
1
When r, is
( )
very much greater than w,'M' /r[= k'RD(Li /L,)] we have simply
I, gm . eo (45)
I, gm . ec . Mw, /r, (46)
and e, /et = A, ti gmw,MQ, = gm . kRD1/Ll /La (47)
If the inductances L, and L, have the same ratio of diameter to length, or form
factor, and the turns are N1 and N, respectively, then 1/L, /L, in the above formulae
may be replaced by N, /N,. The plate resistance r, in parallel with the primary is
reflected as a series resistance M' w,' /r, into the secondary. If the value of this
reflected resistance is greater than say 5% of r, its effect should be taken into account
when computing the selectivity of the secondary. This selectivity, then, is deter-
mined by means of the formula
z/r' = 1/1 + Q,'l(f/fo f./f12
where r' = r + M'w,'
i
/r - (48)
and Q$ = w,L,
r' (1 /Q,) + (k'L,wo /r,)
Note that Qs and r' must not be used when calculating the gain A,.
FIG. 9.6
FIG. 9.7
Fig. 9.6. Amplifier stage using a high frequency transformer with a tuned secondary.
Fig. 9.7. Amplifier stage using a double tuned high frequency transformer.
of view by Aiken (Ref. C2). For the design procedure see Chapter 26 Sect. 4. In
this case the resistance in parallel with the primary at resonance is
LI /Cira + LI /CIR1 = LI'CIR', where 1 R' = 1 /r, 11R1 ;
and the resistance of the secondary is L, /C,R so that the resistance reflected into
the primary is w °2M2C,R,;'L2. The primary current I, is, therefore,
g,,,e, 1 (49)
I' - w °C1 LI /CIR' + WO,M2C,R2'L2
The secondary current I, can readily be shown to be
I. = (Mw °C,R,'L,)Ip, (50)
and the induced voltage e, across L, and C, to be
e, = L,w,I, = (MC,R,w °2)I °.
Therefore the voltage gain at the resonant frequency, A °, is given by
A° - Li, R'R,C,gj'1w°
- w°2M2C1 /L,
(51)
where, as before, k = MIA/L,L and Q', Q2 have their usual meaning, i.e. R' /L,w°
and R, /L, w ° respectively.
The expression for calculating the selectivity is lengthy and complicated and a
graphical treatment described later is preferable (Sect. 7).
As k is increased from low values, the gain increases until k\/Q'Q, = 1, after
which it decreases. This value of k = 1 /1/Q'Q, is known as the critical coupling
factor (k,).
(iv) Coupled circuits of equal Q
When the primary and secondary circuits are identical, and the coupling factor is
equal to k,, and the plate and grid return resistances are very high, we see that the
voltage gain obtained is exactly half that with a single tuned circuit. The critical
coupling factor in this case, of course, is
k, = 1/Q where Q = Q' = Q2.
For values of k less than k,, the response curve (gain versus frequency) has a single
maximum at f °, the resonant frequency of each of the circuits. When k exceeds
the critical value, however, the amplification curve becomes double -humped, i.e.
there are two frequencies of maximum response, and these are separated by equal
amounts above and below f °. The distance between these two peaks increases with
k, and very approximately we find that
(12 - f1) /f°
\/k2 1 /Q2 - (54)
s. 1/k2 - k,2, -
f,) is the band
(55)
when f, and f2 are the frequencies for maximum response, i.e. (f5
width between peaks. The amplitude of these two peaks is substantially the same
as the maximum possible gain g,,,R /2, where R = R,
Frequently, an approximate formula for band -width is used :
(f2 -f,)lf° k ti
of the valley atf and
(56)
hence
While k largely determines the band -width, the depth °,
the uniformity of the response in the pass band of frequencies, is determined by the
relation of Q to k. For a constant value of k (above critical coupling) the dip be-
comes more pronounced as Q is increased, while the frequency separation between
peaks becomes greater ; conversely as Q is decreased the dip becomes less pronounced
and the frequency separation between peaks becomes less. The ratio of the response
at f° to that at the two peaks is found to be 2.b /(1 + b'), where b k /k°.
416 (iv) COUPLED CIRCUITS OF EQUAL Q 9.6
A value for k in the order of 1.5 times critical, i.e. kQ = 1.5, is often used for i -f
amplifiers requiring band pass characteristics. However, the exact value chosen for
kQ depends of course, on the bandwidth requirements.
Further points on the resonance curve can be obtained from the result that the fre-
quency band width between the points on either flank of the resonance curve, at which
the response is equal to the minimum in the " valley " between the two peaks, is A/
times the peak separation. It can be shown also that, in general, the gain at any
frequency f is given by
A, r / Q2Y2 2 r 2QY 2 l (57)
A .v L1 1 + k2Q2 + Li + k2Q2J
where Y = f/f, -
f, /f.
This expression may well be solved graphically according to a procedure developed
by Beatty (Ref. C7) ; this procedure will be described below.
In this table A ,D, is the optimum value of the gain, i.e. the gain at resonant frequency
with critical coupling (= gm1/R,R2 /2)
The gain at f, when k> k, is given by [2b /(1 + b2)]A,,,, where b = k /k,,, as
in the case of equal Q's ; as the gain at the peaks is less than A,,,, however, the re-
sponse curve is flatter in the present case.
See also Sect. 11 Summary of Formulae.
there occurs the maximum value of the response, and the gain falls away on both
sides. This curve may be computed from eqn. (24), namely
AO /A =1/1 +Q2Y2,
and tan d' = + QY, from eqn. (25)
where Y = (f/fo - fO /f).
These formulae, however, lend themselves to a simple graphical treatment as in-
dicated in Fig. 9.9.
The ratio of the gain A. at resonance to the gain A at any other frequency may be
plotted as a vector quantity, OP in Fig. 9.9, having both magnitude and phase. At
resonance, when the frequency is f o, it becomes OP o in Fig. 9.9, where OP, is of unit
length since A o/A is then equal to unity. At any other frequency f, the ratio A, /A
is then given by OP where the point P is fixed by the relation
length P0P = Q(f/f, -,f, /f).
The phase angle is the angle POOP.
Near resonance, when f is nearly equal to f this may be approximated by
length P,P ,.2Qdflf0 where df = f f, -
so that df measures the amount of detuning. Thus, near resonance, the length P
nearly proportional to the amount of detuning.
Po P
OPO I
OP=Áo
FIG. 9.11 OR
P,P=p((f
f,
-l1f J
WHERE :- Po
2Tr f,L
------
Y° [7,-1J
I K. COEFFICIENT OF
I COUPLING
graph to determine the gain, we first compute the quantity OX = 2QY /(1 + k2(2t),
then from X draw a line perpendicular to OX to cut the curve at P. The line OP
represents the gain (A, /A) while P OOP is the phase angle.
The form of the parabola depends upon the magnitude of kQ. It is found that
when OT < -/2, corresponding to kQ > 1, there are two frequencies of maximum
gain as shown by the two vectors marked OP' in Fig. 9.10. When the attenuation
is plotted against df as in Fig. 9.11, it becomes clear that a much flatter top may be
obtained by using coupled pairs of circuits than by using single tuned circuits. Fig.
9.11 serves also to show the variation in band width with variations of OT (i.e. changes
of kQ). It will be seen that the shape of the skirt of the curve is practically inde-
pendent of the value of kQ.
See also Sect. 11 Summary of Formulae.
418 (i) MUTUAL INDUCTIVE COUPLING 9.8
K=
IFx
C,C=
Fig. 9.12. High impe- Fig. 9.13. High impe- Fig. 9.14. Low
dance capacitive coup- dance inductive coup- impedance capa-
ling. ling. citive coupling.
A fifth type is link coupling shown in Fig. 9.16A, in which a relatively small
coupling inductance L1' is coupled to L, and similarly L2' to L2 and L1' is connected
directly in series with L2'. The behaviour of this circuit is the same as that to be
described for Fig. 9.15.
" The coupling between two circuits, from a general point of view, is the relation
between the possible rate of transfer of energy and the stored energy of the circuits "
(Ref. C5, Sept. 1932).
From this definition it follows that for low impedance coupling
k -
X. X m when X1 and X2 > Xm
/ (X1
+ Xm)(X2 + Xm)
v
rte
V X1X2
and that for high impedance coupling
V l + Xm)(X2 + Xm) AV X1X2
Xm Xm
when X1 X2» Xm
where Xn, is the coupling reactance and X1 and X2 are the effective reactances of either
the coils or the condensers* with which the circuits are tuned.
X1 and X2 must be of the same " kind " (i.e. either inductive or capacitive) as X m
9.8 (ii) MISCELLANEOUS METHODS OF COUPLING 419
The effective reactances X, and X, in the high impedance case are calculated by
regarding the actual tuning reactances (Liu), and L2 o or 1/C, w o and 1/ C2 w 0) as
CO
being in parallel with the coupling reactance Xm ; while in the low impedance case,
X, and X, are calculated by taking X m to be in series with the actual tuning reactances.
FIG.9.16A
K'ITT;
Fig. 9.12 Cm Cm
when Cm < (C CO
V(C, + Cm)(C2 + Cm) 1/C1C2
(at w,), while the other peak moves to one side : the shift of the semi-stationary
peak depends upon the series resistances of the two circuits and decreases with them,
being zero in the ideal case r, = r2 = 0 : the second peak is lower in frequency in
Figs. 9.12 and 9.15, but higher in Figs. 9.13, 9.14 and 9.16A. The selectivity and
bandwidth (between peaks) may be calculated from the formulae (57) and (55) re-
spectively-already quoted for transformer coupling-provided the appropriate value
of k is used. It is theoretically possible, although seldom convenient, to combine
two types of coupling in equal amounts to give symmetrical separation of the two peaks.
Normally, when this is required simple mutual inductive coupling is used.
It may be shown that for all types of coupling the centre frequency is determined
by the effective reactances obtained by taking the coupling reactance into account.
Thus, for example, for Fig. 9.15, (L2 + L,)C2w,2 = 1, where w, corresponds to
the frequency of the minimum between the peaks ; while for Fig. 9.16A,
(L2 -M,2 ¡L,)C2w,2 = 1.
In practice, a reasonably constant band width over the tuning range can be obtained
by a suitable combination of the types of coupling already described. Two common
arrangements are shown in Figs. 9.16B and 9.16C.
For the circuit shown in Fig. 9.16B the coupling reactance X,, = wM + 1/ wC,,, ,
while for that shown in the Fig. 9.16C it is
1 CH
wCL + wC,C2
The corresponding coupling factors are respectively
M 1 /w2Cm
ti (k < 0.05),
VL,Ls
and k N CH I/C,Cs
(k < 0.05).
1/C,C, CL
Aiken (Ref. C2) gives a practical design procedure for obtaining the values of the com-
ponents which give the best average results over the whole tuning range.
When k has been determined, the band width and selectivity may be calculated
from eqns. (55) and (57) given earlier. Also, for circuits with the same values of Q,
a suitable value of k is given by kQ = 0.5 -as in the case of transformer coupling
when the greatest possible selectivity without notable loss of gain is desired.
-
9.9 RESPONSE OF STAGES IN CASCADE 421
In Fig. 9.16D there is a combination of mutual inductive coupling and top capacitive
coupling, as commonly used in aerial and r-f coils, and, effectively, in i-f transformers.
The analysis is given in Chapter 26 Sect. 4 (vii).
See also Sect. 11 Summary of Formulae.
tween the response at the bottom of the valley and that at the two peaks. It is then
good practice to add another stage employing a single tuned circuit which substan-
tially removes the " valley " of the preceding circuits. The procedure by which the
best results may be obtained is described by Ho -Shou Loh (Ref. C13). In this
manner a nearly flat response may be obtained over a range of frequencies 10 Kc /s
to 20 Kc /s wide, with very sharp discrimination against frequencies 20 Kc/s or more
away from the centre frequency, 450 to 460 Kc /s.
See also Sect. 11 Summary of Formulae.
..\'.. 9.10
\
and that Q is reasonably high 0.4 °
2 4 6 8 IO 12 14
6
(> 25 say). Very low values of 2n(nf/fo .2a/(Irof1
\
0.5
(2K°a 2 _ (i 'a 2] á.4
.
Q require a different curve for 07 Ì.I.\.1=1111=1 b2
I+a MN
06
each value, but the effect is only 09 EZ= WHERE
RE a - I
Q2
\N".,..
FOR
.
b.
,\"",,,,
curves without altering appreci- 2
6
ably the main portions.
As an illustration of the use of 4
ma\o
o\\-\\
....
SINGLE
CÌRCUIT
.
9.5
2
.`.`,,
/ (/5
11M1111\`C1M111111=1111M 4
iz
Fig. 9.17 consider the example A 6 156db
7 6.9
f0 = 1000 Kc /s, Q, = Q2 = 200, 6 8
9
k /k, = 2 = b: we see that the Ó !0
.....,
20
dfo = 4 Kc/s, and that the gain 6
.-`_91i5IN'
50 4
results agree well with those 60
70
B
calculated from the formulae 90
$
100
already given, namely FIG 9-17
2 4 6 D 8
UNIVERSAL SELECTIVITY CURVES FOP
10 12 14
2Q(pf
I
D= 2a/ (i+a) I
180°
b2° [(2KQa2- G-4) 2]/ (I+a)2
/
a= QI/Q2 150c
I I 1I111
I%..
%'i/',
/Ib20
I I
2D
TAN 6a
IIII 1 -D2+ b2 baO'22- 120°
PHASE CHANGE >QO 71 5bbZ32:045
=9±90°OR 9' =I0
b-17/Ì%ill
ó l 9
TAN_
60°
90
'/''Co Any
ONE
/I%/11 TUNED'
CIRCUIT
_--.r,;í:%11111 30°
a = damping factor
S logarithmic decrement
A wavelength in metres
i current at frequency f (in amperes)
io current at resonant frequency f o
A gain at frequency f
Ao gain at frequency f
Q magnification factor.
FIG.9.19
-6 -5 -4 -3
Selectivity
-2 -I
for Two
Curves
f
O
O foo
or f f- o
159 200
V LC
Kc/s, where L is in microhenrys
and C is in micromicrofarads.
(4)
Magnification factor
Ratio of total circulating current to input current = Q (18)
(7) SELECTIVITY
(a) Series Resonant Circuit
= A0/A = 1/l + Q2(f/f° f°lf)' = 1/1 +
2 + df o/f
- Q2Y2. (19)
Q.
+-ry k2
. woL1
+ k1/(2'QQ
where 1 /R' = 1/r,,
1
r
A
Qsys
1s r 2Qy 1s (39)
°/A 1 1 + k'Q2] + L1 } k'Q2]
Gain at peaks, .f, and fl, is very closely equal to the optimum value A, (max.).
Circuits of unequal Q
Transitional coupling factor, k,, is
kt = 1/i(1 /Q'' + 1 /Q2'). (40)
Band width between peaks is
(fß
Selectivity
-
LW. = Vk' kt'. - (41)
A.
'
/r Q'Q2Y1 1 [2YVQ'Qs 1' a
(42)
A 1 1 + k'Q'Q2-J + L1 + k'Q'Q,J
>
- 84f0'/fok' <
When k'Q' 1 and dfo /fo 1,
Ao /A 1/1 (43)
(10) COUPLING COEFFICIENTS
High impedance coupling*
Cm
k for capacitive coupling (Fig. 9.12) (44)
SECTION 12 : REFERENCES
(A) BOOKS DEALING WITH RADIO TUNED CIRCUIT THEORY
Al. McIlwain, K., and J. G. Brainerd, " High Frequency Alternating Currents " (John Wiley and
Sons Inc., New York, Chapman and Hall Ltd., London, 1931).
A2. Harnwell, G. P. York and London,
" Principles of Electricity and Magnetism" (McGraw-Hill, New
1938).
M. Terman, F. E. " Radio Engineering" (McGraw-Hill Book Company Inc., New York and London,
3rd ed. 1947).
A4. Everitt, W. L. " Communication Engineering " (McGraw -Hill Book Company Inc., New York
and London 2nd ed. 1937).
A5. Nilson, A, R., and J. L. Hornung, " Practical Radio Communication " (McGraw -Hill Book Com-
pany Inc., New York and London, 2nd ed. 1943).
A6. Henney, K. " Principles of Radio " (John Wiley and Sons Inc. New York, 4th edit. 1942).
A7. Glasgow, R. S. "Principles of Radio Engineering" (McGraw -Hill Book Co., New Yorkand London,
1936).
A8. Henney, K. " Radio Engineering Handbook " (McGraw -Hill Book Company Inc., New York
and London, 4th ed. 1950). London).
A9. " Admiralty Handbook of Wireless Telegraphy " (His Majesty's Stationery Office,
A10. Sturley, K. R. " Radio Receiver Design " (Chapman and Hall Ltd., London, 1943) Part 1.
All. Terman, F. E. " Radio Engineers' Handbook " (McGraw -Hill Book Company, New York and
London, 1943).
Al2. Reich, H. J. ' Theory and Applications of Electron Tubes " (McGraw -Hill Book Company,
New York and London, 1944).
A13. Sandeman, E. K. " Radio Engineering " (Chapman and Hall Ltd., London, and1947).
A14. Welsby, V. G. " The Theory and Design of Inductance Coils " (Macdonald Co., London,
1950).
A15. ' Reference Data for Radio Engineers " (Federal Telephone and Radio Corp. 3rd ed. 1949)
pp. 114 -129.
CALCULATION OF INDUCTANCE
based on original chapter in the previous edition by L. G. Dobbie, M.E.
429
430 (ii) SOLENOID WITH SPACED ROUND WIRES 10.1
o
2
OMMEMMEMEMMOMMEMOMMOMMOOMMOMMOMMEWAIMMEMMMOOOMMOMMEMEMEMEMMII
lOOMMOMMOIVOOMOOMMUMMUOMMOMMIPídOOMMOMOMIUMMOMUMOMOMMI
.1)
o
t
==
o
r.
umummumumnunummeom .
imiriummmommnnumommmmmommummomminomm
a
=
-7;
== = =
I-
Li
cc
<
--- o
N
IIMMEMEMEMEMEMMEMMOMMEMOOMMOMMEMMEMMEMEMMIMMEMMEMMMIMEMOMOMME
HEMMEMM211111M MMMMMMMM OMMEMMOMMOMMEMEMOMEMOMMOMMEMMEMMEMOMOOMMO
O IIMMOMMOMMEMOMMEMEMOOMEMMOMMOMMEMOMEMOMMEMMIMOMMOMMEMMEMMO o
ul
o o
r)
6
al
(5 6 o
[si ] INVISNOD S,VNOVOVN FIG. 10.1
Fig. 10.1. Nagaoka's Constant (K) for a wide range of d/l.
10.1 (ii) SOLENOID WITH SPACED ROUND WIRES 431
This formula for L 0, together with the values of A and B, have been taken from the
above quoted Bulletin of the Bureau of Standards. The value thus obtained for L o
is given as correct to one part in a thousand.
The equation given above for L o can be expressed in the alternative forms :
Q
iNEMunNnonMMENNEMEoxrNNOnuuuNE noEMMEM
mom. -
O!!!!NO!!!lOpEE!!V!!!!pN!lEONE\\NEMENE!!!!!!!!E!!t!
imummuumum ln\.ummommgmt;mpmmomm
IO_ : ----___=--=_--
__
___ -__
!!!!/!!!!!!Mt/!tN\mmN \'
M!!!!!!!!!!!!!!!!!!!, \!!!!!!!!!!!!!! N!!!!!!M!!!!!!!!!!\III!!!!!!\!!!!tf \!!!!!!!!!!!!M \M M ME N O MONEE
ME!lNENN!lMEMEEM{lNENEEMONE!!N!!!!M !!NE!!!!E!!INI!lENOM!!!!!!\\tfY!!!lO MEMO
MOW ENEMIES WNW NMONMYMMONOMOINI!!!lMIMMM\\!!/N!lIIIMMO
Fig. 10.2. Constants A and B as used in the formula for the correction of " current
sheet " formulae for application to round wire unth spaced turns (eqn. 4).
432 (iii) APPROXIMATE FORMULAE 10.1
B)1
Lo = L,[1 - aNK
1(
- L11 - (A + B)] (5)
wh ere P, the pitch of the winding in inches, is equal to I/N.
As an example of the use of these formulae and curves, let us consider the following
case : a coil of 400 turns of round wire of bare diameter D = 0.05 inch, is wound
with a pitch of 10 turns per inch, on a form of such a diameter that the mean radius
to the centre of the wire is 10 inches. Then,
a= 10,1= NP =40, N = 400, P = 0.1, D/P = 0.5.
The value of K corresponding to 2a/1 = 0.5 is 0.8181 from Fig. 10.1. Therefore,
L, = 0.10028 (400)2 x (100/40) x 0.8181
= 32 815 microhenrys.
From Fig. 10.2 with D/P = 0.5, N = 400,
A = 0.136 -
B = 0.335
Therefore A + B = 0.199.
The correction is, therefore, 0.0319 x 10 x 400 (0.199)
= 25.4 microhenrys.
The total inductance L o = 32 815 -
25 = 32 790 microhenrys, and the error which
would be introduced by calculating as a current sheet inductance is less than 0.08 %.
(c) For solenoids whose length is small compared with the diameter
a2N2
- - (a/51)] a +
L'
[9
which is accurate 2%
101 (10)
to for all values of 2a/1 up to 20. The error approaches + 2%
when d/1 =
d/1 = 10 to 12.
2.0 to 3.5 and at d/1 = 20. The error approaches -
2% in the range
Let us apply Wheeler's and Esnault -Pelterie's formulae to the solenoid already
considered, namely :
a = 10 in., 1 = 40 in., N = 400.
102x4002
Wheeler's formula gives L, =
90 + 400
32 650 microhenrys. -
Esnault -Pelterie's formula yields
102 x 4002
L, = 0.1008
40 + 9.2 -
32 790 microhenrys.
These results agree, within the stated limits, with the value 32 815 obtained pre-
viously.
10.1 (iv) DESIGN OF SINGLE LAYER SOLENOIDS 433
L, = 0.004 50
A3 /2
(K
2a
P
microhenrys j (11)
This gives the inductance in terms of the length of wire, the pitch P, and the shape
2a/1, as K is a function of 2a/1.
From the graph of the quantity K/2a /1 against 2a/1, its maximum value is found
to occur at 2a/1 = 2.46. Thus, for a given length of wire, wound with a given
pitch, that coil has the greatest inductance, which has a shape d/1 = 2.46
approximately ; or to obtain a coil of a required inductance, with a minimum re-
sistance, this relation should be realized. Further, the inductance diminishes rapidly
for coils longer than this optimum value, but decreases only slowly for shorter coils.
The optimum value of d/1 can also be obtained roughly from the approximate
expression for K:
K = 1 /(1 + 0.45 d11).
Therefore (d /1),Dr = (1/0.45) = 2.2 approximately (12)
(B) Given the diameter of the coil, the pitch and inductance, to determine
the length of the coil.
A suitable form of the equation is obtained by substituting for N its value 1/P in
the formula :
f
200 x (0.01)2
0.200 x (0.5)3
- 0.8.
434 (iv) DESIGN OF SINGLE LAYER SOLENOIDS 10.1
':a
FIG.10.3
a. ;
1":::1491: ... . a.
uu nn:...........
...........:. .: ....
...
° :
:::: ,
z:
- :::iEB:@i
:::::¡;;:a;:-
:
Fig. 10.3. Variation off and loglof with (2a /1). Curves derived from U.S. Dept. of
Commerce Circular C74 " Radio instruments and measurements ".
From Fig. 10.3, we see that the value of 2a /1 is 0.9, and hence
1 = 1/0.9 = 1.1 inches.
Formula for determining the coil length when the coil diameter and winding
pitch are known for coils such that d < 31.
As an alternative to the use of Fig. 10.3, we derive a simple result for solenoids such
that 2a /1 < 3. Here we have
K ti 1 /[1 + 0.45 (2a /1)] (15)
I 1 1
Hence, f = K 2a
-2a 1 + 0.45 (2a/1)
(16)
Therefore, 0.9(2a/1) = -
1 + 1/l + 1.8 /f (17)
Thus, in the first case above, where f = 3.2,
0.9(2a /I) = - 1 + 1/1 + 1.8/3.2 = 0.25,
Therefore,
P = V0.200 x 0.688/500 = 0.0167 inch,
equivalent to 60 T.P.I.
Simple procedure for cotnputing the length and total turns when the coil
diameter and pitch of winding are known for coils such that d 31.
Substituting in equation (17),
0.9(2a/1) = -
1 + + 1.8/f,
f
where = 5L,P' /a' = 5L, /n'a', n being T.P.I.
Putting F = 1.8/f and y = 0.9(2a/1),
we obtain
y= - 1 + V1 + F, (20)
where F = 0.36n'a3 /L
and 1 = 1.8a /y.
As an example, let us consider a coil of 2 inches diameter, wound with 33 T.P.I.
of required inductance 380 microhenrys.
(a) F = 0.36n2a' /L, = 0.36 x 332 x 1'/380 = 1.031
(b) y = - 1 + V1 + F = -
1 + x/2.031 = 0.426
(c) 1 = 1.8a/y = 1.8/0.426 = 4.23 inches
(d) N = nl = 33 x 4.23 = 139i turns.
Another method applicable to the same type of problem is due to Hayman.
From Wheeler's formula (equation 8) :
a''
L'ti9a +101'
the length I is eliminated by the substitution 1 = N /n, n being the turns per inch ;
and then the resulting expression is solved for N, yielding :
0.36a'n
N= [1 + 1 + (21)
na' L, 2]
For convenience in computation, the quantity x = 5 /na' is introduced, so that
finally
N = xL,[1 + V1 + (9 /ax'L,)] (22)
As an example, we take the same problem as above, namely, a = 2 inches, n = 33,
L, = 380 microhenrys. The procedure is as follows :
(a) x = 5 /na' = 5/(33 x 1) = 0.151,
(b) x' = 0.0227,
(c) 9 /ax'L, = 9/(1 x 0.0227 x 380 = 1.042,
(d) N = 380 x 0.151 x (-/2.042 + 1) = 139 turns,
(e) 1 = N/n = 139/33 = 4.2 inches.
(C) Given the diameter and length of coil, value of capacitance and fre-
quency of resonance, to determine the number of turns.
See Chapter 38 Sect. 9(v)B.
2 3 .4 5 6 8 i 2
310
e °=_=°__° °°°°_ =-=/4f .7 r7f/= `e
t r=Mttt.
UM
.
r.imsm.....
rl.11.......
..
MEE::13110=
lií.l
IP.1111.1u'.uu1.'
'111'.111111111
alMMEN IMM1 IIUI IIIIIIIIIIIíIII:
1
IÌ/1r.=I.
11.1111111=r/ /MN .
IÌÌ1111.I
111.4111 .111Y.31111
111'.111
1Í11.11111;//11;1111111':111111111111/11111
ammo 111111111111I41111111111111
A111:111111i111111,/1111U111IIP
...ip Ául
mni!a1Oa1oIIII 1.0
11
1
CR :ie
ii
3
:.Ï....
r...IM..I...
rAlu.
MIN.484W
Wi1143111/....
/4/41W/M1>.
.
. ..rrrr11111Ì.
4111 1111111111
1==
i31==. 1i
z
411/431,41111% O'Z
I/LIAVAN MMM1I11I1111S1I1111111111111111111111 './111111111111111111 11ii
11111111=MME11p111111 i
0
r.r.rl.a=Mrl
r.r.IIIM...I. 1u111u1u11111Dm11 1111l11111111rr.r./.=.
-
MINNr.. III =111114= 1
=1=131.
IMMIIIM. A
I1111111111 11111111111r11.1==
111111111111111111111111111111111111MIONIMMII
1
1
1
allIMM .
moI.a
2
11111111111111111111111111111111111111111MMMIN
1111111111111111
3 .4
11111111111113MM
sammp111111111111111111IIIIIIIIII11111111u 11111111I1 IIIII
.5 6 0
tom Or6mNnp111111 -01
2
1 3
LENGTH OF COIL IN INCHES
FIG . 10.4
Fig. 10.4. Curves for the determination of Current Sheet Inductance (L,) for small
solenoids, plotted against length of coil (I). For other pitches refer to Fig. 10.5.
10.1 (v) DIFFERENCE BETWEEN L, AND La 437
-
L. = L, 0.0319aN(A + B),
we have, under such conditions,
-
L. Iv L, 0.0223aN, (23)
with an error not exceeding 20% of 0.0223aN ; and thus in many cases not exceed-
ing 0.2% of L,.
"
(vi) Curves for determination of the current sheet " inductance
(A) Method of using the curves
Figure 10.4 applies to a winding pitch of 10 turns per inch only ; for any other
pitch the inductance scale must be multiplied by a factor, which is easily determined
from Fig. 10.5. The diameter of a coil is considered to be twice the distance from
coil axis to centre of the wire.
5 10 20 100 200
mime
I
Inu1u1111111111n11111
11111111111
11111111111....1
=___----- ...
.M.1111111111.1
111111111111111111111.111111111111111111111111.==........
1111111111
IIIIIIIII III IIIIIIIIIIIIIIIIIIIIIIIIII IIIII
11111111111111111111111110ggg11111n1u11111u111 u111111n111uu11111n111111nI11111UMI11111
100 Illll llllllqllllllllllll111111111111111111111111111111111p1Ip11111111111111111IIIII
r.1111111111 100'
.L..
---------
111111111111111111111111..sINI.U1
... _==_e:::::
,IiB:;,;?;;
1111111111111
11111111;
... 1
111
1111111111==--.:.1
AllnalllllnIIIIIIIIIIIIIIIL..
WIIIIIIIII 11.....
11111 1111111111 ..
11111111111
NMI luullnlla111I11..... 111'.11111111
1111111111111111111111111.....I111111111111111111111I1C11111111111111111111111111IIII11111..IIu11
.111111111111111111P111IIIII111I111II1 IIIII11I111I 111111NE111111
IIIIIAYIIIIIIIUINWII. .111_1_ 10
1111111111IIIIIII11111111
IIIIIIIIIIIIIIIIIIiIIIIII/01111111111111111!íIÍIIIIII
..: 7111uII111l111I1f1
.-
.
1
ais_s1
-
IL
uuuuw.u...
iiiliiiiiii
11111111111111... 111111 111111 I I u 111111111111111111111.....
IIl1111111111111111111111....1
IIIIIIIIIIIIIIIIIIIIIIIII..., 11111111111 111 111111
11111111111111111111111111111111.....mis1
1111111111111111111111111.'I..MINII1IIII11111111111111 111
illllllllllllllllllllllllrI111111U11111111111111111111 11111111111111IIIIIIIIIIIIIIIIIINNI1.111S1
111111II II I IIII II I II II
I I I i11111IIIIIIII
IIIIIIII I II 111111.11I1111111111111111III111_
:
111111
ill i{liui+4Iiii II#ú1f -
nlirl111I11iiIlIlÉ11111F1M1'
..,..; .==--za11. .-.- --....
-
_--------_
==
10 20 50 100 200
TURNS PER INCH FIG. 10.3
Fig. 10.5. Winding pitch correction for Fig. 10.4.
438 (vi) CURVES FOR " CURRENT SHEET " INDUCTANCE 10.1
(B) To design a coil having required " current sheet " inductance
Determine a suitable diameter and length, and from Fig. 10.4 read off the " current
sheet " inductance for a pitch of 10 T.P.I. The required inductance may then be
obtained by varying the number of turns per inch. The correct number of turns
may be found by calculating the ratio of the required inductance to that read from
Fig. 10.4 and referring it to Fig. 10.5 which will give the required turns per inch.
Alternatively if the wire is to be wound with a certain pitch, a conversion factor
for that pitch may first be obtained from Fig. 10.5, and the required inductance
divided by that factor. The resultant figure of inductance is then applied to Fig. 10.4
and suitable values of diameter and length determined.
(C) To find the " current sheet " inductance of a coil of known dimensions
Knowing the diameter and length, determine from Fig. 10.4 the inductance for a
pitch of 10 T.P.I. Then from Fig. 10.5 determine the factor for the particular pitch
used, and multiply the previously determined value of inductance by this factor.
only for very short solenoids ; they indicate how the change in inductance of long
solenoids in spherical screens may be calculated, but it is clear that no simple formula
can be thus obtained.
B:'
:? :
_ =ÏIl
_
:
á;
.
riiÏ i° =
1li::::::::: ïii:iI::i:::::::::nauu..1
. E6E
'cü:ü"1:irli'l:i:i::
ÌÌ1Illlqp 111:IÌÌ,IUII IIIIII 11Ia/lLp IN N N :.:i1:1l:::i1::::lñii:l
Ñ .ur+.'i:
JC'
%i
ñ
li Ii 111111111111111aniirillfi
1111111111IÌlÌlllÌi i Ì:iIILIIIIIIii
11 111ÌI lÌll11
ill11u11111111
.:i
Ilfllllllllll
IIIalO..1.11111
"fLlYrlffl Ü
li 1
, ridi "
f:1llbrf
. 1ÌÌCÌ h IÌÌÌ
Iu NC C:C'.
C Ì {
-';?a9iii$alEIll[;si:"i;IiirAiiiiiilg_á2!áúI!1!`4':'',:1é'!i!:3tFi1
1
11111111 1111
g'_== - -_ 0 i:iá=
m 72410 ii694II li'ü'l' l9tE 9F5_ .1$1it:.;"L
:
'
:11E
¡
_
-__
=-_ O
Mil F?éé ?1i'2115E3
:? ÌiR?!i
E 14 DI
EN pk r..::::ii!iii1l:E?m1i1iR°3__
141 Mi RH EEi: i15T iE:tE.A5-
:zEms á
1 1
áil . .... :_
:.u{üi3,,,üliie!:i.::'iiniy?0i!'
'
:t59.119I_:i 5 =f _i=o
__r3?':r::;::7.; ...m s!! 1?:?: _:::_
ee =
:.::z -_
1ir:i'-°.iii^!ü:ii°i::ï?:::
_
__
sg
. _
_ _
---
________
Q
::n ::::::
..
. '
Iñd i!iü'9äE°.'..::11
.,.:.. iE51_-.
::u....:a::Í1¡jj'i'::k11::::::^.":.^:6:::IIa.....
..... __.._.I:,:::::au:::::::.:a:ï
__.._ ................_.....
.
1
en
...I... ...( .....1
..... .:.::: ..... ......:::
..1.. .. .....:'.
1:1 pp,I..
,p@
Wbw
{:.19r'
..:............'
,1
.,.,.1nuw,.
11111,:::::,':::::;:::::::::i:
. .....
..... iii!
11l Ll11 , .... li
ff\\ !N
011111111111 IF:61 i11
1,1111111.1111 IIInI 111lI1111111 ,ÌII
111:11111111111111111ppIn11111111111'11111!111111' ,1a.1n1\IIII.91111111Ì1111191ÍIÍÌÌÍÌÌÌÌÌÌÌi1' ` wl
D14 IIIIl11111111111111111111m11.1111111, ,1 .'Illlllllllllllllllh'1111111111111111.I111111111n1
11111111111111111111111Ilnlll 1111111111111111111'
O.\fnlf11111111111111111111116'1111111111111111`
1111111111111111 Ilirig 11Vffl'
11111111111111'11111, \M11111111i uuC.aM 1111111111111111'111111111111111:
1111111111!UII1h 1111111111111111'1I1111V11111'1111CIIUI\IIf lia& IIIIIIIIIl111111911111111111111111ItfC1uG\
WVIIIIIIII'1111111111111111L
IIIIIIIIII!IVIIIInIII!11111111111'111I111111IU'11111111111711I7f1MIIIIIIIIIII'11111111I111111111111111111111111:1111I111111 flf1s11
-
'
.7i üif:'=
- z?
'!
IS .;
-,. ..-
c.ttiié : ..Scc _ -__ `_= iiïiE'!:ü'-il3.¡i'r.'
_ -
:
°, >: _ :._
:
l:i"::t::.::s:.:::.
1
a
-:°:i3;i¡ÍIEiC ákl11i1,:p::=6'r-===_3i-====`=_=''';i..:Iu,::k,-.11¿:s14:iÌ,'si.;;,-.
l.¡! l'-
E....::.-:::_.=á:":il3:";;.i,:::
:x;:r::Fi ï!iiÍié: i'üilal. ,ii1i:ii _ - :'-:aI:?i_-:._!ñ...
_===='ir
I
is:EEEr'- .: =
_
'
_
_
======
-°°===
-==°
-----
_-==== á
i!P.'19i^!ril?g.':in6,'F.i_üF::u:::..e:::, z :.
11Ei tu Re'!EL 3
lag
14;!!i'!ii!I11'i1.:iiiilmi1t`S
lit !14!p;:ñ'' ú $$:`I$úll`ti4ái141iii
!!; 111:.1d:-'. 1
1 0
. 1T:
ii..xES e;, It= e:eua = r ó
m
ó
4;
1 111 1:1'111111111'.d,111!1\
1111111111111111111111111161'Ihlllllllllll u11u11,11i111111911 !Illlllllh'11IU1libE'I11111JI11I llll1111111'111111111I1U111111S\
d lllllll _.....' °----_-=- Q3E2_ F _ d
.....:......
.. c_=.illill :1!
- Eal
d
vtd
c Ip g a á
R
N
FIG. 10. 6
Fig. 10.6. Curves for determination of decrease in inductance produced by a coil shield
(reprinted by kind permission of Radio Corporation of America).
10.1 (vii) EFFECT OF CONCENTRIC SCREEN 441
small values of l/g where Bogle's formula should not be used. In the tables below
comparisons are made in a number of typical cases :
Table 1 : Comparison between estimates of the percentage reduction in inductance
caused by a cylindrical screen for various coil shapes when the ratio of coil diameter
to screen diameter = 0.7.
(Lo - Ló) /Lo
1/d R.C.A. Bogle gl1
20 49% 48.2% 0.0107
10 48% 47.3% 0.0215
5 46% 46.0% 0.0430
1 35% 36.8% 0.215
0.2 20% 18.3% 1.075
0.1 15% 11.3% 2.15
where 1 = coil length, d = coil diameter, g = distance between coil and shield
Table 2 : Comparison between estimates of the percentage reduction in induct-
ance caused by a cylindrical screen for various ratios of coil diameter to screen
diameter.
(Lo LoWLo -
1/d = 10 1/d = 1 1/d = 0.2
a/b R.C.A. Bogle R.C.A. Bogle R.C.A. Bogle
0.9 80% 80.6% 72% 74.5% 54% 56.5%
0.7 48% 47.3% 35% 36.8% 20% 18.3%
0.5 23% 23.2% 13.6% 14.0% 6.7% 5.7%
0.3 7.8% 7.6% 3.0% 3.2% 1.4% 0.9%
where a = coil radius, b = shield radius.
When the dimensions 1 and c are small in comparison with a (see Fig. 10.7) the in-
ductance is given closely by Stephan's formula, which, for 1 > c, takes the form :
L, = 0.031 93aN2 [2.303(1 + 32a2 + ) log 8a yi + (3)
962 g
- a]
while for 1 < c, that is for pan -cake coils, the formula becomes :
f
The formulae of this section have been put in the form °I
(9a
+ 101) ; (7)
9a 1+ 101 [1 10a1
a2N2 caN2
or L, N (8)
9a + 101 lOnl
z
3.62
when or lies between 0.35 and 1 (10)
1 3.5 log's V1 +c 2 c 1
z
4.02a
022
when or lies between 0.15 and 0.35 (11)
13.5 logic/1 + c c 1
aN z 4.55a when - or ç
hes between 0 and 0.15 (12)
13.5 1/12 _} c2 1
These results are accurate to 5% as 1 and c approach a, and are increasingly accurate
as 1 and c decrease compared with a. When 1/a and c/a are both very small, it is
sufficient to use the approximation :
aN' 4
N log"
L' 13.5 Vi z +c z
where A is the wire length, P is the distance between centres of adjacent wires and G
is a function of the shape of cross section (1/c) and of the shape ratio of the coil (c /a).
The quantity G is represented by means of curves. This equation is used, in cases
where the ratios 1/c and c/a have been decided, to determine the necessary wire -
length for a given pitch, or vice versa. The mean radius of the winding is then
obtained from
3 A c Pa (18)
a
2a 1 (c /a)a
and the total number of turns is given by N = A /27ra or lc /Pa.
It can be shown that, for a given resistance and coil shape, the square cross section
(1 /c = 1) gives a greater inductance than any other form ; and, further, for a square
cross section, the inductance for a given length of wire is a maximum for c/a = 0.662.
(2) Stephan's formulae may also be expressed in the form :
L, = 0.031 93aN2g, (19)
where g is a known function of 1/c and c /a.
This form is useful when a, ^ and 1 have been decided to give N, and thence P.
We have
le
P =Ñ
and A = 27taN.
(B) Universal coils
The usual problem is the calculation of the inductance of a given number of turns
before winding the coil, with only former size and wire size known. It is stated in
Ref. 20 that the inductance of universal coils is about 10% greater than that of normal
multilayer windings of the same external dimensions.
A suitable procedure is to determine the gear ratio from the formula,
gear ratio = ¿n(P + 1) /P (20)
as described in Chapter 11, Sect. 3(iv). The radial depth of winding, c, can then be
obtained from the formula
c = Nqu /(P + 1) (21)
where N = number of turns in coil,
q = number of crossovers per winding cycle
[Refer Table 1 in Chapter 11, Section 3(iv)]
w' = diameter of wire plus insulation,
and P = an integer defined in Chapter 11, Section 3(iv).
Since the length of the winding, 1, is equal to the sum of the cam throw and the wire
diameter, the dimensions of the coil are then known and the inductance can be cal-
culated from the methods previously described.
Example : A coil of 500 turns of 38 A.W.G. enamelled wire (0.0044 inch) is to be
wound on a 0.5 inch diameter former with a 0.1 inch cam. From Chapter 11 Sect.
3(iv), P = 43 and q = 2. Therefore
c = 500 x 2 x 0.0044 /44 inch = 0.1 inch
1 = 0.1 inch + 0.0044 inch = 0.1044 inch
winding, a is the radius of the turns of the winding and N is the total number of turns.
for a put ao + (N -
1)P/2, where 2a is the distance (Fig. 10.8) across the
centre of the innermost end of the spiral.
The correction for cross section is given in the Bureau of Standards FIG. io.e
Circular No. 74 in the form :
Lo = L, -
0.0319aN(A, + B,), T
where A, and B, are tabulated functions. The accuracy obtained T
for Lo is 1%. °
a'N'
L' 8a +llc
The first of these formulae is the more accurate as the ratio c/a decreases, while the
second is the more accurate the closer c/a approaches unity. The agreement between
them is quite fair :
L. .
while from Wheeler's formula
aN2 /(8 + 2.2) = 0.098aN2.
For c/a = 0.5, we obtain, respectively,
(aN2 /13.5) log 9.4 = 0.97aN2/13.5 and
L, aN2 /(8 + 5.5) = aN2 /13.5.
There are few simple formulae which can be used for the more common practical
cases, such as are possible with self inductance. The following exact method may be
used for two windings on the same former with a space between them, both windings
being similar in pitch and wire diameter (Fig. 10.9).
These inductances may be calculated from the formulae given in earlier sections of
this chapter.
References to accurate methods : Refs. 2, 3, 5, 13, 14, 15, 18.
s.0 DI 111
-
.2
.4
.1
.6
K
s1
12
1.4
1.6
.01
2.2
2.4
2.6
2.
s.
.001
0 2 3
L
FIG. I0. I0
Fig. 10.10. Chart giving coefficient of coupling for a specfied spacing between two
coaxial solenoids having identical dimensions. The quantities S and L are measured in
terms of coil diameters. (Ref. 56.)
448 (ii) APPROXIMATE METHODS 10.5
SECTION 7 : REFERENCES
EXACT METHODS OF CALCULATING SELF AND MUTUAL INDUCTANCE
1. de Holzer, R. C. " Calculating mutual inductance ; co-axial and co -planar multi-layer coils " W.E.
25.300 (Sept. 1948) 286.
2. Bureau of Standards Circular No. 74 (1924).
3. Bureau of Standards Scientific Paper No. 169 (1912).
4. Nagaoka, H. " The inductance coefficients of solenoids " Journal of the College of Science, Tokyo,
Vol. 27, Art. 6 (August 15, 1909) p. 1.
5. Grover, F. W. " Tables for the calculation of the mutual inductance of any two coaxial single layer
coils," Proc. I.R.E. 21.7 (July 1933) 1039 (includes further bibliography).
6a. Reber, G. " Optimum Design of Toroidal Inductances " Proc. I.R.E. 23.9 (Sept. 1935) 1056.
6b. Moullin, E. B. " The use of Bessel Functions for Calculating the self-inductance of single -layer
solenoids " Proc. I.E.E. 96. Part 3.40 (March 1949) 133. See also Ref. 55.
Also Refs. 15, 25b.
APPROXIMATE FORMULAE FOR SELF AND MUTUAL INDUCTANCE
7. Simon, A. W. " Calculating the inductance of universal -wound coils " Radio, 30.9 (Sept. 1946) 18.
8. Ricks, J. B. " A useful formula for solenoid inductor design -determining coil dimensions with
known capacitance," Q.S.T. 31.5 (May 1947) 71.
9. A. I. Forbes Simpson " The design of small single-layer coils " Electronic Eng. 19.237 (Nov. 1947)
353.
10.7 REFERENCES 449
10. Simon, A. W. " Wire length of universal coils " Elect. 19.3 (Mar. 1946) 162.
11. Dwight, H. B. " Self inductance of a toroidal coil without iron " A.I.E.E. Trans. (Nov. 1945) 805 ;
discussion supplement A.LL.E. Trans. (Dec. 1945) 999.
12. Blow, T. C. " Solenoid inductance calculations " Elect. 15.5 (May 1942) 63.
13. Di Toro, M. J. " Computing mutual inductance " Elect. 18.6 (June 1945) 144.
14. Maddock, A. J. " Mutual inductance : simplified calculations for concentric solenoids " W.E.
22.263 (Aug. 1945) 373.
15. Grover, F. W. (book) " Inductance calculations : Working Formulae and Tables " (New York,
D. Van Nostrand Co., Inc., 1946). Includes extensive bibliography.
16. Dwight, H. B. (book) " Electrical coils and conductors : Their Electrical Characteristics and
Theory " (McGraw-Hill Book Co., Inc. New York, 1945).
17. Amos, S. W. " Inductance calculations as applied to air -cored coils " W.W. 47.4 (April 1941) 108.
18. Turney, T. H. " Mutual inductance : a simple method of calculation for single -layer coils on the
same forther " W.W. 48.3 ;March 1942) 72.
19. Everett, F. C. " Short wave inductance chart " Electronics, 13.3 (March 1940) 33.
20. Zepler, E. E. (book) " The Technique of Radio Design " Chapman and Hall, 1943.
21. Jeffery, C. N. " Design charts for air -cored transformers -co -axial windings of common diameter
and pitch " A.W.A. Tech. Rev. 8.2 (April 1949) 167.
22. Wheeler, H. A. " Simple inductance formulas for radio coils " Proc. I.R.E. 16.10 (Oct. 1928) 1398
and discussion Proc. I.R.E. 17.3 (March 1929) 580.
23. Esnault- Pelterie, M. R. " On the coefficient of self inductance of a solenoid " Comptes Rendus,
Tome 205 No. 18 (Nov. 3, 1937) 762 and No. 20 (Nov. 15, 1937) 885.
24. Bunet, P. " On the Self Inductance of Circular Cylindrical Coils " Revue General de l'Electricite,
Tome xliii, No. 4 (Jan. 22, 1938) 99.
25a. Hayman, W. G. " Approximate Formulae for the Inductance of Solenoids and Asiatic Coils "
E.W. and W.E. 8.95 (Aug. 1931) 422.
25b. Welsby, V. G. (book) " The Theory and Design of Inductance Coils " (Macdonald & Co. Ltd.,
London, 1950).
NOMOGRAMS AND CHARTS FOR SELF INDUCTANCE
26. Sabaroff, S. (letter) " Nomogram for Rosa inductance correction " Proc. I.R.E. 35.8 (Aug. 1947) 793.
27. Pepperberg, L. E. " Coupling coefficient chart " Elect. 18.1 (Jan. 1945) 144.
28. Data Sheets 32 and 33 " The inductance of single layer solenoids on square and rectangular formers "
Electronic Eng. 15.173 (July 1942) 65.
29. Data Sheets 26, 27 " Skin effect " Electronic Eng. 14.170 (April 1942) 715.
30. Data Sheets 23 -25, 28 " The shunt loaded tuned circuit" Electronic Eng. 14.169 (March 1942)
677, and 14.170 (April 1942) 715.
31. " Inductance chart for single layer solenoids " G.R. Exp. 15.3 (Aug. 1940) 6.
32. Purington, E. S. " Simplified inductance chart " (inductance of multi -layer coils) Elect. 15.9 (Sept.
1942) 61.
33. " Instructions for use of Federal Band Pass Nomographs A to C and 1 to 3," Federal Telephone
and Radio Corporation.
34. " A convenient inductance chart for single layer solenoids " G.R. Exp. (based on Circular 74, U.S.
Bureau of Standards) 15.3 (Aug. 1940) 6.
35. Data Sheets 12-17 " Inductance of single layer solenoids on circular formers " Electronic Eng.
14.164 (Oct. 1941) 447 and 14.165 (Nov. 1941) 495.
36. Shiepe, E. M. " The Inductance Authority " published by Herman Bernard, New York, 1933
(contains 50 large sized charts based on formulae and tables of Bureau of Standards Circular No. 74).
37. Seki, H. " A new Abac for single layer coils " Wireless Engineer, 10.112 (Jan. 1933) 12 (nomo-
gram based upon Hayman's formula).
38. Maynard, J. E. " Multilayer coil inductance chart " Elect. (Jan. 1939) 33.
39. Beatty, R. T. " Radio Data Charts " published by Iliffe and Sons, London.
40. Nachod, C. P. " Nomogram for coil calculations " Elect. (Jan. 1937) 27 -28.
40a. Sabaroff, S. " Multi-layer coil inductance chart " Comm. 29.12 (Dec. 1949) 18.
40b. Sulzer, P. G. " R.F. coil design using charts " Comm. 29.5 (May 1949) 10.
40c. " Reference Data for Radio Engineers," Federal Telephone and Radio Corp., 3rd ed. 1949, p. 75.
40d. Wheeler, H. A. " Inductance chart for solenoid coil Proc. I.R.E. 38.12 (Dec. 1950) 1398.
THE EFFECTS OF METAL SCREENS ON COILS
41. Davidson, C. F. and J. C. Simmonds " Effect of a spherical screen on an inductor " W.E. 22.256
(Jan. 1945) 2.
42. Anderson, A. R. " Cylindrical shielding and its measurement at radio frequencies " Proc. I.R.E.
34.5 (May 1946) 312.
43. ' Effect of shield can on inductance of coil " Radiotronics No. 108 p. 67.
44. Sowerby, J. McG. " Effect of a screening can on the inductance and resistance of a coil " Radio
Data Charts (2) W.W. 48.11 (Nov. 1942) 254.
45. Bogle, A. G. " The effective inductance and resistance of screened coils " Jour. I.E.E. 87 (1940) 299.
46. Reed, M. " Inter-circuit screening -survey of the general principles " W.W. 48.6 (June 1942) 135.
47. Davidson, C. F., Looser, R. C. and Simmonds, J. C. " Power loss in electromagnetic screens "
W.E. 23.268 (Jan. 1946) 8.
48. R.C.A. Radiotron Division Application Note No. 48 (June 12, 1935) reprinted in Radio Engineering
(July 1935) 11.
49. Sowerby, A. L. M. " The modern screened coil " W.W. Sep. 23, 30 and Oct 7, 14 ; 1931.
50. Howe, G. W. O. editorials, W.E. 11.126 (March 1934) 115 and 11.130 (July 1934) 347.
51. Hayman, W. G. " Inductance of solenoids in cylindrical screen boxes " W.E. 11.127 (April 1934) 189.
52. " Formulas for shielded coils " Radio Eng. 16.12 (Dec. 1936) 17.
53. Levy, S. " Electro- magnetic shielding effect of an infinite plane conducting sheet placed between
circular coaxial coils " Proc. I.R.E. 24.6 (June 1936) 923 (contains further bibliography).
54. Moeller, F. " Magnetic screening with a plane sheet at audio frequencies " Elektrische Nachrichten
Technik, Band 16, Heft 2 (Feb. 1939), 48.
55. Phillips, F. M. " A note on the inductance of screened single -layer solenoids " Proc. I.E.E. 96. Part
III, 40 (March 1949) 138 ; R. G. Medhurst, discussion, 98. Part III, 53 (May 1951) 248. Sequel
to Ref. 6b.
CHARTS FOR COEFFICIENT OF COUPLING
56. Sulzer, P. G. " Coupling Chart for solenoid coils " TV Eng. 1.6 (June 1950) 20.
CHAPTER 11
SECTION 1 : INTRODUCTION
The space available for coils in a radio receiver is invariably limited, and part of the
design work on each coil is the obtaining of maximum Q in a minimum volume.
In the case of i -f transformers the volume is clearly defined by the shield used, and the
same applies to r -f coils in cans, but where shielding is not used with r-f coils the
increased Q obtainable from larger diameter formers is offset against the increased
damping from components and magnetic materials, such as the receiver chassis, ad-
jacent to the coil.
The resulting form factor for each individual winding requiring maximum Q and
minimum self-capacitance is usually such that the length of the winding is very ap-
proximately equal to its diameter. To obtain this shape for i -f, broadcast and short-
wave coils the winding method is varied.
I -F transformers use universal windings, and where the Q and distributed capacit-
ance requirements are not severe a single coil wound with a large cam (perhaps I in.)
will be satisfactory, even without an iron core, if litz wire [see Sect. 5(i)J] is used.
However higher Q values are usually required than are obtainable in this way and in
such cases each winding consists of two or more pies, perhaps with an iron core.
The same types of coils are satisfactory in the broadcast band, although in the case
of signal frequency coils which are tuned over a ratio of more than three to one it is
essential to use very narrow windings or two or more pies to reduce the distributed
capacitance of the winding. Litz wire and iron cores are commonly used to give a
secondary Q of 100 or more.
Another method of decreasing distributed capacitance is the use of progressive
universal windings in which the winding finger travels along the former in addition
to moving to and fro. Such coils are rarely built up to a greater height than five wire
thicknesses, and are equally useful with or without iron cores. Solenoids are also
used either with iron cores or on comparatively large diameter formers if an iron core
is not used.
450
11.1 INTRODUCTION TO DESIGN 451
and H depends only on the length /diameter ratio of the coil. A table of values of H
is given below and when used in conjunction with a capacitance correction for the
" live " lead, Medhurst states that the accuracy should be within 5%.
TABLE 1
Length ) H ( Length ) H Length 1 H
( Diameter/ `Diameter/ `Diameter/
50 5.8 5.0 0.81 0.70 0.47
40 4.6 4.5 0.77 0.60 0.48
30 3.4 4.0 0.72 0.50 0.50
25 2.9 3.5 0.67 0.45 0.52
20 2.36 3.0 0.61 0.40 0.54
15 1.86 2.5 0.56 0.35 0.57
10 1.32 2.0 0.50 0.30 0.60
9.0 1.221.5 0.47 0.25 0.64
8.0 1.12 1.0 0.46 0.20 0.70
7.0 1.01 0.90 0.46 0.15 0.79
6.0 0.92 0.80 0.46 0.10 0.96
Lead capacitance can be determined separately and added to the coil self- capacit-
ance. Fig. 11.1 (from Medhurst) can be used for this purpose.
6
FIG. 11.1
0
O 5 10 15 20 25 30 35 40 45 50
LENGTH OF W RE (cm)
outside the coil can. Unless a good joint is made between can and chassis this trouble
will be further aggravated.
If the peaks are not symmetrical, undesired couplings are probably responsible.
All traces of regeneration must be removed and, because the stability requirements
are more severe than usual, unusual effects are liable to be uncovered. For example,
regeneration may occur due to coupling between a loop formed by the generator input
leads and the output of the i -f amplifier (twisting generator leads will cure this) or
coupling may occur between first and second i -f transformers within the steel chassis.
When this happens, rotating one or both transformers will probably give cancellation
but leave production receivers susceptible to the trouble. Additional spacing between
transformers is advisable with each primary and secondary wired for minimum re-
generative coupling.
The switch used for the tertiary winding should preferably be of the " break before
make " variety. A " make before break " switch momentarily short circuits the
tertiary winding during switching, giving a sudden reduction and increase in sensitivity
which can be heard as a click.
Even after a satisfactory i -f selectivity curve has been obtained, the over -all curve
may be too narrow if a r -f stage is used in the receiver. In such cases the r-f stage
should also be expanded [Chapter 35 Sect. 5(iii)].
WIRE 4
A
COIL CAM FOLLOWER
o
o
FIG, 11.4
Symbols
d = former diameter (inches)
c = cam throw (inches)
n = nominal number of crossovers per turn (see Table 1)
q = number of crossovers per winding cycle, i.e. before wire lies alongside
preceding wire (see Table 1).
y = nominal number of turns per winding cycle (see Table 1)
R = gear ratio = former gear/cam gear
w = modified wire diameter (inches)-see Note below
P = qc /(w + x) = an integer
x = smallest amount necessary to make (P + 1) /v an integer (inches).
Note. For fabric covered wire, w = (measured diameter of covered wire) x 8/7.
If the wire is enamelled only, the same formula is used but the bare wire diameter is
multiplied by 8/7.
Procedure
(A) From n (2d /3c) determine the largest convenient value for n. Do not use
values of n less than 2 for bare enamelled wire. Obtain values of q and y from Table 1
below for the value of n chosen.
TABLE 1
n 4 2 1 2/3 f 1/3 f
q 4 2 2 2 2 2 2
v 1 1 2 3 4 6 8
(B) Determine w from information given in the note above.
(C) Calculate P from P = qc /(w + x).
(D) Obtain R from R = f n(P + 1) /P.
Example 1
Given d = I in. and c = 0.1 in. determine the gears to wind a coil with 42 S.W.G.
enamelled wire.
(A) (2d /3c) = 1.0/0.3. Take n = 2, giving q = 2 and y = 1 from the table.
(B) The diameter of bare 42 S.W.G. wire is 0.004 in., so w = 0.004 57 in.
(C) P = qc/(w + x) and (P + 1) /v must be an integer, i.e. P must be an integer
since v = 1.
qc /w = 200/4.57 = 43.7.
But P = qc/(w + x) = an integer.
Therefore P = 43.
(D) R - f n(P + 1) /P = f x 2(43 + 1)/43 = 44/43.
Example 2
Given d = f in. and c = in. determine the gears to wind a coil with 0.016 in.
litz wire.
(A) 2d/3c = 2/3. Take n = 2/3, giving q = 2 and v = 3.
(B) w = 0.016 in. x 8/7 = 0.0183 in.
(C) qc /w = 500/18.3 = 27.3.
But (P + 1) /v = an integer.
Therefore P = 26.
(D) R = fn(P + 1) /P = 1/3 x 27/26.
To obtain suitable gears :-
R= 2/3 x I x 27/26 = (28/42) x (27/52).
When it is known that n will be 2, as is the case with the majority of coils, the method
reduces to dividing the modified wire diameter into twice the cam throw (ignoring
any fractions in the answer). This gives P and the required ratio is (P + 1) /P.
For further information on universal coil winding see Refs. I 1 to I 8.
458 (v) MISCELLANEOUS CONSIDERATIONS 11.3
(D) F -M i -f transformers
For details of F -M i -f construction and other i -f information of practical interest
see Chapter 26 Sect. 4(vi).
(E) Other ferromagnetic materials
The development of a non -metallic ferromagnetic material named Ferroxcube
has been announced by Philips (Refs. A28, A29, A30, A31, A32, A33).
Several grades of Ferroxcube are manufactured. They have in common a high
specific resistance of 10' to 10" ohm cm. and a high initial permeability of from 50
to 3000 depending on the type. Ferroxcube IV, which is useful to 40 Mc /s, has a
permeability of 50, and Ferroxcube III with an upper frequency limit of about 0.5
Mc /s has a minimum permeability of 800.
11.3 (v) MISCELLANEOUS CONSIDERATIONS 459
Because of the closeness of the Curie point to room temperature (the Curie point
of Ferroxcube III is 110 °-160 °C) some change of permeability occurs with changing
temperature. For instance the permeability of Ferroxcube III can be almost halved
by an increase in temperature from 20 °C to 80 °C. However, the permeability of
Ferroxcube V is decreased less than 10% by a similar temperature change under the
same conditions. Between 10° and 40 °C the change in permeability averages 0.15%
per 1°C for the various types of Ferroxcube.
In cases in which the magnetic circuit is normally completely enclosed, the high
permeability of Ferroxcube can be used to minimize losses by means of an air gap.
If the gap is such as to reduce the effective permeability to one tenth of its original
value (which could still be high), losses and the effect of heat on effective permeability
will also be reduced in the same proportion.
The properties of Ferroxcube allow considerable reductions to be made in the size
of such items as i -f transformers or carrier- frequency filters, and it is an excellent
material for magnetic screens or for permeability tuning. However its saturation
point is rather low and it is not used for power transformer or output transformer
cores.
Ref. A27 describes research on ferromagnetic spinels by the Radio Corporation
of America. These spinets are ceramic -like ferromagnetic materials characterized
by high permeability (up to greater than 1200), high electric resistivity (up to 108
ohm cm), and low losses at radio frequencies. Wide ranges in these and other pro-
perties are obtainable by varying the component ingredients and methods of synthesis.
Ferrospinels are being used increasingly in electronic equipment operating in the
frequency range of 10 to 5000 Kc /s. At power and low audio frequencies the ferro-
spinels are not competitive with laminated ferromagnetic materials, and at very high
frequencies the losses in ferrospinels are excessive when high permeability is re-
quired. It is possible however to produce a ferrospinel, with low permeability, useful
at frequencies in the order of 100 Mc /s. The application of ferrospinels as core bodies
in the deflection yoke, horizontal deflection transformer and high voltage transformer
for television receivers is now finding wide acceptance. The ferrospinels are especi-
ally suited to television video frequencies as their use in these components results in
improved performance at lower cost and with smaller space.
In the standard broadcast receiver, the ferrospinels are expected to be used in the
radio frequency circuits as " trimmers " and as permeability tuning cores. With a
properly designed coil it is possible to tune a circuit, by the movement of a ferrospinel
rod, from 500 to 3000 Kc /s, or to cover the standard American broadcast band (540
to 1730 Kc /s) with only three eighths of an inch movement of the rod.
By using a ferrospinel with a high electric resistivity as the core body for radio
frequency inductances, the wire body may be placed on the ferrospinel without
additional insulation. In fact, the conductor may be affixed by the printed circuit
technique for some applications.
compensate for differences between individual receivers will not seriously affect the
coupling between primary and secondary.
To avoid losses due to adjacent components with high Q coils, iron cores can be
used which completely enclose the winding. These cores give higher Q's than the
more usual slugs with a similar winding and reduce the external field of the coil to
such an extent that in extreme cases it may be difficult to obtain sufficient coupling
between the secondary and an external high impedance winding.
An advantage of using an iron core in the oscillator coil is that, apart from giving
increased Q, it can greatly increase the coupling between primary and secondary,
thereby decreasing the primary reactance for a given amount of coupling.
(iii) Permeability tuning
The increasing use of permeability tuning is an indication of the extent to which
the design and production difficulties associated with this type of tuning are being
overcome.
(i) Design*
Much has been written on the subject of designing coils suitable for use on short
waves. A number of references will be found at the end of this chapter. The work
by Pollack, Harris and Siemens, and Barden and Grimes is very complete from the
practical design viewpoint. The papers by Butterworth, Palermo and Grover, and
Terman are basically theoretical. Austin has provided an excellent summary and
practical interpretation of Butterworth's four papers. Medhurst's paper gives the
results of measurements which in some cases disagree with Butterworth's theoretical
values. For coils whose turns are widely spaced the measurements of high- frequency
resistance are in good agreement but for closely-spaced coils the measured values are
very considerably below those cf Butterworth. Theoretical reasons are given for
these differences.
(A) Pollack (Ref. G4) summarizes the procedure for the optimum design of coils
for frequencies from 4 to 25 Mc /s as follows :-
1. Coil diameter and length of winding : Make as large as is consistent with the
shield being used. The shield diameter should be twice the coil diameter, and the
ends of the coil should not come within one diameter of the ends of the shield.
2. A bakelite coil form with a shallow groove for the wire, and enamelled wire may
be used with little loss in Q. The groove should not be any deeper than is necessary
to give the requisite rigidity. The use of special coil form constructions and special
materials does not appear to be justified (except for the reduction of frequency drift
due to temperature changes).
3. Number of turns : Calculate from
N = VL(102S + 45)/D
where S = ratio of length to diameter of coil,
D = diameter of coil in centimetres,
and L = inductance in microhenrys.
4. Wire size : Calculate from
do = b /N/ = optimum diameter in centimetres.
where b = winding length in centimetres.
This is a revision and expansion of the chapter on this subject in the previous edition by L. G.
Dobbie.
464 SHORT -WAVE COILS 11.5
(2) Single layer discs (pancake) : Winding depth equal to one -quarter of the
external diameter.
(E) Butterworth's paper (Ref. D1) deals with the copper loss resistance only, and
insulation losses must be taken into account separately. Insulation losses are
minimized by winding coils on low loss forms, using a form or shape factor which pro-
vides the smallest possible self-capacitance with the lowest losses. Thus air is the
best separating medium for the individual turns, and the form should provide only
the very minimum of mechanical support. Multilayer windings in one pie have high
self-capacitance due to proximity of the high and low potential ends of the winding.
The same inductance obtained by several pies close together in series greatly reduces
the self-capacitance and associated insulation losses. Heavy coatings of poor quality
wax of high dielectric constant may introduce considárrable losses.
(F) In the section of Medhurst's paper (Ref. D6) dealing with h.f. resistance he
states that so long as the wire diameter is less than one half of the distance between
wire centres, Butterworth's values are applicable. An increasing error occurs as the
ratio of wire diameter to distance between centres increases, the values being 190%
too high when the ratio is 0.9.
A graph is presented giving variation in optimum wire spacing with variation in
the ratio of coil length to coil diameter (Fig. 11.6A).
It is shown that a good approximation to the high frequency Q of coils of the type
measured is given by the simple expression
Q = 0.15Rç4
where R = mean radius of coil (cm),
f = frequency (c /s)
and depends on the length /diameter and spacing ratios.
4,6
(a) The ratio of wire diameter to wire spacing must approximate to the optimum
shown in Fig. 11.6A.
(b) The formulae apply only for very high frequencies because of skin effect con-
siderations. However the following table, giving the thinnest wire for which the
formula applies within ± 10 %, shows that most practical solenoids will be covered
f
(Mc /s) 1.0 4.0 16
-
wire 22 S.W.G. 28 S.W.G. 37 S.W.G.
(c) The formulae do not hold for coils of very few turns (or extremely short coils).
(d) Dielectric loss is not allowed for. This is unlikely to be serious except where
the coil has a rather poor dielectric (bakelite or worse) and is used in a circuit having
a low parallel tuning capacitance.
Io
O.
o
O
O.
l0 1 2 3 4 S 6 7 B 9 II
In turns near the centre of a solenoid the current concentrates on the surface of
each turn where it is in contact with the form, i.e. at the minimum diameter. In
turns at either end of a solenoid the maximum current density occurs near the
minimum diameter of the conductor, but is displaced away from the centre of the
coil.
Thus most of the conductor is going to waste. Multi -strand or litz (litzendraht)
wires have been developed to meet this difficulty. Several strands (5, 7, 9, 15 being
common) are woven together, each being of small cross section and completely in-
sulated by enamel and silk covering from its neighbours. Owing to the weaving of
the strands, each wire carries a nearly similar share of the total current, which is now
forced to flow through a larger effective cross section of copper. The former tendency
towards concentration at one side of a solid conductor is decreased and the copper
losses are correspondingly reduced.
Litz wire is most effective at frequencies between 0.3 and 3 Mc /s. Outside of
this range comparable results are usually possible with round wire of solid section,
because at low frequencies " skin effect" steadily disappears while at high frequencies
it is large even in the fine strands forming the litz wire, and is augmented by the use
of strands having increased diameter.
Screens placed around coils of all types at radio frequencies should be of non-
magnetic good -conducting material to introduce the least losses. In other words,
the Q of the screen considered as a single turn coil should be as high as possible. 1E11
addition, the coupling to the coil inside it should be low to minimize the screen losses
reflected into the tuned circuit. For this reason the screen diameter should, if pos-
sible, be at least double the outside diameter of the coil. A ratio smaller than 1.6 to 1
causes a large increase in losses due to the presence of the screen.
The design of coils for use with iron core materials depends mainly upon the type
of core material and the shape of the magnetic circuit proposed. Nearly closed core
systems are sometimes used with high permeability low loss material. More com-
monly, however, the core is in the form of a small cylindrical plug which may be
moved by screw action along the axis of the coil and fills the space within the inside
diameter of the form. The main function of the core in the latter case may be only
to provide a means of tuning the circuit rather than of improving its Q. When im-
provement in Q is possible with a suitable material, the maximum benefit is obtained
by ensuring that the largest possible percentage of the total magnetic flux links with
the core over as much of its path as possible ; the ultimate limit in this direction is
of course the dosed core.
An excellent series of charts for the design of single-layer solenoids for a required
inductance and Q is presented by A. I. Forbes Simpson in Ref. G17. These charts
and the instructions for their use are reprinted by permission of " Electronic Engineer-
ing " and the author (Figs. 11.7 to 11.13 inclusive).
The use of the charts is simple and gives direct answers. As so many unascertain-
able factors govern the final inductance of a coil in position, no attempt has been made
to achieve an accuracy better than 1 per cent. The inductances indicated by the charts
assume that the leads to the coil are of the same wire as the coil and perpendicular
to it, and are each a coil diameter long.
Coils of this form have a low self-capacitance which is largely independent of all
save the coil diameter and to a lesser extent the wire gauge.
The capacitance may be expressed as
C = D(0.47 + a) µµF
where D is coil former diameter in cms. and a is a constant depending on the gauge
of wire, lying between 0.065 for 42 S.W.G. and 0.11 for 12 S.W.G. wire, where one
lead of the coil is connected to chassis.
Use of charts
The charts shown may be used as follows :
If the desired Q is known, then reference to Fig. 11.8 will suggest a suitable diameter
of former.
(i) DESIGN 467
11.5
1114
III 11111111111111111[P
ilü uI
NN The Q in air of a coil wound to this information,
when tuned by capacitances between 45 and 500 µµF
may be read directly from Fig. 11.8.
o
The Q values shown in the chart have been cor-
rected for coil capacitance and are somewhat higher
E
than those indicated by the usual Q meter. The
accuracy of these curves is rather less than that of
the inductance charts, but various values have been
checked by several methods which have given sub-
stantial agreement.
Example
11111rIe11
III As an example* of the use of the curves, suppose
have a variable condenser of 450 µµF swing and
11111
nnuununuuu MEMMM we
0 ogram to illustrate use of Charts desire to find a coil with which it will tune from
FIG. 11.7
12 Mc /s to 4 Mc /s. If the total capacitance
at
Fig. 11.7. Diagram to illus- 12 Mc /s. be C µµF then
trate use of charts (Ref. G17).
(C 450)/C = 122/42 = 9 and C = 56.25 µF
L = 1/497.2f2C
1012. 102 µH = 3.15 µH.
x 122 x 10" x 56.25
= 47r2
Inspecting Fig. 11.8 we see that the Q at 4 Mc /s. will lie between 116} for in. and
}
former, while that at 12Mc /s. will lie between 182 for a in. and 265
168 for a in.
for in. former.
in Fig. 11.10
If we choose a in. former we find that the line corresponding to thisT.P.I. at 23.8
cuts the L line at a wire diameter of 0.025 in. This line intersects
and the " No. of turns " is 18.5 with a wire gauge of 22 S.W.G.
Referring to Fig. 11.8-we find for L = 3.15 µH :
100 200 400 500
C µµF 45
212 195 175 150 142
Q
Z, is readily obtained by writing Z, = QVL /C.
This example is presented in a different form from that in the original article.
468 SHORT -WAVE COILS 11.5
If, however, we had used the i in. former we would find that 18 turns of 26 S.W.G.
at 37 T.P.I. would give
C µµF 45 100 200 400 500
Q 181 165 147 123 117
(ii) Miscellaneous features
(A) Matching
Short wave coils are frequently matched by altering the position of the last few turns
on the former. This sometimes leads to damage to the coil, particularly if there is an
interwound primary.
sss!
An alternative method, which is applicable to coils terminated by passing the wire
sssssl
r ir.=.
diametrically through the former, is to push sideways this terminating wire in the
300
200
Q
C 8Y
.-.j.
t11!ii._"I"__1_'_'_'1."III"'1'
FORMER
aöäiáá===_?a.
MRa.\It.ItIMîiè`
....---.==--_ç___
.Nm._-_-_
-__-
---7iS7O-
z__-- _ -
I
100 mammimmi..mmi.mImmolmemno..mmE.
MMOMOIMemim.m.Mmimmem-im...mIli
..:... -
mom mnomm..mm..I:
80 MMIEN1MMONIM.MBEmmwm8===..mM EmliMMIM.MOMi.
Iimulmoi..mmm..
60 IMMIIMMINI...INIII..MMIMINI11111..1MMII111=1111.ThWO
slIMIIIIMIMIIIMMIIEMIIIIMII1MIIIIIMMMIIIIIMMIIIIM
300
200 ia!sesi1<!a!1i.i=a
' FORMER
....Az....á
-...0- .
--._..-...
-------__-_
..ra::_-.____-_c=_
s.=.
_____IIIIIIIII.
i-
100
eo
60
iiiiiiiiiiiiiiiiiiiii:
IIIMMIIIIMEI...11.1=1111= MII.NI..11111181.memmimmimm..
IMINIMIIMI111111111111=11111111111111111111MIMMIIIIIII
IIIIIIIIIIIIIIIIIIIIfl1111111111
3pQp
200
100
_--- =-a.
__ ---=Sáx1:s13'msèlziá?s9==1==1
c=--__._
====óóéóeóe°__--__-
MMailsfslMIi=
nCo-=7-
asas=.-Z-Mse!-..
__ _--- _ =__=-__e
=_°_'= pp -
i--'_=.
m__m=a..mo..oIInm1Il ......0
1MMu ===
I...
-M
MIME pp
-
MIMI= 1
MI
MMEMiiiMNiiiMM MEiiiiiiiiiiìi:SSZ
NIM
45
80 !1 =.11.11E M11=M8-1111.12. 1C 00
60 IMIEMME..IMIIEMINIMEM..M.IMMINIMIM..111110.:S. 200
Q 1REflfl!!!!'I!!!!I!I!!'II!!I!ii!IIIIu : 300
500
o
..MMiii=NEM..rftrx_M
__1
!00
m-m.o-m.u-m.o-m.. a,MN:mMii
80 INNIIMEMINIMMI.MENI..EMMIMMIMINE..ItMIN...`.%0\mit.
\.. -
YYF
2\rA\. -45
60 MOMININIM...IIII..NIIMMIMM.MMIMMEMNIMMQSZSZSIMCS,WZ,
MNIMMINIIIMIMIIIMINIIMIIIMINIIIIIIII\\N -100
eo IMMEEMEMINOMMEMENUM11UM11NEMEMini1gRFa -200 -ISo
- 400
500
1.1111111111111111111111111111111111111111.111111111111111
1 2 .3 4 '5 8 2 3 4 5
1
8 10 20. 30 40 50 80100
1. (Mk roAtnria) SIG. II.8
Fig. 11.8. Q/L chart for obtaining suitable size of former (Ref. G17).
11.5 (ii) MISCELLANEOUS FEATURES 469
middle of the coil. In this way the last half turn of the coil can be increased almost
to a complete turn or decreased to a small fraction of a turn. Both the first and last
half turns on the coil can be treated in this way, thereby providing a possible induct-
ance adjustment equivalent to approximately one complete turn of the winding.
When matching short wave coils it is essential that associated primaries, and any other
coils on the same former, be connected in a similar manner to their ultimate connection
in the receiver, i.e. earthed ends of coils should be earthed, and hot ends either earthed
or left open depending on which is more suitable.
MO
90
1.
......
......
.9:211Z.;"..9:12:11= ===C:...::.g;
..NrJ u..l
.===......1
....MNMMM
.. //........=..11116
Iw..::weemMI
....I.N1.
MEuu.=uN=1
.....NM
M
...
a0
70 01....I111f
1111... sum mom .Nm.mm MEN
.Il.......1.IIMM1II. .N1.r11 MINIM
....
11.11......mm11N1.1...IIIIIP .......11.MIM111
1111......1111111....EN.A111...../ :.11..S/
60 MIN ....11111111.. WEN ..E11=IIHgY.
MM MIME
I111.....um 4'
SO
::::::::......=::iI.su..i.ó:::i
....
ü::ii:::.
1s1s....
....... ....M .s....N .../II ....IM.MINIMMIII
I.........__
........
IMO ssuME..I.rI
q...N1111M..'/.
.uME.....s.1.........
...MINN
a..==imi VIM
MIME
MIN
Nall
r
.r/1'.
1111
111I...... .. .
..111.NIMN...1IrI
uII N MIME .....MII1M..
V..111111111.........MIM
;=
1....1111111111.....UM=
II11......11III11111........== I1I.r/1.unman 111....u.00I.
II
11111
II
20 111111I...IIIIIIIIII.......=MN11I./...IIIIIIIII.I......III=1
IIIII TINI IIII qIIIIMM N 1I D/ t11111111111g MMEIN 200
111111I1
111111111111111p1111\MINEIIIII4riI1IIII111:1 EMII1IIIIIIIIIIIN
111111111111111111n111111u11RI,,111111111C 1II11111NMN1111 ó
II11111111111111111111111M11fl411111. 11111 111111111
l0
9
6
111111111111111111111W1111:111M111111111n1_1
.....°:
Is::ii : .:Ó...1m....s.s..Isusr.....I.
.u.....
.....
:::========.%1:::::::E r1
.ME...........a
..O..Nr/..1/.....II'.I
...... .I.1NM1N//..
MM
.....n..
J....11:I AIM.
...p.....11'M :11......7;
...=_
.....I....
°
MINIM
.a.
....=
90
BO
7
II1I..II/1M..../.r
1111......
1111.1... 11.....1.....r//' III...Ii1111111II....NIMMI
M1IiIM.'ii+ .1......MIMMI
7O
60
ró'
6
3
x
:i::::::::
.IIl:iii:iiliilii:iM
uI1N..
I11...... ..rL1/
:::
:'.:::::: ..... ...: a.:
:iii_
.I.W..I/MI7.1
..
.C:Ç:Ç'GS:
-__"i'%i.r .:::::
..
..
M r.a. Ii:iii::i
- --=.CC
MIN NOB IIIIIIIII/--MBE _/f.''M1111111r.1V1111111111I111NI
:: wimmem, O
...N12p
mu
.... I.........
iC
..=
so
.::::O
.7
111111111 5II11111111!{ 1I1I11111111111111III11111111i1
111111iii1111I1I1M11111I111IIIIII1111IIII111III1111111MIN
:::::::::':I:'.:?';:::Ñ:::C
I ir::......zal.l:.:
9
11....1.
r1I
ME .==s.=.. .....
- "=======w
::::
1.......
....
Mn
....
.... llu
----'IAn
::::::::C==C to
ME
..........IIMMEMINII I/ M..1.Illll
.I.....u.l.u.Il
.
....
...=N=
...sm..
MN M=66=1=1
ll.....ul.
.......=1111M1111
92 s
7
í1i1..II
jl 1 ./, r
G
'%G1111 ....11
I . MII1...1.11.11llull
...NIU1u...u1/1 .. wain ..N..
.N1..1
I. ...u.l.
Emma 3
007I1
Iluul IIIIIN
1
111 M
of
II111
21111 11111N11111.....mNNll....11N111 11.1. ..Mm2
1f iIIllllllllll......u.m11MIl..\.111111l11.111.=..
11111/,r,I1IIII111111I.M.1I1111.I1111111II1N...
[OM EN MID ME] T aim
IIIGG11I11111111111111111MIMMIIIIIIIIIIIIIIIIIIIIIIIIMU
IIii111111111111IIIII111111 II II II II II IIIIIIIIII11111
R111111111IIIIIII111111111111IIII1I11111IIIIIIIII1111111111111
-005 005 -004 -003 -003
I
Iron cores are widely used for adjusting the inductance of short wave coils. Be-
cause at higher frequencies increased losses in cores offset the advantage of increased
permeability, the cores in some cases do not increase the Q of a coil, but provide a
convenient method of inductance adjustment.
.
In shortwave oscillator coils an iron core also has the advantage of increasing the
coupling between primary and secondary for a given winding. Thus increased grid
current may be obtained although Q is not increased.
120
loo :CCSI::III.::::C:-------......CS..I:Y' .. -----
....:::_
90
o.------.---u. .,r
Iiiiil:=::11:mlll::::::=_===Efi:EëEèCI::
.... ...............
...- -----
80 .
70
60
50
11.:111111.:::::----:ÌItÌ
111.11111111......111111111111..
11111..IIIIII.......=III... / IIII
am
111...1111 %
.
r
//....._--
40
30
°--
....-
.......-
.... / .. -------
.-_---r-
._.=-m
°O:%nl:aiirm
...-1.
...-
.r1...
..
.._.--.-..rn..
111.11111111....MN/I%'I....1111111.....
1/
----
---_---
.---
..__--
-----
....__-
..._.=
iiiii.iiilliiiiiiill::C:::::=1i::C::IIIIIIIIIIIC::::::
IIIIIIM IIIII IIIIIMIMINEW/ MI 11111111.M
111111// 1i 1111111111111 U111111I lilld!!' IMI1i IM
III
UIIIIIIIIIIIIIIIIIUIIIIIIIiIIIIIINI!iilP
i-
No. of Turns
l0
ó:aaeOl.WIüiSLi.C.eiIOs:STiüi'.Cii11í' dBII.aCiCEM 100
9
:::::ri1:::::::::.--v -=:i: .%i s::i!.iw.x..::: ii 90
0
8
7 ::::1:::I::i::::::::::::::::":as:1! IIU1ti ;,11111111::.=:i::C3 70
I=illili1:i111111iiliii7:MUMNP_:6il.1, :::III:Ir11.CI:C= 60
6
5 iiiiiiiiiiiiiiiiiiiiiiiii
spap
üiiiiiiiii:iiiiii=Ilii
I1:xg.- rs I®=1;.iI'IiIII=1::::::'_C;C:::
i=°:áiiïi
50
:Ci=:I:ilalll'11',_C-
! p
i::...-_ ......... C_:S____
.... 3i5 s-JG r" .-Ile...... ...-----
iiiii:::iiiiiiiiiiiì.ä
.. .,.._
á9:y:=.
::111:1:="""'=1111/8=:=
-..._.......................----
1I1IIIIiIIIa:3:11::
moue
i:::::'C1:1:::=.=.
2 11111.. IIIIIIIiI,j:. ,'r,..IN=1111..1111111... 20 j
_
Iii1II1C=ZIIIII;-,Ii;::::::iIIiII::::IIIII11:iiI=::CC
1
..
1111111 u:i1111/.1IIN111111:::111111111:1:::111
I1111115IllliOA ;1111111111111IIII1111111IIIIIIII111111
"
:IIIIII.'::1II%i!i11111111IIIIIIIIIIII
:-'---°C==::::::::: IIIIIIIIIIIIIII_
/
C=S-1;' ---:C:
==
10 i
i4I ...-------......
. ....-
.9 9
8 1C:':::I. viL. .__---
u
V' .._.--.>_-....
.._---- ....1111111.._.---
ri',T.if11::.".._:
//.111//111....111...11111111.......
raOC::::::::::.....i:::1111:òá 8
.
.._.
7
I riI
I1111ii
6 111 qr' 1/Ì
I ....
..IIIIIIIIUC_...S.IIIII...IIIIIIIIIIi1. , 6.
---- 'i'
ii 7 1: ilII_3SiIiII_I_I
111
s
"i'::i!!"r,ï===IIIIIIIIII.._:ZTS=: .=9iI.1 s
Ilium
VIII::iI11111111:..C111111111IIU1::CC
20 22 24 49 42
1ll
0.1 ,08 .0605 04 03 02 01 -008 006005 004 003 002 001
WIRE DIAMETER IN INCHES FIG. II.10
III1111111111n11111111111r7IIIIII11111! !11111111111101111
i ..--
...
r
11111111111111111111111i11I111111riî!11111111111111EE
lo.-..--.--
9
8
6
;NEI
iiii.iiiï
......N....... E.i.--a
...-
t.a::::I::I:::CC=C:C:== 100
iñiiiiiiiiiii.-_-__
.
Illll.l..I..I/..1.11 /I....
71.111.....111u1.
I.........
.r.......
11r, ...n.l..i
gql1Ill....0so
90
I N.tl.I.
80
s.IIIIIIINNN11111111111N1N
-----------os:A''::II:EE:m
4I:.IiNigÌKM1111111IIRIláEÌarrtl1III BIM C:I::N1:.::::C:C =1:1= 40
---."ao o==so
;::1 I:9E 11'i:ä
3 HI N..NIÌM
é .i'i.Wii
':::C:CC:::i":I::C:Pi:CC...
MENU
:?
..'/1:1111:::a1N'°
-S
Ell
- .. =EC T.C.:::
1:: .: ::::::::IE
:%=E i:::81 I1I1111I1 I E='.'.1:EEE 30
==-=
OE
i
..M 11...M
ri'==ilÌié:::IIIII:=I1::
11II.:::1:I1iV
i!'f'd
21II1I1I111
IIIIIII/IIII!
ii!' /M.II
; IIIIIIINIl!PA:-.._1I'i1NN11111III Ì11.1
11 IIIIINIIIIIIIIIIIIINM
1111111111tf'líl1'í'1
VA ,/
11111111111111111111u1 1
1111111.
1111111I11IINMEW
20 s'
1111111111111!!'.1111I111MMICIIIII11111I11IfII1111111Inn'
...--
m
!liilii111_IIIII1:IIIIIIIIIi11111111111111
I
iiÌl
,6
u.... ME
,71._.,,,,,ir,N:.111
1I.II.iii-.=IIIIIIiiiil11111Hiiiiiii.-=
I.. IIINEM
:Ñ1:.
..M-.2r IIMEN
MN
11
1
7
6
,511iÍI1011111111111INNINNIIIIIINNNIIIIIIII.IIIIaNN
..wÑ 5
_-:-i.
4ii.::::s=
- ..N....-.....::=:A....::. ...-omm 4
1:::::: ::M:.. .-MMM
...-....NEN.E. ....n ---mMl
I.N......
.31......N.. .NN.....--E.E.N..ENEEN..
N4.E....-EE.....NEE. N-.IM..I-N.E.
..N.N....-EEEEN-
MEE
..-...-EEE.E. ..NENE..
11.1......II.111111. E....-E.11111.....
IIN.N.E.E. N...E....-...E.
.. ......--m
3
II IN MU
I
11111I1IN NN 1111111111.1
N. 11111
..M
In
II II II II N IIIII111I I III..
II -.M E-Ntl 11111111.I
..M.
11...1IIIN.111.11II11.N ..NINNNIIIIINNNIIIIII111IIINNN..== 2
ZIIIIIIIINIIIIIIIIIIlIMIIIIIIIIIIII11111II1II1111111111IIINummum
®ENmLIIIm0EMM0MEIMI1.TlOQ7MIME_
111111111111I1111111111IIM1111111111111111111111IMINI
111111111111111111111111111111111111111111111111111111111111111111111
liiiilll111iiiiiiiillElIIIiEEllliil:IlliiiiiiillllllllME
008
0.1 06 05 404 -004 -003
-03 02 -01 005 .002 -001
WIRE DIAMETER IN INCHES FIG. IIIII
(B) Self-capacitance
To reduce capacitances in a coil with an interwound primary winding without
noticeably decreasing the coupling, it is possible the spacing factor of the secondary -if
winding allows it -to wind the primary against the wire on the low potential side of
each space between secondary wires. A definite advantage is gained from this method
of winding and at the same time the cause of appreciable random deviations in coil
capacitance is removed.
It is usually advisable to start an interwound primary winding just outside a tuned
secondary winding for a desired coupling with minimum secondary capacitance.
Best results will probably be obtained with the primary started between one half
l00
90
.mi.titttIlllllltttitt11111/.;/tI.......--
. .. : :::=CC:C:1=.
------- - 1111:f1
0
70
--
-=111. iIIIii:.. ...
-m 1.irtilltIIIItItlllret1111!d ttIIIIIIIIttettClEMEC
ll.i....I.-LI
.- 1.111.111111..11.
1:®ml!..
-III1...- I ..
I
60
so
1.1111111..
11111111IIIIIIII.III1G.i.Íln.ii:1111111i.11I.1
==::EE=:'.:;sj:;ll::::::--- Ï 11 :: j 1a._a=e,== _1i=.._' -.- -._-==..1
40--......---:.. 1:-::-xasl.f.ilEl..l=_9:i..:--
i: -I... ... --_:_e._==
=
elj¡ljllllllljliljlljjSiÌ .iixx==x.
äjlstijjjFiliIIIIlliljlit----
....-.
.
2- . . .. ....1s1-
.. 11
. ---------
rat.---
30--. .=i:i11.. /.1191...il1i=ii.1=111111.
r,'r.- .
1111..
II.1....r.t.1
IIIIIi11/í1i
I.11111. II
Ir./.'/.rI.I.II.1I.11IIl:l.I..II. 1 11
C.
. s..
MN= 11111111111111Es1
I-rr
20 200
n
I
IIIlei ,r,jíiÌiÌ
iiiiiii
10
8
_IÌ
-_ ÌIÌIIíIIIIIIÌÌÌÌÌÌ/1111111
t: n i Ir a utl111tniiiiJ imo111:1'itIt1111111
--.:ter.lIItIIt:.atar.xxa--C
.. -- Itt:;:ttIllltlet11C.------
O:C ii:i'iÍt1i:ii:=i1i ilis:::11:11t0W1f1tt11lllliiii"
. C..=2; e0 if
1111.
------
--iiiiZE=
100
90
7
ZE'í::11111Iiiili"" ;i1i:11mPí1i111111Illllill....::::EC= 6070
I.-
6
5 I?iiíiiiilillillliiiP.11i11C'wi:,IIi11i:1i11ii:iiiii::i:iiiZC50
r=éi13:1-.. ,.
:=
t:Illiliiéié`!iilii::s
1II11=Eeai::=?=____
:x:::::ii-
'::
.
4 =-
: ..v.::
3 V::::::::::::I.i :::=====....:::-
--.. --IMI -------
,1 i=..ì...=---..i.i..-:
$:i O= ::ii=::__ 40
...=======
==l:iiii4l.iii1lt
...
M1 /1.1I!ill .._. ..m. I .1mm
Ill .....I.1.1MOM
-INu1...
.o1 i1.ItI14111 In'II..n G...=-M
..E= ni 1 nn.....---
SO
..-----MIIII 30
uM
1lIÌIi111111:11:.I_
t tt
1
M_11111111;1111111iI1I.ME
M 1 1 11
1 11
IU I 1
x
1111111I:IIIII111111111111IIIIIIIIIIIIIIIIIIIII:::II111
.+
6
1
e x:::::
.9
.7
-rr..I
-......-I....
..1.. ....
.
.IIIIII;::III1I111111111111111111111:IIIIIIII11111111.
I:.-.
--..::.'::=iI::t::::::::::x=====:te::::.-.:IIRI:::rl::::Cm-
----I.-.1.121----xx=:=w_
-,.... ... . .-
..
..---.-- .
..1....C--.i......
-r:í.,...1.11/
..
1 Ii1i1111=.11111111IIIIlIlIn1l:Í
...1
....-
-------
.II.
......-
.111
111I1I IIIIi
...mim-mo e7
1 11111 II III
--- 10
9
. ii
.1111 6
.5 il 1 aomulum
AMIN.
¡;;::::==!111I1:i:5==:1i1:111:::11111iiliijjj 5
.4 e?:+í s:e:áulmnulil i
...-::-_
i:I3I::::r:Eàxl.....:éEEEBEòé"á=--9:iii:il=élil:i::
tiCi::
4
...ii
:l. CC I::: t:::: II:I:Iltrl::t
te:::::=:II11I:::::
.3
:=i:101j !¡ u i:::::n:oC
:::1:IC
3
.2
1 Illil i11
i11.
i©.
11..1111811
i1
111
2
.r....:.:
10 8
:: :::..
..1......
...
...
6
':::M__
5
:_ó
4
_._MIM.
...____MMM.
..._IMMEMEII
3 2
100
.1111......11111111.1.......
...1......11..1111 .....MMEN
u......uu.u......MM. //
:'Z' -- :aC,.
HIM NM 11111111111NEII :i 11111 I
50
........
_......._
:::::::::==="'W
:::ZCmC: :::::::::.'::CC:E _..
.. N.____I.,MEM ...
u........1..............
I
MINN
1.........11........ //M
IIIM -
1111......1111111..1
.1......111.111... Pm=
.//M=
111111 IIIII IOII II I: /IMMI
1.N111111111111xnrJAMI
11111111111111111111./
....._
IIIIIIIIIIIIlIIIIIIIIIIP!
..i:iio._ ...ur/.iü_ 10
........_
......._ ...
MIA __
uu...Nr.___
........
1...... ..../...=
../.._
8
11...... /.....
11
11111111.1.1/....
1..1......11.1.1..1.. 6
111111 11111111111,/1 s
::::.:.:::=::::::11: ',::EZ=
"""*"'::1117." "':::7/::CZ EC
i.......i.
.... _ ...t. /......
Ammo _MI
......i Ìii Ì/Ì
.1........111111M4 MEN
Ili MO NM MU MN NM
..
111111
1111111 IMO 11111 MOM= I
2
I11 II
......... . :M:::__
imm
....
l= ..
........I,.
111.NIMO/ /.111.1.1... Winn
......... ..
111111 I 11111111111111111M NM
/.
MM 11111111111
:......._.
:EEEi:,=__.. _:::____
...____
_.uri
u_.d .,_..
. u..iiii...o
OiC:,::5:: ENTI
III, ::11 =1111111M ::C:C::
11 AIM .111.11..........
11' 11111111111111
11 1111111111111111I
.I
10 8 6 5 4 3 2 I
Fig. 11.13. Effect of circular and square screening cans on inductance (Ref. G17).
474 (ii) MISCELLANEOUS FEATURES 11.5
and one and one half turns outside the cold end of the secondary, close spacing being
used for parts of the primary outside the secondary and the external primary turns
being wound as close to the last secondary turn as possible
of its self -capacitance, i.e. the impedance is inversely proportional to frequency and
eventually reaches values low enough to render the choke useless.
Thus to obtain high impedance over a wide band of frequencies a choke requires
maximum inductance and minimum self-capacitance. For lower frequencies universal
windings are used, usually split into two or more spaced pies with each pie as thin as
possible for minimum self- capacitance.
To keep the impedance high over a band of frequencies it is necessary to avoid any
tuning of sections of the choke by other sections. A typical case of such tuning
would occur in a r-f -c consisting of two pies of different sizes. The impedance -
frequency curve of such a choke would show two maxima at the self-resonant fre-
quencies of the two sections, but between these peaks there would be a serious drop
in impedance at the frequency at which the inductive reactance of one pie became
series resonant with the capacitive reactance of the other. At such a frequency there
would also be a sudden change in the effective reactance of the choke which would
give tracking errors.
Series resonance of the type described occurs whenever pies with different self-
resonant frequencies are connected together. To eliminate them it is not sufficient
to wind several pies with the same numbers of turns because the effects of mutual
inductance in different sections of the coil still give different self -resonant frequencies
to adjacent pies. To avoid this, Miller (Ref. H6) recommends winding progressively
smaller pies towards the centre of a multi-pie r -f-c to make the self resonant frequency
of each pie the same. He gives figures which show a considerable increase in uni-
formity of impedance over a range of frequencies when this method is adopted.
In addition to the need for individual pies to be resonant at the same frequency,
Wheeler (Ref. H5) gives the requirement (for two -pie chokes) that the pies should
be connected so that mutual inductance opposes the self-inductance of the pies.
This almost entirely removes minor resonances which occur at approximately har-
monically related frequencies, and gives a very smooth curve of apparent capacitance
for the r-f -c. The spacing between pies and the number of turns on the pies should
be experimentally adjusted for best results with the choke located in the position in
which it is to be used.
(ii) Baking
Irrespective of the moisture proofing treatment which is to be applied to a coil,
the first requirement is to remove all moisture. This is done most conveniently
by baking at a temperature above the boiling point of water. Materials used in coil
winding are liable to be damaged at temperatures greatly in excess of 100 °C and a
satisfactory oven temperature is from 105° to 110 °C.
The time required depends on the components being treated, and it is essential
for all parts of a coil to be raised to oven temperature and maintained at that tempera-
ture at least for a short time. Baking for a quarter of an hour is a minimum for simple
coils, and half an hour is a more satisfactory time.
11.7 (ii) BAKING 477
One detail of oven design which must not be overlooked is that provision must
be made for removing the water vapour as it is expelled from the coils. A current
of fresh air passing through the oven is satisfactory for this purpose.
Whether the baking is followed by impregnation or flash -dipping only, the next
process should begin before the coils have cooled.
(iii) Impregnation
Although vacuum impregnation gives the best penetration of the impregnant and
best removal of moisture vapour and gas, a " soaking " treatment is often used and
gives satisfactory results for commercial requirements.
A typical soaking treatment calls for the coils to be immersed in wax which is main-
tained at 105° to 110 °C for a period of from one quarter to one half hour, depending
on the type of winding, e.g. a thick universal winding needs more time than a spaced
solenoid. In any case the coil should be left in the wax until no more bubbles are
given off.
The temperature of the wax-tank needs accurate adjustment as apart from damage
to coils which may result if it is too high, the characteristics of the wax may be altered
by excessive heating.
Soaking in varnish (air -drying or oven -drying) is also possible but it is necessary
to keep a close check on the viscosity.
(v) Materials
The first requirement for an impregnating compound is good moisture resistance.
It is also important that its losses be low at the frequencies concerned, and it is quite
possible to obtain impregnants which can be used without noticeably affecting the Q
of the impregnated winding.
Because it is usually desirable to have minimum self-capacitance in a coil, the dielec-
tric constant should be low, and the temperature coefficient should also be known.
Waxes can be obtained with dielectric constants having a negative temperature co-
efficient and this is useful in offsetting the temperature coefficient of the coil itself,
which is positive in all usual cases.
Apart from the resistance of a moisture proofing compound to moisture absorption,
its surface properties are important. A compound which does not wet easily is de-
sirable and tests on moisture proofing should be designed to separate the effects of
surface leakage from those of absorption.
478 (v) MATERIALS 11.7
Where flash- dipping only is carried out, some of the properties mentioned above
are not of such importance but they cannot be ignored completely.
Other points which must be considered when choosing a moisture -proofing com-
pound are that it must be capable of maintaining its properties over the range of tem-
peratures to which it is liable to be subjected. Melting of waxes is the trouble most
likely to be encountered, but crazing (and the consequent admission of moisture)
may be experienced at low temperatures.
Waxes and varnishes do not themselves support fungus growth, but they are
liable to collect coatings of dust which may do so. Suitable precautions to avoid
this may be desirable. More important is the avoidance in coils as far as possible of
materials which are subject to fungus attack. Some types of flexible tubing fall in
this category and should not be used, and the same applies to some cheap coil former
materials. The fabric covering of wires also may support fungi and should be com-
pletely coated with the wax or varnish to avoid this occurrence.
SECTION 8 : REFERENCES
(A) REFERENCES TO IRON CORES
Al. Nottenbrock, H., and A. Weis " Sirufer 4 " Radio Review of Australia 4.4 (April 1936) 5.
A2. Austin, C., and A. L. Oliver " Some notes on iron -dust cored coils at radio frequencies " Marcon
Review, No. 70 (July 1938) 17.
A3. Editorial " Iron powder compound cores for coils " W.E. 10.112 (Jan. 1933) 1.
A4. Editorial " Iron core tuning coils " W.E. 10.117 (June 1933) 293.
A5. Editorial " Iron powder cores " W.E. 10.120 (Sept. 1933) 467.
A6. Friedlaender, E. R. " Iron powder cores : their use in modern receiving sets " W.E. 15.180 (Sept.
1938) 473.
A7. Editorial " Distribution of magnetic flux in an iron powder core " W.E. (Sept. 1938) 471.
AS. Welsby, V. G. " Dust cored coils " Electronic Eng. (1) 16.186 (Aug. 1943)15.180
96 ; (2) 16.187 (Sept.
1943) 149
; (3) 16.188 (Oct. 1943) 191 ; (4) 16.189 (Nov. 1943) 230 (5) 16.190 (Dec. 1943) 281.
Correspondence by E. R. Friedlaender and reply by V. G. Welsby Electronic Eng. 16.192 (Feb.
1944) 388.
A9. Cobine, J. D., J. R. Curry, C. J. Gallagher and S. Ruthberg " High frequency excitation
cores " Proc. I.R.E. 35.10 (Oct. 1947) 1060. of iron
A10. Buckley, S. E. " Nickel -iron alloy dust cores " Elect. Comm. 25.2 (June 1948) 126.
All. Polydoroff, W. J. " Permeability tuning " Elect. 18.11 (Nov. 1945) 155.
Al2. Bushby, T. R. W. " A note on inductance variation in r -f iron cored coils " A.W.A. Tec.
6.5 (Aug. 1944) 285. Rev.
A13. White, S. Y. " A study of iron cores " Comm. 22.6 (June 1943) 42.
A14. Friedlaender, E. R. " Permeability of dust cores " Correspondence, W.E. 24.285 (June
A15. Fetherston, N., and L. W. Cranch " Powder metallurgy and its application to radio 1947) 187.
Proc. I.R.E. Aust. 5.9 (May 1945) 3. engineering "
A16. Tucker, J. P. " A permeability tuned push button system " Elect. 11.5 (May 1938)
A17. " Powdered-iron cores and tuning units " Comm. 21.9 (Sept. 1941) 26. 12.
A18. Friedlaender, E. R. " Magnetic dust cores " J. Brit. I.R.E. 5.3 (May 1945) 106.
Al9. Foster, D. E., and A. E. Newlon " Measurement of iron cores at radio frequencies " Proc.
295 (May 1941) 266. I.R.E.
A20. Jacob, F. N. " Permeability -tuned push -button systems " Comm. 18.4 (April
A21. Fetherston, N. " Magnetic iron and alloy dust cores " R. and E. Retailer (1) 22.21938) 15.
18 ; (2) 22.3 (12 Oct. 1944) 20. (28 Sept. 1944)
A22. Polydoroff, W. J. " Coaxial coils for FM permeability tuners " Radio 31.1 (Jan. 1947)
A23. Polydoroff, W. J. " Ferro-inductors and permeability tuning " Proc. I.R.E. 215 (May9.
A24. Information supplied by Mr. L. W. Cranch of Telecomponents Pty. Ltd., Petersham, 1933) 690.
A25. Samson, H. W. " Permeability of dust cores " W.E. 24.288 (Sept. 1947) 267. N.S.W.
A26. Bardell, P. R. " Permeability of dust cores " W.E. 24.281 (Feb. 1947) 63.
A27. Harvey, R. L., I. J. Hegyi and H. W. Leverenz " Ferromagnetic spinels for radio
R.C.A. Rev. 11.3 (Sept. 1950) 321. frequencies "
A28. Strutt, M. J. O. " Ferromagnetic materials and ferrites " W.E. 27.327
A29. Snoek, J. L. " Non -metallic magnetic material for high frequencies "(Dec. 1950) 277.
Philips Tec. Rev. 8.12
(Dec. 1946) 353.
A30. Ferroxcube" Philips Industrial Engineering Bulletin, Australia, 1.5 (Jan. 1951) 18.
Additional references will be found in the Supplement commencing on page 1475.
(B) REFERENCES TO INDUCTANCE CALCULATION
Bl. Parington, E. S. " Simplified inductance chart " Elect. 15.9 (Sept. 1942) 61.
B2. Maddock, A. J. " Mutual inductance : simplified calculations for concentric solenoids
22.263 (Aug. 1945) 373. " W.E.
B4. " The inductance of single layer solenoids on square and rectangular formers " Data Sheet,
tronic Eng. 15.173 (July 1942) 65. Elec-
B5. Amos, S. W. " Inductance calculations " W.W. 47.4 (April 1941) 108.
B6. Everett, F. C. " Short wave inductance chart " Elect. 13.3 (Mar. 1940) 33.
B7. Turney, T. H. " Mutual inductance : a simple method of calculation for single -layer coils on
same former" W.W. 48.3 (March 1942) 72. the
B8. " The inductance of single layer solenoids " Data Sheets Electronic Eng. (1) 14.164 (Oct.
447 ; (2) 14.165 (Nov. 1941) 495. 1941)
B9. Blow, T. C. " Design chart for single layer inductance coils " Elect. 16.2 (Feb. 1943) 95.
B10. Grover, F. W. (book) " Inductance Calculations " (D. van Nostrand Co. Inc. 1946).
B11. Blow, T. C. " Solenoid inductance calculations " Elect. 15.5 (May 1942) 63.
11.8 REFERENCES 479
Section Page
1. Introduction ... ... 481
2. Resistance -capacitance coupled triodes 482
3. Resistance -capacitance coupled pentodes 496
4. Transformer -coupled voltage amplifiers 517
5. Choke -coupled amplifiers ... 521
6. Methods of exciting push -pull amplifiers 521
7. Push -pull voltage amplifiers 527
8. In -phase amplifiers ... ... 529
9. Direct -coupled amplifiers ... 529
10. Stability, decoupling and hum ... 535
11. Transients and pulses in a-f amplifiers 540
12. Multistage voltage amplifiers ... 541
13. References ... 542
SECTION 1 : INTRODUCTION
(i) Voltage amplifiers
A voltage amplifier is one in which the voltage gain is the criterion of performance.
To be strictly correct it is not possible to have voltage without power since infinite
impedance does not exist in amplifiers, but for all ordinary purposes a " voltage
amplifier " is one in which a " voltage " output is required. Voltage amplifiers
generally work into high impedances of the order of 1 megohm, but in certain cases
lower load impedances are used and there is no sharp demarcation between " voltage "
and " power " amplifiers. In cases where transformer coupling is used between
stages, the secondary of the transformer may be loaded only by the grid input im-
pedance of the following stage and the numerical value of the impedance may not be
known. In such cases the transformer is usually designed to operate into an infinite
impedance, and the effect of normal grid input impedances on the transformer is
very slight compared with the primary loading.
It is important to bear in mind the reversal in polarity which occurs in any valve
when used as an amplifier with a load in the plate circuit. As a consequence, the
a -f voltage from grid to plate is
Eav = Eck + Ekv = Eok(1 + A) (I)
where A is the voltage gain from grid to plate.
481
482 (i) CHOICE OF OPERATING CONDITIONS 12.2
e
in Sect. 2(ix).
-411. The plate supply voltage (Ebb) should
FIG. 12.1
generally be as high as practicable pro-
vided that the maximum ratings are not -
exceeded. Plate supply voltages up to (P
14
µ not greater than 40, a grid resistor up to 5 megohms may be used, provided that the
total resistance of the plate load resistor and the plate decoupling resistor is not less
than 0.1 megohm.
If grid -leak bias is used, R, may be from 5 to 10 megohms.
The output grid resistor (Rv,) may be the maximum recommended for the
following stage by the valve manufacturers-usually 0.5 megohm for power valves
with cathode bias -or as determined by eqn. (6) or (7) in Chapter 3 Sect. 1(v)d.
Lower values are desirable to reduce the effects of reverse grid current in the follow-
ing stage, and there is no appreciable advantage in either gain or distortion through
the use of a resistance greater than 4 R L. Calculation of the maximum grid resistance
for use with power valves is covered in Chapter 13 Sect. 10(i).
(ii) Coupling condenser
The coupling condenser (C) may be selected to give the desired low frequency
response. The loss of voltage due to C may be calculated by the use of a vector dia-
gram or by the following equation, which applies to the circuit of Fig. 12.1 provided
that the input resistance of V, is very high.
E, /E = R /IZI (la)
where E, = signal voltage on grid of V,
E = signal voltage across RL
R = R,, rPRL/(ra + RL)
or as an approximation, R R,! if the plate resistance of V, is
small compared with R0,,
IZI = magnitude of series impedance of R and C
/R' + X,''
and X, = 1 /0,C = 1 /27.íC.
For example, if R = 1 megohm and f = 50 c /s, the following results will be ob-
tained
db loss
- E, /E X ,/R X, C
1 0.891 0.51 0.51 megohm 0.006 24 µF
2 0.794 0.76 0.76 megohm 0.004 19 µF
3 0.708 1.00 1.0 megohm 0.00318 µF
The phase angle shift is given by
= tan-1 1 /wCR = tan-' 1 /27rfCR (lb)
In certain cases a low value of C is adopted intentionally to reduce the response
to hum arising from preceding stages. However, the use of a low coupling capacit-
ance, when the following grid resistor is 0.5 megohm or more, increases the hum
contributed by the following valve through the a.c. operation of its heater. A low
coupling capacitance should therefore not be used on a low -level stage. In high
fidelity amplifiers a fairly large value of C is generally adopted, thus not only improving
the low frequency response but also reducing phase shift and possibly also improving
the response to transients. However, excessively large values of C are undesirable.
The following table gives the approximate values of C for certain selected con-
ditions. Note that R must be as defined below eqn. (la).
COUPLING CONDENSER
Attenuation 1 db at 12.5 25 50 100 200 c/s
2 db at 8.5 17 34 67 134 c/s
3 db at 6.5 13 26 51 102 c/s
=ii-_-
0 --M_'
=
...
--
'... u.....
4 ... 4 4_-u
.......-----
==ae- m-;.'::::.fil:
- EE:iBEi; ..EB E::E28E9i9;
ss
59:EiEBE;E=_EEE?ee E===EE
10 20 50 00 200 SOO IQ00 2P00 SÍ700 IOpC0
FREOUENCY C/S
Fig. 12.3A. Frequency characteristics due to cathode by -pass condenser with general
purpose triode having t = 20, r,,, = 10,000 ohms, RL = 0.1 megohm, Rv, = 0.5
megohm, R k = 2700 ohms.
If positive grid current is flowing the bias will be increased by I,R,,.
The cathode by -pass condenser Ck is only fully effective at high frequencies, and
it becomes increasingly ineffective as the frequency is decreased-under these con-
ditions there is degeneration (negative current feedback -see Chapter 7). The
effect of Rk and Ck on the voltage gain at any frequency is given by the equation
¡A' 1 + (WCkRk)'
= (R (2)
IA Ll + +1 rRkJ
D
+ (co CkRk)2
where A' = f
stage voltage gain at frequency with self bias resistor Rk by- passed
by condenser Ck
A = stage voltage gain with Rk completely by- passed
= mid -frequency voltage gain
w = 2irf = 2,r x frequency of input signal
*Also known as reverse grid current.
12.2 (iii) CATHODE BIAS 485
RL =
= r valve plate resistance in ohms at the operating point
resistance of plate load resistor in ohms
12,2 = resistance of following grid resistor in ohms
and R' = RLR,z /(RL + R22).
The derivation of eqn. (2) is given in Ref. B11 ; see also Refs. All, A13.
The attenuation characteristics of a typical general -purpose triode with cathode
bias are given in graphical form in Fig. 12.3A. It will be seen that all curves have
the same shape, but are shifted bodily sideways depending on the value of Ck. The
maximum slope of the curves in Fig. 12.3A is 1.4 db /octave ; the slope does not
normally exceed 3 db /octave when R5 is the optimum value to provide bias.
The limiting loss of gain at zero frequency due to Rk is given by
A' R' + r (3)
A R' +r +(µ
+1)Rk
Examples of limiting loss of gain at zero frequency (based on equation 3)
Type RL R,2 R' Rk µ* r,,* A' /A loss
6J5 0.1MS2 0.5MS2 0.08MS2 2700 18 17 000 0.65 3.7 db
6Q7 0.25M S2 0.5MS2 0.17MQ 3000 68 70 000 0.53 5.5 db
6SQ7 0.25MS2 0.5MS2 0.17MS2 3900 100 100 000 0.40 8.0 db
*At operating point.
Volis of e
1.5
25
12 t
Ñ
14 9
oi
10
Pr 0.5 Rk/xk 5 10 SO
1e
20
100
, ...,
s 10 SO 100 S00 1,000
Frequency ch FIG. 12.38
and at higher audio frequencies. The maximum value for any normal r.c.c. amplifier
does not exceed about 30 °.
Universal curves showing phase angle shift are given in Fig. 12.3C in which the
symbols have the same meaning as in Fig. 12.3B.
(C) Choice of cathode bias resistor
The best method is to determine the grid voltage as for fixed bias -see (iv) below
and then to calculate Rk. A fairly satisfactory approximation is to select Rk by'the
-
following table.
Con d itions : Plate supply voltage 200 to 300 volts. Values of /A are the published
values.
+6d
41111k1
,.',"'Itt,VALUES OF
B
.'/,G101111MINW11IIMII
/í'/.:1111112WI`IMINI\
0I
/i11111\'\ti\
0.2
.101M101a\21.0l01111C.
/%1111Mmà111116\001M411\
0.4 0.6 0.8 I 2 pk/xk 4 6 a 10 20 40 60 80 IOÓ
..
ió
1
SO 1Ó O I
Soo 1,000
Frequency cis FIG. 12.3C
Fig. 12.3C. Universal phase angle shift curves with cathode bias (Ref. A 1 1 Part 4).
(D) Cathode bias loadlines
Cathode bias loadlines may be drawn on the plate characteristics as on the mutual
characteristics [Chapter 2 Sect. 4(v) and Fig. 2.27] but they will no longer be straight
lines. Fig. 12.4 shows the loadline and plate characteristics for a high -mu triode.
The cathode bias loadline may very easily be plotted for any selected value of Rk,
e.g. 10 000 ohms. In Fig. 12.4 each curve applies to increments of 0.5 volt in E,,
so that (with Rk = 10 000 ohms) the successive increments of lb are 0.05 mA-
E, = 0 -0.5 -1 -1.5 -2 -2.5 -3 volts
/1, = 0 0.05 0.1 0.15 0.2 0.25 0.3 mA
The intersection of the selected Rk loadline with the RL loadline gives the operating
point (Fig. 12.4).
Alternatively, the intersection of the R L loadline with each grid curve may be
marked with the corresponding value of Rk ; e.g. in Fig. 12.4 at point A, E, =
and lb = 0.47 mA, therefore Rk = (1.5/0.47) x 1000 = 3200 ohms. Similarly at
1.5 V -
point B, Rk = 7800 ohms. If greater accuracy is required in determining an in-
(iii) CATHODE BIAS 487
12.2
200 300 R,
PLATE VOLTS
Fig. 12.4. Plate characteristics of high -mu triode (6SQ7) with E b b = 400 volts and
RL = 0.5 megohm. Cathode bias loadlines for selected values of cathode bias resistance
(R,) have been drawn in.
Fig. 12.5. Resistance-capacitance coupled triode with fixed bias from voltage divider.
k2
ti 100 R1, for high -mu valves with cathode bias.
Consequently, with high -mu triodes, the resistance of the grid resistor
made 1.5 times that with fixed bias for the same effect on the operating -point.
may be R
'These values of k1 are only typical, and are likely to be exceeded by some valves. In each case, a
value should be determined for the valves being used.
is approximately true when the reverse grid current is nearly all due to ionization-see Ref.
GI, Fig.Bs t
12.2 (v) GRID LEAK BIAS 489
Rk
RL- K(RL +r9)
N (6a)
RL
ti RL +r, L
+µR5 (6b)
Relation between K and grid voltage
-
E, [RL K(RL + r,)[(Ebb / /RL)
Eqns. (6a), (6b) and (6c) are exact for linear characteristics, but only approximate
(6c)
in practice.
Example µ = 20, r, = 10 000 ohms, RL = 0.1 megohm, Ebb = 250 volts
:
K = 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
E, = -12.5 -11.1 -9.7 -8.4 -7.0 -5.6 -4.3 -2.9 -1.5 -0.1 +1.25
Distortion
The percentage second harmonic distortion-see Chapter 13 Sect. 2(i) -is given by
EQ -QF
ga% 2(EQ + QF) x 100
(10)
Fig. 12.7. " Ideal " linear dynamic characteristics (A) for general purpose triode
with u = 20, r, = 10 000 ohms, RL = 0.1 megohm and Ebb = 250 volts ; (B) for
high -mu triode with µ = 100, r, = 0.1 megohm, RL = 0.25 megohm and Ebb = 250
volts.
Point C at E, = 0 has a plate current Ibi = Ebb/(RL-I- r,) which in this case is
0.91 E bb/RD This point and point A may be used as the two basic points for plotting
the loadline. Alternatively, point B may be plotted, since FB = Ebb /g mR L. Point
G is the commencement of grid current (E, = - 0.5 volt) and in this case its 162 =
0.86 Ebb /RL. The highest operating point cannot exceed 0.85 Ebb /RL, and this is
only possible for extremely small input voltages. The lowest useful operating point
(for high-level operation) is D, which is the mid point of AG, with 163 = 0.43 Ebb /R L.
Thus the operating point must be within the limits 0.43 and 0:85 x Ebb/RL.
The dynamic characteristic in Fig. 12.7B applies to a high mu triode with µ = 100,
r, = 0.1 megohm, RL = 0.25 megohm and Ebb = 250 volts, The grid current
is taken as commencing at E, = - 0.75 volt. The usable part of the characteristic
extends from A to G, and the operating point must be within the limits 0.25 and 0.5
multiplied by Ebb /RL. The upper limit would be somewhat extended if the load
resistance were increased to 0.5 megohm.
(ix) Maximum output voltage and distortion
It is difficult to lay down any limit to the maximum voltage output, since over-
loading occurs very gradually. It is assumed that in all cases the grid bias is sufficient
to avoid positive grid current.
492 (ix) MAXIMUM OUTPUT VOLTAGE AND DISTORTION 12.2
(1/x) Jr 2S+
where r,, = plate resistance at point C as given by slope of FC)
S = Im1nr, /(Ebb
x = R,2 /r,
-
e1)
c'
Q
R1
¡min
O e, G Ebo H D
Fbb
Eo -
PLATE VOLTAGE FIG. 12.76
Fig. 12.7C. Approximate method for determining conditions for maximum output
voltage (Ref. A16).
The value of y for low distortion in a single valve is 0.5 ; values appreciably higher
than 0.5 are possible with push -pull operation. The value of e1 is a function of the
valve characteristics, the load resistance and the plate supply voltage ; it is usually
less than 10 volts for RL not less than 0.1 megohm and Ebb not greater than 300 volts.
Example : Type 6J5, r, = 10 000 ohms, e1 = 5 volts, Ebb = 250 volts, Rv2 = 0.5
megohm, I,,,,, = 0.2 mA, y = 0.5. We obtain x = 50, S = 0.008 17, y/x = 0.01,
and from the equations, RL N 70 000 ohms optimum ; E. = 186 volts peak -to-peak.
Output voltage and distortion
The maximum output voltage for a general purpose triode (e.g. type 6J5) for 14%
intermodulation distortion* and Ebb from 180 to 300 volts is given approximately by :
Load resistance Following grid Maximum output
resistance voltage r.m.s.
0.1 megohm 0.1 megohm 0.155 x Ebb
0.2 megohm 0.19 x Ebb
0.4 megohm 0.215 X Ebb
0.25 megohm 0.25 megohm 0.20 X Ebb
0.5 megohm 0.24 x Ebb
1.0 megohm 0.27 x Ebb
'For details of intermodulation distortion see Chapter 14 Sect. 3.
12.2 (ix) MAXIMUM OUTPUT VOLTAGE AND DISTORTION 493
For lower values of I.M. distortion, the preceding values of maximum output
voltage should be multiplied by :
I.M. RL = 0.1 megohm RL = 0.25 megohm
10% Factor = 0.8 Factor = 0.72
5% 0.48 0.36
2.5% 0.26 0.18
With a high -mu triode the voltage output for 14% intermodulation distortion is
only about 0.17 Ebb to 0.21 Ebb for RL = 0.25 and R,2 = 0.5 megohm.
Note that with intermodulation distortion the ouput voltage is taken as the arith-
metical sum of the component voltages -see Chapter 14 Sect. 3(ii).
The intermodulation distortion of a triode and a pentode are compared in Fig.
12.16A [Sect. 3(ix)]. At low output voltages the pentode gives less distortion while
at high output voltages the triode gives less distortion. The triode is, however, less
critical than the pentode, and the distortion increases at a lower rate than with the
pentode when the bias is made more negative than the optimum value.
and ¢ = angle by which the voltage across the load impedance leads the equival-
ent voltage acting in the plate circuit (¢ will be positive for an inductive load and
negative for a capacitive load)
When the load is capacitive, as usually with a r.c.c. amplifier, R' is positive and
there is some slight additional loading of the input circuit. When the load is induc-
tive, R' is negative and self oscillation may occur in an extreme case.
(xii) Equivalent circuit of r.c.c. triode
The exact a.c. equivalent circuit of a r.c.c. triode is given in Fig. 12.8 where the
valve V, is replaced by a generator of voltage µE, in series with r,,. The plate load
resistor RL is shunted by Co, which includes the valve output capacitance plus stray
capacitance. The grid resistor of V2 is shunted by C, which includes the input
capacitance of V, (including the Miller Effect capacitance from the plate) plus stray
capacitance. Any additional condenser connected from plate or grid to earth should
be added to Co or C, respectively.
FIG. 12.E
G P C G.
v, rP
t
E.
v,
Et R9 Fig. 12.8. Exact equivalent circuit
of r.c.c. triode.
The cathode bias resistor is here assumed to be adequately by- passed at all signal
frequencies.
This equivalent circuit is the basis of the calculations of gain and phase angle
shift at all frequencies.
(xiii) Voltage gain and phase shift
It has been shown by Luck and others that the voltage gain /frequency characteristic
of a r.c.c. amplifier has the same mathematical form as a tuned circuit except that
the Q is very low (never greater than 0.5). There is a mid -frequency at which the
gain is a maximum (A,,) and the phase shift zero. At both lower and higher fre-
quencies the gain falls off, and when the gain is Ao /V2 the absolute values of the
resistive and reactive components are equal. These two reference frequencies, which
correspond approximately to 3 db attenuation, are the basic points on the attenuation
characteristics, and, with the mid -frequency gain A,,, are sufficient to determine the
whole frequency characteristic. The value of Q is Riven by
Q = f./(f. -f,)
where fo = mid-frequency (at which voltage gain is Ao)
(14)
fb = (high) reference frequency at which voltage gain is 0.707 Ao
and f = (low) reference frequency at which voltage gain is 0.707 A,.
The low reference frequency f, is given by
f, = l /2irCR' (15)
where R' = R, r,RL /(r, -F RL)
and the constants are as in Fig. 12.8.
The high reference frequency (f b) is given by
fb = 1 /2i(Co C,)R" (16)
where R" = r,RIR, /(r,RL r,R, RLR,)
When f, and f have been determined, the attenuation at any other frequency may
b
10
11 AWL.
,,I4111111111101.
yi
Sd
2d
1
1111 ,,a'i/
.
''',f+"
.1
1io;.
g 12
aorfp '°
P
d ,.!íUi\
«14
16
ö
á 7d F,A°ecaiu_
aorfp i_9i
a99.,
18
ed
II ii
20 9ÓI
0 I 2 3 .5 7 1.0 2 3 5 7 10 2 3 5 7 I.0 2 3 5
Fig. 12.9(A). Attenuation in decibels for r.c.c. triode at low and high frequencies
(B) Phase shift at low and high frequencies. Values of phase shift for frequencies beyond
the limits of the curves may be estimated with reasonable accuracy by taking the angle
as proportional or inversely proportional to the frequency.
gain at frequency
cos tan -1
a f - f- (17)
gain at mid -frequency
while the phase angle shift is given by
tan 4' = fa /f (18)
f
where a = low reference frequency (3 db attenuation).
Similarly if the high reference frequency is known, the gain at any frequency f
is given by
gain at frequency
cos tan-1
f - f (19)
gain at mid -frequency
-
f
while the phase angle shift is given by
tan 4' =,f/fb (20)
where f b = high reference frequency (3 db attenuation).
Screen supply A series resistor from the plate supply voltage is generally pre-
:
ferred. Valves in which the screen current has wide tolerances (e.g. some remote
cut-off types and most tetrodes) should be supplied from a voltage divider.
Following grid resistor As for triodes.
:
Va
I Et. +
The effect of incomplete screen by- passing on gain is given by eqn. (1) on the
assumption of complete cathode by- passing (Fig. 12.10), the derivation being given
in Ref. B3 (see also Refs. All, Al2, A13, B2).
12.3 (iii) SCREEN BY -PASS 497
A' /1 + R,2w2C,2
A ,j Bz + R,2w2C,2 (1)
where A' = stage voltage gain at frequency f with screen resistor R, by-passed by
condenser C,
A = stage voltage gain with R, completely by- passed
= mid -frequency voltage gain
w = 2nf
R,g ,.
B + 1 = 1 + R,g m a (la)
mp., (1 + RL /r9) mµt
m = I, /I,, (assumed constant)
g,,, = mutual conductance at operating plate current [see Sect. 3(vii)]
g,,. d = slope of dynamic characteristic at operating plate current
and µt = triode mu (with screen tied to plate).
If the screen voltage is obtained from a voltage divider (RI, R2) then R, should be
a ken as RIR2 /(R, + R2).
-2
-4
6
-o
8Z
0
jr.:
12 6-
14á
16 á
18°
2
200
J
-22
R'/Xs -24
-26
0.1 0.2 04 0-6 0-81 4 6 8 10 20 40 60 80100
FIG.12II(A) 5....10 r . MOVING SCALE
500
FREQUENCY C/S
Fig. 12.11A. Curves for the frequency response of a resistance-coupled pentode
with screen dropping resistor R, and screen by -pass reactance X, (= 1 /wC,).
The value of B is given by eqn. (la). The frequency for R, /X, = 1 is given by
f
= 1/27r R,C,. The frequency scale should be traced and the tracing moved hori-
zontally until f
corresponds to R, /X, = 1 (method after Sturley).
Eqn. (1) is plotted in Fig. 12.11A for selected values of B versus R,/X, which is
equal to 2nfR,C, and therefore proportional to the frequency.
For most purposes it is sufficient to use the approximation, for a loss not exceeding
1 db at a frequency : f
where B
and m =
C,
.
B /irfR,
1 + Rag. /mí'tt
(2)
For example, type 6J7 with RL = 0.25, R, = 1.5 megohms, g = 850 µmhos, ,
f
= 20 and m = 4 has B ti 16.5. For 1 db attenuation at = 50 c /s, C, 0.07 F.
The limiting loss of voltage gain at very low frequencies is 1/B ; in the example
above this is 0.06 (i.e. 24.3 db). If the screen had been supplied from a voltage divider,
the loss of gain at very low frequencies would have been much less.
498 (iii) SCREEN BY -PASS 12.3
An unbypassed screen resistor of g,,,r, ohms gives the same degree of degeneration
as a1 ohm cathode resistor
\ _!/
where r, = dynamic screen resistance (ae, /ai,)
triode plate resistance.
N
For example, when g,,, = 2000 micromhos and r, = 10 000 ohms, an unbypassed
\11111 m/iriim%
"IIÌümosÍl1r%/w%%/ -
\"1111
'N11111=111Ii//I/I/
1/1111
'15I/111,111% I/I1
IMMI111l./1111II/%111
111;MM1111,1II% II
11111\
1
//11I1//.al
IIIII=1i.111%I1111
1111I11N Jo 1111I1NI1
01
FIG.1241(B)
02
1
5
111111=\®1I'
11111 ROx
0.4 0-6043
''''110
1
I
'
'`l'
' 1
50
6
''''1
100
8
VALUES OF B
10 20
I 1 I
500
40 60 80100
1 1 1
1000
1 I MOVING !GALE
FREQUENCY CJs
Fig. 12.11B. Phase angle d'splacement corresponding to Fig. 12.11A.
o
- 4
80°
c
60° gr?
o
- 24 - rr 411.
- 28
screen resistor of 20 ohms will give the same degree of degeneration as a 1 ohm cathode
resistor.
The effect of the screen by -pass capacitance on phase shift is indicated by Fig
12.11B. Under normal conditions with a high resistance dropping resistor, the
angle does not exceed 65 °. This angle may be considerably reduced by reducing
the effective value of R
as for example with a voltage divider. Such action is norm-
ally only necessary with negative feedback.
(iv) Cathode bias
(A) Effect of incomplete by- passing on gain
Provided that the screen is adequately by- passed at all frequencies of operation,
the procedure is as for triodes (Sect. 2) except that the d.c. current flowing through
(Lag) (Lead)
-80° -60° -40° -20° 0 +20° +40° +60° +80°
t:IN.a
°
A /íi:
rtChAti tera.
/IN11i1 ti r°
o 'ó
2
'
*:
"6
4, W, t
AUMF ARP1;140.011
N111.11'
a6
i
9o
6
8 8
o
12
o
mosIu.Iu m1ti1ii%Iem1/
14
16 ---+16
_simmeIl1I.lmmiraimmo iÌ%I +16 16
-+18 +18- 18
+ 201
9 80°
%11:iiiill l SriiiiaNI io°
1 +20-
0. 90.
20
50° 40° 30° 20° 10° 50 60°
0 10° 20° 30° 40°
(Lead) Phase ofl(R4 g,nl n
1
(Logs
22 22
Rk is now Ik = I6 + Ira. The effect of Rk and C,, on the voltage gain at any fre-
quency is given by the approximation (derived from Sect. 2 Eqn. 2)
1 + (wCkRk)Q
(3a)
(1 + gmRk)2 + (wCkRk)Q
on the assumption that the screen is completely by- passed.
When the screen is not completely by- passed, so that both cathode and screen
circuits are attenuating simultaneously, the cathode attenuation characteristic is
affected by the screen circuit. Under these circumstances it is possible to use the
method due to Terman (Refs. Al2, B2) whose curves are reproduced in Figs. 12.11C
and 12.11D.
Terman expresses the relationship in the form
Actual output voltage 1
(3b)
Output voltage with zero bias impedance 1 + R,gmr1ß
-= +1(f/.Î4)
f actual frequency in c/s
f, = 1 /2nC,R, = frequency at which the reactance of C, equals bias re-
sistance R,
A'
and ß =
A
f
contributed by the screen circuit at frequency as given by eqn.
(1) (Note that ß = 1 for complete by- passing).
The procedure in a practical case is
1. Determine ß = IA' /AI from equation (1).
f
2. Knowing and f4, calculate /f,. f
3. Apply this value of f /f, to Fig. 12.11C, thus determining the magnitude of ,l
in db, and also its phase angle.
4. Calculate the value of R,gm70, in db above unity.
5. Apply the value of R,g,,, ß and the phase angle of n to Fig.12.11D to determine
the loss in gain (in decibels).
It will be seen that incomplete screen by-passing results in a smaller value of attenua-
tion by the cathode impedance than would occur with complete screen by- passing.
Attenuation curves for a typical r.c.c. pentode are given in Fig. 12.12 for com-
plete screen by- passing.
As a useful rule of thumb, sufficient for most design purposes other than for
amplifiers incorporating negative feedback over 2 or 3 stages,
fCk 0.55gm for 1 db attenuation (4a)
fCk 0.35 gm for 2 db attenuation (4b)
where Ck is in microfarads,
and gm = mutual conductance in micromhos at the operating plate current.
The effect of incomplete cathode by- passing on phase angle displacement is ap-
proximately the same as for a triode, and the curves of Fig. 12.3C may be used, but
the value of B may be taken as roughly
B 1 + g,,,Rk
on the assumption that is very large and that R' is small compared with
R' = RLR4Z /(RL + Ros).
r where
The combined effect of incomplete cathode and screen by- passing and of a grid
coupling condenser on gain and on phase angle displacement may be determined by
adding the attenuations in decibels and the phase angles in degrees, provided that
eqn. (3b) is used for calculating the attenuation due to the cathode impedance.
.5 4
.y S
I
< 7
2
10
_
..I...
/
II////1 __
FIG. 12.12
c .soµF
..l..
MIr.,,,-.,,
.
ANIZILVAMIlarglir
%!5!J%.i
20
Jm
MIME MEN
50
(iv) CATHODE BIAS
100 200
.
SOo
Frequency (c /i)
.--!d
501
Fig. 12.12. Attenuation characteristics for selected values of cathode by -pass condenser
with valve type 6J7 and Ebb = 250 volts, RL = 0.25, R, = 0.5, R, = 1.5 megohms,
Rk = 2000 ohms : screen adequately by- passed.
effect under normal operation, provided that the screen voltage is adjusted to give the
.orrect operating plate current.
Optimum values may be calculated for any specific case by first finding the fixed
bias and then calculating Rk = Ecl /(Ib + Ic2).
Alternatively, the value of Rk may be determined by eqn. (5f) below.
The procedure for the calculation of the cathode bias resistor when a series screen
resistor is also used, is given in Sect. 3(vi)C, while the graphical method is described
in Sect. 3(xiii) and (xiv).
(C) Effect of reverse grid current with cathode bias
When the screen voltage is maintained constant from a low impedance source,
the following approximate relationship may be derived from Chapter 3 Sect. 1 eqn (6) :
R
R
for cathode bias
for fixed bias
.
ti 1 + Risk 1 + Rkgmlk /Ib
ratio approximately and
(4c)
for
For a typical pentode, type 6J7, the is 1.7, 2.1 2.6
RL = 0.1, 0.22 and 0.47 megohm respectively. Now the value of R for fixed bias
is derived in (v)a below for a reverse grid current of 1µA, so that for type 6J7 the
maximum grid resistor under this condition with cathode bias and fixed screen voltage
is approximately
0.56 megohm for RL = 0.1 megohm,
0.91 megohm for RL = 0.22 megohm,
1.5 megohms for RL = 0.47 megohm.
With a series screen resistor the following approximate relationship may be
derived from Chapter 3 Sect. 1 eqn. (6).
R,1 for cathode bias PgkR`,
R,1 for fixed bias
+ Rkgk + µ,1c2 (4d)
The first two terms are the same as in eqn. (4c), while the third term is the same as
the second term of eqn. (5b) below, so that the ratio in eqn. (4d) will normally exceed
7 : 1, for RL = 0.1 megohm and will be considerably greater than this value for
RL = 0.22 megohm or higher.
However, it is generally advisable to limit the grid resistor to 2.2 megohms maximum
for amplifiers having a reduced frequency range and 1 megohm or less for amplifiers
having a maximum frequency of 10 000 c/s or more.
(y) Fixed bias
The optimum bias is the smallest which can be used without danger of positive
grid current, provided that the screen voltage is adjusted to give the correct operating
plate current.
502 (v) FIXED BIAS 12.3
RL o.se
(
RL \ 1000
100 000 5.75
220 000 7.8
470 000 10.4
From which the following values may be calculated :
Type RL = 0.1 0.22 0.47 megohm
6J7 Rv1 = 0.3 max. 0.4 max. 0.54 max. megohm
6SJ7 R01 = 0.33 max. 0.45 max. 0.6 max. megohm
6AU6 R,, = 0.33 max. 0.44 max. 0.59 max. megohm
We may therefore conclude that, for a maximum reverse grid current of
1 µA, the value of R,1 with fixed bias and fixed screen voltage should not
exceed
0.33 megohm for R L = 0.1 megohm,
0.43 megohm for R L = 0.22 megohm,
0.56 megohm for R t, = 0.47 megohm.
Where the variations of reverse grid currents are such that the large majority of
valves would have values below half the maximum value -i.e. in this case below
0.5 µA- double these values of R,1 would be satisfactory. This would also apply
in any cases where the maximum value of reverse grid current is specified as 0.5 µA.
(b) With series screen resistor
When fixed bias is used in conjunction with a series screen resistor supplied from
the plate voltage source, it may be shown from eqn. (9) of Chapter 3 Sect. 1 that
R,1 for series screen resistor PgkR,,
R for fixed screen voltage µ,1,2
(5b),
where P = I,t /Ik
gk = g,a(I k/I b) at the operating point,
R,2 = resistance of series screen resistor in ohms,
and µ01,a = " triode " amplification factor.
Ratios calculated from eqn. (4c) for typical valves, with a load resistance of 0.1
megohm or more, exceed 6 times, so that when a high- resistance series screen
resistor is used, the grid resistor may be at least 6 times the value quoted
above for fixed screen voltage.
However, it is generally advisable to limit the grid resistor to 2.2 megohms maxi-
mum for amplifiers having a reduced frequency range and 1 megohm or less for am-
plifiers having a maximum frequency of 10 000 c/s or more.
FIG. 12.13
X
t
rz
10 0
to
g "I Ebb
CMYctérietk .F 8
P 2Ebb á
Output!
Voltage
2 .7 3Ebbi
Ebb 6K 4E666
(Peak to Peak) I
bur Z 4
Ì S SEbbJ
a .E.4 . .6Ebbi
W . .3 . .7EySW
Grid Input
.2 ..BEbbñ
I.".I¡f ---- Voltage
(Peak to Peak) 1 .9 Ebb
0..
I
D Ebb
GRID VOLTAGE (Eer) O
Fig. 12.13. Dynamic characteristic of r.c.c. pentode for fixed screen voltage
(6) compared with mutual characteristic (a).
504 (vi) DYNAMIC CHARACTERISTICS OF PENTODES 12.3
10
.9
g
.7
6
K
5
Fig. 12.14A. Family of dynamic characteristics for typical r.c.c. pentode (type 6J7
with Ebb = 250 volts, RL = 0.25 megohm).
(C) Determination of series screen and cathode bias resistors when plate
and screen current curves are available
If curves of both plate and screen currents versus grid voltage under resistance
loaded conditions (e.g. Figs. 12.14A and 12.18) are available, the procedure is-
1. Determine E,1-see (v) above.
2. Determine the optimum value of K-see (vi)B above.
3. From the /b curves, determine Ecz to give the required values of E,1 and K ;
also determine I, at the operating point.
4. From the I tz curves, determine
5. Then Rk = Et1 /(I° -{- 1,2)
I
at the operating point.
(5d)
and R, = (E,,2 -
E,2) /I,2 (5e)
I
The increase in grid bias is
dEr = n/I eo
where I,, = plate current at which the reverse grid current is measured,
(6a)
and /,, may be taken as half the maximum specified reverse grid current (say 1 µA) ;
i.e. (say) In = 0.5 µA.
The increased value of cathode bias resistor is given by
Rk = Rk + AR, (6b)
where JRk = dE dEc, m
It, /et m +1
1,
and E is given by equation (6a).
Using the suggested value of I we have ,
0.5
For example, if R
= 1 megohm,
Rk' = Rk + 200 ohms.
I
m +1
= 2 mA, and m = 4, then
(6c)
If the resistance of the following grid resistor R, is limited, the maximum gain
on low input voltages is obtained when the load resistance is approximately equal
to the following grid resistance, i.e. RL R,.
o 61 02 01 01 OS
Load Resistance R. (MOonmil
Plate resistance
The value of plate resistance r, at the operating point may be estimated from Sect.
3(x) below.
Data for types 6J7 and 6SJ7
Detailed values of mutual conductance (g,,,), slope of the dynamic characteristic
(gm a) and plate resistance (r,) for types 6J7 and 6SJ7 are tabulated below. It is
emphasized that there are considerable variations between valves, particularly with the
plate resistance.
508 (vii) GAIN AT MID -FREQUENCY 12.3
RL K gm gmd ry gm gmd r,
MD µmhos mhos M12 mhos µmhos M52
0.1 0.78 1065 980 0.6 1390 1280 0.7
0.76 1080 1010 0.67 1380 1290 0.9
0.70 1130 1080 1.1 1370 1310 1.3
0.62 1120 1080 1.5 1340 1290 1.4
0.55 1100 1060 2.1 1335 1280 1.7
0.25 0.78 896 834 1.6 1050 990 1.7
0.76 880 836 1.9 1040 985 1.8
0.70 860 830 3.0 1000 960 2.4
0.62 820 790 3.3 940 900 3.3
0.55 790 765 3.7 920 845 3.4
0.5 0.78 670 640 3.0 720 660 3.0
0.76 660 630 3.6 710 660 3.0
0.70 647 620 4.5 650 648 3.7
0.62 595 575 5.2 590 580 5.0
0.55 550 540 6.0 519 495 5.4
curve is typical of the shape of all pentode dynamic characteristics which differ mainly
in the slope and the horizontal displacement of the curve. The effects of changes in
RL are largely overcome by the use of the factor K in place of the actual plate current.
Comparison between triode and pentode dynamic characteristics
Both triode and pentode dynamic characteristics are shown in Fig. 12.15 in such a
way as to enable a comparison to be made between them. Over the region from
K = 0.4 to K = 0.6 they appear to be very similar, but in the region from K = 0.15
to 0.4 the pentode characteristic appears less curved than the triode, while in the
region from K = 0.6 to 0.8 the triode characteristic appears less curved than the pen-
tode.
The point of inflexion is the point at which the curvature changes from one direction to the other,
and is the point of greatest slope.
12.3 (viii) DYNAMIC CHARACTERISTICS OF PENTODES & TRIODES 509
1.0
09-09
08-09
0.7 -0.7
06-06
05 -03
Is
A
04 04
0-3. 0-3
02 .02
0-1 oi
o
-1.6 -1 4 -1.2 -1.0 -0.6 -&4 -ò-? o
Eu PENTODE AND TRIODE + SLOPE FACTOR (0.0813) '
FIG. 12.15
Fig. 12.15. Comparison between pentode and triode dynamic characteristics of the
same valve (6SJ7 with Ebb = 250 volts, RL = 0.25 megohm, E,, = 22.5 volts for
pentode, E,, connected to plate for triode characteristics ; triode characteristics divided
by slope factor 0.0813 and superimposed so that points for I , = 0.5 mA coincide).
510 (ix) MAXIMUM VOLTAGE OUTPUT AND DISTORTION 12.3
When R, = 2RL, the factors above should be multiplied by 0.87. When R, = RL,
the factors above should be multiplied by 0.7. The values are for Ebb = 250 volts,
but hold closely over the range from 200 to 300 volts. Optimum operating con-
ditions are assumed.
Harmonic distortion
Type 6SJ7 with Ebb = 250 volts, RL = 0.25 MS2, R, = 0.5 MQ
K = 0.8 0.76 0.72 0.65 0.6 0.56
Ea(r,,a.,.) volts 8 16 30 52 62 78
Eo(, ,.,.) /Ebb 0.03 0.06 0.12 0.21 0.25 0.31
Hs % 0.3 0.9 2.6 3.5 5.4 4.4
H, % 0.24 0.19 1.25 4.2 5.0 12.0
H4 % 0.14 0.11 0.19 1.9 2.2 2.85
Hb % 0.01 0.02 0.18 0.67 0.28 0.17
*Modulation method-r.m.s. sum. For details and for relation between intermodulation and har-
monic distortion see Chapter 14 Sect. 3.
tThe arithmetical sum of the r.m.s. values of the two component waves.
12.3 (ix) MAXIMUM VOLTAGE OUTPUT AND DISTORTION 511
Comments
1. Type 6J7 (with published g,,, = 1225 µmhos) has less distortion than type
6SJ7 (with published g,,, = 1650 µmhos) under the same conditions.
The distortion for a given output voltage increases when a valve is replaced by
another having higher mutual conductance, although there are also differences be-
tween valve types having approximately the same mutual conductance.
2. Load resistance RL = 0.1 megohm provides lower distortion than higher values
of load resistance.
3. The distortion under any given conditions decreases when the resistance of the
following grid resistor is increased.
Comparison between triode and pentode
[Refer Sect. 2(ix)]
The comparison is based on intermodulation distortion with type 6SJ7 as both
triode and pentode, having RL = 0.25 and R, = 1.0 megohm (Fig. 12.16A). Gener-
ally similar results are obtained with type 6J7 and with other load resistances (Ref.
B8).
1. At the level used in the first a -f stage in a typical receiver (E6 = 10 volts r.m.s.)
the pentode gives only about one eighth of the intermodulation distortion given by a
triode, when both are adjusted for minimum distortion.
2. The two curves in Fig. 12.16A cross, and the intermodulation distortion is there-
fore the same for both triode and pentode, at about 31 r.m.s. volts output.
3. At higher output voltages the pentode gives the greater intermodulation dis-
tortion, the ratio being 2.3 1 at 63 r.m.s. volts output.
:
4. The pentode, to give the minimum value of distortion, requires fairly critical
adjustment. Under the working conditions recommended in this section, however,
the distortion with a pentode is likely to be less than with a general -purpose triode at
output voltages up to about 20 volts r.m.s.
''^
G+
K.
The effect of the screen and cathode by- passing has been covered in (iii) and (iv)
above.
Fig. 12.17. Screen In versus E,2 characteristic for type 6J7 ; Ebb = 250
volts, RL = 0.25 M.Q. A screen loadline is shown.
i/Ali! O'2
Fig. 12.18.
-7 -6 -5
2c-4
(VOLTS)
-3 -2 -1 O
0
Screen I,, versus Eel characteristics for type 6J7 static curves are for
;
100 V 150 V
- E,,)I
0.15 mA
1b = 41,,
0.60 mA
80 170 0.17 0.68
60 190 0.19 0.76
40 210 0.21 0.84
20 230 0.23 0.92
The values of lb are then plotted on the characteristics corresponding to the respective
screen voltages (E,,). The screen loadline (Fig. 12.19) is almost a straight line which
may be extended to cut the horizontal axis at approximately
E,, gte, -E s
/µe (11)
and which may be shown to have a slope of approximately
?PIPS ma /(R,gma + mµe) (12)
where µe = valve triode mu,
gma = slope of plate dynamic characteristic at point of interest,
and m = ratio of plate to screen currents.
Alternatively, the point of intersection of the screen loadline with the horizontal
axis may be determined by the point of cathode current cut-off on the triode "
characteristics (if available) where E = E,,,.
The plate current at E,, = 0 is given approximately by
Ib0 ti E
mgmaE /12,gmd + mµe (13)
The value of screen resistor to provide a plate current lb = KEbb /RL at a fixed
R, ti -m
K
f
L
µ,RLE,i
Ebb
lcµ,
gmd
lf
grid voltage E,, when E,,, = Ebb is given approximately by
6J7
PLATE SUPPLY 250Mí Ecz-
PLATE LOAD O.25Mtl 20 VOLIS
SCREEN SUPPLY 250V .A
SCREEN RESISTOR 1640
-6
2
4<
FIG.12.19
-2
o
-e -b -S -4 -3 -2 o 1
En(VOLTSI
Fig. 12.19. Plate dynamic characteristics for type 6j7 with screen loadline.
Variations in ratio between plate and screen currents
Although considerable variations exist, particularly in the region of high plate
current, there is a comparatively large region within which the ratio is constant within
± 5 %, this being fortunately in the region most generally useful for amplification
(e.g. shaded area in Fig. 12.20).
The variations which occur are mainly caused by the variations in the ratio of
plate to screen voltages. It is dear that they will be serious for El, less than E,5
(above broken line in Fig. 12.20) but the variations become greater as E b approaches
Ent. If it is desired to make accurate calculations involving an assumed constant
value of m (= Ib /ICY) it is desirable to maintain Eb /E,2 nearly constant. The use
*A further approximate method is described in Ref. B12.
12.3 (xiv) COMBINED SCREEN AND CATHODE LOADLINES 515
of a series screen resistor with a resistance equal to mR, (or slightly above this value)
assists in maintaining the constant current ratio.
(xiv) Combined screen and cathode loadlines and the effect of
tolerances
(A) Cathode loadlines
The method normally adopted is an approximate one but very convenient, since
it may be used directly with the plate dynamic characteristics. Provided that I IdI c, =
m = constant, the current through Rk will be (m -F 1) Ib, so that it is necessary to
use a " conversion factor " of m/(m -}- 1) in respect to both the current and slope of
the loadline. For example if m = 4, the loadline for a cathode bias resistor of 2000
ohms would have a slope of -
1/2500 mhos. The effect of any error in the value of
m is minimized through the screen current being only a small fraction of the cathode
current.
io
FIG. 12.20
0.6
.6 w
0.4 j
l
n
.O2 0,
_3 _2 _i o
Cc, (V) Ec3
i (VOLTS
Fig. 12.20. Plate dynamic characteristics for type 6J7 with curves of constant
plate to screen current ratio. The shaded area has m within + 5 %.
Fig. 12.21. Cathode loadline (OF) with single plate dynamic characteristic for
type 617. The broken lines show the limits of ± 10% tolerances in Rk.
The method is illustrated in Fig. 12.21 where only one plate dynamic characteristic
is shown, with a cathode bias loadline OF having a slope of -m
/(m -}- 1)R k. The
static operating point is at P, the intersection of OF with the dynamic characteristic.
Owing to the very large plate load resistors commonly employed, there is very little
rise in plate current due to rectification effects, so that P may be regarded as the
dynamic operating point.
The effect of f10% variation in Rk is illustrated by lines OF" and OF' respec-
tively in Fig. 12.21.
It may readily be shown geometrically that + 10% variation in Rk has less effect
on I6 than ± 10% variation in Ec1.
(B) Cathode and screen loadlines and tolerances
When both cathode and screen resistors are used, the point of operation is the
intersection of the two loadlines (Fig. 12.22). In this diagram it has been assumed
that the screen loadline is a straight line, this being closely correct except at very low
and high values of I b. However, any practicable method for determining the screen
loadline may be used -see (xiii) above.
The effect of ± 10% variation in the resistance of R, is indicated by the broken
lines, and the combined effect with ± 10% variation of both R, and Rk is indicated
by the region shown shaded. It is on account of these inevitable variations, together
with valve variations, that it is inadvisable to operate with too low a nominal screen
voltage and consequently very close to the grid current point.
(C) Tolerances in general
All three resistors RL, R, and Rk have partially self-compensating effects which
enable fairly satisfactory results to be obtained with ordinary 10% tolerances in the
resistors. If operation is required for minimum harmonic distortion at low level or
516 (xiv) COMBINED SCREEN AND CATHODE LOADLINES 12.3
FIG. 12.22
",
i
lO
e o
Cathode Loodline 4 K $
R6.2ß00n ±1ol.
6J7 ' e
Plat, Supply 250V
Plate Load O25Mn
11 lui
Screen Supply 250V
6
,
Screen Loadlin,
R, .ISMn ±l0°ó
pr r
2
ifo" ,
- 14 -12 -IO -6
Ec, Volts
-6 - -2 0
o
Fig. 12.22. Plate dynamic characteristics of type 6J7 with screen and cathode loadlines,
showing the effect of + 10% tolerances in both R, and Rk.
for a high output voltage, the resistors may beneficially have closer tolerances or be
selected so that high RL, high R, and high Rk go together, and similarly with low
values.
The supply voltage may be varied by a ratio up to 2 1 in either direction without
:
very serious effects in normal operation and without necessarily making any change
in R L, R, or R k.
52%
-
1.0 1.2% 3.0% 10.5% 34%
Optimum value of K
0.25 MQ 0.8 0.78 0.73 0.63
(i) Introduction
Transformer -coupled voltage amplifiers usually employ general purpose triode
valves with plate resistances about 6 000 to 10 000 ohms. Valves having higher
plate resistance require excessively large transformer inductances, while valves having
lower plate resistances are only used in special FIG. 12.23
applications, with transformers designed to
handle the higher plate currents.
r
(c) Low operating plate current-reduces the direct current through the trans-
former primary and thus increases the effective inductance compared with a
valve having a higher plate current. If necessary, any valve may be over -
biased so as to bring the plate current to a suitable value -5
mA is quite usual.
(d) High maximum plate voltage-only desirable if a high input voltage is required
by the following stage (e.g. push -pull low mu triodes, or cathode follower).
(viii) Fidelity
Distortion caused by the valve is usually very low in the vicinity of the mid -fre-
quency with unloaded transformers of high impedance-about 1% to 2% second
harmonic with maximum grid swing-but it increases as the transformer is loaded,
up to about 5% second harmonic and 2% or 3% third harmonic. At frequencies
below 1000 c/s the impedance steadily falls, while the loadline opens out into a broad
ellipse at low frequencies. The valve distortion may be considerable at high output
levels under these conditions [see Chapter 2 Sect. 4(vi)].
The transformer (core) distortion is usually quite appreciable and may cause serious
intermodulation distortion [see Chapter 5 Sect. 3(iii)].
siderably less than the supply voltage and this may restrict the maximum output
voltage. If a high output voltage is required, the supply voltage may be increased
up to the maximum rating. It is generally desirable to make R L a resistance of 3 or 4
times the plate resistance of the valve. Higher values of R L result in increased dis-
tortion at low frequencies due to the elliptical load-line. Lower values of RL result
in lower gain and increased distortion at all frequencies.
and the following values are suggested
L = 10 20 30
-
The optimum value of C is dependent upon the transformer primary inductance,
50 100 150 henrys
C = 4.0 2.0 2.0 1.0 0.5 0.5 F.
These values of capacitance are sufficiently high to avoid resonance at an audible
frequency. Use is sometimes made of the resonance between C and the inductance
of the primary to give a certain degree of bass boosting. By this means a transformer
may be enabled to give uniform response down to a lower frequency than would other-
wise be the case. It should be noted that the plate resistance of the valve, in parallel
with R L, forms a series resistance in the resonant circuit. The lower the plate re-
sistance, the more pronounced should be the effect [see Chapter 15 Sect. 2(iii)CJ.
It is frequently so arranged that the resonant frequency is sufficiently low to pro-
duce a peak which is approximately level with the response at middle frequencies,
thereby avoiding any obvious bass boosting while extending the frequency range to a
maximum.
For mathematical analysis see Ref. A13 pp. 38-41.
In all cases when making use of any resonance effects involving the inductance of
the transformer primary, it is important to remember that this is a variable quantity.
520 (xii) PARALLEL FEED 12.4
Not only are there considerable variations from one transformer to another, but there
are large variations of inductance caused by the a.c. input voltage (see Chapter 5).
The series resonant circuit presents a low impedance to the valve at the resonant
frequency, thus tending to cause serious distortion, particularly when the valve is
being operated at a fairly high level. Fot these reasons the resonance method is not
used in good design.
Fig. 12.25. Circuit diagram of parallelfed auto - transformer coupled voltage amplifier.
Fig. 12.26. Choke-capacitance coupled voltage amplifier.
(xiv) Applications
The cost of a transformer having linear response over a wide frequency range is
considerable and, since equally good response may generally be obtained by a very
simple resistance -coupled amplifier, the transformer is only used under circum-
stances where its particular advantages are of value. Some of these are
(1) High output voltage for limited supply voltage,
-
(2) Stepping up from, or down to, low- impedance lines,
(3) When used with split or centre -tapped secondary for the operation of a push-
pull stage, and
(4) When a low d.c. resistance is essential in the grid circuit of the following stage.
(i) Performance
A typical choke -capacitance coupled amplifier is shown in Fig. 12.26. The opera-
tion and design are similar to those of a transformer -coupled amplifier (Fig. 12.23)
with a transformer ratio 1 : 1 except that C must be designed as in a r.c.c. amplifier
to avoid additional low frequency attenuation. An amplifier of this type produces
a higher maximum output voltage than a r.c.c. amplifier but less than that with a step -
up transformer.
(ii) Application
It is occasionally used with valves having rather high plate current, for which a
suitable transformer may not be available. It was also used with tetrode valves of
old design, the inductance being several hundred henries, shunted by a resistance of
about 0.25 megohm.
Normally we begin with a single-sided amplifier, and then at some suitable level
a stage may be inserted having a single input and a push -pull output. In a radio
receiver such a stage usually immediately precedes the output stage, but in more
ambitious amplifiers there may be several intermediate push-pull stages. In this
section we consider the methods of exciting push -pull amplifiers.
FIC. 12.27
INPUT
J
- B +
Fig. 12.27. (A) Triode valve (V1) followed by transformer (T) having centre-tapped
Modified form with
secondary and exciting the grids of the push -pull stage (V2, V,). (B) centre -tap.
non-centre- tapped transformer, using two resistors to provide an equivalent
522 (i) METHODS INVOLVING IRON-CORED INDUCTORS 12.6
INPUT
-s+
Fig. 12.28. Triode valve (V1) with parallel feed and centre - tapped choke exciting the
grids of the push-pull stage.
Fig. 12.29A. Modified form of phase splitter giving full gain without degeneration.
The input is from grid to cathode and cannot be earthed. The cathode resistor is by-passed
to provide full gain.
An analysis of the balance with equal load resistors and shunt capacitors is given
in Ref. C27, indicating that under these conditions the balance is perfect at all fre-
quencies. However, in practice, the total shunt capacitances across the two channels
differ slightly and there is a slight (and generally negligible) unbalance at high fre-
quencies ; see also page 330.
If a high gain amplifier is placed between the phase splitter and the output stage,
hum may be troublesome. Part of the hum is due to the difference of potential
between the heater and cathode. This may be reduced by operating the heater of
the phase splitter from a separate transformer winding which may be connected to a
suitable point in the circuit at a potential approximating that of the cathode.
It is normal practice to assume a maximum r.m.s. output voltage (grid-to-grid)
In an experimental test it is important to reverse the phase of the input from the B.F.O. when the
valve voltmeter is moved from one output to the other.
12.6 (ii) PHASE SPLITTER 523
of 0.18E b for less than 2% total harmonic distortion, equivalent to 0.25 E b peak -to-
b
b
peak output. The distortion drops rapidly as the output voltage is decreased. With
a plate supply voltage of 400 volts to the phase splitter, the output is sufficient to
excite push-pull Class A 2A3 valves operating with 250 volts on their plates and -45
volts bias.
A complete circuit with 3 stages incorporating a phase splitter and negative feed-
back is shown in Fig. 7.42.
(B) A modified form of phase splitter (Fig. 12.29A) gives the full gain without
degeneration, but the input is floating and cannot be earthed. For this reason it
cannot generally be used with a pickup, although it may be applied to a radio re-
ceiver. In the latter application, the valve may be a duo -diode triode performing
detection and 1st a -f stage amplification ; a.v.c. may be operated with some com-
plication. This circuit is particularly prone to suffer from hum, owing to the high
impedance from cathode to earth and the high gain. The hum may be minimized
by adjusting the potential on the heater to approximately that of the cathode.
References (C) 1, 3, 7, 11, 12.
(C) It is possible to apply positive feedback from a tapping on the cathode
resistor, through a coupling resistor to the unbypassed cathode of the preceding r.c.
pentode, and thus increase the gain. A direct- coupled version is shown in Fig. 7.51A.
See Chapter 7 Sect. 2(xi).
(D) Another modification (Fig. 12.29B) also gives high gain not from the phase
splitter itself but from the preceding stage. It makes use of the high input resistance
of the phase splitter as the dynamic load on V1, thereby increasing its gain. The full
analysis of its operation is given in Reference C18 and the following is a summary
The effective cathode load on V2 (apart from the cathode resistor R 8) is R2 and R6
in parallel, i.e. 20 000 ohms, which is the same as R4. The cathode resistor R8
is by- passed in order to increase the input resistance.
FIG. 12.29 B + B FIG.12.29C
R.
125MII
V,
65.1
-B
Fig. 12.29B. Phase splitter (V2) using high input impedance to increase the gain of the
preceding stage (V1) by about six times.
Fig. 12.29C. Phase splitter using pentode with unbypassed screen and suppressor grid.
The d.c. load resistance in the plate circuit of V, is R. + R2 = 0.29 megohm, while
the dynamic load is the Input resistance of 172, i.e. [see Chapter 7 Sect. 2(ii)B]
r,' = R, /(1 - A')
where R, = R1R, /(R1 + R,) = 167 000 ohms
and A' 0.9,
so that in this example, r/ = 1.67 megohms.
The gain of V, is given by
g 950
A1
+ (1 /r;') 1/4 + 1/1.67
which is about 6 times the gain under normal conditions.
The circuit is nearly balanced if R. = R , and
E01 µ2R, R,-
E02 µ2R, + r,, + R.
where R, = R2R, /(R2 + R,).
524 (ii) PHASE SPLITTER 12.6
The out -of- balance, being about 1.2% in this example, is negligible.
(E) A further type makes use of a heptode (mixer) valve in which unequal
load resistors are placed in both plate and screen circuits, the push -pull output being
taken from plate and screen. The input is taken to grid No. 3.
Ref. C20.
(F) A further modification makes use of a pentode with an unbypassed
screen. The two output voltages are taken from plate and screen, and the sup-
pressor grid is maintained at a negative potential with respect to the cathode (Fig.
12.29C and Ref. C23).
(iii) Phase inverter (Fig. 12.30A)
This is a popular arrangement with twin triode valves, either general purpose or
high -mu. It is not self -balancing, and requires individual adjustment for accurate
balance both during manufacture and after the valve has been replaced. It is slightly
out of balance at very low frequencies owing to the two coupling condensers operating
in the lower channel, but C, may be made larger than C1 if desired. It gives a gain
(to each channel) equal to the normal gain of one valve.
If it is preferred to avoid individual balancing, the value of R2 is given by
R, = (R, + R,) /A where A is the voltage gain of valve V2. If R, and R, both have
+ 10% tolerances, the maximum possible out of balance will be nearly 20% due
to the resistors alone, plus valve voltage gain tolerances.
Separate cathode resistors, each by- passed, are helpful in reducing valve gain
tolerances, but require independent cathodes. If a common cathode resistor is used,
it may be unbypassed, thus introducing negative feedback for out-of- balance voltages.
The hum level is quite low.
N.B. This circuit was originally named Paraphase, but the latter name covers a
large number of different circuit arrangements and cannot therefore be used to dis-
tinguish one from another.
References Cl, C3, C7, C11, C14.
FIG. 12.30 A
C
FIG.12.30
Fig. 12.30A. Conventional form of phase inverter in which V, excites V,, and V,
excites V4, the grid of V, being connected to a tapping on the grid resistor of V,.
Fig. 12.30B. Phase inverter with pentodes using the suppressor grids for self-balancing
(Ref. C19).
(iv) Self-balancing phase inverter (Fig. 12.30B)
In this circuit V, and V, are pentodes, and any unbalanced voltage appears across
the common plate resistance R, and is fed to both suppressors through the blocking
condenser C thus causing degeneration in the valve producing the larger signal
output, and regeneration in the other. Ref. C19.
(v) Self-balancing paraphase inverter
(A) Floating paraphase (Fig. 12.31)
This circuit is, to a considerable extent, self-balancing thereby avoiding any neces-
sity for individual adjustment except in cases where a very high accuracy in balancing
is required.
12.6 (v) SELF -BALANCING PARAPHASE INVERTER 525
In order to visualize the operation of this circuit consider firstly the situation with
V, removed. Resistors R, and R9 in series form the load on valve V and the voltage
at the point X will be in proportion to the voltage at the grid of V3. When V2 is
replaced, the voltage initially at point X will cause an amplified opposing voltage to be
applied to resistors R, and R9. If resistor R, is slightly greater than R5, it will be
found that the point X is nearly at earth potential. If the amplification of V, is high,
then R, may be made equal to R5 and point X will still be nearly at earth potential.
The point X is therefore floating, and the circuit a true Paraphase ; the derivation
of the name " Floating Paraphase " is obvious.
FIG.12.32
FIG.12.31
V,
INPUT
Fig. 12.31. The Floating Paraphase self-balancing phase inverter with cathode bias.
Fig. 12.32. The Floating Paraphase circuit with fixed bias in the following stage.
The degree of
E,
Ez R5
Rv
+ A,-
1 \ + R5
1
is given by
Ro
+ R,
R5/
R9
(1)
where A, = voltage gain of V, into plate load resistor R. and following grid resistor R,.
If R, = R, = R9, then EI /E, = 1 + 3 /A, (2)
If V, is type 6J5 (or half type 6SN7 -GT) R, = 0.1, 125 = R, = R9 = 0.25 megohm,
then A2 = 14 and E, /E, = 1.21 which is too high to be acceptable. In such a case
R, may be increased to, say, 0.3 megohm giving E, /E, = 1.03.
If V, is type 6SQ7 with R, = R, = R, = R9 = 0.25 megohm, then A, = 48 and
E, /E, = 1.06 which is generally acceptable.
If V, is a pentode (e.g. type 6J7) with R, = R. = R, = R9 = 0.25 megohm, then
A, = 104 and E, /E2 = 1.03, which is very close.
The gain from the grid of V, to the grid of V3 is only slightly less than the gain
with R, = 0.
F16.12.33 If fixed or partially -fixed bias is employed, it is
necessary to couple the grid of V2 to point X through
a suitable condenser (C in Fig. 12.32). In addition,
a hum filter (R10, C,9) may be required, because
INPUT
most partially -fixed bias sources contain appre-
ciable hum voltage ; any hum voltage appearing
across the grid resistor of V, is amplified by V, and
V,. References C12, C14, C16.
va
î-- (B) Common plate impedance (Fig. 12.33)
This follows the same principle as the Floating
Paraphase, except that the common impedance is
Fig. 12.33. Common plate
impedance self -balancing in the d.c. plate circuit instead of in the shunt a.c.
phase inverter. (following grid) circuit. Here similarly
526 (v) SELF -BALANCING PARAPHASE INVERTER 12.8
FIG. 12.34 R, R4
EZ R4 + A2 \1
+ R 6
+ R>
8
(3)
while if R3 = R4 = R6, then El /E2 = 1 + 3/A2 (4)
where A, = voltage gain of 173 into plate load re-
sistance R4. Ref. C12.
(C) See -saw self-balancing phase inverter
(Fig. 12.34)
This is merely another form of the common plate
impedance circuit, with two separate resistors.
When R3 = R4 = R6,
then El /E2 = 1 + 3/(A2 + 3) (5)
For perfect balance,
e-
R4 = R3 [1 + 6/(2A2 3)]. - (6)
Fig. 12.34. " See -saw " It is desirable for A, to be greater than 40, thereby
self-balancing phase inverter giving an out -of- balance less than 7 %. References
(Ref. C9). C9, C10, C17.
III
6C8-G
5,000 OOIyF
9 B
Fig. 12.35. Modified " see-saw " self- balancing phase inverter (Ref. C10).
Fig. 12.36. Practical " Schmitt " common cathode impedance self-balancing inverter
using type 6C8-G twin triode (µ = 36). C, and C2 should have high insulation resistance.
(i) Introduction
A push-pull voltage amplifier stage is one having push -pull (3 terminal) input and
push -pull output. Two separate valves (or one twin valve) are required. The two
valves are each treated as for a single -ended amplifier, whether resistance- or trans-
former- coupled.
centre- tapped primary, or the so-called " phase compressor " -see (v) below. The
same also holds true when feeding into a single -ended stage.
Many push -pull transformers are designed to operate with an out -of- balance plate
current not exceeding about 1 or 2 mA, thus necessitating matched valves, or adjust-
ed bias, or parallel feed.
and cathode of V2. The voltage drop across Ry1 will normally be more than half
the voltage between A and C.
A circuit requiring only a single source of voltage is Fig. 12.42 ; the two dividers
are desirable to avoid interaction (degeneration) between the two stages unless the
bleed current is very high. The output terminal is returned to a point on the voltage
divider having the same potential.
Another circuit requiring only a single source of voltage is Fig. 12.43 ; this has only
a single voltage divider. The voltage drop across R 1 must equal the plate voltage
of V1 minus the grid bias of V2. There will be degeneration caused by the unby-
passed cathode resistors, which may be avoided by a push -pull arrangement with
common cathode resistors for both stages.
FIG. 12.42 FIG. 12.43
VI
INPUT
III 111
Fig. 12.42. Direct- coupled amplifier with only one source of voltage (Ebb).
Fig. 12.43. Direct -coupled amplifier with one source of voltage and one voltage divider.
Pentodes may be used, if desired, in all these circuits by making suitable provision
for the screen voltages. In Fig. 12.43 the screen of V1 may be taken to a tap on R12,
provided that either the screen is by- passed to cathode, or R12 is by- passed ; the
amplifier would then be limited to audio frequencies only (Ref. D40).
When the first amplifier stage is a pentode, its load resistor may be increased to
values much greater than those conventionally used, provided that sufficient negative
voltage feedback is applied to secure an acceptable high frequency response (e.g.
Ref. D40).
A pentode may also be used with another pentode as its plate resistor. By this
means a gain of several thousand times may be obtained, but this is only useful in
electronic measuring instruments.
Such circuits (Figs. 12.40 -12.43) are generally limited to two stages. If any in-
crease is made, there is distinct danger of slow drift occurring in the direct plate
current, due to variations in battery voltages and valve characteristics. These may
be minimized by voltage regulators and controlled heater voltage or current, or may
be avoided by one of the special methods described below (viii to x). See also Refs.
D36, D39.
Negative feedback may be applied to any d -c amplifier in the normal manner.
(ii) Bridge circuit
The bridge circuit (Fig. 12.44) may be used with any number of stages in cascade
from a single B supply. The basic design equations are
-
:
OUTPUT
INPUT
e
Fig. 12.44. " Bridge " circuit, Fig. 12.45. Three stage d -c ampli-
direct -coupled amplifier. fier with V2 and V, as cathode -
coupled twin triodes.
(iii) Cathode- coupled
The fundamental form of a cathode -coupled amplifier is covered in Chapter 7
Sect. 2(viii)B.
Fig. 12.45 shows a conventional d -c single triode (V1) followed by two twin triodes
as d -c cathode- coupled amplifier stages. For example, V2 has one triode as a cathode
follower with its grid at a fixed voltage from the voltage divider R1R2, while the other
triode operates as an amplifier but sharing the common cathode resistor Rk1.
Fig. 12.46 is an alternative arrangement in which the positions of the two triodes
are reversed. This input circuit was primarily for use with a phototube (Ref. D19)
but could be adapted to any other requirement.
In the form shown, the output circuit has a direct potential difference, being in-
tended for coupling directly to the screen of the 6V6 following.
See also Section 6(vi),
References D1, D19 ; also Chapter 7 Refs. (G).
(iv) Cathode follower
The cathode follower may also be used as a d -c amplifier. One circuit is Fig. 12.47A
in which the total cathode load is Rk + R,R, /(RB + R7). The value of R, is equal
to (Ebb // b) -
R6 so as to eliminate the undesired direct voltage across R G.
References D15, D36.
FIG. 12.47 e
FIG. 12.47 A
+45í4250V
Fig. 12.47A. Cathode follower used as d.c. amplifier (Ref. D15).
Fig. 12.47B. Direct- coupled amplifier with r -c pentode exciting phase inverter (Ref. D14).
532 (v) PHASE INVERTER 12.9
65.17 R11 Ru
INPUT + 1 -`
Saxon OUTPUT
- Ecc
+45V +250V 111111
Fig. 12.48. Two stage amplifier with the plate of the first stage directly- coupled to the
screen of the second stage (Ref. D14).
Fig. 12.49. Two stage amplifier with gas tube coupling from the plate of the first stage
to the grid of the second stage.
FIG. 12.50
OIM
6Q7
Fig. 12.50. Improved two stage am-
plifier with gas tube coupling, in
which o cathode follower is introduced
for better performance.
1 r, + RL 1 (4)
gmRL + RL µ(µ + i)
Eqn. (4) may be compared with the ordinary form for expressing amplification,
namely
1
A _
1 1
gmRL + µ
and it will be seen therefore that from equation (4)
µ(µ + 1)RL
(5)
r,+RL
gm =
and therefore
gm
r,' = - (µ + 1)
r ,R L
r, +
(6)
(7)
gm RL
It is preferable to make RL very much greater than in order to avoid distortion inr
V, due to the low load resistance into which it works ; a value of 2µr is satisfactory
For example, consider a twin triode with µ = 20 and = 10 000 ohms under r
resistance- coupled conditions. A suitable value for the load resistance is 2µ x 10 000
= 400 000 ohms -say 0.5 megohm.
510 000 = 412 -
From eqn. (6)
From eqn. (7)
:
= µ.
: gr'
From eqn. (5) µ' = 20 x 21 x 500 000
/r
= 20/10 000 = 2000 µmhos.
= 21 x 10 000 x 500 000 = 510 000 = 206 000 ohms.
:
A high -mu triode would show even higher values of amplification factor and plate
resistance, resembling those of a sharp cut -off r -f pentode.
Curves have been drawn for some typical twin triodes
operating as cascode amplifiers ; they resemble the
curves of pentodes except that the rounded knee has FIG. 12.515
oUrpUT
been replaced by a nearly straight, sloping line (Ref. V,
D35). _
One special application is as a low -noise r -f amplifier
(Refs. D33, D35). °b
Another application is as a voltage stabilizer (Refs. -.
D32, D34). INPUT
The cascode amplifier has been used as a direct cur-
rent amplifier responding to zero frequency (Ref. D35).
Two cascode amplifiers have been used in a " float-
ing paraphase " push-pull amplifier operating with
single -ended input, to deliver a balanced output. In
this application, two high -mu twin -triode type 6SL7 -GT
valves were used each as a cascode amplifier, to deliver Fig. 12.51B. Funda-
an output of abut 30 volts peak, each side. Only one mental circuit of cascode
coupling capacitor was used in the whole stage, thus amplifier.
simplifying the design of the feedback circuit (Ref D41).
References to cascode amplifiers D32, D33, D34, D35, D36, D41.
:
and third stages in phase, and the total signal current through R, is ID, + ID3 -I.
A three stage r.c.c. amplifier (Fig. 12.52) has the signal plate currents of the first
Since (ID, + 143) is greater than ID3, the resultant signal current through R, will be
in the direction of I,p3, thus causing across R, a signal frequency voltage drop E,.
As a result, the signal voltage applied to the grid of 173 through RL, and C, will be in
phase with the normal signal on the grid of 172, thus giving positive feedback. If
the gain of the amplifier is high, there may be sufficient positive feedback to cause
oscillation. This effect may be prevented by the use of decoupling.
On the subject of stability, see Chapter 7 Sect. 3 (stability with feedback), also
Ref. El (requires a high standard of mathematics).
FIG. 12.53
between the path through C and the path through the B supply in the ratio
I, Rd +R4 (1)
la Xd
where R4 = resistance of '13 supply
and X, = 1 /c,C = 1 /(271C).
If X, is very much less than Rd + R4 then almost the whole of the signal current
will pass through C, thus reducing the coupling through the B supply. If the value
of R4 is unknown, it may be neglected as an approximation, since Rd is usually much
greater than R4.
The decoupling circuit comprising C and R d is actually a resistance -capacitance
filter [for theory see Chapter 4 Sect. 8(ii)]. The frequency at which the current
divides equally between the two paths is given by
fi 1 /2nRdC (2) L- 1
and the value of K (for low level operation) may be taken as approximately
K = 0.75 for general purpose triodes and pentodes
K = 0.6 for high -mu triodes
[for more exact values see Sects. 2 (vi) for triodes and 3 (vi) for pentodes].
A good general value of Rd for most purposes is one fifth of RL. If the stage is
operating at low level, Rd may be increased up to about the same resistance as RL
to give better decoupling. If the stage is operating at high level, Rd should be reduced
as much as possible provided that sufficient decoupling can be maintained. This
may be assisted by increasing C, but the cost and size of paper condensers may set
a limitation. Two stages of decoupling can sometimes be used to advantage. Elec-
trolytic condensers should be used with caution, since their leakage currents are
appreciable and they tend to cause noise if used in low level stages. They should
never be used with values of Rd above 50 000 ohms, and very much lower values
are desirable.
The plate decoupling circuit of Fig. 12.53 has the effect of increasing the bass
response, since at extremely low frequencies the total plate load becomes very nearly
(RL + Rd). If V1 is a triode, and RL is greater than 5 r9, then the effect is slight.
If V1 is a pentode, or a triode with RL less than 5 r9, the effect may be appreciable.
The increase of gain and the phase angle are both identical in form (except for the
sign) with the loss of gain and phase angle caused by the cathode bias resistor and by-
pass condenser -see Section 2 (iii) for triodes and 3 (iv) for pentodes ; also Ref. E3
for curves.
The limiting increase in gain is given by
= 1 +
(RL' /Rd)(1 /rn + 1 /R,) + RL /r, + + 1 /Rd)
or if rD ) RL as for pentodes,
(1 1 /R9 (4)
1 + R + R d+ R,) (5)
12.10 (iii) PLATE CIRCUIT DECOUPLING 537
Fig. 12.54. Special circuit giving plate decoupling without decoupling condensers
(Ref. E4).
FIG. 12.56
(B) Hum from the screens may also be reduced by screen decoupling. The
hum from the screen is out of phase with the hum from the plate, but the screen hum
predominates. For amplifiers using negative feedback, see Chapter 7 Sect. 2(ix).
(C) Hum from the grid bias supply may be reduced in some cases by grid
decoupling. This hum voltage may be either in phase, or out of phase, with the
plate circuit hum, depending on the source.
(D) Hum neutralization
In a r.c.c. pentode the hum voltages from the plate and screen are out of phase,
so that it is possible to neutralize the hum by a circuit such as Fig. 12.57. Here the
screen is fed from a series resistance Rv2, and the hum voltage is applied to the screen
by the capacitance divider C1C2. If V1 is type 6J7 with RL = 0.25 and R,2 = 1.5
megohms, minimum hum is obtained when C1 = 0.05 and C2 = 0.5 µF, the hum
then being about 14% of that without neutralization. C1 and C2 may have ± 10%
tolerances in quantity production. For perfect neutralization it is necessary to
balance both resistive and capacitive elements.
An alternative form (Ref. E2) is shown in Fig. 12.58 in which, for perfect neutraliza-
tion at all frequencies,
R, R1( Rt11
Ck
C1 R,
1 +
r,,,
j and also C,
C1
-`1
= R, + rq1 f (12)
where rD1 = plate resistance of V1 (preceding stage) and the other values are marked
on the diagram. This circuit gives neutralization both for hum and for low frequency
regeneration.
There are other methods which may also be used for hum neutralization or re-
duction (see Chapter 31 Sect. 5).
FIG. i 2.57
For example, if X is required to be 20 (i.e. the amplitude of the square top falls by
5% at the end of the pulse) and if T = 0.01 second, then RC ti 0.2 second. If
R = 0.5 megohm, then C = 0.4 µF.
If the pulses are repeated periodically with the length of pulse equal to the time
between pulses, and if the applied voltage is zero during the period between pulses
(as Fig. 12.59),
then RC X /2f1 (5)
where fl = frequency in cycles per second.
With sine -wave input, the frequency (fo) at which the response of a single-stage
r.c.c. amplifier falls by 3 db is given by
fo = 1 /(2isRC) (6)
But with " square- wave " input, RC .. X /2f1 (7)
where fl = frequency of square wave in cycles per second.
Therefore fo = f1 /orX (8)
By using a square wave input, and noting on a C.R.O. the frequency fl at which
there is (say) 10% drop at the end of the pulse (i.e. X = 10), it is possible to calculate
the frequency for 3 db attenuation with sine -wave input :
f, = Xnfo (10)
where X = E/(E - e).
Equations (9) and (10) may also be used in connection with multi -stage amplifiers.
SECTION 13 : REFERENCES
(A) REFERENCES TO RESISTANCE -CAPACITANCE- COUPLED
(including general articles) TRIODES
Al. Mitchell, C. J. " Miller effect simplified " Electronic Eng. 17.196 (June
A2. Sowerby, J. McG. " Radio Data Charts -18 : Transmission 1944) 19.
W.W. 51.3 (Mar. 1945) 84. and phase shift of rc couplings "
A3. Sturley, K. R. " The frequency response of r.c. coupled amplifiers
17.209 (July 1945) 593. -Data Sheet " Electronic Eng.
A4. Design Data 1 " Cathode Bias -effect on frequency response
" W.W.
A5. Luck, D. G. C. " A simplified general method for resistance-capacity 52.1 (Jan. 1946) 21.
Proc. I.R.E. 20.8 (Aug. 1932) 1401. coupled amplifier design "
A6. Cowles, L. G. " The resistance-coupled amplifier " Supplement
Trans. A.I.E.E. (June 1945) 359.
Screen
A7. Seletzky, A. C. " Amplification loci of resistance coupled amplifiers
1936) 1364 discussion 56 (July 1937) 877. " Trans. A.I.E.E. 55 (Dec.
A8. Sowerby, J. McG. " Radio Data Charts (16) Voltage gain of resistance
50.7 (July 1944) 209. coupled amplifiers " W.W.
A9. Thurston, J. N. " Determination of the quiescent operating point of
bias " Proc. I.R.E. 33.2 (Feb. 1945) 135. amplifiers employing cathode
A10. Data Sheets 45 and 46 " Performance of resistance-capacity coupled
15.181 (March 1943) 421. amplifiers " Electronic Eng.
All. Sturley, K. R. " Low frequency amplification " Electronic Eng. (1) 17.201 (Nov.
(2) 17.202 (Dec. 1944) 290 (3) Frequency response for low and high 1944) 236
335 (4) Cathode bias 17.204 (Feb. 1945) 378 ; (5) Anode decoupling frequency 17.203 (Jan. 1945 )
(6) 17.205 (Mar.
decoupling 17.206 (Apr. 1945) 470 ; (7) Increasing l.f. response 17.207 (April1945) 429;
Al2. Terman, F. E. (book) " Radio Engineers Handbook." 1945) 510.
A13. Sturley, K. R. (book) " Radio Receiver Design " Part 2.
A14. Roorda, J. " Improved analysis of the r -c amplifier " Radio, 30.10 (Oct.
A15. Pullen, K. A. " Using G curves in tube circuit design " Tele-Tech 8.7 1946) 15.
A16. Diamond, J. M. " Maximum output from a resistance -coupled triode (July 1949) 35.
I.R.E. 39.4 (April 1951) 433. voltage amplifier " Proc.
Additional references will be found in the Supplement commencing on
page 1475.
(B) REFERENCES TO RESISTANCE-CAPACITANCE- COUPLED
Bl. Staff of Amalgamated Wireless Valve Company Ltd. " Resistance -coupled pentodes PENTODES
(Sept. 24, 1937) 308. " W.W. 41.13
B2. Terman, F. E., W. R. Hewlett, C. W. Palmer and Wen -Yuan Pan, "
Calculation
resistance coupled amplifiers using pentode tubes " Trans. A.I.E.E. 59 (1940) and design of
B3. Baker, W. G., and D. H. Connolly " Note on the effect of the 879.
ponse of a single stage " A.W.A. Tec. Rev. 4.2 (Oct. 1939) 85. screen by -pass capacity on the res-
B4. Hammond, C. R., E. Kohler, and W. J. Lattin " 28 volt operation of
(Aug. 1944) 116. receiving tubes " Elect. 17.8
B5. Haefner, S. J.'" Dynamic characteristics of pentodes " Comm. 26.7 (July 1946) 14.
B6. Adler, R. " Reentrant pentode a -f amplifier " (using CK511X valve)
B7. Terlecki, R., and J. W. Whitehead " 28 volts H.T. and L.T. ? " Electronic Elect. 19.6 (June 1946) 123.
157. Eng. 19.231 (May 1947)
B8. Langford-Smith, F. " The choice of operating conditions for resistance-
pentodes " Radiotronics No. 132 (July /Aug. 1948) 63. capacitance -coupled
B9. Edwards, G. W., and E. C. Cherry " Amplifier characteristics at low frequencies,
reference to a new method of frequency compensation of single stages " Jour. with particular
1940) 178. I.E.E.'87.524 (Aug.
B10. Crawford, K. D. E. " H.F. pentodes in electrometer circuits " Electronic
227. Eng. 20.245 (July 1948)
B11. " Cathode by- passing -derivation of mathematical formula " Radiotronics
B12. Shimmins, A. J. " The determination of quiescent voltages in pentode No. 113 (June 1941) 39.
Eng. 22.271 (Sept. 1950) 386. amplifiers." Electronic
See also A6, All (Parts 1, 2, 5, 6) Al2, A15.
(C) REFERENCES TO PHASE INVERTERS ETC.
Cl. McProud, C. G. and R. T. Wildermuth, " Phase inverter circuits" Elect. 13.10 (Oct.
C3. Paro, H. W. " Phase Inversion " Radio Eng. 16.10 (Oct. 1936) 13. 1940) 50.
C7. Cocking, W. T. " Phase splitting in push -pull amplifiers " W.W. 44.15 (April 13th,
C8. Parnum, D. H. " The phase compressor " -a resistance -capacity output circuit 1939) 340.
to the phase splitter -W.W. 51.1 (Jan. 1945) 19. Correction 51.2 (Feb. 1945) complementary
C9. Scroggie. M.G. "The See -saw circuit : a self- balancing phase splitter" W.W. 38.
194. 51.7 (July 1945)
C10. Carpenter, R. E. H. " See-saw or paraphase " -origin of the circuit -W.W.
263, with reply from M. G. Scroggie. 51.8 (Sept. 1945)
C11. Saunders, L. A. (letter) " Phase splitter " Electronic Eng. 18.216 (Feb. 1946) 63. Reference
Chart " Methods of driving push -pull amplifiers " 17.214 (Dec. 1945) 816.
C12. Wheeler, M. S. " An analysis of three self-balancing phase inverters " Proc. I.R.E. 34.2 (Feb.
1946) 671'.
C13. E.M.I. Laboratories " Balanced output amplifiers of highly stable and accurate balance
tronic Eng. 18.220 (June 1946) 189. " Elec-
C14. R.C.A. " Application Note on a self -balancing phase-inverter circuit " No. 97 (Sept.
C16. 28,
Resistance- coupled push pull- Floating paraphase circuit " Radiotronics No. 83 (January 1938).
97. 1938)
C17. E.M.I. Laboratories " Balanced amplifier circuits " Electronic Eng. 17.209 (July 1945)
C18. Jeffery, E. " Push -pull phase splitter-a 610.
new high gain circuit " -W.W. 53.8 (Aug. 1947) 274.
C19. Crawley, J. B. " Self-balancing phase inverter " Elect. 20.3 (March 1947)
2,383,846 Aug. 28, 1945. 212. U.S. Patent
C20. Beard, E. G. " A new high gain phase splitting circuit " Philips Tec. Com. No. 8 (Sept.
C21. Cocking, W. T. "Push -pull input circuits," W.W. (1) General principles W.W. 54.1 (Jan. 1947) 10.
7 ; (2) Cathode follower phase splitter 54.2 (Feb. 1948) 62 ; (3) Phase reversers, 1948)
85 ; (4) The anode follower, 54.4 (April, 1948) 126 ; (5) Cathode- coupled stage, 54.3 (Mar. 1948)
183. 54.5 (May 1948)
12.13 REFERENCES 543
C22. Johnson, E. " Directly-coupled phase inverter," Elect. 21.3 (March 1948) 188.
C23. Sulzer, P. G. " Applications of screen -grid supply impedance in pentodes." Comm. 28.8 (Aug.
1948) 10.
C24. Sowerby, J. McG. " The see -saw circuit again " W.W. 54.12 (Dec. 1948) 447.
C25. Van Scoyoc, J. N. " A cross-coupled input and phase inverter circuit " Radio News 40.5 (Nov.
1948) 6.
C26. E. E. Carpentier, U.S. Patent 2,510,683 described by R. H. Dorf " Audio Patents-Phase in-
verter improvement " Audio Eng. 35.2 (Feb. 1951) 2.
C27. Jones, G. E. " An analysis of the split-load phase inverter " Audio Eng. 35.12 (Dec. 1951) 16.
(D) REFERENCES TO DIRECT COUPLED AMPLIFIERS
Dl. Miller, S. E. " Sensitive d.c. amplifier with a.c. operation " Elect. 14.11 (Nov. 1941) 27. " Stable
dc amplification " W.W. 48.5 (May 1942) 111.
D2. " A novel dc amplifier " Electronic Eng. 14.170 (April 1942) 727.
D3. Hay, G. A. " High sensitivity dc amplifier- another application of the C.R. tuning indicator" W.W.
49.1 (Jan. 1943) 9.
D5. Mezger, G. R. " A stable direct -coupled amplifier " Elect. 17.7 (July 1944) 106.
D6. Lawson, D. I. " An analysis of a d.c. galvanometer amplifier " Electronic Eng. 17.198 (Aug. 1944)
114.
D7. Goldberg, H. " Bioelectric-research apparatus " Proc. I.R.E. 32.6 (June 1944) 330.
D8. Artzt, M. " Survey of D -C Amplifiers " Elect. 25.8 (Aug. 1945) 112, with extensive bibliography.
D9. Williams, J. A. " Crystal -driven modulator for d -c amplifiers " Elect. 18.12 (Dec. 1945) 128.
DIO. Middleton, R. G. " An analysis of cascode coupling " Radio 30.6 (June 1946) 19.
D11. Lampitt, R. A. " A d.c. amplifier using a modulated carrier system " Electronic Eng.18.225
(Nov. 1946) 347.
D12. Noltingk, B. E. " D.C. amplifiers " (letter with reply from R. A. Lampitt) Electronic Eng. 18.226
(Dec. 1946) 389.
D13. Iannone, F., and H. Baller " Gas tube coupling for d -c amplifiers " Elect. 19.10 (Oct. 1946) 106.
D14. Yu, Y. P. " Cathode follower coupling in d-c amplifiers " Elect. 19.8 (Aug. 1946) 99.
D15. Ginzton, E. L. " D.C. amplifier design technique " Elect. 17.3 (March 1944) 98.
D16. R.C.A. " Application Note on special applications of the type 79 tube" No. 28 (Nov. 9th, 1933)
-Fig. 5, d -c amplifier.
D18. Mezger, G. R. " Feedback amplifier for C -R oscilloscopes " Elect. 17.4 (April 1944) 126.
D19. Scully, J. F. " A phototube amplifier " Elect. 18.10 (Oct. 1945) 168.
D20. " Balanced modulator," U.S. Patent, 1,988,472 (Jan. 1935).
D21. Whitaker and Artzt " Development of facsimile scanning heads -Radio facsimile " (R.C.A. In-
stitutes Technical Press, Oct. 1938).
D22. Black, L. J., and H. J. Scott " A direct- current and audio -frequency amplifier," Proc. I.R.E.
28.6 (June 1940) 269.
D23. Goldberg, H. " A high-gain d -c amplifier for bioelectric recording " E.E. 59 (Jan. 1940) 60.
D24. Richter, W. " Cathode follower circuits " Elect. 16.11 (Nov. 1943) 112.
D25. Shepard, W. G. " High -gain d -c amplifier " Elect. 20.10 (Oct. '.947) 138.
D26. Aiken, C. B., and W. C. Welz " D -C amplifier for low -level signals " Elect. 20.10 (Oct. 1947) 124.
D27. Anker, H. S. " Stabilized d -c amplifier with high sensitivity " Elect. 20.6 (June 1947) 138.
D28. Gall, D. C. " A direct-current amplifier and its application to industrial measurements and con-
trol " Jour. I.E.E. 89 Part 2, 11 (Oct. 1942) 434.
D29. Offner, F. F. " Balanced amplifiers " Proc. I.R.E. 35.3 (March 1947) 306.
D30. Lash, J. F. " Feedback improves response of d -c amplifier " Elect. 22.2 (Feb. 1949) 109.
D31. Bishop, P. O. " A note on interstage coupling for d.c. amplifiers " Electronic Eng. 21.252 (Feb.
1949) 61.
D32. Hunt, F. V., and R. W. Hickman " On electronic voltage stabilizers " Rev. of Sci. Instr. 10.1
(Jan. 1939) 6.
D33. Wellman, H., A. B. Macnee and C. P. Gadsden, " A low -noise amplifier " Proc. I.R.E. 36.6
(June 1948) 700.
D34. Sowerby, J. McG. " The cascode amplifier " W.W. 54.7 (July 1948) 249.
D35. Sowerby, J. McG. "The cascode again W.W. 55.2 (Feb. 1949) 50.
D36. Valley, G. E., and H. Wellman (book) " Vacuum tube amplifiers " (M.I.T. Radiation Laboratory
Series, McGraw -Hill Book Co. New York and London, 1948).
D38. Harris, E. J., and P. O. Bishop " The design and limitations of d.c. amplifiers " Electronic Eng.
21.259 (Sept. 1949) 332 ; 21.260 (Oct. 1949) 355.
D39. Sowerby, J. McG. " Reducing drift in d.c. amplifiers " W.W. Part 1 : 56.8 (Aug. 1950) 293 ;
Part 2 : 56.10 (Oct. 1950) 350.
D40. Volkers, W. K. " Direct- coupled amplifier starvation circuits " Elect. 24.3 (Mar. 1951) 126.
D41. Passman, B., and J. Ward " A new theatre sound system " Jour. S.M.P.T.E. 56.5 (May 1951) 527.
Additional references will be found in the Supplement commencing on page 1475.
(E) REFERENCES TO STABILITY, DECOUPLING AND HUM
El. En -Lung Chu " Notes on the stability of linear networks " Proc. I.R.E. 32.10 (Oct. 1944) 630.
E2. Wen -Yuan Pan " Circuit for neutralizing low frequency regeneration and power supply hum "
Proc. I.R.E. 30.9 (Sept. 1942) 411.
E3. Sturley, K. R. " Low frequency amplification -Part 5, The anode decoupling circuit " Electronic
Eng. 17.205 (March 1945) 429.
E4. British Patent 567021, Furrehill Laboratories Ltd., 27/7/43 " Decoupling Circuits " Review W.W.
51.9 (Sept. 1945) 288.
E5. Design Data (1) " Cathode bias-effect on frequency response " W.W. 52.1 (Jan. 1946) 21.
E6. Zepler, E. E. (book) " The technique of radio design " (1943) pp. 206 -219, 241.
E7. Retch, H. J. (book) " Theory and applications of electron tubes " (1944) pp. 132, 209-211.
E8. Zakarias, I. Reducing hum in pentodes " Elect. 21.11 (Nov. 1948) 170.
(F) REFERENCES TO PULSE AMPLIFIERS
Fl. Moskowitz, S. " Pulse amplifier coupling," Comm. 25.10 (Oct. 1945) 58.
(G) GRID CIRCUIT RESISTANCE
Gl. Crawford, K. D. E. " H.F.. pentodes in electrometer circuits " Electronic Eng. 20.245 (July 1948)
227 -also gives additional references.
See also Chapter 2 Ref. Hl, Chapter 3 Ref. 41.
CHAPTER 13
544
13.1 (i) TYPES OF A-F POWER AMPLIFIERS 545
The analysis of a power supply shunted by a condenser and supplying sine -wave
signal current does not appear to have been published.
An analysis has been made (Ref. B6) for the simpler case where the signal current
is of rectangular form. The minimum frequency of rectangular waveform for a
specified variation of supply voltage over the half -cycle is given by
f2.
K(RL /R,) + 1
4.606KRLC logo A
where K = Ibo /(I bo + ID)
R, = resistance of voltage source
RL = Ebo /Ibo
C = shunt capacitance
(1 -K)
A
(1 - K) - [K(RL /R,) + l]x /100
Ebo = direct voltage across C
Ibo = direct plate current of valve
ID = peak signal current through valve plate circuit
and x = fall of supply voltage between beginning and end of positive half-cycle,
expressed as a percentage of Ebo.
The only approximation in eqn. (1) is that the voltage across the condenser at the
commencement of the positive half-cycle is taken as being Ebo.
Eqn. (1) has been applied to the following example-
E" = 250 V, Ibo = 40 mA, I, = 30 mA, C = 16µF, x = 2,
to give these results :
When RL /R, = 5 10 20 30 35
then f, = 170 157 127 96 49 c/s
This example is a severe one since a rectangular waveform is more severe than a
sine waveform with the same peak current, while the value of x only allows 2% drop
in voltage over the half-cycle. If 5% voltage drop is permissible, f, becomes 39 c/s
for RL /R, = 10. The importance of good power -supply regulation for satisfactory
performance at low frequencies is demonstrated.
This effect does not occur to any appreciable extent with push -pull Class A am-
plifiers, and is not so pronounced with Class AB amplifiers as it is with single -ended
Class A or with Class B.
Amplifiers should always be tested for frequency response and distortion
at maximum power output, as well as for frequency response at a lower level.
The maximum grid swing which can be used is 20 volts, that is from A (E2 = 0)
to B (E = -
20 volts) since the E = 0 curve is the border of the grid current region.
Actually a slight grid current usually flows at zero bias, but this is generally neglected.
FIG. 13.2
D
FlfwIfJ
PLATE DISSIPATION (w)
-E,+ - E +
`-min E. Emoz E
3II.
25
AQ /QB -1
x 100 (7b)
SSSSSSSSSSiSiiiii:í
2(AQ /QB + 1) MEIM EMMEM MIMI! AMIN=
The equation (7b) has been plotted graphic- 20
ally in Fig. 13.3. u
.....5.....'........
""P...im
The voltage gain of a power amplifier 81
is given by
M = (Emax
= 2.82V/PR L
-
Emin) /(2E2 )
Wa
MIIMMri=
t MMMMMM
MMMMMM WIMIWIN
=MO MMMMMM MIIIIIoIS
The peak values of the fundamental and
second harmonic components of the signal SïSSSSSSSSSïSSSSSSS
plate current are AWN. MMMMM ME an.i.
Fundamental
:
Second harmonic 'hi = «Imax + 'min Fig. 13.3. Curve of second har-
- 2IQ)the second
(10) monic distortion plotted against the
If AQ /QB = 11/9 = 1.22, then ratio AQ /QB for the two portions
harmonic distortion will be 5 %. This of the loadline (Fig. 13.2).
550 (i) SIMPLIFIED GRAPHICAL CONDITIONS 13.2
is the principle of " 5% distortion rule " reproduced in
Fig. 13.4. The rule has
each division to the left of 0 a length of 11/9 or 1.22 of the
length of a corresponding
o division to the right of O. It may be made with each left -hand
division
11 millimetres and each right -hand division 9
millimetres. Each of
these divisions may be divided into 10 equal subdivisions. It may
be
used by placing the " 0 " of the distortion rule at any likely
operating
point and tilting the rule gradually until the reading on the rule corres-
ó
Q ponding to the zero bias curve is the same as the reading corresponding
to the curve of twice the grid bias at the operating point.
AQ /QB
á Ñ will then be 11/9 and the second harmonic distortion corresponding
a to AQB as a loadline will be 5 %.
Y There are also methods for the graphical determination of any
ó degree of second harmonic distortion. One of these was originated
by Espley and Farren (Ref. BI) and is illustrated in Fig. 13.5 ;
? xi must be drawn on transparent material. It is applied to the curves it
in a manner similar to the 5% distortion rule, except that a loadline
5'
ó is first drawn in, and then the Harmonic Scale is moved so
9 that DBC
is parallel with the loadline. For example, with the loadline EFG
the
u distortion is 20% while with E'FG the distortion is 10 %. The scale
may be constructed on the basis of :
a LL 2nd Harmonic Distortion % 0 5 10 15 20
° DB /BC 1.0 1.22 1.5 1.86 2.33
W
If the valve is to be operated with a known plate voltage, the usual
procedure is to take as the operating point (Q) the intersection of
the
vertical line through Eb with one of the E, curves such that Q is either
on, or below, the " maximum plate dissipation curve " (Fig. 13.2).
0 If the latter is not included in the published curves, it may readily be
plotted over a small range. This dissipation curve only affects the
operating point Q, and there is no harm if the loadline cuts the curve.
In general, if the plate voltage is fixed, a triode gives greatest output
when it is operated at the maximum permissible plate current, with
the limit of Ib = lE, /r9 (see eqn. 22).
If there is no predetermined plate voltage, a triode gives increasing
power output as the plate voltage is increased, even though the plate
current is limited in each case by the dissipation.
(ii) General graphical case, power output
and distortion
In practice there are several additional factors which should be
taken into account.
(A) Resistance of transformer primary
Provision may be made for this resistance by giving the line QE,
a slope of - 1 /R' where R' is the resistance of
the transformer
primary or choke (Fig. 13.6). By this means the plate voltage of
point Q is less than E, by the value R'IQ. This does not affect the
slope of the loadline, which is determined solely by the effective im-
pedance of the load reflected across the primary of the transformer.
It is assumed that the final filter condenser has sufficient capacitance
to supply the varying current over each cycle without appreciable
change in voltage.
(B) Effect of primary inductance
The reactance of the transformer primary gives an elliptical load-
line [see Chapter 2 Sect. 4(vi)j. This may be reduced to any desired
extent by increasing the inductance. Obviously this effect is only
serious at the lowest signal frequencies ; suitable values of inductance
are given in Chapter 5 Sect. 3(iii)c.
Fig. 13.4. 5% distortion rule for use in calculating the power
output of triodes.
(ii) GENERAL GRAPHICAL CASE 551
13.2
O
B C O PLATE VOLTAGE Eb
202 152 10%5%0%
Fig. 13.5. Alternative form of distortion rule for application to triode loadlines (Ref. B11.
Fig. 13.6. Plate characteristics of triode exaggerated to show rectification effects.
(D) Rectification effects
Owing to the second harmonic component of the signal frequency plate current,
the average plate current under operating conditions is greater than with zero signal.
The increase in plate current is given by
dla = +(1max + /min 2IQ) - (11)
which is also equal to the peak value of the second harmonic current (Ina) as shown
by eqn. (10). This may be put into the form
d1a = 1(IA -- IB)
IQ and IB = (IQ -
(12)
where IA = Im,x Imin)
This is illustrated in Fig. 13.6 where AE = QB and EQ = AQ - QB. Obviously
AQ /QB oc /A//B. The line EQ is then divided into four equal parts ; C is the
centre point of AB and D is the centre point of CQ. The plate current of point D
is greater than that of point Q by the amount «(IA - IB) which is equal to d1 b. Point
D is the only point on the loadline which fulfils the conditions regarding current,
while point Q is the only point which similarly fulfils the conditions regarding voltage.
The condition is therefore an impossible one, and the loadline must shift upwards
until the point of average current lies on the line QE p.
Unfortunately, as the loadline moves upwards, the relationship between IA and IB
changes, and the simplest procedure is to draw a second loadline A'B' (Fig. 13.7)
parallel to AB, then to determine its points E'F'C'D' as for the original loadline.
Points D and D' are joined by a straight line
FIG.13.7 DD', and the intersection of this with QEb
i gives point D" which must be on the working
A I loadline. Through D" draw a loadline parallel
to AQB, and A "D "B" is then the maximum
signal dynamic loadline. The only error is
through assuming that DD' is a straight line,
Á__°
whereas it is actually the locus of point D ;
the error is small if DD' is short.
It is evident that the change from the no-
signal quiescent point Q to the maximum
signal dynamic loadline A"B" will be a gradual
process. As the signal increases, so will the
loadline move from AQB to A"P "B ". At
£b
O PLATE VOLTAGE
intermediate signal voltages the loadline will
be intermediate between the two limits. The
Fig. 13.7. Method for graphical de- average plate current for zero signal will be
termination of " shifted " loadline due
to rectification. I Q but this will rise to /b" at maximum signal.
552 (ii) (D) RECTIFICATION EFFECTS 13.2
It will be observed that D" does not correspond to the intersection
and the static bias curve. Point P" is not the quiescent point of the loadline
be described as " the point of instantaneous zero signal voltage (which is Q) but may
line." Point P" must therefore be used in the calculations foron the dynamic load-
at maximum signal. The loadline A"D"P"B" provides harmonic distortion
the data necessary for the
calculation of power output, second harmonic distortion,
and average direct current.
All these will, in the general case, differ from those indicated
by the loadline AQB.
Summary :
1. Power output is calculated from the loadline
A"B" in the usual manner.
2. Second harmonic distortion (per cent) at maximum signal
x 100. = (2D"P"/A"B")
3. Average d.c. current at maximum signal = lb" =
IQ + Mb" = current for
point D ".
The graphical method above is accurate, within the limits of graphical
but rather slow. Approximate results may be calculated from the construction,
by the equations : original loadline
7
Ib,.
- -
IQ +
3(Imnx + Im1, 21-0(1 + Ry /r,) (13)
d lb" + Imaz + Imtn 2IQ)(1 + Ry /r,)
Alternatively the rise in current may be calculated from the second (14)
tortion (H,) and power output : harmonic dis-
1
44" /2H21/Po /Ry(1 + Rt /rP)
In another form, the rise in current may be calculated from the harmonic (15)
distortion,
IQ and Im;n :
or less accurately by
-
dIb ti (IQ Imin)(1 + R y /r,,)[H, /(1 2H2)] - (16)
dIb" (0.78 IQ)(1 + Ry /r,)[H, /(1
(on the assumption that Im,,, = 0.22 IQ).
-
2112)] (17)
Eqn. (17) is interesting, since it does not involve any data beyond
published. If H2 = 0.05 it may be reduced to those normally
dIb" 0.043 IQ(1 + RL /r,)
Eqns. (13) to (18) give results slightly lower than the graphical (18)
For rectification effects with cathode bias see below. method.
(E) Cathode bias
Cathode bias loadlines may be drawn on the mutual characteristics
Sect. 4(v)] but the position is complicated by the rise in plate [Chapter 2
rectification. The simplest approach is to assume a voltage betweencurrent caused by
(E b), determine the plate current under maximum signal conditionsplate and cathode
as for fixed bias
616.1 3.6 (Ib ") and then to calculate the cathode bias
resistor (R,,) from the equation R,, =
E,/Ib ". It is then necessary to check for
plate dissipation at no signal ; this may be
done by the method of Fig. 2.27, drawing
a cathode loadline having the correct slope,
and noting the plate current (Ibi) at which
the cathode loadline intersects the mutual
characteristic for the desired value of Eb.
The dissipation with no signal input is
then Eb/b,. There is a slight error
through assuming that Eb remains con-
stant from zero to maximum signal, but
this is usually negligible. Alternatively,
the maximum value of 1br may be calcu-
eb lated from the maximum dissipation, and
bEe.
marked on the corresponding Eb curve ;
Fig. 13.8. Determining change of plate through this point a cathode loadline may
current with fixed and cathode bias, be drawn whose slope will indicate
the
having operating conditions identical at minimum permissible value of cathode
maximum signal. bias resistor.
13.2 (ii) (E) CATHODE BIAS 553
Alternatively, cathode loadlines may be drawn on the plate characteristics, but they
will be slightly curved [for method see Chapter 12 Sect. 2(iii)].
Rectification effects with cathode bias may be determined entirely from the
plate characteristics (Fig. 13.8). The loadline corresponding to maximum signal
conditions is A"B" (as for Fig. 13.7) while point P" is the point of instantaneous
zero signal voltage. The maximum signal plate current is I b" as for fixed bias. For
the sake of comparison, the zero signal loadline AQB for fixed bias has also been
added. From D" a straight line is then drawn to Ebb where Ebb is the total supply
voltage across both valve and R5; this line will have a slope of - 1/R k.
The cathode bias loadline must pass through the point R on the E1 curve at which
the plate current is I b", because Rh = E ci/ I b". The Rk loadline also passes through
0 as shown ; it intersects the sloping line through Ebb at point S with plate current
lb'. This point S is the quiescent operating point with cathode bias and has a plate
voltage Eb, which is slightly greater than E,,1.
The change in average plate current from no signal to maximum signal is given by
b' for cathode bias and ab- for fixed bias. It is obvious that JI,,' is always less
than A/ b". More specifically,
r, E,1 (19A)
AL" Rk Ebl
As a typical example,
ohms, Rk = 750 ohms ;
for type 2A3,
b'/L1/b"
E,, =
0.2.
250 volts, -
E,1 = 45 volts, r, = 800
In words, the change of plate current with fixed bias is five times that with cathode
bias.
If the value of the cathode bias resistor and the total supply voltage (Ebb=Eb -FEe)
are known, the following procedure may be adopted (Fig. 13.9A).
If it is desired to make allowance for the transformer primary resistance, the slope
of D "Ebb in Fig. 13.8 should be -
1/(R' + R k) where R' = primary resistance.
In general, any single Class A triode may be operated either with fixed or cathode
bias as desired. The maximum value of grid circuit resistance frequently depends
on the source of bias. When it is permissible to use a value of cathode bias resistor
which provides the same grid bias as required for fixed bias, the load resistance,
distortion, and power output will be identical. When it is necessary, on account of
plate dissipation, to use a higher value of Rk, then the conditions in the two cases will
be different.
Cathode by- passing -It
is important for the by -pass condenser to be sufficiently
large to maintain the bias voltage constant over each cycle -any fluctuation leads to
increased distortion and loss of power output. When operating at low levels it is
usually sufficient to ensure that the reactance of the by-pass condenser, at the lowest
frequency to be amplified, does not exceed one tenth of the resistance of the cathode
bias resistor.
At high operating levels it is necessary to ensure that the direct voltage from B +
to cathode remains substantially constant over each half- cycle. This may be treated
in the same manner as the by-passing of the power supply -see Sect. 1(iv) and Eqn. (1).
The network comprising the resistive elements R, and R,, and the capacitive ele-
ments C and Ck may be replaced by its equivalent network comprising C, in parallel
with R1, connected from B+ to cathode. In the special case when C/Ck = Rk /R,
20
11EEM
we have Rt = R, Rk and C, = CCk /(C + Ck), but the general case is more in-
volved. Eqn (1) of Sect. 1 may then be applied directly.
MMnEMMmEmummomm
ME1.1MEEMEEMMMMM
MM
N9MR"I
MEE11
MEr1o'/
I, OP
a
sa°°
--
1....
El
10
M!!I
EI.I Bi.MME\EM
Mr16EIIM!!CME
'
'1
EMME
i ssnn;u
RaIl
Qi'IEE%/IMM
E
EMMEM1MIMI,MIM3iiNMEEM
MII
Ei tIE11MC>r
EEMMM
M I
/
o''I 11'112911I,I
I,MI 'I /E
° rIMMIs,I
1 o°' .y
'EN
\[fá9 9IIr1ZfI11E1'I
AWE'
TIMM 6R
Si/M'IMMI,
`II//III
M.VMrl:
..MIIM
' EI
°° 9
UII II
o,MEO/,rIE!!EE
I/n_GiMM
E/llì\IC/a
rfl!'PIEIIIIa IEîill/S / ,s°-eí nail
1
.'3i;IEtm+.
rn
!DíIEaSIMM
s
:.'sarie_Mií
dl[fi..`\'IEMGIEnE'
rs2w7tM,EGIS
EI,RámwI3iHor=j
I,Mr1l.StRM
rewAPMEiiM I7
F.f/M:..ZMQEiP.\E=' d/1d'Li//e=ii
MIEiQ/7Eiawm:'4\m - A" i:MOM
iiáiszíii,ásMé=i=:%:ïr°'=
r.s/.M=c'Az=r.IlepVA* ePl
aar' loo°
-R.aMMLM A `O-
Is0°
a__
°
.IiUM
s°,:iiiiiiiii
=$4zi/%J.T1=d=Dr ee! .+ürv2\MM
/41193ï1iDMiii!==réAi.;sMN-PnIZMM
'
0
0
'!Et::G3t:P.=t0."ss
loo
:!SW-'cfl2:a=tM\iaT200 300 400
PLATE VOLTAGE FIG. 13.90
Fig. 13.9B. Plate characteristics type 6J5 with g,,, and g, curves and loadlines (Ref. B5).
13.2 (ii) (F) METHOD USING Gm AND Gp CURVES 555
g. = mutual conductance
and g, = plate conductance = 1 /r9.
For predominant second harmonic distortion :
A' _ (A, + A,) (19C)
Second harmonic distortion per cent = 25(A1 - A,) /(A1 + A,)
where A' = amplification with maximum input signal
Al = amplification at maximum positive excursion
and A, = amplification at maximum negative excursion.
The values of Al and A, may be calculated, using eqn. (19B), from the values of
gm and g, corresponding to the relevant points on the loadline which may be derived
from Fig. 13.9B for type 6J5. Alternatively, the values of g,,, and g, may be derived
by measurement or graphically at the points of maximum positive and negative ex-
cursion on the loadline. By this means it is possible to calculate amplification and
distortion with high signal inputs from measurements made at low inputs.
This method may be used for cathode followers by using eqn. (19D) in place of
(19B) :
A = gmRk /[1 + (gm + g,)Rk] (19D)
It may also be used for degenerative amplifiers by using eqn. (19E) :
A = -
g,,,RL /[1 + (gm + g,)Rk + g9RL] (19E)
References B5, B7, B8, H17.
(iii) Optimum operating conditions
The preceding graphical treatment enables the optimum load resistance to be de-
termined to give maximum power output for limited distortion, provided that valve
curves are available. It is frequently desirable to be able to make an approximate
calculation without going to so much trouble. The optimum load resistance is a
function of the operating conditions (see Refs. A). The following treatment is based
on Ref. A14, and relates to " ideal " (linear) valve characteristics.
Case 1 Grid current and distortion zero, fixed signal input voltage, no
:
°
-
(2r, + RL)2
l+Eb1ERL (21)
FIG.13.10
FIG . 13.1 1
Ebt Eb Ebi Eb
Fig. 13.10. " Ideal " plate characteristics of triode to illustrate choice of optimum
loadline.
Fig. 13.11. " Ideal " plate characteristics of triode with fixed plate voltage to illustrate
choice of optimum loadline.
Case 3 Grid current and distortion zero, fixed plate voltage and maximum
:
FIG.13.13
Fig. 13.12. Plate characteristics of imaginary triode with " ideal" characteristics
above I. and curved characteristics below this line.
Fig. 13.13. Ideal triode characteristics with RL = 2rp.
When a triode is being operated at a plate current which is less than the optimum,
then Rt will be greater than 2r for maximum power output. Refer to Case (3),
eqns. 23, 24 and 25.
FIG.13.14
ó74
4
0c",
4G
2
H a.
1S1'NC
20,000 30,000 5q000Ì
(OHMS)
6
8000
00 IOC
0
RATIO Ra'ry. 2 3 4 5 6 8 ro 25 50 60
Fig. 13.14. Power cutput and distortion of type 45 triode plotted against load resistance.
The curves were derived graphically, with allcwance for shifting loadline. The operating
conditions are : E, = 250 volts, E, _ -50 volts, peak grid signal voltage 50 volts,
I c = 34 mA (zero signal), r, = 1610 ohms.
13.2 (v) PLATE CIRCUIT EFFICIENCY 559
FIG.13.15
Ib
1
r
1 Dissipation Iin
tit
Transformar
Primal
Eb Ebb o
INPUT
o
u Rt
(i) Introduction
Multi -grid valves include pentodes, beam power amplifiers and similar types, and
space- charge valves. All of these have a family resemblance, in that they have higher
plate circuit efficiencies and greater power sensitivities than triodes, but on the other
hand they have higher distortion, particularly odd harmonic distortion. Their
advantages over triodes generally outweigh their disadvantages, particularly as the
distortion may be minimized by negative feedback, and they are almost exclusively
used in ordinary commercial radio receivers. Reference should be made to Chapter 7
for amplifiers incorporating negative feedback.
The basic circuit of a power pentode with transformer- coupled load is Fig. 13.18.
A choke -capacitance coupled load may also be used, as for a triode (Fig. 13.17). The
13.3 (ii) IDEAL PENTODES 561
only essential difference from a triode is that provision must also be made for a constant
voltage to be applied to the screen. In some cases the screen is operated at the same
voltage as the plate, so as to avoid a separate screen voltage supply.
In a practical circuit the screen, the plate return and the grid return should each
be by- passed to the cathode.
21a
ibi
\Ea ¡-15
FIG. 13.71
k b 0.293 Ect_
Ie
------------
{ ro
--------------- --
I
IA
... tQ
___ -.
I
Ebs Eea
Ec1"0_
tcz
Ir
O Emin
- 1.707 Esj_
Eb, Emaa
Fig. 13.21. Plate, screen and cathode current characteristics of pentode.
Fig. 13.22. Calculation of pentode power output and distortion using the five selected
ordinate method.
(F) The screen should be supplied from a voltage source of good regulation ; a
voltage divider, whose " bleed " current is very much higher than the maximum
screen current under dynamic conditions, is the most common source when the screen
voltage is lower than the plate voltage.
(G) The voltage gain is determined as for a triode [Section 2(i) eqn. 8].
(H) If oscillation or parasitics are experienced, a grid stopper resistance of
about 500 ohms may be used. A small mica condenser from the plate terminal to
earth is often beneficial. The bias resistor and its electrolytic by -pass condenser may
be by- passed by a 0.001 µF mica condenser. The shells of metal valves should be
earthed directly. The length of leads should be as short as possible, particularly
between the valve terminals and their by -pass condensers.
13.3 (iii) PRACTICAL PENTODES -OPERATING CONDITIONS 563
With type 807, a stopper resistor of 100 ohms (non- inductive) may be connected
directly to the screen terminal of the valve, and a by -pass condenser (0.01 µF mica)
taken from the end of the resistance remote from the screen, directly to earth.
(I) Overload characteristics -Owing to the crowding together of the charac-
teristics at and below the knee of the curves, a pentode tends to give a smoother over-
load characteristic than a triode. There is no point at which the distortion begins
sharply, but rather a gradual flattening effect on the top, or top and bottom, of the signal
current wave. Pentodes or beam power amplifiers with negative feedback lose most
of this " cushioning effect," and more closely resemble triodes as regards overload.
The distortion due to grid current may be minimized by a low impedance (triode)
preceding stage and a fairly low value of grid resistor ; this will then leave the dis-
tortion due to the characteristics as the predominant feature, and provide a reasonably
smooth overload.
A third harmonic scale may be prepared for reading the third harmonic percentage
(Ref. A14 pp. 71 -72).
(C) Seven equal-voltage ordinate method (Espley)
This is exact for harmonics up to the sixth, provided that higher harmonics are
zero. It is sometimes more convenient than the five ordinate method, when there are
no -0.5 and -
1.5 Ec1 curves, even when the higher harmonics are of no interest.
The symbols have the same significance as in Fig. 13.23 with the subscript indicat-
ing the grid voltage.
Po = (16710 + 25210.33 -
4510.67 + 45/1.39 25211.87 -- -
167/2.0)2 RL /819 200 (20)
1-12% = 25(55910+ 48610.33- 1215/0.67 +340/1.0 121571.33 +48611.67 +559/2.0) /I (21)
1-13%=250(45/0- 3610.99 -6310.67 +6311.99 +36/1.67 -4512.0)/1 (22)
H4 %= 450( 1710 -4210.,9 +15/0.67 +204.o +1511.33 -42/1.67 +1712.0) /I (23)
H6 %= 4050(10 -410.33 +510.67- 511.33 +411.67- /3.0) /1 (24)
H6 %= 2025(10 -6I0.,, +1510.67 -20/1.0 +1511.8a -6/1.67 +13.0)/1 (25)
where 1= 16710 +25210.33- 4510.67+4511.33 -25211.87- 16712.0.
(D) Eleven selected ordinate method
To use this method, it is first necessary to plot the loadline on the plate characteris-
tics, and then to transfer it to a dynamic mutual characteristic (Fig. 13.24) from which
the required values of plate current may be derived.
Po -
= i(0.510 0.573.0 + 1A+,á /Am6)2RL - (26)
- -
= 0.510 0.513.0 + /A ma /5 +6 = fundamental
/Al (27)
/AZ = 1(Io + 13.0 2I1.0 - (28)
/A9= 0.167(210.6 +12.0 -4
211.6)
-
/h, = 1( 4 + 211.0 + 19.0 210.3 211.7)
-
- (29)
(30)
/A6 = 0.1(24., + /0 + 2 40 210.3 - -
211.3 /9) - (31)
Percentage second harmonic = ( /Am1/IA,1) x 100
13.3
Fig. 13.23.
Fig. 13.24.
(iv) GRAPHICAL ANALYSIS
r
at
60
ao
20
FIG. 13.24
c'...'.
olN"IIIMMMMOM 114
=wimmmmm
-Is
-AM Ioa
'MIN
-o -as
-0'1
-os
Calculation of pentode power output and distortion using the five equal-
Calculation of pentode power output and distortion using the eleven selected
ordinate method.
Percentage third harmonic = (Ihm3 /Ihml) x 100 etc.
Note that all currents, including I h m1, I h m2 etc., are peak values. The exact values
of grid voltage are 0, 0.191, 0.293, 0.5, 0.691, 1.0, 1.309, 1.5, 1.707, 1.809 and 2.0.
The approximate values are, however, sufficiently accurate for most purpóses.
Eqn. (26) shows that third harmonic distortion adds to the power output, while
fifth harmonic subtracts from the power output as calculated for distortionless con-
ditions. See also (iv)A above ; Sect. 7(iii).
For theoretical basis see Chapter 6 Sect. 8(iii).
(y) Rectification effects
The general effects are the same as for a triode except that it is sometimes possible
with a pentode to have zero second harmonic and no loadline shift. In all other cases
it is necessary to determine the corrected loadline before applying harmonic analysis or
calculating power output. The method of deriving the corrected loadline is the same
for pentodes as for triodes [Sect. 2(ii)D]. For example see Fig. 13.28.
Iy
O
565
11:2
10.1
Io.1
11.0
I,,
I2o
With a triode, the loadline always shifts upwards into such a position that the dis-
tortion is less than it would otherwise be. With a pentode, the loadline shifts upwards
when the load resistance is lower than a certain value, and shifts downwards when the
load resistance is higher than this value. When a pentode loadline shifts downwards
it causes increased distortion or decreased power output.
(vi) Cathode bias
The operation of cathode bias is the same as with triodes [Section 2(ii)E] except
that the total current flowing through the cathode bias resistor is the sum of the plate
and screen currents [Chapter 2 Section 4(v)]. If the screen is supplied from a voltage
divider which is returned to the cathode, the bleed current must be added to the
cathode current in calculating Rk. This tends to stabilize the bias voltage. Even in
cases where there is no rise of plate current at maximum signal, there will always be
a rise in screen current [Section 3(iii)E].
When a tentative value of Rk has been determined for maximum signal conditions,
it is necessary to check both plate and screen dissipations at zero signal. If the plate
and screen voltages are equal, the simplest method is the use of the " triode " mutual
characteristics, if available [Chapter 2 Sect. 4(v)].
If there is a " bleed " current through R k, the procedure is shown in Fig. 13.25
where the triode mutual characteristic is shown in the upper part, with OA repre-
senting the bleed current in the lower part. The cathode loadline is drawn from A
(instead of from 0) with a slope of - 1 /Rk where Rk is the value determined for
The rectification effects with cathode bias may, like triodes (Fig. 13.8) be determined
from the plate characteristics but an approximation is involved since D "E66 must
have a slope of -m /Rk(m + 1) where m = 'b /',2. The value of I,2 may be taken
as that for maximum signal, and a slight error will then be introduced when deriving
the zero signal condition. The cathode bias loadline should, ideally, be drawn as two
loadlines -one for zero signal, and the other for maximum signal. For most pur-
poses, however, the maximum signal cathode loadline may be used for both conditions.
The zero bias lines are, in each case, above the lines for maximum signal, and the
difference in slope is of the order of 5% to 10 %.
In general, any Class A, pentode may be operated either with fixed or cathode bias,
as desired, except for the limitations on the maximum grid circuit resistance [see
.,
Sect. 10(i)]. It occasionally happens that a condition is permissible only with fixed
bias, owing to the rise in plate dissipation at zero signal.
For back bias, further details regarding fixed bias, and grid circuit resistance see
Sect. 10.
For cathode by- passing see Sect. 2(ii).
FIG.13.26
W
U
I2
3
a
Q10
08 _ Ó
8g
a A
a
o
Cr
3
20.2
06
0.4
0
Iilin--I'a,1/4
4 8
LOAD RESISTANCE
O/sTOas%oN
- 12
KILOHMS
_II 6.
16
4u
z
22
cc
Fig. 13.25. Cathode loadline of pentode with additional bleed current passing through
the cathode resistor, plotted on mutual characteristics.
Fig. 13.26. Power output and harmonic distortion of typical pentode plotted against
load resistance. Valve type 6ÁK6, E b = E,,Q = 180 V, E,,r = -9, peak signal = 9 volts.
shunt filter -a
resistance R in series with a capacitance C. If R is variable, the com-
bination is the simplest form of tone control. If R is fixed, typical values are :
R =
1.3RL ; C = 0.0251.4F for RL = 5000 ohms.
This does not affect the rise of impedance at the bass resonant frequency, which
is a function of the loudspeaker design and the type of baffle -see Chapter 20 Sect.
2(iv) and Sect. 3.
At maximum signal voltage the conditions are somewhat different (Fig. 13.26).
The power output in this case reaches a maximum at RL = 9000 ohms. The second
harmonic reaches a minimum (practically zero) at about RL = 14 000 ohms, and then
rises steadily ; actually it undergoes a change of phase near RL = 14 000 ohms.
The third harmonic rises all the way from zero to the limit of the graph. Minimum
" total distortion " occurs at RL = 10 000 ohms, which is the published typical load,
being a close approach to maximum power output. In this particular case the load
resistance for zero second harmonic is not that for maximum power output.
In the case of a loudspeaker load, the load resistance may rise from the nominal
value to (say) 6 or 8 times this value ; all the variation is in the upwards direction. If
full signal voltage is maintained for all frequencies, the distortion will be very severe
and the maximum power output will be reduced at low and high frequencies. The
only methods of minimizing the trouble are the use of a loudspeaker and baffle with
less prominent impedance peaks, and the use of negative voltage feedback
(see Chapter 7) or reduced signal voltage on the grid. A pentode, operating well
below its nominal power output, is capable of giving reasonable fidelity even on
an ordinary loudspeaker load. In a normal radio receiver, a power pentode with a
nominal maximum power output of 4 or 5 watts can give reasonable fidelity up to
somewhat over 1 watt, but it has the advantage of being capable of delivering its full
power output when the distortion can be tolerated.
The effect of these high impedance loads, which are here assumed to be purely
resistive for the purpose of illustration, is shown by the beam power amplifier plate
characteristics in Fig. 13.27. The effect may be minimized by reducing RL slightly
below the optimum value.
fc,.o
Fig. 13.27. Plate characteristics of 6L6 or 807 beam tetrode with loadlines of optimum
resistance also twice, four times and eight times optimum. The loadlines have been corrected
for the rectification effect.
The impedance of a loudspeaker is, however, far from being resistive (see Chapters
20 and 21), having a reactive component varying with frequency, which must be com-
bined with the shunt reactance of the transformer primary at low frequencies and that
of the shunt capacitance from plate to earth at high frequencies. The combined
reactive components increase the distortion and reduce the power output [see also
Chapter 5 Sect. 3(iii)c and Chapter 2 Sect. 4(vi) ; Ref. C4].
The published values of power output apply to highly efficient output transformers.
The available power output from the secondary of a normal power transformer is
equal to 77 times the published value, where i is the efficiency [see Chapter 5 Sect.
2(ii) and Sect. 3(vi)]. Typical efficiencies are from 70% to 95% (depending on the
price class) for well-designed transformers.
568 (ix) EFFECTS OF PLATE AND SCREEN REGULATION 13.3
electrode voltages,
=
increase in screen current, with constant electrode voltages,
dEb =
change in plate voltage,
JE,, =
change in screen voltage,
Eb' =
plate voltage at maximum signal,
and E,,,' =
screen voltage at maximum signal.
Then Et,' = E5 dEb -
(33) EC,' = L. 1E,, - (34)
where AEI, = RI dIb (35) dE,, = R,dIr, (36)
E0 0 Eco 3O0
FIG. 13.213
200
Ed.-10
lbw' Ica
NA)
Ec,.-I
IOO C
sEcs4-20
Ee -25
Ec..O Ea.-30
B W' 13
Fig. 13.28. Plate characteristics of 6L6 or 807 illustrating method of correcting the
loadline for rectification. The corrected loadline is A" B ".
On the plate characteristics (Fig. 13.28) AB is the uncorrected loadline with C as
its average current point, A'B' is a parallel loadline with C' as its average current
point, A "B" is the corrected loadline. The average current may be determined from
the point on A"B" which has a plate voltage 300 volts ; from this the rise of current
dl b may be calculated. The value of dt ,, may be determined from eqn. (9), JE,,
may be calculated by eqn. (36) and E,,' by eqn. (34). In the example of Fig. 13.28
let it be assumed that the plate, screen and grid voltages are all reduced proportion-
ally, F ,* being 0.9 :
1. Plot a new curve for Est = 0, E,,' = 270 volts, by the use of conversion factors *.
This is drawn on Fig. 13.29. For F, = 0.9, F; = 0.86.
2. Plot a curve for Est' = -
13.5, Er,' = 270 volts similarly. This will be the
-
15 volt curve (for Er, = 300 volts) with the current ordinates multiplied
by the factor 0.86.
'For conversion factors see Chapter 2 Sect. 6 and Fig. 2.32A.
13.3 (ix) EFFECTS OF PLATE AND SCREEN REGULATION 569
EcI:O FIG.13.29
Ec,-300 volts
Ev EM 270
R
Eci=-10
I y0cict
2
(mA)
Ec.-15
IOO
Ect.270
E
Ecl.-25
E1,--30
S S' S rIt-- Eci -27 E,,.270
100 200 300 400 500 600
Ep(VOlts)
Fig. 13.29. Plate characteristics of 6L6 or 807 giving graphical method for deriving
the effect of internal resistance in the voltage supply source.
3. Plot a curve for E,l' = - 27, Eßß' = 270 volts. This will be the 30 volt -
curve (for E,2' = 300 volts) with the current ordinates multiplied by 0.86.
4. Mark the point Q at E6 = 270, En = 270, E,1 = - 13.5 volts.
5. Through Q draw the uncorrected loadline RQS.
6. Draw the corrected loadline R"D "S" by the same method as previously (Fig.
13.28).
7. Determine the rise of plate current from Q to D" (intersection of R"S" with
E 6 = 270 volts). This should be approximately the same as derived in the first
case (4f b). If there is an appreciable error, some adjustment should be made.
_\N
46E\
..0
14
2
r l¡.
-.ONE
Ó
II, ov
C.4i-.y,
I _ e.
uo/
MEW,
u.i I
65
MEMO a MMMMMM_BM._ , 1.
MEN i ..!., IU u
4
v
L /AIM
i 70MMM0 /NNE. a
21
MMM!
!:
0o immiNMMMMM
sp0
1,000 2,000 3,000 4,000
LOAD RESISTANCE- OHMS
FIG. 13.30
Fig. 13.30. Power output and distortion of type 6L6 beam power amplifier plotted
against load resistance, for E6 = E,, = 250 volts, E,, = -14 volts, peak signal 14 volts.
1.The two valves should be placed closely together, with very short leads between
grids and plates (and screens in the case of pentodes).
2. A grid stopper should be connected directly to one or both grids. It is usually
cheaper -and just as effective-to have one grid stopper of, say, 200 ohms than
two stoppers each of 100 ohms.
3. Screen stoppers (50 or 100 ohms for each screen) are very helpful, particularly
with types 6L6 or 807 [see Sect. 3(iii)H].
N.B. Plate stoppers are less helpful, are wasteful of power, and are generally
unnecessary.
The advantages of parallel operation lie principally in the elimination of the phase -
splitter or input transformer required with push -pull operation. The disadvantages
are :
1. The necessity for handling the heavy direct plate current. This necessitates
either a special output transformer or a choke (say 20 henrys) with parallel feed
to the output transformer.
2. The higher distortion-this is not serious if negative feedback is used, and in
any case is no worse than that of a single valve.
3. The attenuation of lower frequencies at maximum power output due to the
limited size of the by -pass capacitor. This effect is also a function of the plate
supply regulation -see Sect. 1(iv).
Parallel operation may be used with a cathode- follower stage, permitting the use
of two smaller valves with lower plate voltage in place of one valve with higher plate
voltage, and thereby reducing the difficulties of grid excitation.
(i) Introduction
(A) Fundamental principles of push -pull operation
The fundamental circuit of a push-pull power amplifier is Fig. 13.31. A balanced
(push -pull) input voltage must be applied to the three input terminals, and a balanced
FIG. 13.31 iy,
(push -pull primary) output transformer must
be connected to the two plates with its centre-
tap connected to Ebb +. For the best results
the input voltage must be exactly balanced,
the valves must have identical characteristics
and the output transformer must be exactly
balanced between the two sections of the
primary, with perfect coupling between them.
' r7 Under these conditions, any even harmonics
introduced by the valves will be cancelled,
but the odd harmonics will not be affected ;
Fig. 13.31. Fundamental cir- the flux in the core due to the d.c. plate
cuit of push -pull power amplifier. currents would be zero.
The load resistance R2 is connected across the secondary, and the reflected resist-
ance* across the whole primary is
RL = 4R2(N12 /N22)
ance-see Chapter 5 Sect. 3(iii)c -or by reducing the rate of change of current. The
latter is accomplished in limiting Class A operation owing to the avoidance of the
sharp bend or " discontinuity " in the characteristic at the cut-off point. Valves
of the 6L6 or 807 class, when connected as triodes, have a slower rate of cut-off than
normal triodes, and are therefore particularly adapted to Class AB, operation, which
merges closely into limiting Class A.
The use of a resistive network to pass a steady current through the primary of the
transformer does nothing to reduce the rate of change of current, although it may help
in damping out any parasitics which may occur.
(D) Application to
Composite characteristics
characteristic curves - Ibo
by the same voltage that the other is decreased. This may be applied graphically
as shown in Fig. 13 32 ; the upper half includes the E, characteristic for V1 (ACDE)
while the lower half includes the E, characteristic for V, but inverted and placed left
to right (FGHJ). Point C on the V1 characteristic is the quiescent operating point,
and point B corresponds to the plate voltage E, b The V, characteristic is placed
so that H, the quiescent operating point, comes below C ; then since BC =
BH = Ibe
the point B is on the composite characteristic. At any other plate voltage N, the point
L on the composite characteristic is found by subtracting PN from KN, giving LN
as an ordinate. Valve V, cuts off at point J, so that the amount to be subtracted
from
the V, ordinate is zero, giving D (and its opposite number G) as points on the com-
posite characteristic. The composite characteristic is therefore FGRBLDE which
may be compared with the straight line FBE. Sudden bends occur at D and G, but
the portion between D and G is fairly straight ; the latter includes the whole Class A
operating region. It is obvious that the non -linearity of the composite characteristic
becomes worse as the quiescent operating point is moved towards the foot of the char-
acteristic-that is as I bn becomes less. Thus we have the Class A condition (including
limiting Class A) with nearly straight composite characteristics, the Class AB/ con-
dition which includes the kinks at the points of plate -current cut-off, and finally the
Class B condition with quite considerable non -linearity.
The composite characteristic of Fig. 13.32 is that for zero signal input voltage
(e, = 0). Other composite characteristics may be drawn by a somewhat similar
method, except that the (E, + e,) characteristic of V1 must be combined with the
- -
(E, e,) characteristic of V, to give the + e, composite characteristic. For example,
if E, = 60 volts, we may take e, in increments of 10 volts, giving :
e, 0 +10 +20 +30 +40 +50 +60 volts
E, + e, -60 -50 -40 -30 -20 -10 0 volts
E, - e, -60 -70 -80 -90 -100 -110 -120 volts
A family of composite characteristics is shown in Fig. 13.33 in which Ebb = 300
volts and E, = -
60 volts, with values of e, in accordance with the table above. The
Eel (voLTs)
600
VI
t
N" 4d0
T
x
Ji)
o
43
O
}O WORKING BIAS
fIG.13.33
o
/ / Y
`p o
'p ó
n
P
%
'143
/ / '/ ^ '
OO
$
/ / / / i`/ o a$
% t
i/ / // // / // // /,//O
loo
o41" /- / / / / //
/ ///// /
°` /,+/ j /j
' 4r/ / ó,.
/ //
/
/
/
I Ee
/% %
i d , Vg
° 03
Fig. 13.33.
500
t- 400
Eb=(VOLTS)
300 200
Fig. 13.34. Method of deriving the loadline on an individual valve (AQV) from the
composite characteristic AOB.
The loadline on one valve (e.g. V,) is determined as Fig. 13.34 in which the com-
posite characteristics, the composite loadline AB, and the individual V, characteristics
are the same as in Fig. 13.33. At every intersection of a composite characteristic
with the composite loadline (e.g. point C) draw a vertical line (e.g. CS) to cut the
corresponding V, characteristic. Then join these points APQSTV etc. with a smooth
curve, which is the loadline on a single valve, Q being the quiescent operating point.
It is obvious that this loadline is curved, although it is less curved with Class A than
with Class AB, operation.
Equivalent circuit for push -pull amplifier
There are various forms which an equivalent circuit might take, but the one adopted
here (due to Krauss) has some special advantages for the purpose (Fig. 13.35). The
two valves are assumed to have equal constant amplification factor (p), and plate
resistances (rand r9,) which are functions of the plate current.
13.5 (ii) THEORY OF PUSH -PULL AMPLIFICATION 577
The impedance seen by V1 is the impedance of the circuit to the right of the line
ab, which is called r,1.
It may be shown (Ref. E7) that
raó = RL'(1 + r,1 /r,2) (11)
The dynamic plate resistance of the composite valve may be expressed
ra = de61 /ia (12)
From (10) i a = di b1 + di 62.
Now dibl = debl /rD1 (13)
and di b2 = de b2 /r,2 (14)
From (12), (13), (14), ra = rvlr,2 /(r,1 + r,2) (15)
This indicates that the plate resistance of the composite valve at any instant is equal
to the parallel combination of the individual plate resistances of V1 and V2. For
Class A operation r a is very nearly constant, so that
ra r,0 /2 (16)
where r,,, = plate resistance of V1 or V2 at the quiescent operating point.
Now if RL' = rd (the condition for maximum power output)
RL' _ ra ti r,,0 /2 (17)
The load impedance seen by a single valve (V1 or V2) in Class A operation is given by
[From (11), (15), (17)], rab ti r, (18)
so that each valve at any instant is working into a load resistance approximately equal
to its own plate resistance.
The value of load resistance to provide maximum power output may be determined
approximately from the plate characteristics of one valve (Fig. 13.36). Since the
E, = 0 characteristic approximately follows the 3/2 power law, it may be shown
(Ref. E5) that maximum power output occurs when the loadline intersects the E, = 0
curve at 0.6Ebb. The plate current at this point is I6,,, and the other values are :
RL = 1.6E b b /Ibm plate to plate (19)
Po = 0.2E b o// b, for 2 valves (20)
If it is desired to determine the optimum value of Rt in a particular case, several
half-loadlines may be drawn as in Fig. 13.36 but radiating from B. The power
output for each may be calculated from the expression
Po = *Ibm2RL (21)
or Po = iIbm(Ebb -
Emin)
Eqns. (21) and (22) may be used with reasonable accuracy for Class AB since the
(22)
third harmonic distortion is usually less than 3 %. A useful rule is to multiply the
power output, as indicated by these equations, by the factor 10 000 /(100 -
H, %)2
to obtain a close approximation to the actual power output. Values of this factor
for various third harmonic percentages are given below :
H, 1% 2% 3% 5% 7% 10%
Factor 1.02 1.04 1.06 1.11 1.15 1.23
It is here assumed that fifth and higher order odd harmonics are negligible.
The loadline slope is unaffected by the grid bias, but the two extremities are slightly
affected-as the grid bias is decreased, the point A (which is really the intersection
of the loadline with the composite characteristic) moves slightly towards B. Thus
with Class Al the output will be slightly less than indicated by eqns. (20), (21) and (22).
FIG. 13.36
0.6 Ebb
(E,,,,)
Fig. 13.36. Method of deriving the approximate power output and load resistance of a
Class AB, amplifier from the characteristics of a single valve. The method may also be
used with poorer accuracy for Class A1.
Fig. 13.37. Calculating third harmonic distortion with balanced push -pull. Curve A
is the dynamic characteristic.
The dissipation at maximum signal may be calculated from the product of the plate
supply voltage Ebb and the average plate current [see (iv) below], minus the power
output. The normal procedure is firstly to select a grid voltage such that the plate
dissipation is slightly below the maximum rating at zero signal, then to adjust the
loadline so that it approaches as closely as possible the value for maximum power
output, without exceeding the plate dissipation limit at maximum signal.
Simple method for calculating third harmonic distortion
This method is only accurate in the absence of all distortion other than third har-
monic ; it is a close approximation under normal conditions provided that all even
harmonics are zero.
The procedure is to draw the loadline on the plate characteristics, then to transfer
this to the mutual characteristics in the form of a dynamic characteristic (Fig. 13.37).
13.5 (iii) POWER OUTPUT AND DISTORTION 579
Curve A is the dynamic characteristic while B is a straight line joining the two ends
of A and passing through O. d1b is the difference in plate current between curve A
and line B at one half of the peak grid voltage. The percentage of third harmonic
is given by
I,.
-20 O
composite characteristics. The total plate -E 9 tEg
current may then be plotted as in Fig. a lla
13.38 as a function of the signal grid volt- 13oá
17
O.
age. In order to find the average current d
it is generally most convenient to take equal 150° 30
angle increments over the cycle, for ex- l6Ó
ample, every 10° as shown in Fig. 13.39. 2d
The plate current should then be noted at Id
each point corresponding to 10° increase IS o
in angle over the whole 360 °. The aver-
s°
age plate current is then the average of
these individual values. 34d
This method is illustrated in Fig. 13.40 which shows the total plate current for both
valves over one -quarter of a cycle. Grid voltages are shown as fractions of the peak
grid voltage. The plate currents corresponding to grid voltages of 0, 0.17, 0.34, 0.5,
0.64, 0.77, 0.87, 0.94, 0.98 and 1.0 times the peak voltage are shown as Io, 10, etc.
The average plate current (Il.) is then given by
4
I4o= á(1o+ I7 +Io.34-1-Io.5+- Io.46+Io77+4.97 +10.94+4.,+ IIo) (26)
I,.Q
19e
V itl 1.94
IN7
1.,7 Fig. 13.40. Plate current (both
i
'--11 _.4111011111 I6o
I44
I!4
I .r,
Io
valves) plotted over one quadrant
for the calculation of average plate
current.
1=
O 1 2 3 4 .5 4 7 1 .9 FO
TIMES PEAK GRID SWING
ated under single valve conditions as regards grid bias. There will be only a slight
effect on the maximum power output or the odd harmonic distortion, but there will
be some second harmonic distortion and some out -of- balance flux -producing current
in the transformer. The second harmonic distortion will normally be small, par-
ticularly if the plate -to-plate load resistance is not much less than 4r D,,, and for most
purposes it may be neglected with valves of ordinary tolerances. The maximum
out -of- balance plate current should be provided for in the design of the output trans-
former and this additional cost should be compared with the alternative additional
cost of using valves which have been stabilized and matched for quiescent plate current
only. There is normally no real need for matching for any other characteristic.
In order to demonstrate the effects of abnormal mismatching, composite charac-
teristics have been drawn in Fig. 13.41 for two valves of entirely different types. It
has been shown by Sturley (Ref. E27) that the method of deriving composite charac-
teristics also holds with mismatching.
Type 2A3 45
Amplification factor 4.2 3.5
Mutual conductance 5250 2175 µmhos
Plate resistance 800 1610 ohms
There is a difference of 12% in amplification factor, while there is a ratio exceeding
2 1 for the other characteristics. The selected operating conditions are E b = 250 :
-
:
volts, E,, _ 50 volts, RL = 5800 ohms (plate -to- plate). It will be seen that the
composite characteristics (dashed lines) are not quite parallel, although they are very
nearly straight. On account of the unmatched condition, rectification occurs, leading
to a shift of the loadline, and the corrected loadline may be derived by the method
of Sect. 2(ii)D and Fig. 13.7. The second harmonic distortion on the corrected load -
line is only 5 %.
FIiS.13.41 EbPolt>I)
O 100 200 300 400 500
200
ISO
I b.
(mA)
100
50
SO
bt V2 45
(mA)
100
500 400 300 200 OO O
Ebt(VoItt)
Fig. 13.41. Composite characteristics for two different valves (2A3 and 45) in Class
A,, push -pull. Ebb = 250 volts, E,, _ -50 volts, R 5800 ohms plate-to- plate.
582 (v) MATCHING AND THE EFFECTS OF MISMATCHING 13.5
The fact that two valves so different in characteristics can be used in Class
pull to give what is generally classed as good quality (5% second harmonic) A, push-
indicates
the wide latitude permissible, provided that the bias is retained at the value
for a single
valve and also that the load resistance is not much less than the sum
of the single
valve loads (or 4 times the average plate resistance).
For Class A, triodes it has been suggested (Ref. E8) that the following are reasonably
satisfactory limits in conjunction with ordinary commercial tolerances
for valves :
Max. unbalance
Signal input voltages on the grids of V, and 172 : 5%
Phase unbalance at high and low frequencies :
Quadrature component = 3%
These are easily achieved by attention to the phase splitter or other source (see
Chapter
12 Sect. 6). When testing for balance with a C.R.O. on the grids of V,
and V2, it is
important to reverse the connections of the B.F.O. when changing from one grid
to
the other.
200
I 131
(m A)
00
100
1
b2
(mA)
200
Fig. 13.42. Composite characteristics for push-pull Class Al beam power amplifiers
type 6L6, Ebb = 250 volts, E92 = 250 volts, E91 = -15 volts.
584 (ii) POWER OUTPUT AND DISTORTION 13.8
harmonic is so small that its effect on the power output
may be neglected, so that
Po = bm(Ebb - Emtn)
which is the same as for triodes. With pentodes the power
(1)
higher than indicated by eqn. (1) owing to the third harmonic output will be somewhat
the third harmonic distortion is slightly less distortion. In general,
FIG. 13.43
than half that with a single valve, owing Eae,0
to the effect of the lower load resistance. t
If it is desired to calculate the harmonic
distortion, it will be necessary to plot at
least portion of the composite characteris-
tics.
The effect of a higher load resistance is Ib
to increase rapidly the odd harmonic dis-
tortion, while the effect of a lower load re-
sistance is to decrease the power output. It
is therefore advisable, with a loudspeaker
load, to adopt a nominal impedance rather
less than the value for maximum power a
output.
Emin E Ebb
The power output of Class A push-pull b
R, and Rs are grid stoppers. All other components are normal. Rk may be by-
passed if desired.
Fig. 13.44. Push -pull circuit using phase inversion in the power stage.
Fig. 13.45. Push -pull circuit using screen resistance coupling from V, to the grid of Vg.
The signal voltage on the grid of V2 must first pass through V, where it is dis-
torted, then through V2 where it will be distorted again. Thus the second harmonic
will be the same as for a single valve, and the third harmonic will be approximately
twice the value with balanced push -pull. The balance, if adjusted for maximum
signal, will not be correct for low volume, owing to the third harmonic " flattening."
(B) Screen resistance coupling (Fig. 13.45)
This is a modification of (A) being an attempt to obtain from the screen a more
linear relationship than from the plate. No comparative measurements have been
published. R, may be about 1500 ohms for type 6V6 -GT or 2500 for type 6F6 -G,
with E, = E,.2 = 250 volts -the exact value should be found experimentally ; C,
may be 0.002 µF. For better balance an equal screen resistor might be added for V2.
Ref. E10.
(C) Common cathode impedance (Fig.
13.46)
R, and R2 in series provide a common cath-
ode coupling impedance [see Chapter 12
Sect. 6(vi)]. R, may have a value of, say,
1000 ohms to give an approach towards bal-
ance, but necessarily must carry the plate cur-
rents of both valves -say, 70 or 80 mA-and
will have a voltage of, say, 70 to 80 with a dis-
sipation around 6 watts. Care should be
taken to avoid exceeding the maximum heater -
cathode voltage rating. Fig. 13.46. Push -pull circuit with
See Reference E23. common cathode impedance coupling.
586 (viii) PHASE INVERSION IN THE POWER STAGE 13.6
Fig. 13.46A. Basic circuit of Parry Cathamplifier, (B) Modified circuit, (C) With
both a.c. and d.c. balancing, (D) Keeps circulating screen current out of cathode circuit
(Ref. E30).
13.6 (ix) EXTENDED CLASS A 587
(i) Introduction
A Class B amplifier is an amplifier in which the grid bias is approximately equal
to the cut -off value, so that the plate current is approximately zero when no signal
voltage is applied, and so that the plate current in a specific valve flows for approxi-
mately one half of each cycle when an alternating signal voltage is applied.
Class B amplifiers are in two main groups- firstly Class B. (otherwise known as
quiescent push -pull, see below) in which no grid current is permitted to flow, secondly
Class B. (generally abbreviated to Class B) in which grid current flows for at least
part of the cycle.
Class B. amplifiers have inherently high odd harmonic distortion, even when the
utmost care is taken in design and adjustment. This distortion frequently has a maxi-
mum value at quite a low power output, making this type of amplifier unsuitable for
many applications. They are also comparatively expensive in that a driver valve and
transformer together with two output valves form an integral part of the stage, and
together give only the same order of sensitivity as a pentode.
The one outstanding advantage of a Class B. amplifier is in the very high plate -
circuit efficiency, although the current drawn by the driver stage should be included.
The principal applications are in battery- operated amplifiers, public address systems
and the like,
The grid bias must be fixed (either battery or separate power supply) and special
high -mu triodes have been produced to permit operation at zero bias to avoid the
necessity for a bias supply. In the smaller sizes, twin triodes are commonly used.
For the best results, accurate matching of the two valves is essential. If they are being
operated at a negative bias it is possible to match their quiescent plate currents by
adjusting the bias voltages separately. If the dynamic characteristics are not matched,
a difference of 10% in the two plate currents, measured by d.c. milliammeters under
operating conditions, will produce roughly 5% second harmonic distortion (Ref. E18).
The matching of valves is covered in Sect. 5(v).
Well regulated plate, screen (if any) and bias supplies (if any) are essential.
The rate of change of current in each half of the output transformer at the plate
current cut-off point is considerable, often resulting in parasitics
Parasitics in the grid circuit may be eliminated by the use of a driver transformer
having low leakage inductance [Chapter 5 Sect. 3(iii)a] and, if necessary, by connect-
ing a small fixed condenser from each grid to cathode -a typical value is 0.0005 p.F-
see (iv) below.
588 CLASS B AMPLIFIERS -(i) INTRODUCTION 13.7
Parasitics in the plate circuit may be eliminated by the use of an output trans-
former with low leakage inductance -see Chapter 5 Sect. 3(iii)c- together with
a
series resistance -capacitance network connected across each half of the primary or
alternatively from plate to plate.
Typical values for connection across each half of primary :
Load resistance (p -p) 6000 10 000 12 000 ohms
R 3300 5600 6800 ohms
C 0.05 0.05 0.05 µF
Typical values for connection from plate to plate :
Load resistance (p -p) 6000 10 000 12 000 ohms
R 6800 12 000 15 000 ohms
C 0.03 0.02 0.02 µF
Note --The McIntosh Amplifier may also be
Sect 8(iii).
used for Class B operation see -
(ii) Power output and distortion-ideal conditions -Class B2
(Circuit Fig. 13.31 ; characteristics Fig. 13.36.)
If the input voltage is sinusoidal, the output transformer ideal, the plate charac-
teristics equidistant straight lines, and the valves biased exactly to cut -off, the opera-
ting conditions will be-
Power output (total) Po = il,,,,(Ebb
: Em{- ) (1)
or Po = ilbr'RL = 2(Ebb -
Emr )2 /RL
where Ib,,, = maximum (peak) plate current of either valve (2)
E,,,; = minimum plate voltage of either valve
and RL = load resistance plate -to- plate.
Load resistance (plate-to- plate) : RL = 4(E5b
Load resistance per valve :
-E )-
,{ /lam
RL' = RL /4 = (Ebb Em;,,) /lam
(3)
(4)
Maximum power output is obtained when RL = 4r
i.e. when Em{ = 0.5Ebb
where ry = plate resistance of one valve.
Average plate current (each valve) la = I o,./.77.
:
-
,
0.3181 b m
0.637E b bib,.
(5)
(6)
- E,,
Plate circuit efficiency: = (1
'77 /Ebb) x 0.785 (7)
Plate dissipation : P,, = d.c. power input power output
= Ibm(0.137Ebb -1- 0.5E m, )
Power output in terms of plate dissipation and plate circuit efficiency :
(8)
The various harmonics are unpredictable, but at maximum power output H, usually
predominates, and a purely arbitrary approximation is to take
F = (1 + 0.6H,)2.
Usual values of H, at maximum power output vary from 0.05 to 0.10, so that F
varies from 1.06 to 1.12, a reasonable design value being 1.08. Thus, from eqn. (2),
P, ti 0.135Ibm2Rr (11)
Alternatively, the total power output may be calculated from a knowledge of the
0.251,0)2Rí (12)
The more accurate method is to use the composite characteristics (which are the
same as the single valve characteristics of Fig. 13.48 over the range concerned) and
to apply the " eleven selected ordinate method " [Sect. 3(iv)D] for the determination
of power output and distortion. Although harmonics higher than the fifth are ap-
preciable, it is difficult to calculate them graphically with any degree of accuracy.
PC. 13.47 () (B) 3
o
l b,t FIG.13.48
O
Ebb
; O
1
II
Ebb lrbb
.
EWOlts
II
Emin Ebb
Plate dissipation : P9 .
[Usual values of plate circuit efficiency are from 50% to 60 %.]
Ebb(0.6371bm + 0.5160) -
Po (17)
With a plate circuit efficiency of 60 %, the power output is 1.5 times the total plate
dissipation, whereas with Class A operation and a plate circuit efficiency of 25 %, the
power output is 0.25 times the plate dissipation. Thus, if in both cases the plate
590 (iv) GRID DRIVING CONDITIONS 13.7
dissipation is the only limiting factor, six times more output may be obtained from the
same valves in Class B than in Class A.
features.
In the case of a particular loadline such as
AB in Fig. 13.48, the peak grid current may
be determined by noting the intersection of
the vertical through A and the grid current
curve (shown with a broken line) corres-
ponding to the E, characteristic which
passes through A. This is shown in Fig.
13.49 where - E,, indicates the fixed grid _E Ecm *Ec
bias and the positive grid current corn-
mences to flow at approximately E, = 0. Fig. 13.49. Typical grid mutual char -
The peak grid current is I,,,, and the peak acteristic illustrating Class B operating
signal grid voltage is E,,,, corresponding to conditions.
a positive grid bias voltage E,m. The peak grid input power is then given by
Pu+n = EumIum (18)
and the minimum grid input resistance is given by
r,,,,, = E,m /lam (19)
[Note : The minimum variational grid resistance is derived from the slope of the
I, curve-it is always less than r,,,,;,,.]
The driver valve has to supply this peak power, plus transformer losses, into a half-
secondary load varying from infinity at low input levels to r,m;,, at the maximum
input. The basic driver circuit is shown in Fig. 13.50A, where V, is the driver valve
and T, the step -down transformer with a primary to half- secondary turns ratio N, /N2.
In practice, the transformer has losses, and it may be represented by the equivalent
circuit Fig. 5.9 (omitting CIO).
The efficiency is usually calculated (or measured) at 400 c/s where r r2 and Ro
are the principal causes of loss. An efficiency of from 70 to 80% at peak power is
typical of good practice. The voltage applied to the grid of V2, provided that the
iron losses are small, is approximately [Chapter 5 Sect. 2010].
E22 E,n(N2 /Ni) (20)
where 71 = percentage efficiency ± 100,
and E, = voltage applied across primary.
The load presented to V, is roughly as calculated for an ideal transformer [Chapter 5
Sect. 2(ii)] driving the same grid.
At high audio frequencies, the effects of L, and L2 (Fig. 5.9) become appreciable,
causing an additional loss of voltage during the time of grid current flow, and a tend-
ency towards instability. The instability is brought about by a negative input re-
sistance which some valves possess over a portion of the grid characteristic [Chapter 2
Sect. 2(iii)]. It is therefore important to reduce the leakage inductance of the trans-
former to the lowest possible value. Fortunately the bad effects of any remaining
leakage inductance may be minimized by connecting a small condenser across each
half of the secondary so as to resonate at a frequency about 1.5 times the highest fre-
quency to be amplified (Ref. E18). The condenser and leakage inductance form a
half -section of a simple two- element constant k type low -pass filter [Chapter 4 Sect.
8(vii)].
13.7 (iv) GRID DRIVING CONDITIONS 591
At low audio frequencies the primary inductance Lo causes loss of gain as with any
a -f transformer. It should be designed for an unloaded secondary (Chapter 5 Sect.
3)
because it is unloaded for part of the cycle. Secondary loading resistances should be
avoided with Class A triode drivers (Refs. E17, E18).
In order to avoid excessive distortion due to the non -linear grid current characteris-
tic, the effective impedance looking backwards from the grid- cathode terminals
towards the driver, must be small compared with the effective input impedance of
the valve. The " looking backwards " impedance has as its components (Fig. 5.9) :
" Driver resistance " R' = r$ -I- (Na /N1)2(r01 r1) (21)
" Driver inductance " L' = L2 + (N2 /N1)'L1 (22)
r,,,
rI It is usual to restrict the driver resistance R' to a value less than 0.2
most favourable conditions with high-mu valves and with limited grid drive, where
for the
does not differ seriously from the minimum variational grid resistance ; in other
cases it should be less than one fifth of the minimum variational grid resistance (Refs.
E2, E12). For minimum distortion, R' should be made as low as practicable. Hence
the transformer usually has a step-down ratio from primary to half-secondary.
The driver inductance," if large and uncompensated, produces an effect like a
faint high pitched hiss or scratch that rises and falls with the signal (Ref. E18).
The effect of the varying load resistance on the driver valve is shown by the curved
loadline in Fig. 13.50B. The horizontal portion corresponds to the conditions
without grid current flow, while the slope at any other point corresponds to the
variational grid resistance at that point. The broken line ioining the two extremities
has a slope corresponding to a resistance of r,,I (Nl /N,)s ; this should not be less
than 4r93. The driver valve should preferably be operated near its maximum rated
plate voltage and dissipation.
FIG.13.50 D
MINIMUM
LOAD
OPERATING
INPUT Ib POINT
111--111
FIG. 13.50A
As the output valves are driven harder, the driver valve is called upon to provide
increased power into a decreased minimum grid resistance ; this necessitates a greater
step -down ratio in the transformer. The extreme limit is when the diode line "
is reached. As a guide which may be used for a first trial, the minimum plate volt-
age may be taken as twice the " diode line " voltage at the peak current level. The
diode line is shown in many characteristic curves, and is the envelope of the character -
istics-it is the line where the grid loses control, in other words the minimum plate
voltage for any specified plate current no matter how positive the grid may be.
The cathode follower makes an excellent driver for all forms of amplifiers drawing
grid current, having very low effective plate resistance and low distortion -see Chapter
7 Sect. 2(i).
A cathode follower has been applied as a transformerless driver, one 6SN7-GT twin
triode being used as a push-pull driver to the grids of two 6L6 valves giving an output
of 47 watts (Ref. E22). However, this is not usually the most efficient arrangement
see Chapter 7 Sect. 1(ii).
-
592 (v) DESIGN PROCEDURE -CLASS B2 AMPLIFIERS 13.7
1/7.4P2/R I (23)
5. Calculate the average power input from the plate supply, using eqn. (15).
6. Calculate the plate dissipation from equation (17).
7. Check the plate dissipation and the peak plate current to see that they do not
exceed the maximum ratings.
8. Determine the maximum positive grid voltage E,m from the loadline and Ibm.
9. Determine the peak grid current I,m corresponding to extreme point of the
loadline.
10. Calculate the peak signal grid voltage : E,, == E7,m + E,.
11. Calculate the peak grid input power : P,, = E.,,1,.
12. Calculate the minimum input resistance : r,m,,, = Eom /I .
13. Repeat steps 1 to 12 for several other values of R L and select what appears to
be the best compromise, with a view to the lowest driving power
14. Assume a reasonable peak power transformer efficiency -say 70 %. This value
is used in the following step.
15. Select a driver valve with a maximum power output at least 0.9P,, under
typical operating conditions (this makes allowance for the higher load resistance).
16. Select a driver load resistance twice the typical value or at least four times the
plate resistance.
17. Draw the assumed (straight) driver loadline on the driver plate characteristics
and determine the maximum available peak signal voltage Erm.
18. Determine the transformer ratio from primary to half-secondary :
Ni /N2 = IEvm /Eom
19. Calculate the driver resistance from eqn. (21) using values for r, and r2
indicated* from similar transformers. These resistances are usually less than 1900
and 350 ohms respectively for 10 watt amplifiers, or 2700 and 500 ohms respectively
for low power battery amplifiers.
20. Check the " driver resistance " to see that it is less than 0.2r,m; for high -mu
valves or 0.1r,m; for other cases (this assumes that r,m; is approximately twice the
minimum variational grid resistance).
*Alternatively the driver transformer may be tentatively assumed to have 10%
primary and 10% secondary copper losses, and 10% iron losses. On this basis,
r2 = rum, /7 and r1 ti r,m,(N12 /N22)/9.
voltage is desirable. A high -ratio step -up input transformer is used to supply the
high peak signal voltage, nearly equal to the cut -off bias. Both the input and output
transformers should have low capacitance and leakage inductance. The output trans-
former should have its half-primary inductance, and its leakage inductance, accurately
balanced. Condensers (0.002 to 0.005 µF) are usually connected across each half
primary (Ref. E4).
(i) Introduction
Class AB, amplifiers closely resemble Class B2 amplifiers, but the valves are biased
as for Class AB, operation. Consequently they are less critical with regard to match-
ing, and the distortion which occurs in the plate circuit is much the'same as for Class
AB,. The plate circuit efficiency is intermediate between Class AB, and B, operation
-typical values for triodes are from 40 to 48 %. The variation in plate current from
zero to maximum signal is less than with Class B operation, being of the order of 1 : 2.
The matching of valves is covered in Sect. 5(v).
Pentodes and beam power amplifiers may be used quite successfully in Class AB,
operation.
Fixed bias is essential.
Type 6L6 or 807 beam power amplifiers may be used in Class AB, with plate circuit
efficiencies of about 65% (or 61% including screen losses). The peak grid input
power does not exceed 0.27 watt for power outputs from 30 to 80 watts. The total
harmonic distortion under ideal laboratory conditions is 2% at maximum signal and
less than 2% at all lower output levels (this assumes a low- distortion driver stage).
Well regulated plate, screen and bias power supplies are essential.
Parasitics in plate and grid circuits -see Sect. 7(i).
References E20, E21.
(ii) Bias and screen stabilized Class AB2 amplifier (Ref. E26)
Fig. 13.50C gives a circuit with 807 valves operated in Class AB, with stabilization
of the bias and screen supply provided by the driver valve. Two high tension sources
are used, obtained from separate windings on the power transformer, and neither
R1. This is minimized by a d.c negative feedback circuit giving several times reduction
of the effect, and is offset by a corresponding increase in 807 screen voltage.
(c) Any tendency for the 807 screen voltage to fall and for the bias voltage to increase
with rising 807 screen current due to signal input is minimized by the same negative
feedback circuit.
Referring to Fig. I 3.50C, the voltage divider from the positive side of the 600 volt
supply to the negative side of the 300 volt supply has values to provide a suitable bias
for the triode- operated 6V6 so that it draws about 45 mA with a standing plate voltage
of 250 V. A decrease in the 807 plate voltage causes the 6V6 grid to become more
negative, decreasing its plate current which also flows through R, and so decreases
the 807 bias. The low voltage power supply is deliberately given poor regulation
with resistive filtering so that the decrease in 6V6 plate current appreciably increases
its plate voltage and thus the 807 screen voltage.
The 807 grid current at large output levels increases the negative potential at C
and also at A and B and thus increases the 6V6 bias. This reduces the 6V6 plate
current which makes point C less negative and minimizes the increase in 807 bias,
and at the same time increases the 807 screen voltage thus off-setting the bias increase.
The 807 screen current rises with increasing signal input and this tends to decrease
the screen voltage and increase the 807 bias. However the increased bias is also
applied to the 6V6, reducing its plate current and thus tending to restore the 807 bias
and screen voltages to their original values.
Between no output and full output the 6V6 plate current falls from 45 mA to 25 mA
but the valve is still well able to provide the small power required to drive the 807's.
An additional feature of the circuit is that a very large input reduces the ability of
the driver to overload the output valves, and after full output is reached very effective
limiting is provided.
(iii) McIntosh Amplifier (Ref. E28)
With any push-pull amplifier in which each valve is cut -off during portion of the
cycle, some form of quasi -transient distortion tends to occur at a point in each cycle
at the higher audio frequencies (Ref. E13). This distortion is caused by the leakage
reactance of the primary of the output transformer, which cannot be reduced suffici-
ently by conventional transformer design technique. A completely new approach to
the problem is made by the McIntosh amplifier which incorporates special types of
driver transformer and output transformer, together with many other novel features.
The basic principles are indicated in Fig. 13.50D in which both driver transformer
T, and output transformer T2 have two windings wound together in a bifilar manner
so that the coupling between them is almost unity. It is claimed that it is practicable
to wind coils with a ratio of primary inductance to leakage inductance better than
200 000 to 1, whereas conventional transformers do not nearly reach the minimum
requirement (for low distortion) of 80 000 to 1. This type of transformer is cheaper
to wind than a sectionalized winding as
e used in conventional high quality trans -
e+ formers.
Each output valve works into two
primary sections, one in its plate circuit
and the other in its cathode circuit, but
these have practically unity coupling.
The effective number of primary turns
for each valve is equal to the total turns
for each of the bifilar windings. The
output transformer should therefore be
designed to have a total impedance on
each of its primaries equal to one quar-
13.500
FIG.
ter of the plate -to -plate load impedance.
Fig. 13.50D. Basic principles of In the case of the amplifier of Fig.
McIntosh Amplifier (Ref. E28). 13.50E the plate -to -plate load imped-
13.8 (iii) McINTOSH AMPLIFIERS 595
p
1 OUTPUT PLUG
p
112A %7 5NF OHMS
4
CONNECT
1 -6
TO STRAP
I15
266
n -4 I15
50,00015 I
314
56
16 1 -6 265
INPUT
32 1 -4 315
600 7 -B NONE
613,0000.
OUTPUT
SOCKET
INVERTER
AMPLIFIER
1 6J5
IMll 6B4OOOfL
L FIG. 13.50E
Fig. 13.50E. Circuit diagram of complete 50 watt McIntosh Amplifier (Ref. E28).
ance is 4000 ohms and the total impedance of each primary winding is 1000 ohms ; the
impedance from each cathode to earth is only 250 ohms. These low impedances
reduce the effects of capacitive shunting and thereby improve the high frequency
performance.
The voltage from each screen to cathode is maintained constant by the unity coupling
between the two halves of the bifilar windings, no screen by -pass capacitor being re-
quired. This arrangement, however, has the limitation that it can only be used for
equal plate and screen voltages.
The driver transformer Ti makes use of the same bifilar winding method adopted
in the output transformer. The primary impedance in this design is above 100 000
ohms from 20 to 30 000 c /s, while the response of the whole transformer is within
0.1 db from 18 to 30 000 c /s. These high performances are made necessary by the
inclusion of this transformer in the second feedback path of the whole amplifier.
The method of loading the output stage, with half the load in the plate and half in
the cathode circuit, provides negative feedback as a half-way step towards a cathode
follower. Additional feedback is achieved by connecting suitable resistors between
the cathodes of the output valves and the cathodes of the phase inverter stage. The
complete amplifier (Fig. 13.50E) has a typical harmonic distortion of 0.2% from 50
to 10 000 c /s, rising to 0.5% at 20 c/s and 0.35% at 20 000 c /s, at an output level of
50 watts. The frequency response under the same conditions, measured on the sec-
ondary of the transformer, is level from 20 to 30 000 c /s, -
0.4 db at 10 c/s and
- 0.3 db at 50 000 c /s. The phase shift is zero from 50 to 20 000 c/s, 10° at -
20 c/s and + 4° at 50 000 c /s.
596 CATHODE- FOLLOWER POWER AMPLIFIERS 13.9
The output resistance is one tenth of the load resistance, thus giving good damping
and regulation
These same principles may be applied to any type of push -pull output whether triode
or pentode, with any class of operation, A, ABI, AB2 or B.
The high power output of the McIntosh amplifier as described above is due to the
operation of the screens at voltages greater than 400 volts, which is very considerably
in excess of the maximum rating of 270 volts (design centre) for type 6L6 or 300 volts
(absolute) for type 807. It is unfortunate that the McIntosh amplifier is limited to
operation with equal plate and screen voltages, but these should always be within the
maximum ratings for the particular valve type.
See also Refs. E31, Gl.
The principle of combined plate and cathode loading is applied in some high -
fidelity Class A amplifiers, e.g. Acoustical QUAD, Refs. H4, H6.
flow through the grid- cathode circuit-see Chapter 2 Sect. 2(iii), also Chapter 3
Sect. 1(v)d, and Chapter 12 Sect. 2(iii)E, (iv)B, (iv)C, (iv)D ; Sect. 3(iv)C, (v).
Typical values (e.g. type 6V6 -GT) are :
With cathode bias 0.5 megohm (maximum)
With fixed bias 0.1 megohm (maximum).
If the heater is operated, even for limited periods, more than 10% above its average
rated voltage, the grid circuit resistance must be reduced considerably ; this holds in
automobile receivers.
If partial self-bias operation (" back bias ") is used, the maximum value of grid
circuit resistance may be found from the relation
Rom = Rol + P(Roe -
Rol)
where Roy = max. grid circuit resistance (fixed bias)
:
(1)
Ro, = max. grid circuit resistance (self bias)
and P = ratio of cathode current in output valve to the total current flowing
through the bias resistor.
In cases where the maximum value of grid resistance for a particular application
is not specified, the procedure of Chapter 3 Sect. 1(v)d should be followed, using
eqn. (6) for pentodes or tetrodes, or eqn. (7) for triodes. The value of maximum
reverse grid current (M ,,) should be obtained from the specifications or from the valve
manufacturer ; failing this, a value of 2µA may be tentatively assumed for valves of
the 6F6, 6V6 class, and 4µA for valves having higher cathode currents.
If either the value of Ro, for cathode bias, or that for fixed bias is specified, the value
for the other may be calculated from Chapter 3 Sect. 1 eqns. (8) to (13).
Voltage stabilized grid bias supplies are sometimes used (see Chapter 33).
It is sometimes convenient to have two plate voltage supplies, one to provide for
the plates of the output stage only, the other for the earlier stages together with the
screens of the output stage and grid bias. The second supply is loaded by a heavy
current voltage divider (say 150 mA total drain) tapped near the negative end and the
tapping connected to the cathodes of the output stage ; one section provides grid bias,
and the other section provides the positive potentials.
Back bias is intermediate between cathode bias and fixed bias, so far as its con-
stancy is concerned. With this arrangement, the whole return current of a receiver
-E
or amplifier is passed through a resistor between the cathodes (which are generally
earthed) and (Fig. 13.52). The value of R, is given by Ee, /I where En, is
598 (ii) GRID BIAS SOURCES 13.10
FIG. 13.52
25yF
Fig. 13.52. Circuit of portion of a receiver using back bias. V, is the power amplifier,
Vz is the rectifier.
the bias required by V, and I is the total current passed through R1. Bias for earlier
stages is obtainable from a tapping or tappings on R1. A hum filter is necessary for
each separate bias system. [Note that the curved plate of an electrolytic condenser
represents the negative terminal ; while on a paper condenser the curved plate re-
presents the outside electrode.] Back bias works best in large receivers where the
total current is much greater than the cathode current of the power amplifier.
An alternative form having fewer components is shown in Fig. 13.53. Here the
field coil is in the place of R, in Fig. 13.52. The bias on V, is equal to the voltage
drop across the field coil multiplied by R, /(R, 1 0.5 megohm). The filtering in
this circuit is not so complete as in Fig. 13.52.
FIG.13.53
Eice Coll
Fig. 13.53. Alternative form of back bias, with the speaker field coil
in the negative lead.
(i) Introduction
While every amplifier must be built up of individual stages, the purpose of the design
is to produce a complete amplifier having specified characteristics in regard to over-
all gain, maximum power output, frequency range, output resistance, distortion, hum,
noise level and special features such as overload characteristics and tone control.
A designer's job is to produce the required results at the minimum cost within any
limitations imposed by space or by the availability of valve types.
FIG . 13.54
SO
OUTPUT
3rd
2nd
4
O O
10 f0 102 'O° IO' IO'
FREQUENCY C/S
Fig. 13.54. Harmonic distortion of amplifier having type 6J7 pentode coupled to 2A3
triode loaded by a loudspeaker on a flat baffle. The distortion was measured by a wave
analyser connected across the voice coil. The bass resonance frequency is 70 c/s.
SECTION 12 : REFERENCES
(A) REFERENCES TO OPTIMUM LOAD RESISTANCE AND DISTORTION
Al. Brain, B. C. " Output chara:teristics of thermionic amplifiers " W.E. 6.66 (March 1929) 119.
A2. Notthingham, W. B. " Optimum conditions for maximum power in Class A amplifiers " Proc.
I.R.E. 29.12 (Dec. 1941) 620.
A3. Editorial " Optimum conditions in Class A amplifiers " W.E. 20.233 (Feb. 1943) 53.
A4. Hadfield, B. M. (letter) " Optimum conditions in Class A amplifiers," W.E. 20.235 (April 1943) 181.
A5. Sturley, K. R. " Optimum load, Ra or 2Ra " ? W.W. 50.5 (May 1944) 150.
A6. Sturley, K. R. (letter) " Optimum conditions in Class A amplifiers " W.E. 20.235 (April 1943) 181.
A7. Bradshaw, E. (letter) " Optimum conditions in Class A amplifiers " W.E. 20.237 (June 1943) 303.
A8. Benham, W. E. (letter) " Optmmum conditions in Class A amplifiers " W.E. 20.237 (June 1943) 302.
A9. Gladwin, A. S. (letter) " Optimum conditions in Class A amplifiers " W.E. 20.240 (Sept. 1943) 436.
A10. Date, W. H. (letter) " Optimum load," W.W. 50.7 (July 1944) 222.
All. Good, E. F. (letter) " Optimum load " W.W. 50.8 (Aug. 1944) 252.
Al2. Hughes, E. " Optimum valile load-unified treatment for different operating conditions " W.W.
51.8 (Aug. 1945) 246.
A13. Foster, H. G. " Load condiions in Class A triode amplifiers " Electronic Eng. 19.227 (Jan. 1947)
11.
A14. Sturley, K. R. (book) " Radio Receiver Design " Part II pp. 56 -64 (Chapman and Hall, London,
1945).
A1S. Warner, J. C., and A. V. Loughren " The output characteristics of amplifier tubes" Proc. I.R.E.
14.6 (Dec. 1926) 735.
A16. Kilgour, C. E. " Graphical analysis of output tube performance " Proc. I.R.E. 19.1 (Jan. 1931) 42.
A17. Lucas, G. S. C. " Distortion in valve characteristics " W.E. 8.98 (Nov. 1931) 595. Letter P. K.
Turner 8.99 (Dec. 1931) 660.
A18. Nelson, J. R. " Calculation of output and distortion in symmetrical output systems," Proc. I.R.E.
20.11 (Nov. 1932) 1763.
A19. Ferris, W. R. " Graphical harmonic analysis for determining modulation distortion in amplifier
tubes " Proc. I.R.E. 23.5 (May 1935) 510.
A20. Mouromtseff, I. E., and H. N. Kozanowski " A short-cut method for calculation of harmonic
distortion in wave modulation " Proc. I.R.E. 22.9 (Sept. 1934) 1090.
A21. Hutcheson, J. A. " Graphical harmonic analysis " Elect. 9.1 (Jan. 1936) 16.
A22. Harries, J. H. O. " Amplitude distortion " W.E. 14.161 (Sept. 1937) 63.
A23. Kemp, P. " Harmonic analysis of waves up to eleventh harmonic " Electronic Eng. 23.284
(Oct. 1951) 390.
See also References to Graphical Harmonic Analysis Chapter 6 Sect. 9(C).
(B) GENERAL REFERENCES TO CLASS A AMPLIFIERS
Bl. Espley, D. C., and L. I. Farren, " Direct reading harmonic scales," W.E. (April 1934) 183.
B2. R.C.A. " Application Note on design of audio systems employing type 2A3 power amplifier triodes "
No. 29 (Dec. 29, 1933).
B3. R.C.A. Application Note on short-cut method for determining operating conditions of power
output triodes " No. 42 (Sept. 5, 1934).
B4. R.C.A. " Application Note on use of the plate family in vacuum tube power output calculations "
No. 78 (July 28, 1937) -Reprinted Radiotronics 80 (Oct. 11, 1937) 72.
B5. Pullen, K. A. " The use of G curves in the analysis of electron -tube circuits " Proc. I.R.E. 37.2
(Feb. 1949) 210.
B6. Langford- Smith, F., and W. J. Steuart " Effect of power supply on power amplifiers "-unpub-
lished analysis.
B7. Pullen, K. A. " G curves in tube circuit design " Tee -Tech 8.7 (July 1949) 34 ; 8.8 (Aug. 1949) 33.
B8. Pullen, K. A. " Use of conductance, or G, curves for pentode circuit design " Tele-Tech 9.11
(Nov. 1950) 38.
602 REFERENCES 13.12
(C) REFERENCES TO MULTI -GRID AMPLIFIERS
Cl. Brian, W. S. " Experimental audio output tetrode " Elect. 20.8 (Aug. 1947) 121.
C2. Pickering, N. " Space -charge tetrode amplifiers " Elect. 21.3 (Mar. 1948) 96.
C3. Schade, O. H. " Beam power tubes " Proc. I.R.E. 26.2 (Feb. 1938) 137.
C4. Jonker, J. L. H. " Pentode and tetrode output valves " Philips Tec. Corn. No. 75 (July 1940) 1.
C5. Advantages of space- charge-grid output tubes " (Review of paper by N. C. Pickering) Audio
Eng. 31.9 (Oct. 1947) 20.
C6. Hadfield, B. M. " Amplifier load impedance reduction," Tele-Tech 7.5 (May 1948) 33.
See also B5.
C7. Hafler, D., & H. I. Keroes " An ultra -linear amplifier" Audio Eng. 35.11 (Nov. 1951) 15.
(D) REFERENCES TO PARALLEL AMPLIFIERS
Dl. Jones, F. C. " Parallel tube high fidelity amplifiers " Radio 29.10 (Oct. 1945) 27.
D2. Jones, F. C. " Additional notes on the parallel tube amplifier " Radio (June 1946) 26.
(E) REFERENCES TO PUSH -PULL AMPLIFIERS
El. Thompson, B. J. " Graphical determination of performance of push-pull audio amplifiers " Proc.
I.R.E. 21.4 (April 1933) 591.
E2. Barton, L. E. " High audio output from relatively small tubes " Proc. I.R.E. 19.7 (July 1931) 1131.
E3. M.I.T. Staff (book) " Applied Electronics " (John Wiley and Sons, New York ; Chapman and
Hall, London, 1943) pp. 433-448, 548 -558.
E4. Sturley, K. R. " Radio Receiver Design " Part 2 (Chapman and Hall, London, 1945) pp. 84-97.
E5. " Push-pull triode amplifiers " Radiotronics No. 79 (Sept. 1937) 64 ; No. 80 (October 1937) 78.
E6. Adorjan, P. " Power amplifier design," Radio Eng. 16.6 (June 1936) 12.
E7. Krauss, H. L. " Class A push -pull amplifier theory " Proc. I.R.E. 36.1 (Jail. 1948) 50.
E8. Cocking, W. T. " Push -pull balance " W.W. 53.11 (Nov. 1947) 408.
E9. R.C.A. Application Note No. 54 " Class AB operation of type 6F6 tubes connected as pentodes,"
(Dec. 1935).
E10. Williams, W. N. " Novel 10 watt amplifier for P.A." Radio and Hobbies, Australia (April 1942) 37.
Ell. Heacock, D. P. " Power measurement of Class B audio amplifier tubes " R.C.A. Rev. 8.1 (Mar.
1947) 147.
E12. Barton, L. E. " Application of the Class B audio amplifier to a-c operated receivers " Proc. I.R.E.
20.7 (July 1932) 1085.
E13. Sah, A. Pen-Tung " Quasi transients in Class B audio -frequency push -pull amplifiers," Proc.
I.R.E. 24.11 (Nov. 1936) 1522.
E14. Nelson, J. R. " Class B amplifiers considered from the conventional Class A standpoint " Proc.
I.R.E. 21.6 (June 1933) 858.
E15. Butler, F. " Class B audio -frequency amplifiers " W.E. 24.280 (Jan. 1947) 14.
E16. R.C.A. Application Notes Nos. 5, 14, 18, 33 and 54 giving operating data on types 79, 53, 19, 800,
54 respectively.
E17. Barton, L. E. " Recent developments of the Class B audio- and radio-frequency amplifiers "
Proc. I.R.E. 24.7 (July 1936) 985.
E48. McLean, T. " An analysis of distortion in Class B audio amplifiers " Proc. I.R.E. 24.3 (March
1936) 487.
E19. Strafford, F. R. W. " Join-up distortion in Class B amplifiers," W.E. 12.145 (Oct. 1935) 539.
E20. R.C.A. Application Note No. 40 " High power output from type 45 tubes " (June 1934).
E21. Reich, H. J. " Theory and applications of electron tubes " (McGraw -Hill, 2nd edit. 1944).
E22. Greenwood, H. M. " Cathode-follower circuits " Q.S.T. 29.6 (June 1945) 11, especially Fig. 20.
E23. Amos, S. W. " Push-pull circuit analysis- cathode -coupled output stage " W.E. 23.269 (Feb.
1946) 43.
E24. Edinger, A. " High- quality audio amplifier with automatic bias control " Audio Eng. 31.5 (June
1947) 7. Based on patent by Lincoln Walsh.
E25. Minter, J. " Audio distortion and its causes " Chapter 6. " Standard FM Handbook " (FM
Company, Great Barrington, Mass., U.S.A. 1st edit. 1946) -gives intermodulation test results
on Lincoln Walsh amplifier.
E26. Patent application, inventors C. G. Smith, W. Storm, E. Watkinson, assignee Philips Electrical
Industries of Aust. Pty. Ltd.
E27. Sturley, K. R. " Push-pull A.F. amplifiers-load curves for Classes A, B and C conditions "
W.E. 26.313 (Oct. 1949) 338.
E28. McIntosh, F. H., and G. J. Gow " Description and analysis of a new 50-watt amplifier circuit "
Audio Eng. 33.14 (Dec. 1949) 9.
E29. Parry, C. A. " The cathamplifier " Proc. I.R.E. Aust. 11.8 (Aug. 1950) 199.
Parry, C. A. "The cathamplifier " Proc. I.R.E. 40.4 (April 1952) 460.
E30. " The Cathamplifier Handbook " (Mingay Publishing Co., Sydney, Australia).
E31. Sterling, H. T. " Extended Class A audio " Elect. 24.5 (May 1951) 101.
See also A13, A18, A20, B5, C3.
E32. Peterson, A., and D. B. Sinclair "A single-ended push -pull audio amplifier" Proc., I.R.E. 40.1
(Jan. 1952) 7.
(F) REFERENCES TO COMPLETE AMPLIFIERS
Fl. Williamson, D. T. N. " Design for a high- quality amplifier " (1) Basic requirements : alternative
specifications, W.W. 53.4 (April 1947) 118 ; (2) Details of final circuit and its performance, W.W.
53.5 (May 1947) 161.
F2. Baxandall, P. J. " High- quality amplifier design-advantages of tetrodes in the output stage "
W.W. 54.1 (Jan. 1948) 2; Correction 54.2 (Feb. 1948) 71.
F3. Gibson, W. F., and R. Pavlat " A practical cathode-follower audio amplifier " Audio Eng. 33.5
(May 1949) 9.
F4. Williamson, D. T. N. " High -quality amplifier -new version " W.W. 55.8 (Aug. 1949) 282 ; 55.10
(Oct. 1949) 365 ; 55.11 (Nov. 1949) 423.
F5. " The Williamson Amplifier," Reprint of articles published in the Wireless World (Iliffe and Sons
Ltd., London).
(G) GENERAL REFERENCES
01. Sulzer, P. G. "A survey of audio- frequency power- amplifier circuits" Audio Eng. 35.5 (May
1951) 15.
Additional references H will be found in the Supplement commencing on page 1475.
CHAPTER 14
SECTION 1 : INTRODUCTION
(i) Fidelity (ii) Types of distortion (iii) Imagery for describing reproduced sound.
(i) Fidelity
True fidelity is perfect reproduction of the original. In the case of an amplifier,
true fidelity means that the output waveform is in all respects an amplified replica of
the input waveform.
In acoustical reproduction, true fidelity is achieved if the listener has the same aural
sensations that he would have if present among the audience in the studio or concert
hall.
In practice, true fidelity can only be regarded as an ideal to be aimed at.
The concept of " hole in the wall " listening put forward by Voigt (e.g. Ref. A49)
is a useful device. The listener, surrounded by his listening room, is imagined as
being in the concert hall and able to hear directly, through his open window, the
sound coming from the stage, together with echoes. Echoes from his own room will
come from all directions, but echoes from the concert hall only come through the open
window. This is probably the nearest approach to true fidelity which there is any
hope of approaching with single channel transmission. See also Ref. A50.
The purpose of high fidelity reproduction of music is to satisfy a particular listener,
who is primarily interested in the emotions arising from what he hears. The complete
process involves sensations and emotions which cannot be treated objectively and must
bring in personal preferences and differences of opinion.
It is manifestly impossible to reproduce at the two ears of the listener an exact
equivalent of the sounds which he would hear in the concert hall. The greatest
deficiencies are probably the single point sound reproducer (the loudspeaker) and the
603
604 (i) FIDELITY 14.1
(i) Non-linearity
A distortionless amplifier has an input voltage versus output voltage characteristic
that is a straight line passing through the origin (OA in Fig. 14.1), the slope of the
line indicating the constant voltage gain. When non-linearity occurs, as may happen
with curvature of the valve characteristics, the input -output characteristic becomes
curved as in OB, thereby indicating that the gain of the amplifier is a variable quantity.
606 (ii) HARMONICS 14.2
(ii) Harmonics
One effect of this non -linearity is the production of harmonic frequencies in the
output when a pure sine -wave input voltage is applied, hence the name harmonic
distortion. For example, if the input voltage is a pure sine wave of frequency 100 c /s,
the output may consist of a fundamental frequency 100 c /s, a second harmonic of
200 c /s, a third harmonic of 300 c /s, and so on. Only the fundamental frequency
is present in the input, and the harmonics are products of the non -linearity (for mathe-
matical treatment see Chapter 6 Sect. 8).
-
Fig. 14.1. Linearity characteristic (or transfer characteristic)
OA for distortionless amplifier ; OB for non -linear
amplifier.
Harmonics themselves are not necessarily displeasing, since all musical instruments
and voices produce complex sounds having many harmonics (or overtones). But all
sounds have certain relationships between the fundamental and harmonic frequencies,
and it is such relationships that give the sound its particular quality*. If certain
harmonics are unduly stressed or suppressed in the reproduction, the character of the
sound (i.e. the tone) will be changed. For example, it is possible for a displeasing
voice to be reproduced, after passing through a suitable filter, so as to be more pleasing.
The critical ability of the human ear to distinguish harmonic distortion depends
upon the frequency range being reproduced and also upon the volume. Thus, with
wide frequency range, the limit of harmonic distortion which can be tolerated is
noticeably lower than in the case of limited frequency range, for the same volume level.
If the amplifier does not amplify the harmonics to the same extent as the funda-
mental, the effective harmonic distortion will be changed. Bass boosting reduces
the harmonic distortion of bass frequencies, while treble boosting increases the har-
monic distortion of fundamental frequencies whose harmonics are in the frequency
range affected by the boosting.
The shape of the audibility curves for the ear, which has, maximum sensitivity at
about 3000 c/s (Fig. 19.7), indicates that there is some effective boosting of harmonics
compared with the fundamental for harmonics up to about 3000 c /s, and an opposite
effect for fundamental frequencies above 3000 c /s. This effect, however, is quite
small at frequencies up to 1500 c /s, at the usual maximum listening levels (say 65 to
80 db on a sound level meter). At lower listening levels the effect will be appreciable,
but the distortion is normally fairly low at these levels in any case. With fundamental
frequencies above 3000 c/s there is an appreciable reduction of harmonic distortion
at all levels (this effect has been referred to by Ladner, Ref. E14). For example,
with a fundamental frequency of 3000 c /s, at a loudness level of 80 phons, the second
harmonic will be attenuated by 9 db and the third harmonic by 15 db.
Harmonic voltages or currents are usually expressed in the form of a percentage
of the fundamental voltage or current. For example, if the fundamental voltage is
100 volts, and there is a second harmonic voltage of 5 volts, the second harmonic
percentage is 5 %.
Some harmonics are dissonant with the fundamental and (unless at a very low level)
are distinctly unpleasing to the listener as regards their direct effect, quite apart from
'These remarks apply to sounds which have slow attack and
orchestral instruments this effect is largely overshadowed by the recovery times. In the case of most
Many sounds (e.g. vowels) have an inharmonic content in additionmore prominent effects of transients.
to the harmonic content (Ref. E14).
14.2 (ii) HARMONICS 607
the secondary effect on intermodulation. The following table gives the relevant
details when the fundamental frequency (C) is taken for convenience as 250 c /s.
Harmonic Harmonic Musical Effect
frequency scale*
2nd 500 Ci
3rd 750 G
4th 1000 Cii
5th 1250 E
6th
7th
8th
1500
1750
2000
G
Cii i
- dissonant
G
dissonant
--
13th
14th
15th
3250
3500
3750
-
B
dissonant
dissonant
dissonant
16th
17th
4000
4250 -
Cin
dissonant
18th
19th
4500
4750 -
D dissonant
dissonant
20th
21st
22nd
5000
5250
5500
-
E
dissonant
dissonant
23rd 5750 -- dissonant
24th 6000 G
25th 6250 G # dissonant
'Natural (just) scale. See Chapter 20 Sect. 8(1v).
Harmonic
level
percentage
2nd
- 30
3.16
3rd
- 50
0.3
4th
-0.03
5th
70 - $
db
With push -pull operation, the third harmonic becomes the dominant harmonic
and, if the grid excitation is increased to give the same value of total harmonic dis-
tortion as with the single valve, the higher harmonics become more significant. With
Class AB, operation, the higher harmonics are still further increased and, when
weighted in proportion to the order of the harmonics, odd harmonics up to the thir-
teenth are prominent, and harmonics up to the twenty fifth may be appreciable.
tThese conditions require the distortion to be similar to that produced by a Class A triode with not
more than about 5% second harmonic distortion and without running into positive grid current.
Mess than 0.01%.
608 (iii) PERMISSIBLE HARMONIC DISTORTION 14.2
With a single pentode the weighted values of the higher order harmonics are quite
appreciable. The values of distortion -columns 1 and 2 -are derived from Olsont
for type 6F6 (Ref. E3).
Harmonic level percentage weighting weighted
(db) factors distortions
2nd -20 10 1 10%
3rd -25 5.6 1.5 8.4%
4th -35 1.8 2 3.6%
5th -45 0.56 2.5 1.4%
6th -62 0.08 3 0.24%
7th -62 0.08 3.5 0.28%
8th -63 0.07 4 0.28%
9th -65 0.06 4.5 0.27%
10th -73 0.022 5 0.11%
11th -74 0.02 5.5 0.11%
12th -73 0.022 6 0.13%
13th -67.5 0.04 6.5 0.26%
14th -74 0.02 7 0.14%
IThese are based on the ideal theoretical condition for equal intermodulation distortion ; they may
differ considerably from the true effect on the listener.
With any system of amplification other than Class B, the percentage of total har-
monic distortion, as defined in (iv) below, decreases as the power output level is re-
duced. Moreover, the percentages of the highest order harmonics decrease more
rapidly than those of the lower order harmonics, as the power level is reduced.
The use of negative feedback* merely reduces all harmonics in the same propor-
tion and does not affect their relative importance, except when the overload point is
approached.
Comparative tests have been carried out by Olson (Ref. E3) which indicate
1. That slightly greater distortion is permissible with speech than with music.
2. That a higher value of total harmonic distortion is permissible with a single
triode than with a single pentode, for the same effect on the listener.
3. That the permissible distortion decreases as the cut -off frequency is increased.
However the rate of change varies considerably with the three categories of distortion,
as given below.
4. Perceptible distortion (for definition see below).
Distortion becomes perceptible when the measured total harmonic distortion
reaches the level of 0.7% for music, and 0.9% for speech, with a frequency range of
15 000 c/s and a pentode valve. Even with the very limited frequency range of
3750 c/s a total harmonic distortion of 1.1% on music, and 1.5% on speech, is per-
ceptible with a pentode valve. The difference between pentode and triode is negli-
gibly small.
5. Tolerable distortion (for definition see below).
The measured total harmonic distortion to give tolerable distortion on music, with
either triode or pentode, increases about four times when the frequency range is re-
duced from 15 000 to 3750 c /s. The value of tolerable distortion with a pentode is
1.35% total harmonic distortion with a frequency range of 15 000 c /s, and 5.6% with
a frequency range of 3750 c/s on music, and 1.9% and 8.8% respectively on speech.
The amount of permissible total harmonic distortion with a triode is greater than that
with a pentode for the same level of tolerable distortion, in the approximate ratio of
4 to 3.
The effect of frequency range on tolerable distortion is much greater than on per-
ceptible distortion.
6. Objectionable distortion (for definition see below).
Objectionable distortion, on music, with either triode or pentode, increases about
5.5 times when the frequency range is reduced from 15 000 c/s to 3750 c/s ; the
tBy kind permission of the author and of the publishers and copyright holders, Messrs. D. Van
Nostrand Company Inc.
It is assumed that the feedback network is not frequency -selective.
14.2 (iii) PERMISSIBLE HARMONIC DISTORTION 609
effect of frequency range on objectionable distortion with music is thus greater than
its effect on tolerable distortion and much greater than on perceptible distortion. The
values of total harmonic distortion to provide objectionable distortion are 2% with a
frequency range of 15 000 c/s and 10.8% with a frequency range of 3750 c/s for
music, and 3% and 12.8% respectively for speech, with a pentode. The value of
total harmonic distortion with a triode to give the same level of objectionable distortion
as a pentode is approximately in the ratio 5/4.
Definitions -Perceptible distortion is defined as the amount of distortion in the
distorting amplifier which is just discernible when compared with a reference system
of very low distortion. Tolerable distortion is the amount of distortion which could
be allowed in low -grade commercial sound reproduction. Objectionable distortion
is the amount of distortion which would be definitely unsatisfactory for the reproduc-
tion of sound in phonograph and radio reproduction. The two latter are dependent
upon personal opinion.
Test conditions -Information derived from these tests has been tabulated below,
the conditions being
Triode single 2A3, power output 3 watts.
:
The sound was reproduced in a room with acoustics similar to a typical living room
with a noise level of about 25 db. The tests were performed with a limited number
of critical observers.
Test Results (Ref. E3)
Cut -off frequency 3750 5000 7500 10 000 15 000 c/s
Objectionable distortion
Music Triode 14.0 8.8 4.8 3.4 2.5%
Pentode 10.8 6.0 4.0 2.8 2.0%
Speech Triode 14.4 10.8 6.8 5.6 4.4%
Pentode 12.8 8.8 6.4 4.4 3.0%
Tolerable distortion
Music Triode 6.8 5.6 4.4 3.4 1.8%
Pentode 5.6 4.0 3.2 2.3 1.35%
Speech Triode 8.8 7.2 4.8 3.6 2.8%
Pentode 8.8 5.2 4.0 3.0 1.9%
Music Triode
Pentode
1.2
1.1
-
Perceptible distortion
0.95
0.95
0.75%
0.7%
Speech Triode 1.4 1.15 0.9%
Pentode 1.5 1.2 0.9%
Subjective More
Classification R.M.S. drastict
System of Sum of Weighted* Weighted Linearity
Distortion Harmonics Distortion Distortion Curvet
output. Tests should also be made on the same amplifiers for intermodulation dis-
tortion, using both r.m.s. and peak sum methods, with suitable test frequencies.
In all cases, any distortion measurements should be supplemented by
(1) An oscillographic inspection of the waveform, and
(2) An oscillographic inspection of the linearity (input versus output) characteristics
for sharp " kinks."
From these results it should be possible to derive a system of harmonic weighting
such that the weighted harmonic distortion is a true criterion of the subjective effects
of the distortion.
(i) Introduction
Intermodulation distortion is one of the effects of non-linearity when more than one
input frequency is applied. It is evident to the listener in two forms, amplitude-
modulation of one frequency by another, and the production of sum and difference
frequencies. See also Chapter 2 Sect. 9(iv).
An example of amplitude- modulation is the effect of a non -linear amplifier on the
reproduction of a choir with a heavy organ bass accompaniment. The choir is am-
plitude- modulated by the organ-a displeasing effect to the listener. This effect is
negligibly small in amplifiers if the total harmonic distortion is less than say 2 %, but
it is apparent with some loudspeakers (see Chapter 20 Sect. 7).
The formation of sum and difference frequencies is the second and more serious
form of intermodulation distortion. These frequencies are normally in- harmonic
and are the principal cause of the distorted reproduction noticed by any listener-
sometimes described as harsh, buzz, rough or unpleasant.
In actual operation there are very many input frequencies applied simultaneously,
but it is possible to make comparative tests using only two frequencies. Tests for
612 INTERMODULATION DISTORTION -(i) INTRODUCTION 14.3
this handbook unless otherwise indicated), and the stronger is usually lower in fre-
quency than the other *. The sum and difference frequencies are usually expressed
in the form of the percentage modulation of the weaker (high frequency) fundamental
voltage. For one pair of sidebands the modulation percentage is given by
2 x sideband amplitude x 100 /fundamental amplitude.
When testing, in order to make a fair comparison with single- frequency conditions,
the input should be adjusted so as to give the same peak output voltage as under
single frequency conditions. With the 4 1 ratio of the two input voltages, the
:
equivalent single- frequency power output is 25/17 or 1.47 times the indicated
power output under I.M. conditions (Ref. B14). The indicated power output
is that calculated from the r.m.s. voltage across the load resistance.
Since there are many of these sum and difference frequencies, plus harmonics of
both applied frequencies, it is necessary to take account of their magnitudes. This
may be accomplished, in accordance with this method, in special equipment for the
measurement of total intermodulation distortion by reading the r.m.s. sum of all
extraneous frequencies (Refs. B5, B10). For measurement see Chapter 37 Sect. 3(ii).
There is no simple relationship between harmonic and intermodulation
distortion. For example, it has been shown that in record manufacture the excessive
polishing of masters greatly increases intermodulation distortion but does not much
affect harmonic distortion (Ref. B6). See also Refs. A45 (correspondence), B7, B11,
B19, B20.
If all harmonics are within the frequency range of the amplifier it may be shown
that -
1. If only second harmonic is present (a condition which never occurs in practice),
the ratio of total intermodulation distortion to second harmonic distortion is
3.2 (Refs. B7, B11).
2. If only third harmonic is present (another condition which never occurs in prac-
tice), the ratio of total intermodulation distortion to third harmonic distortion
is approximately 3.84 at low values of distortion (Refs. B7, B11).
3. If the distortion is small, the intermodulation sidebands are approximately given
by (Ref. B8) :
Modulation percentage of first ti 2 x second harmonic distortion percent -
intermodulation sideband age.
Modulation percentage of second 3 x third harmonic distortion percent -
intermodulation sideband age.
Thus intermodulation distortion is automatically weighted by the order of the dis-
tortion.
As a very rough approximation, the ratio I.M. /H.D. may be taken as :
3.2 for single -ended triodes
and 3.8 for push -pull (triodes or pentodes)
where I.M. = total intermodulation distortion (r.m.s. sum)
and H.D. = total harmonic distortion,
With a -f systems with peaked response in the middle -to -high frequency range, such as high effici-
ency speech systems and hearing aids, better results are obtained by having the higher input frequency
stronger than the lower frequency (Ref. B13). The two methods of measurement will, of course, give
different results.
14.3 (ii) MODULATION METHOD - R.M.S. SUM 613
provided that the operation is restricted to the normal low- distortion region (Refs
B7, B11). The ratio tends to increase as the distortion increases. These ratios may
increase suddenly as the amplifier reaches the overload point. They are also affected
if a lower I.M. frequency is selected which gives appreciable attenuation. The
ratio is increased if there is any second harmonic cancellation in successive stages in
the amplifier or any attenuation of harmonics. In practice therefore, with many
disturbing factors, the ratio may vary from less than 1 to more than 6-see summary
Ref. B19.
The usual test frequencies are 40, 60, 100, 150 or 400 c/s and 1000, 2000, 4000,
7000 or 12 000 c /s. It is helpful to make tests at two low frequencies -one of these
(40 or 60 c /s) should approximate to the low frequency limit of the amplifier ; the
other may be 100 or 150 c /s. The distortion at the lower of these two frequencies
is largely influenced by any iron -cored transformers, while that at the higher of these
two frequencies gives a more normal overall value. The upper frequency may ap-
proximate to half the upper frequency limit of the amplifier -this does not, however,
give a stringent test of the distortion at high frequencies.
Intermodulation distortion may be visually observed with a C.R.O. (Refs. B9, B19)
-see also Sect. 3(vi) below.
Reliable intermodulation measurements may be made in the presence of consider-
able noise since the latter is excluded by filters to a greater degree than with harmonic
distortion. It may also be successfully applied in the case of restricted frequency
range where the harmonics would be outside the range of the amplifier. In such a
case it is important to remember that the distortion is not zero merely because the
harmonics are not reproduced.
Permissible intermodulation distortion
The following is an arbitrary grouping which may be useful (Modulation method
r.m.s. sum) :
-
Extremely high fidelity a -f amplifier (40 c /s) I.M. less than 2%
High fidelity a -f amplifier (40 c /s) I.M. less than 4%
Good fidelity a-f amplifier (60 c /s) I.M. less than 8%
Fairly good fidelity a -f amplifier (60 c /s) I.M. less than 20%
-a
Typical radio receiver -f amplifier only (150 c /s) I.M. less than 40%
The values quoted are only a rough guide, since so many factors are involved. They
apply to the equivalent single frequency maximum power output, and the lower test
frequency is given in brackets.
See also comments in (vii) below.
Tests with 1 : 1 voltage ratio
More stringent tests at the higher frequencies may be made by the use of a 1 : 1
voltage ratio.
(iii) Difference-frequency intermodulation method
In accordance with this method, two frequencies f, and f2, of equal amplitude, are
applied to the input and the relative amplitude of the intermodulation component
-
at the difference frequency (12 f1) is then considered a measure of the inter-modula-
tion distortion, which may be expressed as a percentage of either output voltage.
A wave analyser may be used to measure the difference frequency. Typical lower
frequencies are 1000, 5000 and 9000 c /s, while difference frequencies range from 50
to 500 c/s (Ref. A51) or even higher (Ref. B22). For the precise interpretation of
results, a fairly low difference frequency is desirable.
Equipment has been designed so that the frequencies of the two input voltages can
be varied over the a -f range with a constant difference frequency maintained between
them (Refs. B3, B22).
See also comments in (vii) below.
References A51, B3, B12, B13, B22.
If all significant sidebands are measured, the distortion may be recorded in accord-
ance with the peak sum modulation method -see (v) below.
(v) Modulation method of measurement -peak sum
This newer method of defining and measuring intermodulation distortion (Ref.
B16) has outstanding advantages over the older method based on the r.m.s. sum. This
peak sum method will measure the arithmetical sum of the amplitudes of the modula-
tion products involved, with no discrimination against the weaker modulation pro-
ducts.
In accordance with this method, the percentage intermodulation is defined as the
arithmetical sum of the amplitudes of the " in phase " modulation products divided
by the amplitude of the high frequency carrier,
i.e. percentage intermodulation Al
-
A _, A, A _' {
x 100
Where Al and A 1 are the peak amplitudes of the modulation products of frequencies
Wz ± wl,
A, and A _ z are the peak amplitudes of the modulation products of frequencies
wz ± 2w1, etc.,
A is the peak amplitude of the high frequency carrier,
col is the angular velocity of the low frequency input,
w2 is the angular velocity of the high frequency input.
The input amplitude of WI is 12 db below that of wl.
In order to measure the sum of these voltages accurately, it is necessary to use a
peak -reading voltmeter. Ref. B16 describes such an analyser primarily for use with
frequencies of 400 and 4000 c/s for testing pickups, but also capable of use with input
frequencies less than 400 or greater than 4000 c /s.
See also comments in (vii) below.
(vi) Le Bel's oscillographic method
The use of an oscilloscope to give a qualitative indication of intermodulation dis-
tortion is well known -a
good description is given in Ref. B9. A quantitative method
has subsequently been developed by Le Bel (Ref. B19). The two voltages of different
frequencies are applied to the input of the amplifier in the usual way, the output is
passed through a high -pass filter to an oscilloscope with the sweep adjusted to cover
one cycle of the low frequency. If there is no intermodulation, the high frequency
wave has constant peak amplitude and the envelope is rectangular as in Fig. 14.1A.
FIG. 14.IC
FIG. 14.IA FIG. 14.IB
Notch -- A
1 a
-1--
(A)
f Notch
(B)
Notch Depth f(tn Percent)
Fig. 14.1A & B. Envelope of oscilloscope 'mages without intermodulation (A), and
with
intermodulation (B). Ref. B19.
Fig. 14.1C. Definition of notch depth (Ref. B19).
Intermodulation results in an envelope such as Fig. 14.1B. Normally, the inter-
modulation causes a " notch " or notches in the envelope and the notch depth is de-
fined by Fig. 14.1C. The experimental relationship which has been determined
between notch depth and intermodulation (using the normal modulation method of
measurement-r.m.s. sum--as in Ref. B2) is shown in Fig. 14.1D. This curve is
practically linear below 50% notch depth (i.e. 10% intermodulation), so that a scale
may be used directly on the screen for measuring the value corresponding to each
notch. If there is more than one notch, Le Bel defines total notch depth as the arithmetical
(vi) LE BEL'S OSCILLOGRAPHIC METHOD 615
14,3
100
z
80
a
-J
ó60
o
amro.o............................0000007/(
2
et
1- 40
z
I-
Z 20
o
rc
O 300
0 50 100 150 200 250
PER CENT NOTCH DEPTH FIG. 14.ID
Fig. 14.1D. Relation between notch depth and per cent. intermodulation (Ref. B19).
FIG. 14.1E FIG. 14.IH
20
16
16
14
Fig. 14.1E. Envelope of notch pattern
with insufficient bias, for single-ended
stage (Ref. B19). FIG. 14.1F 12
10
I
8
1M
20 30 40 50 60 70
sum of the individual notch depths. Each notch in the low frequency cycle on top
and /or bottom of the envelope is counted. Some typical envelope patterns are shown
in Figs. 14.1E, F and G. If a peak instead of a notch occurs at any point, this in-
dicates regeneration.
The harmonic and intermodulation* distortion characteristics of a push -pull
amplifier are shown in Fig. 14.1H. At low output levels the intermodulation is less
than the harmonic distortion, while at high output levels the reverse is true. This
In accordance with Le Beys oscillographic method.
616 (vii) COMPARISON BETWEEN DIFFERENT METHODS 14.3
indicates that the method is a much more sensitive indication of the approach towards
the overload point than is total harmonic distortion.
See also comments in (vii) below.
4400 c /s. It would therefore be possible to reduce the frequency range of an ampli-
fier from 15 000 c/s (say the limit of hearing in a particular case) to 11 000 c/s without
a non -critical listener noticing the restriction. The same principle also holds in re-
verse, when widening the frequency range. (See also Ref. A9).
(iv) Sharp peaks
One particularly objectionable form of frequency distortion is that due to sharp
peaks in the output, as may be caused by loudspeaker cone resonances, especially
in the 2000 -3500 c/s range. Sharp troughs are relatively unimportant through their
direct effects, although they may be accompanied by poor transient response.
The effect of restricted frequency range on articulation is covered in Sect. 11.
loo
110.0
>x 50 00 200 500 1000 2000 5000 10000 20000
FREQUENCY. CYCLES PER SECOND
FIG. 14.2
íi
43d5)
---
based on Refs. A3, D6, ITOTAL NOISE LEVEL 33 db)I
D7, E9. - -- ..
111111
100 200 500 1000 2000 5000 10000 20000
FREQUENCY, CYCLES PER SECOND
FIG. 14.3
The maskingt effect of room noise is shown by Fig. 14.3 for average and for very
quiet residences. These curves may be applied directly to the hearing characteristics
of Pig. 14.2. The result is shown in Fig. 14.4 for average conditions and in Fig. 14.5
'The 10 db ratio is given in Ref. A51, quoting Refs. E7, E16. The 20 db ratio is mentioned in Sect.
6(íi), based on Refs. E15 and A3.
tThe masking level is the level of pure tones which can just
The masking level is higher than the noise (spectrum) level at thebesame
perceived in the presence of noise.
frequency by a margin of 15db
up to 1000 c /s, increasing to 28 db at 10 000 c/s (Ref. A53).
14.7 (i) VOLUME RANGE AND HEARING 621
for a very critical listener and low room noise level (these latter are the extreme con-
ditions for high fidelity). In each case the room noise reduces the effective hearing
over a frequency range from about 150 to 6000 or 9000 c /s.
The effect of noises other than room noise is covered in (v) below.
Ito
100
EE
AVERAGE
NOSE LEVEL (43 db)
WJ
W
Fig. 14.4. Effective hearing
c
characteristic for average lis- J
V. 60
tener with average noise level. J
Curves by courtesy of Jensen
N 40
Manufacturing Company. W
Z 20 ss
acto ToE
CRITICAL L STEHER
LOW NOISE LEVEL (33 db)
Fig. 14.6 gives the maximum r.m.s. intensity levels in 1/4 second intervals in critical
frequency bands for certain musical instruments and an orchestra at a distance of 20
feet from the sound source, together with the threshold of audibility curve for average
listeners, indicating that a frequency range from 40 to 15 000 c/s meets all normal
requirements for an average listener to an orchestra.*
WEE111111110111111111111
Fig. 14.6. Maximum r.m.s.
§ .o 111111M4111111111111111 intensity levels in 1/4 second
2 intervals in critical fre-
W so
quency bands for whole
L' 40 1111íi11. IIII1111IIIIIII orchestra and certain instru-
ments. After Fletcher
1111111111111111111111111 (Ref. A3).
IIIIIIII
So
111111II.IIIIIIII.000
FREQUENCY IN CYCLES PER SECOND
FIG. 14.6
Fig. 14.7 (highest curve) shows the same " orchestra " curve as Fig. 14.6, but com-
bined with the threshold of audibility curve for a very critical listener, indicating that
the maximum possible frequency range is from 32 to 21 000 c /s. These are the
most extreme conditions to be taken into account for high fidelity reproduction.
If this orchestra is reproduced at lower levels as indicated by the lower " orchestra "
curves in Fig. 14.7, the frequency range is more restricted, particularly at the lower
end. The values for both average and critical listeners are tabulated below :
Peak value 100 90 80 70 db
Average listener
Lower frequency 45 50 70 100 c/s
Upper frequency 15 000 14 000 13 000 12 000 c/s
Critical listener
Lower frequency 32 40 50 70 c/s
Upper frequency 21 000 20 000 19 000 18 000 c/s
120
CRITICAL LISTENER
100
Fig. 14.7. Maximum
r.m.s. intensity levels in d6
d1lÌiii11
PEAK 100 ORCHEST RA
60
1/4 second intervals in W
Ñ
60
4D
/m11iii._. `111
levels.
- 2010 20 50 100
200
500 1000
2p00
Sp00Iq 24000
FIG. 14.7 FREQUENCY , CYCLES PER SECOND
The bass drum, played solo and very loudly, and the loud organ have frequenci6d hs thyy vicinity of
30 c/s audible to an average listener.
14.7 (iii) ACOUSTICAL POWER AND LISTENING LEVELS 623
The curves of Fig. 14.6 and the orchestra curves used in Fig. 14.7 are only suitable
for determining the audible frequency ranges of the various sound sources -they
cannot be used for calculating the peak power.
Programme
Public Music- Engineers En-
ians gineers
Men Women Men Women
Symphonic music 78 78 88 90 87 88 db
Light music 75 74 79 89 84 84 db
Dance music 75 73 79 89 83 84 db
Speech 71 71 74 84 77 80 db
Individual variations varied from 60 to 97 db for symphonic music, but in all cases
50% of the subjects were within f
4 db of the mean. Increasing age showed a pre-
ference for lower listening levels.
'
Sound-level Normal
Loudness meter reading r.m.s. peak
Very loud 80 db 90 db
Loud (serious listening) 70 db 80 db
Medium (as a background) 55-65 db 65 -75 db
it
100dó PEAK ORCHESTRAL LEVEL
CRITICAL LISTENER
LOW NOISE LEVEL (33db) 30db
DYNAMIC
RANGE
111
20
hill
II, IM,.
.
ORCHESTRAL LEVEL
III Iiirroro
iii iii
o II DUE
NOISE
20
10 20 sD 00 500 1000 5,000 20,000
200 2,000 la000
FIG. 14.8 FREOUENCY,CYCLES PER SECOND
Fig. 14.8. Effective hearing characteristic for very critical listener with low room noise
level, showing 70 db dynamic range for orchestra in a concert hall. Minimum orchestral
level based on Ref. E13.
effect of low room noise. The 70 db dynamic range of an orchestra in a concert hall
is shown, with the minimum orchestral level of 30 db well above the masking level
of the noise. In this case it would be possible to reproduce the full dynamic range
at a level 15 db lower than the level in the concert hall, without loss due to noise.
With an average listener and average noise level, however, the reproduction could not
be more than 5 db below the level in the concert hall without loss due to noise. Any
further reduction in level will result in the masking of the softest passages by the noise.
(B) Volume range broadcast or recorded
The volume range actually broadcast or recorded is frequently less than that of the
original.
Source Volume range
F -M broadcast (direct) at least 60 db (F.C.C.)
A-M broadcast (direct) at least 50 db (F.C.C.)
Transmission over telephone lines 40 -50 db*
ditto (special lines) 50 -75 db*
Lateral -cut disc records :
4
-N---, Fig. 14.9. Loss of bass and high audio
6 frequencies caused by reproduction at a
-e co loudness of 70 phons as compared with
O
io the original at 90 phons (based on
2 Fig. 19.7).
14
On the other hand, if the reproduction is louder than the original, the bass (and to
some extent the treble) will be accentuated. This commonly occurs with the human
voice when reproduced at a high level, unless some correction is applied in the studio
amplifier.
It is for this reason that orchestral or organ music, when reproduced at normal
room volume, sounds weak and uninteresting. The application of the correct amount
of bass boosting assists considerably in maintaining realism and tonal balance. This
application of bass boosting may be made automatically with the adjustment of the
volume control, provided that maximum setting of the control corresponds to maxi-
mum volume from the loudspeaker on musical peaks (Chapter 15 Fig. 15.59). This
provision makes a second (semi-fixed) volume control desirable in reproduction from
records -it
calls for an exceptionally flat a.v.c. characteristic in a receiver, in addition
to the double volume control. If the automatic arrangement is adopted without
these precautions being taken, it is only partially effective, and many users prefer a
manually operated bass -boost. The automatic arrangement is only satisfactory when
the tonal balance is correct at the maximum setting of the volume control ; it merley
boosts the bass when the setting of the control is lowered.
Refs. A7, A29, A30, A33, A34, A56.
Probably the creation of synthetic bass, intermodulation distortion and transient dis-
tortion play important parts in producing this fatigue.
An excessively high background noise level, whether caused by hum, record scratch
or any form of extraneous noise, is also an important factor contributing to listener
fatigue.
Ref. C7.
(iv) Summing up
The tests by Chinn and Eisenberg indicated that, under their conditions of test,
the majority preferred a restricted frequency range. The tests by Olson indicated
that, under his " all acoustical " listening conditions the majority preferred an un-
restricted frequency range. Obviously the conditions of the tests are of vital im-
portance-see discussion (i) above, also summing -up by Olson, C10.
Subsequent amplifier tests by Olson have indicated that an audience was able to
perceive a value of 0.7% total harmonic distortion (see Sect. 2 and Ref. E3). A reason-
able inference is that the reduction in frequency range preferred by listeners in the
Chinn and Eisenberg tests was caused by distortion present in the sound system.
It has been pointed out (Ref. C14) in connection with the Chinn and Eisenberg tests,
that the choice between " wide " and " medium. " (at 60 and 70 db) and between
" medium " and " narrow " (at 50 and 60 db) was complicated by inadequate dis-
criminability.
Columns 2 and 3 are the results of separate tests with a high-pass filter and a low -
pass filter respectively. They cannot be combined to give a pass -band with the
limiting frequencies as tabulated, except as a rough approximation for high percentages
of articulation (> 90 %) ; in this case the two articulation values should be multiplied
together (e.g. 96% x 96% = 92% articulation approximately for a pass band from
570 to 5000 c /s).
There is no loss in articulation with a frequency range (pass band) from 250 to
7000 c /s, while the loss is quite small with a range from 570 to 5000 c /s. With con-
tinuous speech the loss in articulation is noticeably less than with syllables.
Owing to the unnaturally high -pitched tone and obvious lack of balance, these
frequency ranges are extended towards the bass for all communication purposes.
An articulation index has been proposed (Ref. D3) which may be calculated for any
known conditions, including noise. A value of 1.0 is the maximum (for perfect
articulation) while a value of 0.5 or over is entirely satisfactory.
Those who are interested in the intelligibility of speech either with or without
noise interference should read Refs. D10, D11, and D12.
Experiments have shown that intelligibility is impaired surprisingly little by the
type of amplitude distortion known as peak clipping. Conversation is possible
even over a system that introduces " infinite " peak clipping, i.e., that reduces speech
to a succession of rectangular waves in which the discontinuities correspond to the
crossings of the time axis in the original speech signal. The intelligibility of the rect-
angular speech waves depends critically upon the frequency -response characteristics
of the speech transmission circuits used in conjunction with the " infinite clipper "
A word articulation value of 97 was obtained using firstly a filter providing a frequency -
response characteristic rising 6 db/octave, followed by the infinite peak clipper,
followed in turn by a filter with a characteristic falling at the rate of 6 db /octave
(Ref. D13).
References to articulation : A3, D3, D8, D10, D11, D12, D13, D19.
If the frequencies below 500 c/s are attenuated, the distortion for a given loudness
will be much less than for a wide frequency range.
*Very loud public speaking.
630 (v) FREQUENCY RANGES FOR SPEECH 14.11
30% 45%
Total noise -30 -35 -40 -42 dbm
Conditions of test-Including microphone (or other source) and loudspeaker.
This specification does not include some very important loudspeaker characteristics,
such as frequency -modulation distortion, time delay for specified increase in sound
pressure, transient decay characteristics, and sub-harmonics (see Chapter 20 Sect. 7).
These have been omitted owing to difficulties in testing and the lack of standardized
procedure.
This raises the question whether such high standards are essential. The principal
characteristics in question are variation in output, distortion, and frequency range.
Variation in output over the frequency range
A tolerance of the order of ± 5% appears to be generally acceptable for the overall
electro-acoustical performance of a sound system.
Distortion and frequency range
The permissible non -linear distortion in a high fidelity amplifier is dependent on
the frequency range, the type of programme, the critical character of the listener
and the sound level. It has been observed that the sensitivity of the ear to distortion
in music appears to be a maximum for sound levels in the vicinity of 70 to 80 db
(Ref. A51). Hence for sound levels of the order of 90 db (sound level meter indica-
tion), somewhat higher values of distortion than those specified in (i) above would be
permissible-possibly of the order of 2% total harmonic distortion with a cut-off
frequency of 10 000 c /s, and 2.5% with a cut -off frequency of 7500 c /s.
Most direct radiator loudspeakers exceed 2% total harmonic distortion over certain
frequency bands when operated to give a sound level (meter reading) of 90 db in a
fairly large living room. However, the problem is complicated by the fact that many
types of music have maximum intensity levels in the 200 to 500 c/s frequency band
where loudspeakers often have distortion values well below their maximum values.
When the distortion in any individual case is distressing to a listener, the high cut-off
frequency should be reduced until he is relieved from the discomfort. It is better
to have freedom from discomfort than to have a wide frequency response, particularly
when listening for a sustained period. This leads to the conclusion that " the true
measure of the quality of an electro-acoustical system is the maximum bandwidth
which the public finds acceptable " (Ref. C7). Thus all high fidelity amplifiers
should have a choice of high roll-off* frequencies.
In addition, it is necessary to provide the listener with tone controls on treble and
bass, with the choice of boosting or attenuation on each (see Chapter 15). He can then
please himself on the choice of tonal balance and frequency range.
In the writer's opinion, nearly distortionless reproduction of bass frequencies up
to say 400 c/s is far more important than reproduction of frequencies above 10 000 c /s.
The importance of improved bass response in reproduced music is gradually
gaining recognition. Various methods are being adopted to produce acceptable
extreme bass response without requiring an excessively large loudspeaker. Fortun-
ately the extension of the bass range, with or without bass boosting, does not directly
result in any increase in the subjective effects of the distortion such as occurs with
extension of the treble range.
High fidelity reproduction with a level frequency response, when compared with
a distorting amplifier, seems to lack bass. If a fair comparison is to be made between
them, when operating at the same sound levels, the high fidelity amplifier will require
bass boosting.
A direct radiator loudspeaker, or a conventional horn type, cannot give high fidelity
reproduction at frequencies lower than the bass resonance, because the distortion
rises very rapidly. Thus the minimum frequency for high fidelity depends upon the
loudspeaker and any extension to lower frequencies will require another, and usually
larger and more expensive, loudspeaker.
The output transformer should, for high fidelity, have low distortion at the lowest
frequency which the loudspeaker can handle. A transformer to handle high power
at very low frequencies and low distortion is both heavy and expensive, but is not other-
wise difficult either in design or manufacture.
If the output transformer and loudspeaker are incapable of handling frequencies
below a critical value without serious distortion, it is wise to filter out these lower
frequencies and so rid the amplifier of their bad effects: It is better to have a clean
bass limited to say 80 or 100 c /s, than to have a distorted condition resulting in inter-
modulation effects over a wide frequency range.
Sharp cut -off characteristics are undesirable for fidelity. However a roll-off and sharp cut-off may
be combined to give an acceptable result.
632 (iii) THE EAR AS A JUDGE OF FIDELITY 14.12
SECTION 13 : REFERENCES
(A)REFERENCES TO DISTORTION AND FIDELITY- GENERAL
Al. Massa, F. " Permissible amplitude distortion of speech in an audio reproducing system " Proc.
I.R.E. 21.5 (May 1933) 682.
A2. Wheeler, H. A. " High fidelity problems " Hazeltine Corporation Laboratory Report, reprinted
in A.R.T.S. and P. Bulletin No. 7.
A3. Fletcher, H. " Hearing, the determining factor for high- fidelity transmission " Proc. I.R.E. 30.6
(June 1942) 266.
A4. Ebel, A. J. " Characteristics of high fidelity systems " Comm. (1) 23.4 (April 1943) 38 ; (2) 23.5
(May 1943) 24.
A5. Hartley, H. A. " Aesthetics of sound reproduction-high fidelity or judicious distortion ?" W.W.
50.7 (July 1944) 198 ; W.W. 50.8 (Aug. 1944) 236.
A6. " Aesthetics of sound reproduction -replies to queries " W.W. 50.10 (Oct. 1944) 318.
Al. Stevenson, P. " Scale distortion and visual analogies " Electronic Eng. 17.200 (Oct. 1944) 207.
A8. Hanson, O. B. " Comments on high fidelity " Elect. 17.8 (Aug. 1944) 130.
A9. " Frequency range and power considerations in music reproduction " Technical Monograph No. 3,
Jensen Radio Mfg. Co., Nov. 1944.
A10. Discussion " High fidelity reproduction of music " J. Brit. I.R.E. 5.5 (Oct-Dec. 1945) 190.
All. Toth, E. " High fidelity reproduction of music " Elect. 20.6 (June 1947) 108.
Al2. Haynes, N. M. (letter) " Listening tests " Elect. 20.6 (June 1947) 268.
A14. Williamson, D. T. N. " Design for a high -quality amplifier " W.W. 53.4 (April 1947) 118.
A15. Goodell, J. D., and B. M. H. Michel " Auditory perception " (with bibliog.) Elect. 19.7 (July
1946) 142.
A16. Amos, S. W. " Distortion in radio receivers " Electronic Eng. 14.169 (March 1942) 686.
A17. Hanson, O. B. " Down to earth on high fidelity " Radio 28.10 (Oct. 1944) 37.
A18. Van der Ven, A. J. " Output stage distortion " W.E. 16.192 (Sept., 1939) 444.
A19. Webster, N. D., and F. C. McPeak " Experiments in listening " Elect. 20.4 (April 1947) 90.
A20. " What is good reproduction
54.1 (Jan. 1948) 20.
?- Discussion by British Sound Recording Association," W.W.
A21. " Supersonic high fidelity ? " W.W. 54.1 (Jan. 1948) 23 ; Comments by " Diallist " p. 36.
A22. Morgan, R. L. " Noise measurements," Comm. 28.2 (Feb. 1948) 28 ; 28.7 (July 1948) 21.
A23. Mathes, R. C., and R. L. Miller " Phase effects in monaural perception " J. Acous. Soc. Am. 19.5
(Sept. 1947) 780.
A24. Minter, J. " Audio distortion and its causes " F.M. Radio Handbook (1946) 94.
A25. Gannett, D. K., and I. Kerney " The discernibility of changes in program band width " B.S.T.J.
23 (Jan. 1944) 1.
A26. Glover, R. P. " The problem of frequency range in speech and music reproduction" (review)
J. Acous. Soc. Am. 17.1 (July 1945) 103.
A27. Hilliard, J. K. " Audio quality- intermodulation tests " (review of I.R.E. paper) Elect. 19.4
(April 1946) 218.
A28. Lloyd, M G. and P. G. Agnew " Effect of phase of harmonics upon acoustic quality " Nat. Bureau
of Standards Sci. Paper No. 127, 6 (1909) 255.
A29. " Cathode Ray " ` Scale distortion " W.W. 41.13 (Sept. 24, 1937) 318
A30. " Cathode Ray " " Loudspeaker versus orchestra," W.W. 42.10 (March 10, 1938) 210.
A31. Hawkings, J. N. A. " Notes on wide -range reproduction " Audio Eng. 31.11 (Dec. 1947) 19.
A32. Camras, M. " How high is high fidelity ? " Tele -Tech 7.7 (July 1948) 26.
A33. " Cathode Ray " " Scale distortion again " W.W. 54.11 (Nov. 1948) 392.
A34. Le Bel, C. J. ` Psycho -acoustic aspects of higher quality reproduction " Audio Eng. 33.1 (Jan.
1949) 9.
A35. Distortion -does it matter ? " Further discussion by LE .E. Radio Section, W.W. 55.1 (Jan.
1949) 11.
A36. Quality appreciation " W.W. 55.2 (Feb. 1949) 52. Review of paper by E. A. Vetter " Some
psychological factors in quality appreciation."
A37. Licklider, J. C. R. " Effects of amplitude distortion upon the intelligibility of speech " J. Acous.
Soc. Am. 18.2 (Oct. 1946) 429.
14.13 REFERENCES 633
A38. Nixon, G. M. " Higher fidelity in sound transmission and reproduction " J. Acous. Soc Am. 17.2
(Oct. 1945) 132.
A39. What constitutes high fidelity reproduction ? " (Review of papers presented at Acoustical Society
meeting) Audio Eng. 32.12 (Dec. 1948) 8.
A40. Wiener, F. M. " Phase distortion in electro- acoustic systems " J. Acous. Soc. Am. 13.2 (Oct.
1941) 115.
A41. Powell, T. (letter) " Phase bandwidth " Audio Eng. 33.6 (June 1949) 6.
A42. Hilliard, J. K. " Phase distortion " being chapter 13 of book " Motion picture sound engineering "
(D. Van Nostrand Co. New York, 1938).
A43. Institute of Radio Engineers, U.S.A. " Standards on radio receivers -methods of testing ampli-
tude modulation broadcast receivers " (1948).
A44. Information kindly supplied by the Radio Corporation of America.
A45. Shorter, D. E. L. " The influence of high -order products in non-linear distortion " Electronic
Eng. 22.266 (April 1950) 152. Correspondence 22.272 (Oct. 1950) 443 ; 23.281 (July 1951) 278-
279.
A46. Salmon, V. (letter) " Types of reproduced sound " J. Acous. Soc. Am. 21.5 (Sept. 1949) 552.
A47. Canby, E. T. " Record revue-better audio " Audio Eng. 34.7 (July 1950) 30.
A48. Salmon, V. " Imagery for describing reproduced sound ' Audio Eng. (r) 34.8 (Aug. 1950) 14 ;
(ii) 34.9 (Sept. 1950) 14.
A49. Voigt, P. G. A. H. " A controversial idea from England " Audio Eng. 34.10 (Oct. 1950) 40.
A50. Canby, E. T. " The other side of the wall " Audio Eng. 35.1 (Jan. 1951) 24.
A51. Olson, H. F., and A. R. Morgan " A high- quality sound system for the home " Radio and T.V.
News 44.5 (Nov. 1950) 59.
A52. Souther, H. T. " Design elements for improved bass response in loudspeaker systems " Audio
Eng. 35.5 (May 1951) 16.
A53. Fletcher, H. " Auditory patterns " Review of Modem Physics 12 (Jan. 1940) 47.
A54. F.L.D. " Approach to high fidelity " W.W. 57.7 (July 1951) 289.
A55. Snyder, R. H. " Towards a more realistic audio " Audio Eng. 35.8 (Aug. 1951) 24.
Additional references will be found in the Supplement commencing on page 1475.
(B) REFERENCES TO INTIERMODULATION DISTORTION
Bl. Harries, J. H. O. " Amplitude distortion " W.E. 14.161 (Feb. 1937) 63.
B2. Hilliard, J. K. " Distortion -eats by the intermodulation method " Proc. I.R.E. 29.12 (Dec. 1941)
614 ; Discussion 30.9 (Sept 1942) 429.
B3. Scott, H. H. " Audible audio distortion " Elect. 18.1 (Jan. 1945) 126.
B4. Minter, J. " Audio distortior in radio reception " F.M. and T. 6.3 (March 1946) 24.
B5. Hilliard J. K. " Intermodulation testing " Elect. 19.7 (July 1946) 123.
B6. Roys, H. E. ' Intermodulaion distortion analysis as applied to disk recording and reproducing
equipment " Proc. I.R.E. 35.10 (Oct. 1947) 1149.
B7. Frayne, J. G., and R. R. Scoville, " Variable density recording" Jour. S.M.P.E.32 (June 1939) 648.
B8. Sturley, K. R. (book) " Radio Receiver Design " Part 2 (Chapman and Hall, 1945) pp. 79 -84.
B9. McProud, C. G. " Simplified intermodulation measurement " Audio Eng. 31.3 (May 1947) 21.
B10. Instruction Book " T.1 40 Signal Generator and T.I 402 Intermodulation Analyzer " Altec
Lansing Corporation New York.
B11. Warren, W. J., and W. R Hewlett " An analysis of the intermodulation method of distortion
measurement' Proc I.R.E 36.4 (April 1948) 457.
B12. Avina, J. " Intermodulaticn and harmonic distortion measurements " Audio Eng. 32.10 (Oct.
1948) 17.
B13. " A report on the 1949 IRE National Convention -highlights of papers presented by R. H. Tanner
and A. Peterson " Comm. 29.4 (April 1949) 10.
B14. Aston, R. H. " Intermodulation measurements on Radiotron Amplifier A515 " Radiotronics
No. 130 (March /April 1948) 34.
B15. Read, G. W., and R. R. Scoville " An improved intermodulation measuring system " Jour. S.M.P.E.
50.2 (Feb. 1948) 162.
B16. Fine, R. S. " An intermodulation analyzer for audio systems " Audio Eng. 34.7 (July 1950) 11.
B17. van Beuren, J. M. " Simplified intermodulation measurements " Audio Eng. 34.11 (Nov. 1950) 24.
B18. Berth -Jones, E. W. " Intermodulation testing " W.W. 57.6 (June 1951) 233.
B19. Le Bel, C. J. " A new method of measuring and analyzing intermodulation " Audio Eng. 35.7
(July 1951) 18.
B20. Roddam, T. " Intermodulation distortion " W.W. 46.4 (Apr. 1950) 122.
B21. Callendar, M. V., and S. Matthews " Relations between amplitudes of harmonics and intermodu-
lation frequencies " Electronic Eng. 23.280 (June, 1951) 230.
B22. Peterson, A. P. G. " Intermodulation distortion " G.R. Exp. 25.10 (Mar. 1951) 1.
(C) REFERENCES TO FRBQUENCY RANGE AND SOUND -INTENSITY PREFERENCES
Cl. Chinn, H. A., and P. Eisenberg " Tonal -range and sound intensity preferences of broadcast listen-
ers " Proc. I.R.E. 33.9 (Sept. 1945) 571.
C2. Chinn, H. A., and P. Eisenberg " Tonal -range and sound intensity preferences of broadcast listen-
ers" Discussion, Proc. I.RE. 34.10 (Oct. 1946) 757.
C3. Moir, J. " Perfect v. pleasmg reproduction " Electronic Eng. 19.22 (Jan. 1947) 23.
C4. Wells, L. V. (letter) " Acozsttc preferences of listeners " Proc. I.R.E. 35.4 (April 1947) 378.
C5. Powell, T. (letter) " Tons. -range and sound -intensity preferences of broadcast listeners " Proc.
I.R.E. 35.4 (April 1947) 278.
C6. Moir, J. (letter) " Distorticn and acoustic preferences " Proc. I.R.E. 35.5 (May 1947) 495.
C7. Le Bel, C. J. " Psycho-acoustical aspects of listener preference tests " Audio Eng. 31.7 (Aug. 1947) 9.
C8. " High audio frequencies -are they necessary, are they nice ? " W.W. 53.11 (Nov. 1947) 415.
C9. Haynes, N. M. Factors influencing studies of audio reproduction quality," Audio Eng. 31.8
(Oct. 1947) 15.
C10. Olson, H. F. " Frequency range preference for speech and music " J. Acous. Soc. Am. 19.4 Part 1
(July 1947) 549.
C11. Olson, H. F. " Frequency range preference for speech and music " Elect. 20.8 (Aug. 1947) 80.
C12. Somerville, T., and S. F. Brownless " Listeners' sound-level preferences " B.B.C. Quarterly 3.4
(Jan. 1949) 245.
C13. Chinn, H. A. and P. Eisenberg " Influence of reproducing system on tonal -range preferences "
Proc. I.R.E. 36.5 (May 1948) 572. Discussion Proc. I.R.E. 37.4 (April 1949) 401.
C14. Stuntz, S. E. " The effect of sound intensity level on judgment of tonal range and volume level"
Audio Eng. 35.6 (June 1951) 17.
634 REFERENCES 14.13
(D) REFERENCES TO LIMITED RANGE, SPEECH AND NOISE
Dl. " Radio Club hears de Rosa on synthetic high fidelity " Elect. 15.12 (Dec. 1942) 126.
D2. " The effective reproduction of speech " Jensen Technical Monograph No. 4 (1944).
D3. Beranek, L. L. " The design of speech communication systems" Proc. I.R.E. 35.9 (Sept. 1947) 880.
D4. Sanial, A. J. " Acoustic considerations in 2-way loudspeaker communications " Comm. 24.6 (June
1944) 33.
D5. Dunn, H. K. and S. D. White, " Statistical measurements on conversational speech " J. Acous. Soc.
Am. 11.3 (Jan. 1940) 278.
D6. Seacord, D. F. " Room noise at subscribers' telephone locations," B.S.T.J. 12.1 (July 1940) 183.
D7. Hoth, D. F. " Room noise spectra at subscribers' telephone locations " B.S.T J. 12.4 (April 1941)
499.
D8. Fletcher, H., and J. C. Steinberg " Articulation testing methods," B.S.T.J. 8.4 (Oct. 1929) 806.
D9. Sivian, L. J. " Speech power and its measurement " B.S.T.J. 8.4 (Oct. 1929) 646.
D10. French, N. R., and J. C. Steinberg " Factors governing the intelligibility of speech sounds "
J. Acous. Soc. Am. 19.1 (Ian. 1947) 90.
D11. Pollack, I. " Effect of high pass and low pass filtering on the intelligibility of speech in noise "
J. Acous. Soc. Am. 20.3 (May 1948) 259.
D12. Egan, J. P., and F. M. Wiener " On the intelligibility of bands of speech in noise" J. Acous. Soc.
Am. 18.2 (Oct. 1946) 435.
D13. Licklider, J. C. R., and I. Pollack " Effects of differentiation, integration and infinite peak clipping
upon the intelligibility of speech " J. Acous. Soc. Am. 20.1 (Jan. 1948) 42.
D14. Schafer, T. H., and R. S. Gales " Auditory masking of multiple tones by random noise " J. Acous.
Soc. Am. 21.4 (July 1949) 393.
D15. Hirsh, I. J., and F. A. Webster " Some determinants of interaural phase effects " J. Acous. Soc.
Am. 21.5 (Sept. 1949) 496.
D16. Bolt, R. H., and A. D. MacDonald " Theory of speech masking by reverberation " J. Acous.
Soc. Am. 21.6 (Nov. 1949) 577.
D17. Cook, E. " White -noise testing methods " Audio Eng. 34.3 (March 1950) 13.
D18. Gottschalk, J. M. " A white -noise generator for audio frequencies " Audio Eng. 34.5 (May 1950) 16.
Also A3, E4, E5, E6.
D19. Beranek, L. L., W. H. Radford, J. A. Kessler and J. B. Wiesner " Speech -reinforcement system
evaluation " Proc. I.R.E. 39.11 (Nov. 1951) 1401.
Additional references will be found in the Supplement commencing on page 1475.
(E) MUSIC AND ACOUSTICS
El. Tillson, B. F. " Musical acoustics " Audio Eng. (I) 31.5 (June 1947) 34 ; (2) 31.6 (July 1947) 25
(3) 31.7 (Aug. 1947) 34 ; (4) 31.8 (Sept. 1948) 30 ; (5) 31.9 (Oct. 1947) 25 ; (6) 31.10 (Nov. 1947)
31. Correction 32.6 (June 1948) 8.
E2. Wood, A. (book) " The Physics of Music " (Methuen and Co. Ltd., London, 3rd edit. 1945).
E3 Olson, H. F. (book) " Elements of Acoustical Engineering " (D. Van Nostrand Co. Inc., New York,
2nd edit. 1947).
E4. Fletcher, H. " Speech and hearing " (MacMillan and Co. Ltd. London, D. Van Nostrand Co. Inc.,
New York, 1929).
E5. Fletcher, H. " Some physical characteristics of speech and music," B.S.T.J. 10.3 (July 1931) 349.
E6. Snow, W. B. " Audible frequency ranges of music, speech and noise," B.S.T.J. 3.1 (July 1931) 155.
E7. Sivian, L. J., H. K. Dunn and S. D. White " Absolute amplitudes and spectra of musical instru-
ments and orchestras," J. Acous. Soc. Am. 2.3 (Jan. 1931) 330.
E8. Fletcher, H., and W. A. Munson " Loudness, its definition, measurement and calculation " J.
Acous. Soc. Am. 5.2 (Oct. 1933) 82.
E9. Fletcher, H.,, and W. A. Munson " Relation between loudness and masking " J. Acous. Soc. Am.
9.1 (July 1937) 1.
E10. Steinberg, J. C., H. C. Montgomery and M. B. Gardner, " Results of the World's Fair hearing
tests," B.S.T.J. 19.4
Ell. Trimmer, J. D., and F.(Oct. 1940) 533.
A. Firestone " An investigation of subjective tones by means of the steady
tone phase effect " J. Acous. Soc. Am. 9.1 (July 1937) 24.
E12. Goodfriend, L. S. " Problems in audio engineering " Audio Eng. (1) 33.5 (May 1949) 22 ; (2) 33.6
(June 1949) 15 ; (3) 33.7 (July 1949) 20.
E13. Bell Laboratories Record, June 1934 page 315.
E14. Ladner, A. W. " The analysis and synthesis of musical sounds " Electronic Eng. 21.260 (Oct.
1949) 379.
E15. Canby, E. T. " Record revue " Audio Eng. 34.5 (May 1950) 24.
E16. Wolf, S. K., and W. J. Sette " Acoustic power levels in sound picture reproduction "
Soc. Am. 2.3 (Jan. 1931) 384. J. Acous.
CHAPTER 15
SECTION 1 : INTRODUCTION
(i) The purpose of tone compensation (ii) Tone control (iii) General considerations
(iv) Distortion due to tone control (v) Calculations involving decibels per octave (vi)
Attenuation expressed as a time constant (vii) The elements of tone control filters
(viii) Fundamental circuit incorporating R and C (ix) Damping of tuned circuits
(x) Tolerances.
635
636 (i) THE PURPOSE OF TONE COMPENSATION 15.1
4 : 1 0.50
5 : 1 0.43 ,, ,,
6 : 1 0.39 ,,
7 : 1 0.36
8 : 1 0.33
10 : 1 0.30
1000 to
Example -A change of 0.7 db occurs with an increase of frequency from
1250 c /s. What is the rate of change in db /octave ?
Rate of change = 0.7 x 3.10 = 2.17 db /octave.
on page 368.
Note. A table relating frequency ratio, octaves and decades is given
638 (v) CALCULATIONS INVOLVING DB /OCTAVE 15.1
(B) To convert db/octave to db/specified frequency ratio
Frequency Multiply db /octave
ratio by factor
1.2 : 1 0.263 to give db /specified frequency ratio
1.25 : 1 0.322
1.33: 1 0.412
1.5 : 1 0.585
2 : 1 1.00
3 : 1 1.59
4 : 1 2.00
5 : 1 2.33 ,,
6 : 1 2.59
7 : 1 2.81
8 : 1 3.00
10 : 1 3.33
Example-What is the change in level for a frequency ratio of 1.5 to 1 when the
rate of change is 6 db per octave ?
Change in level = 0.585 x 6 = 3.51 db.
Ì
In the particular case where T = 75 microseconds and f is expressed in Kc /s,
this becomes
attenuation in db = 10 log,,, (1 + 0.222f'). -
A de- emphasis curve for a time constant of 75 microseconds is given in Fig. 15.1.
This curve, like all other curves of this class, has an ultimate rate of attenuation
6 db /octave.
of
o 1N 11 ii11....111111111Ì\\
11MI
Ì M
ÌI
FIG IS.
11
I
11/111
11
101111111\11/1111111111
mmII o solllllllillli
11 \E ÌÌ
:Ml
III 11C:
11
llÌl
11
lllllE111M1111M1 1111
nmutn 1111 uI111111111 111
ii1111I1!lÌl1i
1
11111111\\
111ii
MOIiwi
5 _--:\n111u
11 Ì
n1u1111111
noun
1111111
u11n11
11
====iiw
emu
.gn11111
I11l1I1111111nn. Ì iiiiiiiii ii\ouuliii
i!lliiii!liili!!!!!!llBEi
1111
1111
111
11
11
111
lll::..1
Il 111111II1111111
1
1111111111111111111nu1/111IMO 11111
1
11
1\\IIIIII
W
o
1111u11III
=1111111M
-10 1111u11111111
l lll111 l
1111111\\u11111111\\ 1
111111
1 111
11\1
l
111111111111
1111111u111111111u1u1111
11
==
11111l'
I11 ./ 111
1
IMIMII
OEM
1
11ou 111
u1 1111a1 u1111
.. 11111111 nn =MO
11I11
= 11111uuluMp1111.11111111\ 1111111
1111 1111111M
M11 o 11
11111111111I111111p 111
1 1
E ou111uunn1nn1111nun1n111
111
111111\11Inn1..\11III1u111I111
I
I111
1
11
11
mC
Ein 11 1111.1i11
11111111 11I1\
-IS
EOM
111
11M1
o Q 111111
1
11Il1111 1111\
1111
11 1
1 1 1
== 1111111
i
101-1 M111101
11\1111111111 l 11\\
111
d°'-IQLOy1O(I+WY)
T. w271t
11 11
unl
1
111
11u1
111/n11111
11!!nn11R
11111M 11 u11111:
liti
CR u1 11111
11.1
,.''; =\ : ! ilii!lllll=-iiiiiiï 1\III..
11111
-2%
111. lllil;;;IIC:
E::::::!ll==
Ö Ö
I;JOO
100 FREQUENCY C/S
Fig. 15.1.
75 M CROSECONO GE- EMPHASIS
De- emphasis curve with time constant of 75 microseconds, as used in
CURVE ;000
F-M receivers.
(vii) THE ELEMENTS OF TONE CONTROL FILTERS 639
15.1
The resistor Ri is intended to reduce distortion through low a.c. load impedances
connected to the plate circuit of Vi ; it may be omitted if the stage is operating at a
low level or if R, is not less than twice the plate resistance of V,. A suitable value is
not less than 2r, in the case of a triode or not less than 2RL in the case of a pentode.
The resistor R, also has the effect of decreasing the gain at high frequencies more
than at low frequencies, thereby increasing the ratio of gains at low and high fre-
quencies.
It may be shown that
Gain at high frequencies - (1 + r, /RL)(1 + Rt /R') + r, /R'
(1)
/r + /RL)R, + - /R2
+ ?
R (1 1 1 R4
B = ratio of gains = 1 (3)
R4 /7., + /RL)(Rt +
(1 1 + Rz) 1
quencies.
Typical values of resistors are given below :
General case Valve type 6J5 Valve type 6AV6
Plate resistance r, 7 700* 62 500 *ohms
RL 5 r, 50 000 220 000 ohms
R, 2 r, 20 000 120 000 ohms
Rz 20 r, 500 000 1 000 000 ohms
R4 for 6 db boost 2.4 r, 25 000 140 000 ohms
R,, for 10 db boost 1.1 r, 11 600 67 000 ohms
R4 for 15 db boost 0.53 r, 5 500 30 000 ohms
Gain at high frequencies :
6 db boost 0.36 it 7.6 30 times
10 db boost 0.22 p. 4.7 19 times
15 db boost 0.13 p. 2.6 10 times
A typical example incorporating a pentode is Fig. 15.37A (Sect. 8).
The procedure in design is
(1) to determine the desired total bass boost in db from high frequencies to zero
frequency.
(2) to select a suitable valve type.
(3) to assume suitable values for RL, R, and R,.
(4) to calculate the value of R4 to give the desired boost, using eqn. (4).
(5) to determine a suitable value for C (this may be done by following Chapter 12
Sect. 2(ii), assuming R, R, to be the effective value of Rp,).
(6) to determine the value of Cy to give the required position of the boosting curve
on the frequency characteristic, as described below.
Frequency characteristics
The shape of the frequency characteristic is determined solely by the amount of
the total boost in db (Fig. 15.4). It is convenient to consider the frequency at which
each characteristic reaches half the total boost in decibels, and to call this the " half-
'As normal Class Ai amplifier. Higher values are to be anticipated for resistance- coupled conditions.
642 (ii) (A) CONVENTIONAL BASS BOOSTING CIRCUIT 15.2
16
12
o
0.1 2 3 .4 s .6 .8 2 3 4 5 6 8 10 20 30 40 50 60 80 100
RIO
Fig. 15.4. Bass boosting frequency characteristics. These are quite general and may
be applied to any r.c. boosting provided that the value of the total boost in db is
known. See text for value R/X (Ref. 10).
boost point." It will be seen that the frequency of the " half boost point " (f d) in-
creases as the total boost increases.
Total boost 20 15 10 6 3 db
Half-boost 10 7.5 5 3 1.5 db
Boost ratio (B) 10 5.6k 3.16 2.0 1.41
R/X for half -boost* 3.16 2.37 1.78 1.41 1.19
C2f 3.16 /wR 2.37/wR 1.78/ wit 1.41 /wR 1.19 /wR F
* (R /X) = './B at half -boost point.
t C2 = 1 /(wX) = (R /X) /(wR) and w = 217f0
where R = [r,RL /(r, + RL)] + R, + R4 (Fig. 15.3)
and f 6 = frequency at half-boost point.
The slope of the frequency characteristic at the half-boost point, which is
very nearly the point of maximum slope, is approximately :
Total boost 20 15 10 6 3 db
Slope at half-boost point 4.9 4.1 3.0 2.0 1.0 db /octave
If a tangent is drawn to the curve at the half-boost point, it will be seen that the
slope of the frequency characteristic does not fall off to any appreciable extent from
the half-boost point to the three- quarters boost point (i.e. 75% of the total boost in
db). The falling off in slope does not cause a difference of more than 1 db between
the curve and the tangent up to 90% of the total boost in db, the reading of 90%o being
taken on the tangent.
This circuit may be applied to continuously variable bass boosting by using a variable
resistor in place of R4, but this has the effect of varying the amplifier gain at the middle
audio frequencies and hence varying the apparent loudness. One possible alternative
which avoids this defect, is to put a high variable resistor (say 0.5 megohm logarithmic
taper) across C2. This method gives a variation in total boost without much change
in level of the middle frequencies, but the shapes of the frequency characteristics
are not the most desirable for tone control purposes -Fig. 15.5 ; see also (D) below.
15.2 (ii) (A) CONVENTIONAL BASS BOOSTING CIRCUIT 643
Several values of C2 may be selected by means of a tapping switch, leaving the total
boost unchanged, but this merely moves the frequency characteristic horizontally
and is not satisfactory, on its own, for tone control purposes.
FIG, 15.5
1.0 ,
20
\
1
P,50,00011
R4 spoon
R, ßMn R, Cj 0INF
08 18
Fillir INPUT
Il5p,000ft\' - OUTPUT
_ R
Rs2i
R,.Iqo00n
Cl . Ri
15.6
02 6
1
Rà 011 O
O
M1
FREQUENCY C/5
Fig. 15.5. Frequency characteristics with conventional bass boosting circuit, having
variable resistor across C2 (Ref. 9).
A modification which has some advantages is shown in Fig. 15.6. Here R2 has been
moved from the grid circuit of V2 to reduce the shunting effect on R4 and C2. The
total grid circuit resistance of V2 is (R, + R2). Ref. 11.
References to conventional bass boosting circuit -9, 10, 11, 20, 23, 38, 51, 55.
(B) Plate series compensation (Fig. 15.7)
This is a simple method of providing a fixed amount of bass boosting which uses
the plate decoupling circuit. It is generally limited to use with r.c.c. pentodes.
With the values of components shown, the frequency response curves for two values
of C are given in Fig. 15.8.
References to plate series compensation : 38, 52, 53.
FIG. 15.7
B+
BASS 800511
10
ú5
sto
100 1,000 10,000
FREQUENCY (CA) FIG. S 8
Fig. 15.8. Frequency characteristics with the circuit of Fig. 15.7, V, = 6J7,
Ry =0.05 MQ, R1= 0.2 MQ, Rk =2000 Q, Ck =25 µF, R,= 1.5 MQ,
C, = 0.1 µF, C 0.02 µF, R, = 1 MQ.
(D) Improved variable bass boost
A method for obtaining continuously variable bass boost having improved shape of
the frequency characteristics, is incorporated in the bass and treble boost circuit of
Sect. 8(x)K and Figs. 15.55 and 15.56A.
(iii) Methods incorporating resonant circuits
Resonant circuits give greater flexibility than those incorporating only capacitance
and resistance. They are limited, however, to values of Q not greater than 1 (Ref. 54)
and preferably not greater than 0.7-see Sect. 1(ix).
(A) Parallel resonant circuits
A parallel resonant circuit, which may be connected in the plate circuit of a r.c.c.
pentode (Fig. 15.9) provides boosting in the vicinity of its resonant frequency, with
maximum boost at its resonant frequency. For bass boosting the resonant frequency
is often between 50 and 120 c /s, although higher and lower frequencies are sometimes
adopted.
The inductor L, has to carry the greater part of the plate current of V, (several
milliamperes in a typical case) and should have a butt joint or air-gap to reduce the
effect of the plate current on the inductance.
FIG. 15.10
FIG, 15,9
B+ 5+
Fig. 15.9. Bass boosting with parallel resonant tuned circuit in plate circuit.
Fig. 15.10. Bass boosting with parallel resonant tuned circuit in grid circuit.
In all circuits of this type, it is advisable to select a high L/C ratio in order to give
the highest gain for a fixed amount of boosting.
Q = woCR, (5)
For the limiting value of Q = 1,
15.2 (iii) (A) PARALLEL RESONANT CIRCUITS 645
R, 1 /a0C (6)
where R, = equivalent total shunt resistance in ohms
of
= 2710
= frequency of resonance in c/s
LCwa2 = 1
L = inductance in henrys
and C = capacitance in farads.
Typical values for a resonant frequency of 70 c/s are L, = 51.5 H, C1 = 0.1 µF,
:
loudspeaker, and decreasing the stage gain. This device is usually avoided on ac-
count of its short -comings.
FIG. 15.11
I
1I111 1111111 11111
-11111mIlllllIIIIII
IIIII1111111111
`IIII111111111I
11111IIIII11II11
MII IlìllMMIII111111111=1011111
1=1/ ..:!11MINIMIIIIIIIMM11111
FIG. 15.12
I1IIIIIIp1111I
11111/,p11o!MM11111MM11111 in Fig. 15.11. Condition
1IMO
/I11111IMMI11IIIMMI11111
1lylll'1111111111
\1111111111
'/IÌ11Ì\\1111111111
m IWIUm\Illllmom11111
(1) tone control max. resis-
tance (2) three quarters
max. (3) half max. (4) one
I1111
!IIIIYI11111:
EmIi\hì:.!
11111IIIII1111
IIIII
\111111111
.\.!111 :!!!
quarter max. resistance.
The curves have been super-
imposed to coincide at
immIlllllmmuIIIImmIIIIII 1000 c /s.
s ...IllllmmIIIImsom1111111
io
1
50 00 500 1,000
FREQUENCY (C /5) qo0
Another possible circuit is Fig. 15.13 (Ref. 36) in which the feedback is taken from
the plate to the grid through C and R,. The response is given by
(1- a)(1 +Jwo /(0)
where coo = a /[CR2(1 -
a)]
(8)
and a = proportion added to the original signal by the setting of VR,
Typical values are R, = R2 = 1 MS?, R = 0.22 MS2, C = 200 µµF, giving
:
coo/27r = 250 c/s when a = 0.25. The resistance of VR, and VR, should be high,
preferably with VR, greater than VR, and R, = VR,.
Still another possible circuit is Fig. 15.14A in which feedback is taken from the
cathode of the second stage to a resistor at the earthy end of the volume control feeding
the grid of the first stage (Ref. 65). It is stated that this gives 10 db of negative feed-
back at middle and high frequencies and 3 db of positive feedback at low frequencies,
equivalent to 13 db bass boost.
15.2 (iv) (C) DECREASED FEEDBACK AT BASS FREQUENCIES 647
Fig. 15.13. Single stage amplifier giving bass boosting with adjustable frequency
characteristic (Ref. 36).
Fig. 15.14A. Two stage amplifier giving 13 db effective bass boost (Ref. 65).
In Fig. 15.14B a capacitance is inserted in the feedback network from the plate
of V2 to the plate of V1, with r.c. coupling between the stages. This capacitance
provides a fixed amount of bass boosting which is generally limited to about 6 db maxi-
mum (Ref. 68).
FIG. 15.148 .0O5yF 0.75 A411
A circuit using an inductor in the feedback network is Fig. 15.54 -this may be
used for bass boosting only by omitting C, and R4.
A particularly interesting circuit is Fig. 15.15 which incorporates a parallel-T net-
work in the feedback loop (Ref. 57). In the figure the parallel -T network is tuned
to 80 c /s, but it may be tuned to any other desired frequency. The frequency charac-
.
teristic is shown in Fig. 15.16.
There have been many other applications of negative feedback to provide bass
boosting which have appeared in articles and patents, too numerous to give in detail.
i
Some are given in the References (Sect. 15).
...
See also Sects. 4, 8, 9, 10 of this chapter, and Chapter 17, Sect. 5.
iii--..iii
FIG. 15,16
::.
=mamas _-_-_iiiii =
+20
80 ti 80 -180V
..iiiMiiii
2 M11
OINF
-1P411
2M11
OJNF
j`.02NF
C51F
6SV
I05NF
N- igh 2
OUTPUT
1
+10 wu
wMMM..iii
uo =MN =MENE
°ií!_M==%1111.9M
Mou Z.
1=0.MMMI\IIIMIWI
Miiii
::.M ::::
J
MINI
...M
=miM_"graE
m
IMMMIIME Y _."'nM=M11
M..
M.
MMIIMMININIIIMM
2MA
INPUT
IMA 2,20011
10 ::'_
.M
_==
=_Miii MM==iiiB
FIG. 15.15 40 6(G 00 200 500 000 2,000
FREOUENCY (C/5)
OSM71
INPUT
Fig. 15.17. Circuit providing
bass boosting due to a negative
resistance characteristic of the
valve (Ref. 2).
D6
40
A Dashed cc.r
-
,.`
30 1C7 how effect of
rd-
au.
I,IM%.
rlio.',
Rc.d
Z, 9^ D6
20
10
,\
A'CccD.Sµf
.IiiiiM_
A:C. =0.05uI--- -,
30
20
10
-10 I 1
Fig. 15.18 (A and B). Bass boosting characteristics produced by circuit of Fig.
15.17 (Ref. 2).
15.3 BASS ATTENUATION -(i) GENERAL 649
11= NNW
r ouuor,
:
o.nuun
MON 11ii::::: IMO i
-12
10 20 50 100 200 SOO 1,000
FREQUENCY (C /5') FIG. 15.19
Fig. 15.19. Bass attenuation versus frequency characteristics : (1) S.ngle grid
coupling condenser (R = 1 megohm, C = 0.005 µF) ; (2) Two stages each with
identical values of time constant RC ; (3) Three stages each with identical values
of time constant ; (4) R.c.c. pentode with by-passed cathode resistor (6J7, R L = 0.25
MQ, Rk = 2000 d2, Ck = 0.5 µF) and screen adequately by-passed.
-
quency limit of the whole amplifier. In other cases, the cut -off frequencies may be
made to coincide for more rapid attenuation however see Sect. 1(x).
If a transformer primary is resonated with the coupling capacitance, in the case of
parallel -feed, the ultimate rate of attenuation below the resonant frequency is 12 db/
octave.
Owing to the wide tolerances in inductance normally occurring in quantity pro-
duction, and to the variation caused by the d.c. and signal currents, it is not con-
sidered good practice to use the transformer primary resonance for tone control pur-
poses ; a further contributory factor is the increased valve distortion caused by a low
impedance load-see Chapter 12 Sect. 4(xii).
15.3 (v) REACTANCE SHUNTING 651
If it is desired to vary the cut -off frequency of an a -f transformer with parallel feed,
a variable resistor may be inserted to adjust the equivalent source impedance (R, in
Fig. 15.20). The low frequency attenuation (Ref. 51) is
attenuation in db = 10 log10 [1 + (R/X02] (1)
where R = R, [r,RL /(r, + RL)]
X1 = wL1
and L1 = primary inductance in henrys.
(vi) Bass attenuation by negative feedback
A network may be inserted in the feedback loop which reduces the feedback voltage
at middle and high audio frequencies, thus effectively giving bass attenuation (Fig.
15.21, Ref. 23). Condenser C, is merely for blocking the d.c. path, while R,, R,
and C1 form a voltage divider across the output voltage of V,. The voltage drop
across R, and C, is applied across R,, thus applying the feedback voltage between V1
cathode and earth.
5+
FIG. 15.20
FIG .15.21
Fig. 15.20. Amplifier, giving variable low frequency cut -off using a-f transformer
and parallel feed.
Fig. 15.21. Amplifier providing bass attenuation by negative feedback (Ref. 23).
FIG. 15.22
Fig. 15.22. Amplifier providing bass attenuation by Parallel-T network (Ref. 59).
Fig. 15.23. Frequency characteristics of amplifier Fig. 15.22 when V, = typical
r.c.c. pentode, C1 = 0.001 µF, C2 = 0.004 µF, C, = 0.02 µF, R, = R2 = 2 MQ,
RL = 0.2 MQ, R, = 0.5 MQ. The frequency for infinite attenuation is 56 c /s.
Switch position (1) Fidelity (2) Bass attenuation (Ref. 59).
652 (viii) BASS ATTENUATION USING CONSTANT K FILTERS 15.3
(viii) Bass attenuation using Constant k filters
If it is desired to achieve a rapid rate of attenuation and a sharp cut-off, it is neces-
sary to use some form of correctly designed filter incorporating both inductors and
capacitors. The constant k filter [Chapter 4 Sect. 8(vii)] in its simplest form only
incorporates 2 condensers and 1 inductor. An example of its incorporation in an
amplifier is Fig. 15.24 (Ref. 60). This filter is fairly non -critical as regards the induct-
ance of L, a 2 : 1 variation still giving a reasonable characteristic with, of course, a
change in cut-off frequency. The frequency characteristic obtained with balanced
loading is Fig. 15.25 Curve 1 ; when the source impedance is 50 000 ohms and load
0.5 megohm the frequency characteristic is curve 2, showing the typical rise above
0 db due to unbalanced conditions ; curve 3 shows the effect of increasing the
resistance of the inductor L.
If sharper attenuation is required, two or more filter sections may be incorporated.
4 4. FIG. IS. 25
z_ P
4
á
12
-16
100 270 SOO bOOO 2p00
FREQUENCY (C /S)
Fig. 15.24. Amplifier providing bass attenuation by constant k filter (Ref. 60).
Fig. 15.25. Frequency characteristics of amplifier Fig. 15.24 when V1 = typical
r.c.c. pentode, C1 = C, = 0.002 µF ± 5%, Cg = 1µF, L = 70 H with Q = 6 ;
Curve (1) R1 = R, = 0.25 MQ ; Curve (2) R1 = 50 000 Q, R, = 0.5
(unbalanced) ; Curve (3) as Curve 2 with additional resistance in inductor to MQ
give
Q = 3 (Ref. 60).
(ix) Bass attenuation using M- derived filters
One T section of an M- derived filter may be used to provide a more rapid attenua-
tion than that of a Constant k filter -see Chapter 4 Sect. 8(viii). The filter itself
is shown in Fig. 15.26 and the frequency characteristics for balanced and unbalanced
impedances in Fig. 15.27. Here, also, a considerable tolerance is permissible with the
value of L and in the degree of mismatching. With mismatching, the value of R
may be taken as 1/R1/2, (Ref. 60). 410 FIG .15.27
z o
u
?-10
Q
W
-30
-40
50 100 200 SOO
FREQUENCY (C /S)
Fig. 15.26. One T section of a series M- derived filter.
Fig. 15.27. Frequency characteristics of filter Fig. 15.26 ; Curve (1) R, = R,
0.25 MD, C, = 0.0023 µF, C, = 0.024 µF, L = 74 H; Curve (2) R, = 50 000 Q,
R8 = 0.5 MO (unbalanced), C1 = 0.0058 µF, C, = 0.035 µF, L = 73 H(Ref. 60).
15.4 (i) STEPPED CONTROLS 653
20,000n
8yF /?LF
L1
T 0.002SHF
FIG. 15.28 +250V toss car Bass Boost
Fig. 15.28. Step -control giving bass boosting and bass attenuation.
We may determine the value of R, for any desired value of B from the equation
R, = (B - 1)(R2 + R ")
In the case of pentodes, R" approximates to RI,.
(4)
The frequency characteristics for treble boosting have been plotted (Ref. 51 Fig.
9.18 ; also Refs. 9,55) but for most purposes it is sufficient to work on the " half-
boost " points as for bass boosting, the curves being approximately symmetrical S
curves about these points. The values of R/X for the half-boost point are identical
to those for bass boosting, and not the inverse as might be expected.
Total boost 20 15 10 6 3 db
Half boost 10 7.5 5 3 1.5 db
Boost ratio (B) 10 5.62 3.16 2.0 1.41
(R, /Xi) for half boost* 3.16 2.37 1.78 1.41 1.19
C,t 3.16 /wR, 2.37/wR, 1.78 /wR, 1.41 /coR, 1.19 /wR, F
*(R, /X,) = VII at half boost point. X, = 1 /wC,
tC1 = 1 /wX, = (Rl /Xl) /wR, w = 2arfo
where R, and C, are as shown in Fig. 15.29
and fo = frequency at half-boost point.
A variable control may be achieved in various ways (Ref. 9) :
(a) Varying C1-This can only be done in steps. The result is to change the
frequency of the half -boost point (see table).
(b) Varying Ri-This varies the amplification for middle frequencies, and is un-
satisfactory for tone control purposes.
(c) Adding a variable resistor in series with C1 -This is fairly effective but the shapes
of the frequency characteristics are not ideal for tone control. It is used in a slightly
modified form in the combined bass and treble boost circuit of Fig. 15.46.
References to conventional treble boosting circuit : 9, 17, 48, 51, 55.
This forms part of the usual " tone-control " in many radio receivers of the less
expensive class, as illustrated in Fig. 15.31. The control is in the form of a variable
resistance in series with the capacitance which limits its attenuation. A suitable
656 (ii) ATTENUATION BY SHUNT CAPACITANCE 15.6
maximum value for the resistance is 10 times the load resistance. In some cases a
fixed capacitance only is used-typical values are from 0.005 to 0.02 µF for a loud-
speaker load resistance of 5000 ohms. In this application the " effective resistance "
for the calculation of the attenuation will be the loudspeaker impedance in parallel
with the output resistance (R o) of the amplifier.
Tone control by shunt capacitance, when used with discrimination, is fairly satis-
factory with a flat amplifier operating from an equalized source. It is far from satis-
factory in selective radio receivers in which the high audio frequencies are already
heavily attenuated ; this effect is minimized when variable selectivity i -f amplifiers
are used. One reason for the popularity of shunt capacitance tone control is that the
intermodulation frequencies produced by the distortion in the pentode or beam power
amplifier valve are much reduced, and listening thereby made less fatiguing. If
methods are taken to reduce the distortion, e.g. by negative feedback, the tone control
may preferably take the form of bass boosting.
In order to achieve the desired treble attenuation, a shunt capacitance may be placed
across the source (radio receiver, pickup or microphone), from any grid to earth, from
any plate to earth or across the output terminals of the amplifier. In each case the
effective resistance (R in eqn. 4 of Chapter 4 and Fig. 4.38) is the resultant a.c. re-
sistance between the points across which the shunt capacitance is connected. In
the case of a r.c.c. pentode it is approximately the load resistance in parallel with the
following grid resistor, and it does not make any appreciable difference whether the
capacitance is shunted across the plate load resistor or the following grid resistor,
owing to the coupling through the grid coupling condenser. Even if no shunt con-
denser is added, the output capacitance of the valve plus the dynamic* input capacit-
ance of the following valve plus wiring capacitances provide appreciable treble attenua-
tion -see Chapter 12 Sect. 2(xi). The effect on treble attenuation may be reduced
by reducing the plate load resistance ; it may be increased by increasing the plate
load resistance or by adding shunt capacitance. A typical r.c.c. pentode followed by
a similar stage has only slight attenuation at 10 000 c/s when the load resistance is
less than 0.25 megohm. In the case of a r.c.c. triode the effective resistance is the
plate resistance of the valve (under r.c.c. conditions) in parallel with the load resist-
ance and the following grid resistance.
In the general case the effective resistance is given by
R = R, RL'rp /(RL' -f- r,) (1)
where R, = resistance in series with C as part of the tone control,
R1' = RLRv /(RL Rc)
R = load resistance
R, = following grid resistance
and r, = plate resistance of valve.
With a high -mu triode such as type 6SQ7 the effective input capacitance is of the
order of 100 µµF. If this capacitance is shunted across an effective resistance of
0.5 megohm, the attenuation is about 10 db at 10 000 c /s.
The total attenuation at any frequency will be the attenuation at that frequency
prior to the amplifier plus the sum of the attenuations of the stages in the amplifier
at that frequency plus the attenuation by the shunt capacitance at that frequency, all
expressed in db. It is not sufficient to assume an attenuation of 6 db /octave from each
stage in the amplifier unless all the stages have the same cut-off frequency. In the
general case it is necessary, for design purposes, to calculate the attenuation of each
stage or filter for convenient frequencies, e.g. 10 000, 14 000, 20 000, 28 000, 40 000
c/s and then to add the values in db to determine the overall attenuation characteristic.
The latter will, eventually, almost reach a slope of 6 db /octave per stage or per shunt
capacitance, but the knee of the curve will be very rounded unless the cut-off fre-
quencies are identical.
'Under operating conditions, including the " Miller Effect " capacitance from the plate.
15.8 (iii) TREBLE ATTENUATION BY FILTER NETWORKS 657
Fig. 15.33. Power amplifier valve with small condenser C giving negative feedback
at high audio frequencies (treble attenuation).
Fig. 15.34. Power amplifier valve with variable negative feedback at high audio
frequencies (treble attenuation).
658 (iv) TREBLE ATTENUATION IN FEEDBACK AMPLIFIERS 15.6
Fig. 15.35A. Amplifier with variable treble attenuation due to negative feedback.
Fig. 15.35B. Amplifier with negative feedback providing treble attenuation or
treble boosting (Ref. 67).
(C) Fig. 15.35A is a further variation in which R and C form a potential divider
across the output from V2. At increasing frequencies a larger negative feedback
voltage is applied, the amount being limited by the variable resistance R. C, is a
blocking condenser (Ref. 63).
(D) For feedback over more than one stage, see Sect. 9.
Audio _2,
o E )I
27NNF
00 so
01 F
065Mn
120NNF I Mn IpOOn
50,000n
8,20011 o,`
22M11 e+
03
B IO,O00n
68,000n 0',
.01.2NF
OIMn O.39M41
47,0000
C
o
n 8+
i
2
FIG. 15.36
An alternative form using a 5 position switch is shown in Fig. 15.36 (Ref. 33). The
switch positions are
1. heavy bass attenuation and very limited treble attenuation for very distant re-
ception, no feedback.
2. normal bass, heavy treble attenuation, no feedback.
3. bass attenuation, medium treble attenuation, half feedback.
4. normal bass, slight treble attenuation, half feedback.
Note : Positions 1 to 4 inclusive are for use with narrow i -f bandwidth.
5. normal bass, no treble attenuation, maximum feedback, wide i -f bandwidth,
for fidelity reception of local stations.
At low settings of the tapped volume control the capacitance in series with the
inductance provide a dip at middle high frequencies, thus effectively giving bass and
treble boost.
Columbus Radio (Radio Corporation of New Zealand Limited).
660 (iii) TONE CONTROL NOT USING INDUCTORS 15.8
Ct41F
Boss Control
8 NFT 20,00051 Is,OOOlt
8+
Fig.
15.37A. Simple universal tone Fig. 15.37B. Step -type tone control
control not using inductors. not using inductors (Ref. 70).
An improved form is given in Fig. 15.37B and its frequency characteristics in Fig.
15.37C (Ref. 70). The 15 µµF max. trimmer capacitor is for the purpose of com-
pensating in the " flat " position for the loss of high audio frequencies caused by stray
shunt and Miller Effect capacitances when the following stage is a triode.
FIG. IS 37C Increased rate of boost-
ing and attenuation is
obtainable by connecting
111111111111111111111111111 two or more r.c. filters in
P
giving
cascade. A step of 6
db /octave (nominal) is too
to ill great for a flexible tone
'.1" control, so that networks
reduced rates of
Z boosting and attenuation
devised (Fig.
'''f'li iÍIIIIII','U!!II
S ``
to
a to
15.38). be These
combined to give any de-
IiiiiIIIUIIIIIIIIIIIiii! sired steps. A very com-
prehensive example is Fig.
15.39 (Ref. 50) which
cIIIflhINlIIIINIIlIIflOII
uses two six -pole 11 posi-
six-pole
Q s 8 8 tion control switches. Its
n - r a overall gain is zero db at
FREQUENCY C/S 500 c/s at any setting of
Fig. 15.37C. Frequency response characteristics of the the control switches.
circuit of Fig. 15.37B (Ref. 70).
rw. 5.5e
o OOIS ur 6,600 [7,000 39;000, TAP
1114000 15,000
e
MESS
OS EER
60057
017071
O717E611 ATTE
! 06 PER
UATIOM
OCTAVE
© t.6
SASS 60047
D6 PER OCTAVE
OSASS
[ OR
ATT DREAM*
PER OCTAVE
Fig. 15.38. Networks giving selected values of decibels per octave for boosting and
attenuation (Ref. 50).
15.8 (iii) TONE CONTROL NOT USING INDUCTORS 66]
BASS CONTROL P G. 15. 99 TREBLE CONTROL
CATION 6.400 33,000 CATHODE
FOLLONE e 7500 00,000 390,000,.-00006
27,000 50,000
,350.000 ,8,200
O OÌ AA OÌ 88 1DÌiú m 4.700 00
INPUT
7 7.500 A 8®, ; 0®, DO NCI .N4 MJ .NC
áp00
001J 0.0016 t2 EÉa
NG NC 025
0.1 NC 0025 0006 6T,000 ó ó FF
120005E1000 12,000
200
9N 39.000
2 MEO
QI 003 001 MÈG 000:2
O JJ,000 GG
00006' 00016
6.400
33.000
.18
MEG
000 000
4.700 AIIPLIFI
M1
4
MEG
025
I1
.700 AMPLIFIER 0 TO AUDIO
STEPPING
41ME¿ppTROL AMPLIFIER
6 GAIN CONTROL
0.25 EG 270.000
'°-
EG 270,000 22,000
12 22,000 250v
3.900
5250v
.30
mot mown
m1:o`1111IIII
mo11111
30
.20
.10 m1
mmilìi!m1111
i\111
e!nnolll .. ...n::.II111I
1Ii-;
IIIIIm% -
IIIIII/R1.
" i .20
mí
M= .últl 11111
riA.é:
0
IIIIItiZi::
T7
11111
IIIIImgs1ïi
a1ttM 11-',ii 0g:
IIIII
_V/NA CONTROLA GURVES
-20
_ 1C0NTROLEBCURVES
11miuIIII 20
lowlifa1Woo I 200
FREQUENCY IN CYCLES
EMI=
500 1,000 2,000
v 5000 4050
1
Fig. 15.39. Step -type tone control system not using inductors (Ref. 50).
RESP. CONTACT.
CURVE. 8 9 10 11 12
T1 Z AA BB NC CC
T2 NC Z BB NC CC
T3 Z AA NC NC BB
T4 NC Z NC NC BB
T5 Z NC NC NC AA
T6 NC NC NC NC Z
T7 NC NC NC Z DD
T8 NC NC NC Z EE
T9 NC NC NC Z FF
T10 NC NC NC Z GG
T11 NC NC NC Z HH
RESP. CONTACT
CURVE. 1 2 3 4 5 6
B1 3 I A B NC C
B2 K NC I B NC C
B3 L I A NC NC B
B4 M NC I NC NC B
B5 N I NC NC NC A
B6 P NC NC NC NC I
B7 P NC NC NC I D
B8 P NC NC NC I E
B9 P NC NC NC I F
B10 P NC NC NC I G
Bu 1 Q NC NC NC I 11
ii
BNF ,t>/001pF
00025NF / 1500011
/
- Cut Boost
i
Boost Cut
+ 250V G_ -Boss __- L_Tttble - -o' -
Fig. 15.40. Universal step -type tone control using inductors.
boosting. The inductor may be wound with an air -core as under : Former diameter
} in. Length between cheeks } in. Winding wire 40 S.W.G. (or 36 A.W.G) SSE.
Total turns 6740, tapped at 4520 turns. Layer wound. Total radial depth of wind-
ing say 0.6 in.
One defect of this method of treble boosting is the tendency for the inductor to
pick up hum.
(v) Fixed bass and treble boosting
When it is desired to incorporate a fixed amount of bass and treble boosting, as
for example equalizing the response of a pair of headphones, the circuit of Fig. 15.41
may be used. For headphone equalizing (Ref. 7) suitable values are C1 = 0.0003,
C2 = 0.015 µF, R1 = R, = 0.1 MO, R5 = 0 For bass and treble boost for tone
control suitable values are (Ref. 25) -C1 = C2 = 0.0002 µF, R1 = 0.2 MS1,
:
R, = 0.3 Mi?, R, = 0, giving 16 db bass boost (20 c /s) and 7 db treble boost (10 000
c /s) relative to 1000 c /s. If R, = 50 000 ohms, other values being unchanged, the
bass boost is 16 db and treble boost 12 db.
(vi) Step type tone control using negative feedback
Circuits providing universal step -type tone control using negative feedback are
described in Sect. 9 (Fig. 15.58A).
(vii) Continuously -variable controls -general
Continuously- variable controls are in two groups, those which have one control
knob and those which have two or more. A single control knob is obviously limited
in its capabilities -for example it may be used to provide bass boosting when turned
in one direction from the centre point, or treble boosting in the other direction, with
flat response at the centre point. On the other hand, with two control knobs it is
possible to make any desired combination of bass and treble characteristics.
Fig. 15.41. Circuit giving fixed bass Fig. 15.42. Single control continu-
and treble boosting (Ref. 7). ously variable tone control (Ref. 22).
(viii) Single -control continuously variable tone controls
(A) A typical example is Fig. 15.42 (Ref. 22). When C, goes to the grid end of
potentiometer R2, the bass is attenuated by the grid coupling condenser C1, limited
by R1. When C1 goes to the earthed of R2i the treble is attenuated by the shunt
capacitance of C1.
15.8 (viii) SINGLE -CONTROL CONTINUOUSLY VARIABLE 663
FIG. 15.43
(B) Another example is Fig. 15.43 (Ref. 62) which is capable of giving treble boost
c/s and bass attenuation 6 db at 50 c/s in one extreme position ; bass
18 db at 10 000
boost 4 db and treble attenuation 21 db in the other extreme.
(C) An interesting circuit which gives simultaneous bass and treble boosting, linear
response, or simultaneous bass and treble attenuation is Fig. 15.43A (Ref. 86). The
CROSS- COUPLED STAGE
400 R 221.1.11
IMR
F'NF
/
C I MR
1
6SN7
r 6SL7
Eo i ' '
IMnSS . BoOn
.
I MR 400y E,
1,200n CñYMF a 2.500
BALANCED
OUTPUT
x
I,
100
90
BO
70
I
i
FIG. 15.43A
Fig. 15.43A.
MUM A\7III
Tone control circuit using cross coupled .nput stage (Ref. 86).
Al
-, .,.ii\./.tu1ml
B+
E, E
E3.BE2MAX.
MAX.
/,1rr11`
60
a
D 40
o
W
50
30
=rr///I,,1!..,,1t `E . 5E MAX.
E, A EeMAX.
\E,.6E=MAx.
aJ
¢ 20 E../I111111111111DM11 E3+25E2MAX.
Fe 0
10
IIPP" 1I
0
02 .05 .2 .5 2 5 10 20
/fo FIG 15.43B
two twin triodes form the " cross -coupled phase inverter " -see Chapter 12 Sect.
6(viii)-which is here used to amplify the difference between the input voltages to
the 6SN7 grids. The frequency response curves are shown in Fig. 15.43B for differ-
ent values of R,. The maximum output and the point of zero phase shift occur
f
at /f, = 1, where f, = 1 /2nRC. The frequency of the peak in the curve may be
determined by choice of R and C ; in the circuit shown, this frequency is adjustable
from 360 to 4000 c/s by means of a dual 1 MQ potentiometer. A choice of fo = 600
to 800 c/s is pleasing in many cases.
Treble Boost
Fig. 15.44A. Simple duo -control circuit giving indiv'dual control of bass and
treble. Typical values are : R, = 0.25 MQ ; R, = R, = 0.5 MQ ; R, = 0.25
MS2 ; R, = R, = 0.5 MO ; C, = 0.01 µF ; C, = 0.001 µF (Ref. 22).
Fig. 15.44B. Simple tone control giving bass and treble boosting (Ref. 41).
(B) Duo -control circuit incorporating L and C (Fig. 15.45, Ref. 26)
In this filter, R, controls the bass and R, the treble. Hum may be troublesome
on account of L.
Rs
OIMO 01M0
Re 5,0000
INPUT OUTPUT
C.
INPUT OUTPUT
05M0 OO50
Ks
Fig. 15.45. Duo -control circuit in- Fig. 15.46. Cutler duo-control
corporating L and C (Ref. 26). circuit (Ref. 26).
15.8 (x) (D) PARAPHASE TONE CONTROL 665
(D) Paraphase bass- treble tone control (Figs. 15.47A, Ref. 56)
This is an outstandingly flexible circuit which may be designed for any cross -over
frequency-400 c/s in Fig. 15.47A and the curves in Fig. 15.47B. The design pro-
cedure is :
1. Choose a cross -over frequency f,.
2. Choose R > IORL where RL = generator impedance.
3. Make C = 1 /(21f ,R).
4. Make K = 10 ; then KR = lOR and C/K = C /10.
f
Example : Choose . = 400 c /s. Say generator impedance = 1000 ohms, then
select
R = 82 000 ohms, from which
C = 1/(27 X 400 X 82 000) = 0.005 µF ; C/K = 0.0005 F.
+250v
22,00011 22,0001
C/K
0005pF 00%0S NF
20pF I M1 s
Treble ( R
820001$
]]22
82MÁ
KR
465N7GT 1f65N7GT
R KR
Ip F 82p00n 82Mn
1,00011 Ipo0n
min c 10-0O05pF
OOOSpF'
T FIG. 15.47A
:: ...... ..
.g ::'::MÑ::Ñ_
"_
2
i _::_
=i:
!:7
.;.: .T:{II:iCiid :Mini
:1 I:i i5
5:IIcS::i'JI:G=Lfr:Y:::BCa
;'r.:i?:i=
: Z
o
ç
e.._::::::
_=o-2t:6nü
i:i:©::::.ii::ilYiiii..er,.d:.L, s6BLytl.!
x : :::Ñ
2
:::::878:::::r::ra:
...,... . .i.
.....
:.....a.l: :::. iccssii..
.
r. ;
.:,ii :i:,dC-
. t.S $i:!i.....ci:sd::- :.ä:::
...........-
--m= ..... 0' ...
...I;
c°d-
..:......W.'d::..E._..
:
-4 ::: r.:a:: :::::: as's.., g :::__=
:: I;y
.:
U
:::. .:::
-6 EEE:.6ö:_
E:
si}:::,t.°::. ;s.ä,io=_::Mli°0t:BS"ci-o--_
ó i?iéei! :ii!is'fa iBiii.
I I ®sir1i':1.:.
;i%--sr.:.:2:.0 L:ii;
1.::i'::.::_:.-:..
I,agil111=
1:.9l;ii
8
:::; ::I p a:
- 10 Ei .ppgg5ppiii!,pii37BiNÉpllipiá
77r :
ñ
12 !lEE:i:3i
=llph ill,iiiiIIMNIIIIMIOIRMIIIIIIIIxIN
14
l.°Ölll uoüílift®ä6liiii{i.B.'iillig6im
n i:T=ii11iRIN®IYbMOe©lNqii
16
- 18
20
- 2220
7{i::
,ó:=°:VINIDá
C Tffi3`MMENHEIr0pnaii?N.
; =o'r.`
(E) Two-stage bass and treble tone control (Fig. 15.48, Ref. 41)
This employs a twin triode with the treble boost control in the coupling from V1
to V2 and the bass boost control in the coupling from the plate of V2 to the following
stage.
(F) Simple two channel amplifier (Fig. 15.49, Ref. 22)
Owing to the separate amplifier valves, this circuit does not require the series re-
sistors R6 and R6 used in Fig. 15.44 and has thereby less attenuation in the filter
network.
FIG. 15.48
4Z000
.3
FIG. 15.49
68p BASS
INF IMn
Imn 6N7
OUTPUT
Direct to
Grid of Next
Valve INPUT
IMn
Treble Boost 4.7n (No Grid
Resistor)
OOIjFTREBLE
-
Fig. 15.48. Two -stage bass and treble Fig. 15.49. Simple two channel
tone control (Ref. 41). amplifier (Ref. 22).
(G) Wide -range two-channel amplifier (Fig. 15.50, Ref. 31)
In this circuit V, is an amplifying valve common to both channels,
Vs is a bass amplifier only, while V, is a twin triode having one grid fed
from the bass amplifier and the other from V, through a filter network which
only passes the higher frequencies. However two plates of V, are approximately
equal in amplitude for all frequencies, owing to the common cathode coupling, and
opposite in phase, so that the output may be applied directly to a push-pull power
output stage. The bass amplifier V2 is a true bass booster, so that it is not necessary
to attenuate all except bass frequencies. The two controls provide (1) flat response
(2) independent treble boost (3) independent bass boost equal to the gain of V2. If
it is desired to add treble attenuation, the filter circuit of Fig. 15.51 may be used. This
gives a flat response when the moving arm of the potentiometer is at the centre tap.
0+
50,00011 400V
50,0000
8NF ID
OUTPUT
VOLUME O STAGE
CONTROL GR) Dt
50,0000
O.O55F
IMO
OIM11
Y
FIG. 15.50
L,
BASS
L,
TREBLE
INPUT
B+
Fig. 15.51. Modification to Fig. 15.50 to provide treble attenuation (Ref. 31).
Fig. 15.52. Resonant plate loading to provide bass and treble boost (Refs. 26, 54).
plate load resistance R,. Control of the bass is given by R. and of the treble by -R,.
The Q of each of these circuits should not exceed 1 --see Sect. 1(ix).
(I) Negative feedback incorporating L and C in cathode circuit (Fig. 15.53,
Ref. 26)
Feedback provides degeneration across Rl and R2, which determines the gain at
middle frequencies. With L and C connected directly across R. and R the feedback
is decreased at low and high frequencies respectively and the stage gain is consequently
greater at these frequencies. L may require shielding to reduce hum pickup.
IIE
Fig. 15.53. Bass and treble boosting due to negative feedback incorporating
L and C in cathode circuit (Ref. 26).
Fig. 15.54. Bass and treble boosting due to negative feedbaok incorporating
L and C (Ref. 15).
(J) Negative feedback incorporating L and C (Fig. 15.54, Ref. 15)
Vi is a power amplifier valve with L, and C. and their control potentiometers in
the feedback network. L, may have an inductance of about 10 henrys and C, may
be 0.002 µF ; R. = R. may be 0.1 megohm.
(K) Patchett tone control (Fig. 15.55, Ref. 9)
This circuit has been developed to provide continuously -variable independent
bass and treble boost without variation of the middle frequencies and at the same time
with frequency characteristics which are close to the ideal for tone control purposes.
V, is a phase splitter and develops equal voltages across its plate and cathode load
resistances. The former feeds directly through an isolating resistor to the output ;
the latter feeds the bass and treble filters and potentiometers which in turn feed the
grids of the twin triode V. whose plates are connected to the output. The output
voltage at middle frequencies is about one third of the input voltage. An input of
2.5 volts r.m.s. may be applied with negligible distortion. V, may be type 6C5 or
6J5 ; V. may be type 6N7 or other twin triode.
668 (x) (K) PATCHETT TONE CONTROL 15.8
5.,
FIG. 15.565
Incrcme
Bmf
.Increott 7
1\11M i/I
TrcDle
6
V1
v
2
0 ñ5
fY 4
MONOMERS á
3
MEMOIR! w7
ó ó ó g g q Q
ó
FREQUENCY C/S FREQUENCY CIS
Fig. 15.56A. Response characteristics of Patchett tone control circuit with S, open
and S5 closed, and other control at minimum (Ref. 9).
Fig. 15.56B. Response characteristics of Patchett tone control circuit with S,
closed and S2 open (Ref. 9).
Curve A Max. bass, min. treble Curve D Min. bass, f max. treble
B 4 Max. bass, min. treble ,, E Min. bass, min. treble
C Min. bass, max. treble F Max. bass, max. treble
Fig. 15.56A shows the response characteristics without provision for bass or treble
attenuation ; Fig. 15.56B shows the response with bass and treble attenuation.
If desired, step switches may be arranged to give a wider choice of frequency charac-
teristics by changing the values of the filter condensers.
FIG. 15.57A +300V
(L) Two -stage universal tone 50.000lD
control (Fig. 15.57A, Ref. 77) 01NF
+10
10
20
20
Ìi
(x) (L) TWO -STAGE UNIVERSAL
I
FULL
..III_
:í111
FULL
1
..-
50
Fig. 15.57B.
1 f
BASS CUT
I
BASS BOOST
111111\1111111111 :ii
III%r
MID -POSRION
IONS
1 i
(C. P. S.)
2P00
FULL TREBLE BOOST
MID -POSITIONS
111111
1111i:.
FULL TREBLE
5p00
20,000
FIG. 15.57 B
669
10 000 ohms plate -to -plate to 20 ohms voice coil. Total primary inductance not less
than 60 henrys
(C) Modified " straight " feedback amplifier to provide bass boosting
In many cases a " straight " feedback amplifier may have a capacitive impedance
added to the feedback loop to provide bass boosting. This decreases the feedback
at bass frequencies, thereby causing the amplifier to lose some of the advantages of
feedback ; the arrangement is less desirable than a special circuit such as Fig. 15.58A
which has been specially designed for tone control.
For example, a capacitor of 0.1 µF may be inserted in series with the 5000 ohm
resistor in the feedback circuit in Fig. 7.44 (as in Ref. 69) ; alternatively a switch
giving a choice of several values of capacitance may be used.
LOH
FIG. 15.5BA
47,000ft
.25 Mil
1}4F V
o --
o-'-
MAINS
HIGH L.S.GUTPUT
IMPEDANCE 20 D.
INPUT 50y(IF
4 0.256411 IOOy4iF
LOW L
IMPEDANCE
INPUT
25- 60011
F
22000ft
Boss Boost
1,00011 st
Fig. 13.38A. High gain amplifier with overall feedback giving bass and treble
boosting (Ref. 42).
15.9 (ii) (D) FEEDBACK TO PROVIDE SPECIAL CHARACTERISTICS 671
FIG. 15.588
Fig. 15.58B. Use of feedback to provide special attenuation characteristics at
low frequencies. The total effective resistance across Ti primary is taken as 1000
ohms. Switch Sr normal ; S, tone control (low frequency attenuation). Switch S'
controls feedback.
1. Design the amplifier so that the response without feedback and without tone
control is only slightly down at f0, say -
6 db (100 c/s in Curve A).
2. Design the tone control to provide a total attenuation without feedback, of about
3 db at 10 fo (Curve D). o
Mu11111111111111111111111
3. Apply sufficient negative UIi\111 11111111MIP'.aIMMM111111u1 1
r11111lSíl 11
IId1111111111MI1111111111
quency range (Curve C). 12
Nu1M41111 IIIrI11111111111
uP.,uuun MMIUM1luuuul
In the example, 20 db of
feedback is used, and the
combined effect is to give an
is
20 I.1111/111
MI11111111111
MOU111111111 S
%r
iéiri:.,,___
1V%111'.'IIIIIIMMINU1U1111111
II.u111111111411I-uu1O111111
op111
iNi1oiM11
I-1111111111 1
attenuation of 18 db in the 2 24
octave from 100 to 50 c/s. Z 28
I/I111
/u1111117P1
1111i11S 111111111
U111111111
1
/MUUMUU/UM u11IIIIIII
u111111111
32
/111u11uMIII
(b) To provide optional
a //MUUMUU/Mill I11111111
/U11//11111
1
36 M1111111111
bass boosting 40 UNIIIM11GU1111111ru1111111111
INII1T5111111111 1111111111
If now the tone control is MEW
M I/1111111111 111111111I
//11111111 I11111111111
IÌÌÌÌÌÚ
44
switched out of circuit, there 48 FI111111111111 u111111111 I
will be bass boosting as the
result of phase shift in the 52 ëÌIIIÌÌÌÌÌÌÌ
30 50 100 300 500 X00
amplifier (Curve B). Fquancy cl FIG. S.58C
(i) Introduction
Owing to the special characteristics of the human ear, it is necessary for bass boosting
and (to a less extent) treble boosting to be applied to music or speech when repro-
duced at a lower level than the original sound, if it is desired to retain the full tonal
qualities of the original. Provision is therefore often made in the better quality
receivers and amplifiers for this to be done automatically as the volume control is
adjusted. For this to be fully effective, the volume with the volume control at its
maximum setting should be the same as that of the original sound -a condition which
it is rarely possible to fulfil in radio reception. One possible way of achieving an
approach to the true condition would be to fit two auxiliary volume controls -one
with settings for say (1) speech (2) orchestral and (3) solo instrument, and the other
to be adjusted to bring the reading on some form of level indicator to a predetermined
value. The first of these auxiliary controls is necessary because the various original
sounds differ in level, the second control to provide for imperfect a.v.c. in the re-
ceiver and variations in percentage modulation between stations. These complica-
tions are not likely to be popular. However the intelligent listener should have a
second volume control which may be pre-set to give a desired output level for maxi-
mum volume-this is particularly important in reproduction from records.
In practice, some form of simple automatic frequency- compensated tone control
is found beneficial, in spite of its technical imperfections. One reason for this popu-
larity is that it permits a considerable degree of bass boosting to be used at low volume
but not at maximum volume- control setting. It is therefore more fool -proof than
manually- operated bass boosting, and less likely to cause overloading of the power
amplifier.
(ii) Methods incorporating a tapped
potentiometer
A typical application is Fig. 15.59 in which L and C
form a series resonant circuit tuned to about 1000 c /s,
while R is a limiting resistance to reduce the by- passing
at middle frequencies. The tapping point is usually
about one fifth of the total resistance, being equivalent
to about 14 db below maximum. With this method
it is impossible to obtain theoretically correct compensa-
tion.
If treble boosting is not desired, L may be omitted Fig. 15.59. Tone-com-
leaving C and R to provide bass boosting. pensated volume control
References 71, 72, 89. with tapped potentiometer.
15.10 (ii) METHODS INCORPORATING A TAPPED POTENTIOMETER 673
A more satisfactory result over a wider range of volume levels can be obtained by
using a volume control with two tappings (e.g. at one sixth and one third of the re-
sistance) as in Ref. 88.
A more elaborate type of continuously -variable control has been described (Ref. 79)
but requires a special volume control.
Another more elaborate type (Ref. 82) uses three ganged volume controls which
are not tapped.
FIG . 15.60
o o o o o o o
59R, 41R, .59R, 59Rr 41R, 59R, 41R,
20p00n 3o00on 2gp00n 30,0000 20p00n 3operon 20,00011 30p00n 20000n
2C, 2C, 2C, 2C, 2C,
11pF 0066NF 0068NF 0068µF 006814F
RR2 R R, Rit
Fig. 15.60. Step -type tone- compensated volume control for bass correction only
(Ref. 43).
FIG. 15.61
INPUT
O IMA SMA
0.02y1F
F +2 SOV
11
0.1
HIGH -PASS FILTER 5Ma
`
100µµF I00j41F -1 65N7
IMA V I V2
IMA
50, SAM
IMA
F
101+F
_J
46SN7
0-Iy F
%6H6 65J7
00
V3 V
175MÁ
ISM
IMA
IMA + 250V +250V
LOoon 60011
I0)tF Tü01yIF 0-1rF 10 }IF
0.OS)rF
STRAIGHT-THROUGH CHANNEL 6J5
IMa
2MÁ
IMa w + 250V
I Ma
1,25011
IA 5MA
I}IF
LOW -PASS FILTER + 250V IMa SMA
25y000Á 25,000a 6SN7
v, Va 1
IMa
0.05NFm 005y1F IMA R= R3
IMa RI IMA
600A IoylF
IMA
í66N7 =6H6 6SJ7 11µF1
V, VIO Vii/ TO GRID OF
175MÁ NEXT STAGE
15 Ma
!Mil
+250V 250V
1,000 IDyIF 0.010 mOIµFOn2
Fig. 15.61. Inverse volume-expansion circuit with 3 channel amplifier for automatic
frequency -compensated tone control (Ref. 21).
given by
A/A, _ (R' + Ri) /Ri
where R' = r,RL' /(rP -} RL')
and RL' = RIR /(RL Re).
For effective filtering, the maximum attenuation
may have to be 20 db or more. The sharpness of the
attenuation curve depends upon the Q of the tuned
circuit, where Q = w,Li/Ri. A value of Q = 10 is
probably the highest practicable value, and under
these conditions the filter will have a minimum at-
tenuation of 4 db over a bandwidth of 0.64 f, or a
minimum attenuation of 8 db over a bandwidth of
0.38 f, where f, is the frequency of resonance. Fig. 15.62. Whistle filter
This is equivalent to an attenuation of at using resonant circuit.
15.11 (ii) NARROW BAND REJECTION FILTER 675
least 8 db over a frequency range from approximately 8000 to 12 000 c/s when f, _
10 000 c /s. Thus the filter cuts such a serious hole in the amplifier frequency charac-
teristics that it is not a satisfactory solution. It is reasonably satisfactory for a maxi-
mum attenuation of 6 db, but this may not be sufficient to eliminate the whistle.
For theory and curves see Refs. 17, 51, pp. 52 -54.
(ii) Narrow band rejection filter
A more complicated and more effective circuit is Fig. 15.62A (Ref. 811 which, when
tuned to 9000 c /s, gives an attenuation of 2 db at 8000 c /s, over 40 db at 9000 c /s,
8 db at 10 000 c/s and 5 db at 20 000 c/s relatively to the level at low frequencies.
The frequency of resonance may be changed either by adjusting the inductance of
L1, or by varying the capacitance of both C, and C2. The attenuation at low frequen-
cies (insertion loss) is 2 db.
300mH A narrow band rejection filter is described
in Ref. 75 which has the disadvantage of
to PLATE L. To GRID reducing the maximum a -f power output. An
re10,000A 2p00ppF 2p00ppF
improved form for application to a linear
c. C. reflex detector is Fig. 15.63 in which an attenu-
27,000.11. ation of more than 40 db is obtained at 9000 c/s
IMA
while the attenuation is only 3 db at 8400 c /s.
The principle may also be applied to a cathode -
loaded amplifier by omitting C1 and pro-
viding the correct bias for amplification.
FIG. IS.62A
The resonant circuit LC, is tuned approxi-
Fig. 15.62A. Narrow-band mately to the whistle frequency, a vernier
rejection filter (Ref. 81). control for Co being desirable. The value
of C, should be such that its reactance is
very much smaller than R, at the lowest signal frequency and should be very much
larger than R1 at 9000 c /s. The value of C2 should be such that its reactance is
approximately equal to R2 + R, at the low frequency limit of the amplifier, say 50 c /s.
If its reactance is higher, a slight amount of bass boosting will occur. The potentio-
meter R,R, is adjusted to give zero output at the whistle frequency (Ref. 74).
An elaboration of this principle, using two valves, is given in Ref. 83.
12510
2nd Audio
.0035F,
H
4.7
Mn
FIG. 15.64
B+
Another circuit (Ref. 87) has a supplementary channel with a distorting valve (triode
6SF5) functioning effectively only below about 100 c /s. Grid distortion is produced
through zero bias operation. A plate load resistor of 0.5 MQ is used, and the plate
coupled to the grid of the output pentode. It is stated that the loudspeaker bass
resonance should preferably be less than 60 c /s, and that the results sound unnatural
to the ear on music when the loudspeaker resonance is about 150 c /s.
The principles of synthetic bass are given in Chapter 14 Sect. 3(viii).
THE LISTENER AND TONE CONTROL 677
15.13
SECTION 15 : REFERENCES
1. Light, G. S. " Frequency compensating attenuators " Electronic Eng.
16.195 (May 1944) 520.
2. Barcus, L. M. " New bass boosting circuit " Elect. 16.6 (June 1943) 216.W.E. 10.112 (Jan. 1933) 4.
3. Colebrook, F. M. " A note on the theory and practice of tone -correction "
(Aug. 1943) 120.
4. " Universal equaliser provides a -f amplifier design data " Elect. 16.8 Comm. 23.8 (Aug. 1943) 28.
5. Widlar, W. L. " Compensating audio amplifier for three channels " use with conventional
7. Amos, S. W. " Improving headphone quality ; simple correction circuits for
diaphragm-type earpieces" W.W. 50.11 (Nov. 1944) 332. 2,352,931 Radio Craft 16.4 (Jan.1945) 230.
8. Austin, K. B. "A Bass and Treble Booster" U.S. Patent
9. Patchett, G. N. " A new versatile tone control circuit" (1) Basic principles
of tone control W.W.
51.3 (Mar. 1945) 71 ; (2) " Bass and treble lift without variation of middle frequencies " 51.4
(Apr. 1945) 106. " Electronic Eng. 17.207 (May
10. Sturley, K. R. L.F. Amplification (7) Increasing l.f. response
1945) 510. 51.6 (June 1945) 182.
11. Winget, D. (letter) " A new versatile tone, control circuit " W.W.
13. Smith, H. " Tone compensation amplifier " Australasian Radio 1945) World (July 1945) 20.
14. Shepard " Improved bass for small radios " Elect. 18.7 (July 224.
1943) 596 ; (2) 14.11 (August
15. Sands, L. G. " More highs-more lows " Radio Craft (1) 14.10 (July
1943) 662. tones for radio receiving sets " Philips
16. Hendriquez, V. C. " The reproduction of high and low
Tec. Rev. 5.4 (April 1940) 115. " W.E. 9.100 (Jan. 1932) 3.
17. Scroggie, M. G. ' Amplifier tone control circuits " Proc. I.R.E. 26.1 (Jan. 1938)
19. Fritzinger, G. H. " Frequency discrimination by inverse feedback
207. (June 1946) 199.
20. Design Data (5) " Low-frequency correction circuit " W.W. "52.6
Elect. 19.7 (July 1946) 142.
21. Goodell, J. D., and B. M. H. Michel " Auditory perception 1946) 763.
22. Wortman, L. A. " Tone control circuits " Radio Craft (Aug. 1946) 278.
23. Reference Sheet " Fundamental Tone Control Circuits " Electronic Eng. 18.223 (Sept.
678 REFERENCES 15.15
25. Powell, 13. O. " Tone control " W.W. 46.14 (Dec. 1940) 491.
26. Cutler, S. " Audio frequency compensating circuits " Elect. 15.9 (Sept. 1942) 63.
27. Haines, F. M. " High fidelity bass compensation for moving coil pickups " Electronic
(Feb. 1946) 45. Eng. 18.216
28. " Cathode Ray " " Working out tone control circuits " W.W. 43 (Sept. 22, 1938) 269.
29. Najork, J. " Why bass boost " Service 17.6 (June 1948) 18.
31. Hill, J. M. " Wide range tone control-circuit suitable for correction at low volume levels " W.W.
52.12 (Dec. 1946) 422.
32. Case, N. " All purpose F-M A -M receiver " Comm. 7.1 (Jan. 1947) 34.
33. Beard, E. G. " A quality switch in lieu of a tone control " Philips Tec. Corn. No. 2 (Feb.
Based on Philips C.A.L.M. Report No. 7. 1947) 8
34. Stevens, P. " The control of tone " Australasian Radio World 11.8 (Jan. 1947) 13.
35. Moody, W. " Audio amplifier equalization methods " Service 16.1 (Jan. 1947) 11.
36. Ellis, J. " Bass compensation
37.
-a system using negative feedback " W.W. 53.9 (Sept. 1947) 319.
Negative feedback tone control used by R.G.D. ' extract from review
24.290 (Nov. 1947) 335. Radiolympia 1947, W.E.
38. Schlesinger, K. " Low frequency compensation for amplifiers " Elect. 21.2 (Feb. 1948) 103.
39. Banthorpe, G. H. " A simple tone control circuit " Electronic Eng. 20.241 (March 1948) 91.
40. Winder, N. " A bass correction circuit for moving-coil pickups
Electronic Eng. 20.244 (June
1948) 187.
41. Sugden, A. R. " Pre -amplifier tone control " Australasian Radio World 13.2 (July 1948) 11.
42. Whitehead, C. C. " Quiet high -gain amplifier -tone control by negative feedback " W.W.
(June 1948) 208. 54.6
43. Bomberger, D. C. " Loudness control for reproducing systems " Audio Eng. 32.5 (May 1948) 11.
44. Bonaire -Hunt, N. "Direct- coupled amplifier " W.W. 54.7 (July 1948) 266.
45. Dahl, H. M. " RC circuits as equalizers " Audio Eng. 32.6 (June 1948) 16.
46. Childs, P. A. " A 9 Kc /s whistle filter " Electronic Eng. 20.248 (Oct. 1948) 320.
47. Gerry, C. C. " Feedback and distortion " W.W. 54.9 (Sept. 1948) 347.
48. Savory, W. A. " Design of audio compensation networks " Tele-Tech (1) 7 : 1 (Jan. 1948)
(2) 7 : 2 (Feb. 1948) 27 ; (3) 7 : 4 (April 1948) 34. 24 ;
49. Mitchell, I. A. " AF circuit equalizer " FM and T. 8.6 (June 1948) 28.
50. Lurie, W. B. "Versatile tone control " Elect. 21.12 (Dec. 1948) 81.
51. Sturley, K. R. (book) " Radio Receiver Design " (Part 2) (Chapman and Hall, London, 1948)
pp. 41 -55.
52. ' Bass boosting " Radiotronics No. 73 (February 1937) 14.
53. - ynch, W. A. " Video amplifier 1 -f correction " Comm. 23.4 (April 1943) 16.
54. Scroggie, M. G. " Flexible tone control " W.W. 41.11 (Sept. 10th 1937) 263.
55. Cocking, W. T. " Tone control systems " W.W. 44.23 (June 8th 1939) 532.
56. Jaffe, D. L. " Paraphase bass -treble tone control " Radio 30.3 (March 1946) 17.
57. Rogers, G. L. " Simple rc filters for phonograph amplifiers " Audio Eng. 31.5 (June 1947) 28.
58. " Radiotron receiver RC52," Radiotronics No. 117 (Jan. /Feb. 1946) 3.
59. " Applications of a Parallel -T network," A.R.T.S. and P. Bulletin No. 118 (Jan. 1942) 1.
60. " The design of high -pass filters for bass tone control " A.R.T.S. and P. Bulletin No. 117 (Dec.
1941) I.
61. Callender, M. V. " Problems in selective reception " Proc. I.R.E. 20.9 (Sept. 1932) 1427.
62. Tone Compensation and tone control " A.R.T.S. and P. Bulletin Nos. 27 and 28 (April and May
1936).
63. " Tone control," review in W.W. 46.8 (June 1940) 314 ; British Patent No. 516286, June 20th 1938,
Mullard Radio Valve Co. Ltd. and R. G. Clark.
64. " Frequency compensation in audio amplifiers " Radiotronics No. 114 (July 1941) 43.
65. Gustafson, G. E., and J. L. Rennick " Low cost FM-AM receiver circuit " Tele -Tech 7.10 (Oct.
1948) 36.
66. Austin, K. B. " A bass and treble booster " U.S. Patent No. 2,352,931 ; Reviewed Radio
16.4 (Jan. 1945) 230. Craft
67. Williams, W. N. " Two high gain amplifier circuits " Radio and Hobbies, Australia (June 1942) 19.
68. Williams, W. N. " An 807 amplifier " Radio and Hobbies, Australia (April 1947) 42.
69. Williams, W. N. " Using a new amplifier " Radio and Hobbies, Australia (April 1948) 49.
70. Williams, W. N. " A new tone control unit " Radio and Hobbies, Australia (March 1949) 25.
71. Wolff, I., and J. I. Cornell " Acoustically compensated volume control " Elect. 6 (Jan. 1933) 50.
72. D'Oris, P. A., and R. de Cola " Bass compensation design chart " Elect. 10.10 (Oct. 1937) 38.
73. " A synthetic bass note circuit " Radio Electronics 20.1 (Oct. 1948) 37.
74. Childs, P. A. " A 9 Kc /s whistle filter " Electronic Eng. 20.248 (Oct. 320.
75. Krauke, J. E. " A 10 KC suppressor " Radio News 37.3 (March 1947)1948) 46.
76. James, E. J. " Simple tone control circuit " W.W.55.2 (Feb. 1949) 48.
77. Sterling, H. T. " Flexible dual control system " Audio Eng. 33.2 (Feb. 1949) 11 ; letter J. J. Faran
33.6 (June 1949) 31.
78. Winslow, J. " Full-range loudness control " Audio Eng. 33.2 (Feb. 1949) 24.
79. Turner, J. W. " Construction details of a continuously variable loudness control " Audio Eng.
33.10 (Oct. 1949) 17.
80. Edwards, J., and T. J. Parker " An application of frequency selective negative feedback
Tech 8.12 (Dec. 1949) 30. " Tele-
81. Osram Valve Technical Publication TP4 " Radio receiver for use with high fidelity amplifiers
(The General Electric Co. Ltd. of England). "
82. Johnson, E. E. " A continuously variable loudness control " Audio Eng. 34.12 (Dec. 1950) 18.
83. Nevin, R. B. " An effective frequency rejection circuit " Audio Eng. 35.2 (Feb. 1951) 20.
84. Fling, W. D. " A continuously variable equalizer " Audio Eng. 35.3 (March 1951) 16.
85. Dunn, S. C. (letter) " Twin-T circuits " W.E. 28.332 (May 1951) 162.
86. Van Scoyoc, J. N. " A cross -coupled input and phase inverter circuit " Radio News 40.5 (Nov.
1948) 6.
87. Exley, K. A. " Bass without big baffles " W.W. 57.4 (Apr. 1951) 132. Letters 57.7 (July 1951)
264; 57.8 (Aug. 1951) 321 57.9 (Sept. 1951) 369; 57.10 (Oct. 1951) 408.
88. Brooks, W. O. " A two -tap bass and treble compensated volume control " Audio Eng. 35.8 (Aug.
1951) 15.
Toth, E. "The design of compensated volume controls" Audio Eng. (1) 36.1 (Jan. 1952) 13; (2)
36.2 (Feb. 1952) 14.
90. Lavin, C. J. "A scratch filter with continuously variable cut -off point" Audio Eng. 35.11 (Nov.
1951) 6.
Additional references will be found in the Supplement commencing on page 1475.
CHAPTER 16
(i) Introduction
The maximum volume range of any sound reproducer is the difference in decibels
between the maximum sound output and the level of masking by background noise
which latter may include hum, random noise, needle scratch or microphone noise.
The volume range transmitted by a broadcast station may vary from a low value
up to at least 60 db for a F-M transmitter (F.C.C.), the value depending on the type
of programme. In the case of an A -M transmitter the maximum volume range is
about 50 db -see Chapter 14 Sect. 7(iv).
If the original sound has a volume range greater than the maximum volume range
of the transmitter, it is usual to compress it in some way. The compression may be
accomplished manually by the control engineer, or automatically by a device known
as a peak limiter or a volume compressor.
A similar case arises with recorded music, where the maximum volume range may
be as low as 35 db for shellac lateral -cut disc records.
Many types of programme have a maximum volume range less than 35 db, and
therefore do not require compression. However, most broadcast transmitters have
an automatic peak limiter permanently operating, while it has even been found bene-
ficial to use volume compression in F -M broadcasting (Ref. 29).
If reproduced music is to have the same volume range as in its original form, some
kind of volume expander is required. But, as will be shown later, volume expansion
cannot be applied indiscriminately and cannot be perfect.
One advantage of the use of volume expansion is that it reduces the background
noise, other than room noise. This is a considerable advantage when the source is
a disc record.
A serious shortcoming of automatic volume expansion is that the circuit
can never
duplicate the original volume range because it has no way of knowing what these
679
680 (i) INTRODUCTION 16.1
-it
original levels were will make the loud portions louder and the soft portions softer,
but always by the same amount for a given input level. Thus the expander circuit
might increase every signal that is 10 db above the average volume level to 15 db above
the average, but the level in the studio might have been higher or lower than this.
The amplitude ratios of a soft and a loud note when rendered simultaneously can-
not be compressed or expanded by any system. If the loud note causes a certain
compression in the transmitter, the soft note will be compressed in the same ratio
and may fall below the noise level (Ref. 33).
(iv) Distortion
Apart from the harmonic and intermodulation distortion and extraneous noises
produced by the compressor and expander, which may be kept low by good design,
there are some special features which require consideration. Some of these are de-
scribed in Sect. 2(iv) in connection with volume compression.
(i) Introduction
In this Section it is intended to approach the subject from the point of view of the
radio receiver or amplifier designer.
Volume compressors may be divided into three groups :
18
16 Compras sion
In db
o PF
Threshold o i
14
° gain eduction 1
mil
ailliiiiiii
12 la
lo -- -o
10 1-
-
R.
8 goa
..0
W
6
4 dW
R
2
Some static limiter characteristics are shown in Fig. 16.1. Curve OA is the curve
of a linear amplifier. Curve OTD is that of an " ideal " limiter, with point T as
the threshold of limiting. It will be seen that the output does not rise above a level
of 12 db, no matter how great the input may become.
Curve OTB is that of a practical peak limiter, the curve TB indicating by its slope
a compression ratio of 10 to 1 (both being measured in db). The compression in
db at any point is as indicated by the arrows.
Curve OC is that of a peak limiter having " over-control " (Refs. 36, 40, 44) which
does not exceed 100% modulation (line TD) until the compression reaches 12 db.
With peak limiters a volume expander cannot be used to give correct compensation
for the compression.
Most of the earlier and simpler forms of peak limiters have a comparatively slow
" attack " time, that is the time taken for the limiter to operate on a sudden increase
in input level -usual times are 1 or 2 milliseconds. A slow attack time results in
overmodulation and distortion for short intervals following each transient. More
recent examples of good design (Refs. 36, 38, 40, 44, 63) claim to have " attack "
times less than 100 microseconds, while some (Refs. 36, 40, 44) incorporate a time
delay filter in the signal circuit which is claimed to permit almost instantaneous opera-
tion. Tests have been carried out under dynamic conditions to demonstrate the
performance of various types of limiters (Refs. 4, 40, 63). It is good practice for the
" attack " time of a peak limiter to be less than 200 microseconds, while still shorter
times (e.g. 100 microseconds or less) are desirable.
The recovery times in most cases are manually controlled, with times up to 2 or
even 3 seconds, but automatic control has also been used to lengthen the recovery
time so as to avoid " pumping " when several programme peaks occur in rapid suc-
cession (Ref. 36). The recovery time is not made longer than necessary, since it
causes reduced modulation percentage.
One design (Ref. 86) has a subsidiary series CR circuit that has only a slight effect
with a single sharp peak, thereby giving a fairly short recovery time (0.33 sec.) but
with a sustained peak the recovery time increases to 2 seconds.
If a fixed value of recovery time is used, the most suitable range is from 0.5 to 1.5
seconds, and " pumping " may be avoided by restricting the limiting action to 4 or
16.2 (ii) PEAK LIMITERS 683
? 40 (v) Lamps
Metal filament (dial) lamps may be used as
control devices in volume expanders. A
V
z
Ñ 20 typical tungsten filament lamp has a resistance
á at maximum brilliancy of about 10 times that
at room temperature (Fig. 16.4). See Sect.
O 2 3
4(ii) for applications of metal filament lamps
4
VOLTAGE ACROSS FILAMENT in expanders.
Fig. 16.4. Resistance characteristic Note : Carbon filament lamps have a re-
of typical 4-6 volt 0.04 A metal
used in volume ex- sistance which is less when at maximum
filament lamp brilliancy than at room temperature.
panders (Ref. 45).
686 (vi) SUPPRESSOR -GRID CONTROL 16.3
(i) Introduction
Volume expansion is very similar to volume compression, and the same control
methods are used for both, but the control voltage is of opposite polarity. In fact,
an expander /compressor may be designed with a switch to change from one to the
other (Fig. 16.14).
The desirable characteristics which a volume expander should have are :
1. Negligible non-linear distortion.
2. The degree of expansion should be under control.
3. The degree and control of expansion should be independent of the volume level
at which the amplifier is operated.
4. The expansion should result in the upwards expansion of loud passages and
the
downwards expansion of soft passages.
5. The attack time should be short-times from 0.2 to 200 milliseconds are in com-
mon use, but the shorter times are preferable (say not exceeding 20 milliseconds).
6. The recovery time should be adjustable from a fraction of a second to 1 or 2
seconds.
7. There should be no audible thump or transient distortion with sudden
large
transients.
8. There should be no appreciable reduction in maximum power output.
9. The overall gain should not be reduced seriously by the expander.
10. The shape of the expansion characteristic should provide some expansion
at
low output levels, the amount of expansion steadily increasing all the way to
maximum power output.
For most purposes an expansion of 10 or 12 db is satisfactory, although some prefer
up to 15 or even 20 db.
It is preferable to introduce the expander into the amplifying chain so that mini-
mum amplification follows it. It is therefore desirable to select a type of expander
which is capable of a fairly high output voltage. It is preferable for tone controls
to precede the expansion unit.
Electronic methods of volume expansion may be divided into two groups, those
in which the control voltage is derived from the output voltage, and those in which
it is derived from the input voltage, suitably amplified. The former method is cheaper
and employs fewer valves, but it has a less desirable shape of expansion characteristic
and may be unstable. The latter method is used in Figs. 16.10 to 16.15 inclusive,
and is preferable in order to avoid thumps and blocking effects.
When the control voltage is derived from the input voltage, a " side -chain " am-
plifier is used (e.g. Fig. 16.10) with the final stage transformer- coupled to a suitable
full-wave rectifier and load network.
16.4 (ii) EXPANDERS INCORPORATING LAMPS 687
Fig. 16.5. Simplest form of volume Fig. 16.6. Bridge type of volume
expander using a lamp expander using lamps.
FIG. 16.7
be used with
any type
The bridge circuit of Fig. 16.6 is quite practical, and mayhave an efficiency of 66%.
of output valve. At maximum volume this may possibly
capacitors C, and C, (Fig. 16.7)
With the addition of two inductors L, and L, and two levels only (Refs. 50, 51, 52, 61).
it may be used to provide bass boosting at low output leaving the low level condition
Correct load matching is arranged for maximum output,
to look after itself.
filament lamps is Ref. 53.
A suggested application of both carbon and metal
Fig. 16.10. Volume expander incorporating remote cut-off pentodes (Ref. 23).
Controls A = balance, B = volume, C = input level, D = expansion, E = time
constant.
may be omitted, thus eliminating time delay in the adjustment of electrode voltages.
Valves used in each push -pull pair must be very carefully matched for plate current and
mutual conductance at several points over the operating range of grid bias voltages.
Owing to the limited input -voltage which may be used, there should be one voltage
amplifier stage between the controlled stage and the power amplifier.
An example of good design is Fig. 16.10 (Ref. 23). The minimum attack time is
about 10 milliseconds and the recovery time of the order of 1 second. The expansion
curve (db versus voltage) is approximately linear up to a d.c. control voltage of 20
volts, which gives 10 db expansion. The 6SC7 stage is common to both sections.
The 6SK7 push -pull stage is the controlled stage ; this is followed by the 6N7 voltage
amplifier stage which is coupled to the 6V6-GT power amplifier. The signal to feed
the rectifier is taken from the plate circuit of the 6SC7 stage, amplified in two stages
(6F8 -G), the second being a cathode follower, and rectified by the 6H6. The direct
voltage from the 6H6 filter circuit is applied to the signal grids of the 6SK7 valves.
In any expander of this general type, whatever may be the method of controlling
the gain, it is necessary to prevent the transients in the output of the individual ex-
pander valves from becoming so large as to cut off the following stage. This can be
accomplished by transformer coupling, by a direct-coupled phase inverter (as in
Fig. 16.13) or by the use of low values of load resistances and coupling capacitances
(as Fig. 16.10). With the third "method it is necessary to incorporate an equalizing
network to give a flat overall frequency response.
Refs. 22 (Part 1), 23.
(v) Expanders incorporating remote cut-off triodes
Any remote cut-off pentode may be connected as a triode (screen, suppressor and
plate tied together) to form a triode which may be used in a similar manner to a pen-
tode, except that the gain will be lower. It has been stated that a single valve has
0.37% total harmonic distortion unexpanded, 0.64% expanded, predominately
second harmonic, when the output level is -
25 db (i.e. 0.056 volt). For further
details see Refs. 1, 42.
(vi) Expanders incorporating suppressor -grid controlled pentodes
A sharp zut -off pentode of suitable design may be used for controlling the gain
the English Mazda AC/SP1 and the American 6SJ7 are typical examples. One
-
example of this application is Fig. 16.11 (Refs. 7, 10) which has an attack time of about
1 millisecond, and a variable recovery time of about 1 second maximum.. Resistors
690 (vi) INCORPORATING SUPPRESSOR- CONTROLLED PENTODES 16.4
FIG 16 II
00000 - r 00000
I75:I 1IIII
R4 8
O 7 p
ti 4
C
.F 0.1,..F
R
R9
C6
VI R OIMO 1'1
OOF VI 2óOV
R
Mß
R
-. RII
a rw
MÌL
z
t
RAI
v6 O-
0
R2
§ OI
Mß
_
R '4Mß
-
R
oi/.F
d 11 8nRl.
NPV7 R
í2
o
p
T
Br C7
SOYTO ß13
O SMI
R17
35
98. 1
á T4 Y F 1
O CB 35
) 1 1 rF 197
Fig. 16.11. Volume expander incorporating push -pull suppressor -controlled pen-
todes (Ref. 7). V, = MH4, V2 = V, = AC /SP1, V4 = KT41, V, = V6 = AZ3.
R. and R10 prevent the suppressor grids from being driven positive. With the
moving contacts of R1, and R17 at the chassis ends, R, should be adjusted in con-
junction with the volume control of the main amplifier so that the latter will just be
fully loaded with the loudest signal. R17 is then adjusted to give the desired expansion.
R is advanced until the loudest signal just causes the suppressors to be at cathode
potential. Any further alteration in volume level should be made by R,. Some
suggested modifications are given in Refs. 10, 16.
A second example is Fig. 16.12 (Ref. 27) in which the need for push -pull operation
and for a transformer is avoided by an ingenious device. V, is the suppressor -
controlled stage while V, is a " dummy " designed to balance the plate and screen
currents of V, so that the current passing through each load resistor is constant. The
maximum signal input to the expander stage is about 0.25 volt, and the maximum
signal output of the expander is of the order of several volts, for low distortion. The
controlled range is about 15 db. The diode rectifier incorporated in the suppressor
circuit prevents the suppressor from being driven positive. The time constants
are such that 75% of final gain is achieved in the fast position in approximately 20
milliseconds, and in the slow position 60 milliseconds. The recovery times are
about 0.5 and 1.2 seconds.
References 27, 46. FIG. 16.12
-42V
Fig. 16.12. Single-ended surgeless volume expander incorporating suppressor-grid
control (Ref. 27).
16.4 (vii) EXPANDERS INCORPORATING VALVES WITH FIVE GRIDS 691
Fig. 16.13. Volume expander incorporating push -pull 6L7 control valves (Ref. 2 ).
FIG 16 14
TIME CONSTANT
CONTROL
COMPRESS
T Mil
5
AGG CONTROL
CCW
02NF
IN34
54
250 V
FIG 16 IS
10,OOOa
I 45MÁ
boon
10000
6P5
T2
OUTPUT
-
6J5
0 @A
EXPA SION
IND TOR
IM
yy6SL7 6AF6G
25Ma
1' I 6N6 ¢6SL7
3
RELEASE
TIME
O 3MO
2pF G
IOr.F
2, F
al 5
27OOOn 10 n] z
+
20uFT. 10000
+300V
1_1jQ
db0
I
O EX
NO
12-
Q
6db`
Fig. 16.15. Volume expander incorporating plate resistance control (Ref. 43).
far enough in phase to increase appreciably the peak amplitude of the wave. At the
bass end, satisfactory results are obtained if the response following the clipper stage
is attenuated 3 db or less.
The Plex amplifier (Ref. 37, Fig. 3) is capable of 20 db peak clipping before the
distortion becomes serious. The average increase in power level is about 12 db.
An alternative design of speech clipper incorporated in a speech amplifier is given
in Fig. 2 of Ref. 28.
FIG. 16.16
B t
Fig. 16.16. Simple form of speech clipper suitable for an amateur transmitter
(Ref. 28).
A much simpler form of speech clipper, but one good enough for use with amateur
transmitters, is shown in Fig. 16.16 (Ref. 28).
" Infinite " peak clipping has been used successfully, giving articulation from 50
to 90%. If preceded and followed by suitable frequency- tilting filters, the articu-
lation may reach 97% with a quality sounding very much like normal speech. See
Chapter 14 Sect. 11(ií), also Ref. 71 of this chapter.
References to speech clippers 28, 37, 63, 71.
:
cI
H
Ioo,y.F
DIODE
R2
y
270nnO B 270.000
peak limiters A
r
MNJM MMNM
IST A -F
both diodes, the hum may be re- AMPLIFIER
duced by earthing the end of the 1
heater which is closer to the CI 7OOr,.F
C4
diode D2. The hum may still 82.000 56,000 -')1
be troublesome even with this R1 R2 0.01
precaution.
D2
A circuit giving lower hum is Q 01
C3
Fig. 16.18 ; if a limiter on /off R
3 470,000 R5
D'
SECOND
DETECTOR
1 DIODE
CI
It
200,pF 500,000
IMn
VVY
R2
2
NOISE
LIMITER '
)k_
C3
DI ODE
A! F
R3
OUTPUT
500,000
R4 T C2
WAY.i
1 Mn 0.1..!
Fig. 16.19. Series -type noise limiter with threshold adjustment (Ref. 30).
696 (ii) (C) LOW -LOSS SERIES -TYPE NOISE LIMITER 16.7
FIG. 16.20
SECOND
DETECTOR
IMa
(D) Balanced- detector noise limiter (Fig. 16.21)
This functions as a balanced bridge arrangement for detector voltages above the
limiting threshold, with unbalance at all other times. It must be adjusted manually
for each carrier level. This is not a very satisfactory form of limiter.
NOISE
FIG. 16.21 LIMITER
DIODE
CI C2
22O..rF
R1
AUTOMATIC BIASING
DIODE
100.000 2
R6
16.7 (ii) (F) EALANCING -TYPE NOISE LIMITER 697
FIG. 16. 24
NOISE
LIMITER
D2
SOrrF 50rpF
A- F
OUTPUT
Fig. 16.23. Balancing -type noise limiter Fig. 16.24. Triode shunt-type noise
(Ref. 30). limiter (Ref. 30).
(F) Balancing -type noise limiter (Fig. 16.23)
This uses a limiter diode with reversed polarity, shunted across the detector diode.
The modulation distortion is quite high, even on relatively low modulation depths.
This circuit gives effective limiting of noise peaks. It provides about twice the
normally obtainable a.v.c. voltage when the direct potential across R2 is utilized.
(G) Triode shunt-type noise limiter (Fig. 16.24)
This circuit employs the plate resistance of a triode shunted across a portion of
detector diode load, the magnitude of the shunt resistance being controlled by the
grid and plate voltages, which act in conjunction with differential time -constants.
F IG 16. 25 SECOND The percentage of modulation at
DETECTOR which distortion begins depends on
the triode used, the values of RI, RE
and R;,, the time -constants involved,
and the absolute carrier level. The
higher carrier levels produce no dis-
tortion and no limiting. Serious
distortion has been observed with
10% modulation at low signal levels.
Effective limiting action is restricted
to a narrow range of carrier input
16.25. Diode shunt -type noise limiter levels, generally above 10 Mc /s in
Fig.
(Ref. 30). carrier frequency.
This is the simplest form of limiter, but the performance is not very good ; some
type interference on signals above 10 Mc /s.
improvement is evident on pulse
FIG 16.26
B+
SECOND NOISE R -F
AVC DIODE LIMITER CHOKE
SYSTEM AND AMPLIFIER
FIRST NOISE
LIMITER
--
AMPLIFIER
C,
+100V 50,F
RI
50,000 R
600
100.000 1
R3
50,000 1wvw.
100.000 O.OsIca
Fig. 16.27. Degenerative noise limiter, acting between first and second i-f stages
(Ref. 30).
FIG 16 26
V1(6 SJ7) V2(6 K6GT)
A -F AMPLIFIER A -F OUTPUT AMP
CI
0OI OUTPUT
INPUT FROM RI 600 OHMS
DETECTOR OR
1ST A-F
AMPLIFIER
V6
61.16
SECTION 8 : REFERENCES
1.McProud, C. G. " Volume expansion with a triode " Elect. 13.8 (Aug. 1940) 17.
2.Schrader, H. J. U.S. Patent 2,343,207 (1944)- Improved voltage limiter.
3." Contrast expansion etc." (letters) A. H. King W.W. 48.12 (Dec. 1942) 290 ; J. R. Hughes 49.1
(Jan. 1943) 29 ; J. Moir 49.3 (Mar. 1943) 92 ; D. T. N. Williamson, W.W. 49.5 (May 1943) 151 ;
J. R. Hughes 49.8 (Aug. 1943) 240 ; C. E. G. Bailey 49.10 (Oct. 1943) 313 ; J. R. Hughes 49.12
(Dec. 1943) 382 ; A. A. Tomkins 50.5 (May 1944) 152.
4. Hilliard, J. K. " A limiting amplifier with peak control action " Comm. 23.5 (May 1943) 13.
5. Lewis, R. " A peak -limiting amplifier for recording " Q.S.T. 27.9 (Sept. 1943) 26. Letter 28.8
(Aug. 1944) 55.
7. Williamson, D. T. N. " Contrast expansion unit -design giving unequal pick-up and decline
delays " W.W. 49.9 (Sept. 1943) 266. Corrections W.W. 49.10 (Oct. 1943) 315.
9. Miller, B. F. " Sibilant speech sounds " Electronic Eng. 16.185 (July 1943) 69.
10. Williamson, D. T. N. " Further notes on the contrast expansion unit " W.W. 49.12 (Dec. 1943) 375.
12. Herrick, G. Q. " Volume compressor for radio stations " Elect. 16.12 (Dec. 1943) 135.
13. Felix, M. O. " New contrast expansion circuit -applying the principle of the cathode follower"
W.W. 50.3 (March 1944) 92. Letters W. C. Newman 50.5 (May 1944) 152 ; D. T. N. Williamson
50.6 (June 1944) 187.
14. " Simple volume expander circuits " Australasian Radio World (April 1944) 11 -based on R.C.A.
(E. W. Herold) Patent 264, 942 (March 30, 1939) U.S.A.
16. Ingham, W. E., and A. Foster " Variable contrast expansion -control of contrast range without
change of average level " W.W. 50.8 (Aug. 1944) 243.
17. Roddam, T. " New thoughts on volume expansion -contrast should be proportional to size of room "
50.9 (Sept. 1944) 286.
18. Cosens, C. R. " RF volume expansion " W.W. 50.12 (Dec. 1944) 381.
19. Hansen, I. C. " Contrast without distortion " Australasian Radio World (April 1945) 15.
-
20. Crane, R. W. " Suggestions for design of volume expanders " Elect. 18.5 (May 1945) 236.
21. Weidemann, H. K. " A volume expander for audio amplifiers reducing time constant for more
rapid response " Q.S.T. 29.8 (Aug. 1945) 19.
22. White, J. G. " Contrast expansion-the use of negative feedback ; its advantages over earlier
methods " W.W. 51.9 (Sept. 1945) 275 ; 51.10 (Oct. 1945) 309 with bibliography.
23. Ehrlich, R. W. " Volume expander design " Elect. 18.12 (Dec. 1945) 124.
24. White, J. G. " Contrast expansion-some practical results using negative feedback " W.W. 52.4
(April 1946) 120.
25. Moses, R. C. ` A volume expander compressor preamplifier" Radio News 35.6 (June 1946) 32.
27. Butz, A. N. " Surgeless volume expander " Elect. 19.9 (Sept. 1946) 140.
28. Smith, J. W., and N. H. Hale " Speech clippers for more effective modulation " Comm. 26.10
(Oct. 1946) 20.
/
700 REFERENCES 16.8
29. Phillips, W. E. " Volume compression for FM broadcasting- Raytheon volume limiter " F.M. and
T. 6.9 (Sept. 1946) 28.
O. Toth, E. " Noise and output limiters " Elect. (1) 19.11 (Nov. 1946) 114 ; (2) 19.12 (Dec. 1946)
120.
31. Stewart, H. H., and H. S. Pollock, " Compression with feedback " Elect. 13.2 (Feb. 1940) 19.
32. Moorhouse, C. W. " A high fidelity peak-limiting amplifier " Q.S.T. 28.5 (May 1944) 19.
33. Philips, Volume expansion " Philips Tec. Com. 77 (Sept. /Oct. 1940).
34. Hennquez, V. C. " Compression and expansion in transmission sound " Philips Tec. Rev. 3. 7
(July 1938) 204.
35. R.C.A. " Application Note on the 6L7 as a volume expander for phonographs " No. 53 (Nov. 27,
1935).
36. Maxwell, D. E. " Automatic gain-adjusting amplifier " Tele-Tech 6.2 (Feb. 1947) 34.
37. Dean, M. H. " The theory and design of speech clipping circuits " Tele-Tech 6.5 (May 1947) 62.
38. Level- governing audio amplifier " (W.E. model 1126C) Tele -Tech. 6.8 (Aug. 1947) 67.
39. Jurek, W. M., and J. H. Guenther " Automatic Gain Control and Limiting Amplifier " Elect. 20.9.
(Sept. 1947) 94.
40. Maxwell, D. E. " Dynamic performance of peak -limiting amplifiers " Proc. I.R.E. 35.11 (Nov.
1947) 1349.
41. McProud, C. G. " Experimental volume expander and scratch suppressor " Audio Eng. 31.7 (Aug.
1947) 13.
42. Johnson, M. P. " Multi- purpose audio amplifier " Audio Eng. 31.7 (Aug. 1947) 20.
43. Pickering, N. C. " High fidelity volume expander " Audio Eng. 31.8 (Sept. 1947) 7.
44. Dean, W W., and L. M. Leeds " Performance and use of limiting amplifiers " Audio Eng. 31.6
(July 1947) 17.
45. Korn, T. S. " Dynamic sound reproduction " Elect. 21.7 (July 1948) 166.
46. Tomkins, A. A. Surgeless volume expansion " W.W. 54.6 (June 1948) 234. Correction 54.9
(Sept. 1948) 347.
48. Tillson, B. J. " Musical acoustics," Part 5, Audio Eng. 31.9 (Oct. 1947) 25.
50. Weeden, W. N. " Simplified volume expansion " W.W. 38.17 (24th April 1936) 407.
51. Tanner, R. H., and V. T. Dickins " Inexpensive volume expansion " W.W. 38.21 (22nd May 1936)
507.
52. Tanner, R. H. " Lamps for volume expansion " W.W. 39.7 (14th August 1936) 146.
53. Sayers, G. " Notes on contrast expansion " W.W. 39.12 (18th Sept. 1936) 313.
54. Weeden, W. N. " Improving the simplified volume expander " W.W. 40.3 (15th Jan. 1937) 68.
55. Weeden, W. N. " Contrast amplification ; a new development " W.W. 39.25 (18th Dec. 1936) 636.
56. " Contrast expansion unit " W.W. 41.24 (9th Dec. 1937) 590.
57. " Contrast expansion unit ; a correction " W.W. 41.25 (16th Dec. 1937) 621.
58. Stevens, B. J. " Low distortion volume expansion using negative feedback " W.E. 15.174 (March
1938) 143.
59. Amos, S. W. " Distortion in radio receivers " Electronic Eng. 14.169 (March 1942) 686.
61. " Light -bulb volume expander " Elect. 9.3 (March 1936) 9.
62. Barber, A. W. " Plate resistance control in vacuum tubes as audio gain control means " Comm.
17.10 (Oct. 1937) 23.
63. Black, W. L., and N. C. Norman " Program- operated level -governing amplifier " Proc. I.R.E. 29.11
(Nov. 1941) 573.
64. Sinnett, C. M. " Practical volume expansion " Elect. 8.11 (Nov. 1935) 14.
65. Paro, H. " Public address avc " Elect. 10.7 (July 1937) 24.
66. Cook, E. G. " A low distortion limiting amplifier " Elect. 12.6 (June 1939) 38.
67. Grimwood, W. K. " Volume compressors for sound recording " Jour. S.M.P.E. 52.1 (Jan. 1949) 49.
68. Wheeler, L. J. " Contrast expansion " W.W. 55.6 (June 1949) 211.
69. Singer, K. " High quality recording electronic mixer " Jour. S.M.P.E. 52.6 (June 1949) 676.
70. Singer, K. " Preselection of variable gain tubes for compressors " Jour. S.M.P.E. 52.6 (June 1949)
684.
71. Licklider, J. C. R., and I. Pollack " Effects of differentiation, integration and infinite peak clipping
upon the intelligibility of speech " J. Acous. Soc. Am. 20.1 (Jan. 1948) 42.
72. A.W.A. Model G51501 (Amalgamated Wireless Australasia Ltd.).
73. Weller, J. A. " A volume limiter for leased -line service " Bell Lab. Rec. 23.3 (March 1945) 72.
74. Hilliard, J. K. " The variable- density film recording system used at MGM studios " Jour. S.M.P.E.
40 (March 1943) 143.
75. Miller, B. F. " Elimination of relative spectral energy distortion in electronic compressors " 39 (Nov.
1942) 317.
76. Smith, W. W. " Premodulation speech clipping and filtering " Q.S.T. 30.2 (Feb. 1946) 46.
77. Smith, J. W., and N. H. Hale, " Let's not overmodulate
1946) 23.
-it isn't necessary " Q.S.T. 30.11 (Nov.
78. Smith, W. W. " More on speech clipping " Q.S.T. 31.3 (March 1947) 18.
79. Mather, N. W. " Clipping and clamping circuits " Elect. 20.7 (July 1947) 111.
80. Kryter, K. D., J. C. R. Locklider and S. S. Stevens " Premodulation clipping in AM voice com-
munication " J. Acous. Soc. Am. 19.1 (Jan. 1947) 125.
81. Winkler, M. R. " Instantaneous deviation control " Elect. 22.9 (Sept. 1949) 97.
82. U.S. patent appin. 793,916 -H. E. Haynes, Variable gain amplifier (RCV 11688).
83. U.S. patent appin. 794,050 -H. J. Woll Variable gain systems (RCV 11504).
84. U.S. patent appin. 768,319 -W. W. H. Dean Peak clipper and indicator therefor (RCA 25944).
85. Hathaway, J. L. " Automatic audio gain controls " Audio Eng. (1) 34.9 (Sept. 1950) 16 ; (2) 34.10
(Oct. 1950) 27.
86. Singer, G. A. " Performance and operation of a new limiting amplifier " Audio Eng. 34.11 (Nov.
1950) 18.
87. Haynes, H. E. " New principle for electronic volume compression " Jour. S.M.P.T.E. 58.2 (Feb.
1952) 137. Reprinted Radiotronics 17.8 (Aug. 1952) 136.
Additional references will be found in the Supplement commencing en page 1475.
CHAPTER 17
763
7. Noise reduction ... ... ... ... ...
8. Lacquer disc home recording (direct playback) . 766
9. Reproduction from transcription discs 769
10. References to lateral disc recording ...
See Refs.
1. Magnetic recording includes magnetic wire and magnetic tape.
258 (Chapter 29), 272, 316, 318.
films. The Philips -
2. Sound on film fords its principal application in cinema transcription.
film system is used to a limited extent for broadcast
Miller engraved special applications. The
Film is also used for embossed lateral recording for some
use of this medium in most other fields. See
high cost of the film precludes the
Ref. 258.
3. Mechanical groove recording is used in various
forms including
(a) the cylinder (e.g. dictaphone)
in this chapter.
(b) the disc, which is the only type of recording considered
Two methods of recording are used with discs :
701
702 (i) METHODS USED IN SOUND RECORDING 17.1
Peak Amplitude A
row,,
-1 ?
f
A1 plitude A
Needle
Tip
Centre line of I
WAVELENGTH
Groove ! rie
D
o s'
FIG. 17. I
FIG. 17. 2
Fig. 17.1. Unmodulated (left) and modulated (right) groove of lateral recording.
Fig. 17.2. Motion of stylus tip with lateral recording.
If the recorded sound is of sine -wave form, the maximum transverse velocity of the
stylus tip will occur at O and zero velocity will occur at B and B`, the two
of travel. This is an example of simple harmonic motion (see Chapter extremities
6 Sect. 4).
f
The maximum transverse velocity is 27rfA where is the frequency in c/s and
the peak amplitude. The r.m.s. transverse velocity is given by 1.41 orfA. A is
The recorded level may be specified in terms of r.m.s. velocities at 1000 c /s,
or in decibels with a reference level 0 db = 1 cm /sec. lateral r.m.s. stylus velocity.
The following tables are based on this usage.
(A) 78 r.p.m.- Cross-over frequency = 500 c/s
Velocity Peak Comments
Level r.m.s. amplitude*
+ 10 db 3.16 cm /sec 0.56 mil R.M.A. Frequency Test Record No. 1(A)
+ 16 db 6.31 cm /sec 1.1 mils R.M.A. Frequency Test Record No. 1(B)
+
+
18
22 db
db 7.94 cm /sec
12.6 cm /sec
1.4 mils
2.2 mils
f max. velocity (1000 c /s).
-i- 26.8 db 22 cm /sec 3.8 mils Max. instantaneous programme peak.
Over constant amplitude portion.
703
17.1 (ii) PRINCIPLES OF LATERAL RECORDING
--
Level r.m.s.
+ 7.5 db 2.4 cm /sec Columbia Test record RD -103 (1000 c /s).
+ 22.9 db 14.0 cm /sec Max. instantaneous programme peak.
-constant
There are two basic methods of recording sounds of different frequencies
velocity and constant amplitude. " Constant velocity "
refers to the maximum
held constant as the
transverse velocity of the stylus tip at the zero axis, this being velocity recording
frequency changes. A diagrammatic representation of constant
for two frequencies is given in Fig. 17.3.
FIG. 17. 4
FIG. 17.3
It may be shown that constant velocity recording has the following characteristics
for constant power at all frequencies -
1. The peak amplitude is inversely proportional
to the frequency.
all frequencies.
2. The maximum slope of the curve is the same for
velocity at any fre-
In the general case when the power is changing, the maximum " electro- magnetic
quency is proportional to the peak amplitude. With an " ideal the output voltage
pickup, which inherently follows a constant velocity characteristic,
at all frequencies.
for sine waveform is proporcional to the maximum velocity wide frequency ratio,
Constant velocity recording is not suitable for use over a very over a range of 8
owing to the large variation in peak amplitudes. For
example,
is 256 to 1.
octaves the ratio of maximum to minimum amplitude
amplitude is
" Constant amplitude " recording indicates that the maximum output (Fig. 17.4).
held constant when the frequency changes, for constant power
to the frequency.
It may be shown that the maximum slope is proportional but is not satis-
Constant amplitude recording is very suitable for low frequencies, because the transverse
factory with large amplitudes at the highest audio frequencies
becomes excessive, leading to distortion in recording and in
velocity of the needle tip is satisfactory over a
reproducing. On the other hand, constant velocity recording
'Over constant amplitude portion.
704 (ii) PRINCIPLES OF LATERAL RECORDING 17.1
limited frequency range of medium or high frequencies. Therefore most
systems employ an approximation to constant amplitude recording
recording
at low frequencies
and an approximation to constant velocity recording for at least part of
and higher frequency range (see Sect. 5). the medium
It may be shown (Ref. 146) that the minimum radius of curvature at the peak
the curve is given by of
p = 0.025a2/A
where p = radius of curvature in inches (1)
A = wavelength of sine -wave
curve in inches
and A = peak amplitude of curve in inches.
Equation (1) may also be put into the form
A = 0.025 d2 /p.
(2)
Any stylus is unable to fit accurately an undulation in the groove with a
radius of
curvature at the point of contact less than its own radius of curvature. We may there-
fore apply eqn. (2), taking p as the radius of curvature of the stylus, to give
mum amplitude (called the critical amplitude) for correct operation, the maxi-
Aerre = 0.025 d2 /p (3)
where p = radius of curvature of stylus, in inches.
But since the r.m.s. transverse velocity is equal to 1.41 -,rfA we can determine
critical velocity in cm /sec, the
Critical velocity r m.s. = 11.3 fAc,« cm /sec (4)
where A ,,;t = critical amplitude in inches.
(Refs. 255, 254). The average post -war shellac record, taking into account all the
various types of records on the market, both English and American, has a maximum
signal to noise ratio* of about 38 db (Ref. 256).
A new vinyl fine groove record can have an average surface noise, measured on a
system whose response is flat on a velocity basis from 500 to 10 000 c /s, about 56 db
below the peak recording level ; if measured on a system whose response is the inverse
of the recording characteristic, the surface noise is approximately 62 db below peak
recording level. The B.B.C. " D Channel " recording has a weighted signal to noise
ratio greater than 60 db :Ref. 297).
The surfacé noise increases with the use of the record. After 100 playings with a
2 ounce pickup some shellac records showed negligible increase in noise level while
others showed 5 db increase, the average being about 2.5 db (Ref. 181). After 200
playings of a fine groove vinyl record (LP) the surface noise increased by 2 db (Ref.
308).
The cost factor limits the use of polyethylene (Ref. 181) which has a scratch level
even lower than that of vinyl. See also Sect. 9(ii) for the measurement of noise in
accordance with NAB standards for transcription records.
(v) Processing
The processing of shellac pressings is described in Refs. 7, 10 and 25. The pro-
cessing of vinyl (LP) records is described in Ref. 308.
the advantage of not requiring adjustments for different inside diameters. It has the
disadvantage of requiring, in most designs, that the stylus, cartridge and arm operate a
spring tension as they move inwards. This presses the stylus against the outside wall
of the grooves and tends to produce differential wear as well as other undesirable
effects.
The second type of trip mechanism operates when the stylus reaches a specific inside_
diameter, and the eccentricity of the inside groove has no effect. The disadvantage
of this arrangement is that there are large differences in the inside diameters of the
various types of records, so that an adjustable (preferably continuously -variable)
control is required. This method is used in 45 r.p.m. record changers, where there
is no eccentric inside groove.
The third type is the eccentric groove.
Difficulty is often experienced in an attempt to incorporate automatic record
changers in high fidelity equipment owing to rumble and, in some cases, hum pickup.
In addition, some high fidelity pickups are unsuitable for use on any automatic record
changers. The choice of a pickup for use with an automatic record changer is neces-
sarily a compromise, and many high fidelity enthusiasts prefer to use a good quality
turntable with manual operation.
Automatic record changers for 45 r.p.m.
The R.C.A. record changer is described in Ref. 263. The tripping mechanism
requires an extremely small lateral force from the pickup arm, since the work of put-
ting the mechanism into cycle is supplied by the moving turntable.
N M
\
I? O
O
iYi
Ñ
-: e I
ó
I
N
I
NO O
I 0
I-I
N
,
d
I
1
T
Ó
-.9..
b
p, rnp°oó°
,ypoONI
$ó P. d
pN°
N,,,¡t(1000 p
d WIQT1(1
I
1
E
o
.-+ Ch 00 0 Ó .r Q tel .r ;d 46 (V ..4 co E
Ó
oo
..
41
>
V o
Ñ
Q
\ñ
Vf to
0p M ó
N '..
G
Ñ Ó 1 .0
O,
'á
.rQ,NOm . Iïi
ch u,
\
1
CI
M1il
I
00nN
OOÓ Nwn
0°0,myp°1°A
Ó tryN Q
F, ..
1
Ó
°4
00
in ZO -HN
nQ' p° NO .8
u
\ Q
..
I
N O N
I I
N
I I
Ó V1 in
1 1
I
V V1 M
^+Inrnpoa
,i -. O O N poin-.n I íQ. I I
tl
.,ornccd N,.,¡ 1(1 CO
OrI/Tffiti d' (1 °° Ñ
6w ú °U
C
1
> Ó
C
b 1+
m ÿ
..
ó
1:.
v
> °°
1
n
w
U L
c
w G
Ç
Q Ñ
L
II em
> m
n
., ,R
y ... -h.
C
ád °
Ú a0+
u GL 1
.
g e
° 8
Mb ch
qô .. >
8Ñ
w^ ,z
00
Ñ
Q
..
Ñ0 a+ pa,
ó á
GG VW
aa
E.
O ó N
.d Q
¢ D
wá.aÑ cd
.!
708 DIMENSIONS OF RECORDS AND GROOVES
(ii) 17.2
The wavelength for 78 r.p.m. is given by
A = (1.3nd) /f (5)
where A = wavelength in inches for 78 r.p.m.
d = groove diameter in inches
f=
and
Frequency
frequency in c /s.
Equation (5) gives the following values
50
- 250 500 10 000 c/s
Wavelength- outermost groove 0.94 0.19 0.094 0.0047 in.
Wavelength-innermost groove 0.31 0.061 0.031 0.0015 in.
The effective width of the stylus with dimensions as in Fig. 17.5B is about 4 mils,
giving a radius of curvature of 2 mils. Using eqn. (3) of Sect. 1,
Atrt, = 0.025 A2/0.002
and for 10 000 c/s the critical amplitude is-
, r,,r = 0.000 28 in. for outermost groove
A°t = 0.000 028 in. for innermost groove.
The critical velocity is given by eqn. (4) of Sect. 1,
Critical velocity r.m.s. = 31.6 cm /sec at outermost groove
= 3.16 cm /sec at innermost groove.
The critical velocity at the outermost groove is more than sufficient for the highest
level of recording, but as the groove diameter decreases a point will be reached where
the peak recorded velocity is greater than the critical velocity-thus leading to dis-
tortion and loss of high frequency response.
A slightly higher critical velocity could be obtained by the use of a somewhat smaller
stylus radius, since the two values are inversely proportional.
The maximum instantaneous peak recorded velocity is about 15 cm /sec (American
practice) so that there is every likelihood that the critical velocity will sometimes be
exceeded on the smaller diameter grooves. The saving feature in practice is that
maximum amplitude does not normally occur at 10 000 c/s with either speech or music,
except when cymbals are recorded at maximum amplitude.
(B) 45 r.p.m.*
The R.C.A. Victor 45 r.p.m. fine groove records are only made with a nominal
diameter of 7 inches, and are primarily intended for ready use in record changers.
The inner portion of the record forms a collar which is thicker than the playing area,
thus preventing the playing surfaces from touching. The groove dimensions are
Width across shoulders 3.0 mils + 0 - 0.5
Angle 92° ± 3°
Radius at bottom of groove not greater than 0.25 mil.
The recommended vertical stylus force is 5 ± 1 grams.
The grooves per inch vary from 178 to 274. The maximum instantaneous pro-
gramme peak recording velocity is 14 cm /second (about 4 db below 78 r.p.m. records).
6.875°3 oosOIA
1=Y
030=-664.
002Mae. -030+0041
z'
7 inch 45 r.p.m. record t 4
1
(Ref. 254). 3 5%8 s
The cross-sectional view of a record is shown in Fig. 17.5A. The outermost music
groove is 61 inch diameter, while the innermost music groove is 4.875 inches mini-
mum diameter -the latter gives 10% intermodulation distortion and a groove velocity
of 11.5 inches per second.
The maximum playing time is 51 minutes. A lead -in groove extends from the
edge of the record and makes from 1 to 11 turns before entering the first music pitch.
There are from 1 to 11 turns of unmodulated music pitch before the start of music.
There is from 0 to I turn of unmodulated music pitch at the end of music. The
*For R.C.A. 45 r.p.m. Extended Play records see Supplement.
17.2 (ii) DIMENSIONS OF RECORDS AND GROOVES 709
lead -out groove makes from i to li turns from the music pitch to 1 -15/16 inch radius,
beyond which it makes from 1 to 2 turns to the concentric circle (3 -13/16 inch dia-
meter).
The central hole is 1.504 ± 0.002 inches in diameter.
References 254, 263, 264, 265.
(C) 33 -1/3 r.p.m. (LP)
The long -playing microgroove records revolve at 33-1/3 r.p.m. and are made with
outside diameters of 12, 10 and 7* inches.
The.R.C.A. Victor records are identical to the 45 r.p.m. records with respect to
groove shape, recording pitch, recorded level and recommended playback stylus.
Dimension Columbia R.C.A. Victor
References to L-P microgroove : 159, 166, 178, 179, 232, 236, 255, 262.
(iii) Styli
(A) Styli used with shellac pressings are in five principal groups :
(1) Ordinary steel needles which are ground to shape by the abrasive in the
record. The exact shape of the needle point when new is not so important as with
permanent or semi -permanent styli. For examples of record wear see Ref. 10 Part 4.
(2) Semi- permanent needles such as chromium plated steel. The needle shape
is important and " shadow -graph " needles are recommended. For the best results
these should not be used for more than one playing of a 12 inch shellac record (for
photograph of wear see Ref. 274). Alloys such as osmium are also used.
'Columbia only.
710 (iii) STYLI 17.2
(3) " Permanent " styli employ a jewel such as a diamond, sapphire or ruby,
usually in the form of a jewel tip. The tips of permanent needles should be accurately
ground to shape, and highly polished. Diamond tipped needles have a life of several
thousands of playings of shellac records for good fidelity. Tungsten carbide is also
used. For stylus wear see Sect. 2(vi).
(4) Fibre needles are used by some enthusiasts in the belief that they reduce
scratch and record wear. While it is true that they reduce scratch they do so by
attenuating all the higher frequencies, and they do this no more efficiently than an
electrical attenuator. The attenuation of a fibre needle at 4000 c/s is 19 db down
as compared with a loud tone steel needle in a typical pickup (Ref. 7). After the first
few grooves they wear sufficiently to occupy the whole of the groove and thereby
spread the weight of a heavy, stiff pickup of old design. If such pickups are used,
contrary to all good advice, then fibre needles are possibly the best compromise be-
cause a soft needle damps down the pronounced high- frequency resonance of the
pickup. There is evidence to indicate that abrasive particles from the record become
embedded in the fibre and thereby cause wear, even though the fibre itself is softer
than the record. With modern pickups having light weight and high lateral and
vertical compliance the wear from a well polished sapphire point is undoubtedly
less than that from a fibre needle.
For photographs of new and worn fibre needles see Ref. 274.
(5) Thorn needles are very much harder than fibre, but the amount of needle
wear is very much dependent upon the pickup compliance and stylus force. With
suitable lightweight pickups, the needle wear is reasonable up to 6 playings (Ref. 309),
but some loss of high frequency response is inevitable. If sharpened with the use of
very fine glasspaper, there is risk of glass dust becoming embedded in the needle-.
this danger may be avoided by the use of a very fine rotary cutting wheel (Ref. 309,
April 1951).
Since thorn needles are normally in contact with the whole of the bottom of the
groove, they will tend to remove all dust from the groove. It therefore seems pro-
bable that the noise from a record which has been played exclusively with thorn
needles will be no greater at the bottom of the groove than on the sides.
Fine point thorn needles have also been used with fine groove (LP) records
(Ref. 309).
The shape of the stylus tip to give optimum results with most English and
American recordings may be taken as (RTMA Standard REC-126 -A, Ref. 260) :
Radius of tip Angle
Metal point 2.7 (+ 0.2 - 0.3) mils 40° to 50°
Sapphire 3.0 (+ 0.2 - 0.3) mils 40° to 50°
A cross -sectional view of a correct size of sapphire stylus point in a typical groove
is given in Fig. 17.5B. It will be seen that the stylus does not reach to the bottom of
the groove ; this is very important because the abrasive dust collects at the bottom
of the groove. In addition, there should be some allowance for the stylus to wear
on the sides without the bottom of the stylus coming too close to the bottom of the
groove.
FIG. 17.58 In some modern records, as well as in trans-
86.5° scription discs, a tip radius of 2 mils may be used,
but in other (and particularly older) recordings it is
likely to scrape the groove bottom and suffer from
Ramus
0.0030 " groove skating " and single point contact.
One interesting suggestion is to combine a 2 mil
0.0029
Batton, of radius with a " flattened " extremity (actually a
Groove nos
3.5 mil radius). Although more expensive to pro-
Radios 0.001
duce it seems to have distinct advantages (Ref. 244).
Fig. 17.5B. Cross-sectional Oval sapphire styli are available with a minor
view of typical record groove axis of 1.5 to 2.0 mils and a major axis of 2.5 to 3.5
and needle tip. mils. These should be used with the major axis
17.2 (iii) STYLI 711
at right angles to the centre -line of the groove, and will permit better response to high -
amplitude high- frequency recordings.
The N.A.B. " secondary standard " stylus (permanent point) has an angle of 40°
to 55° and a bottom radius of 2.5 ± 0.1 mils. It provides a compromise suitable
for the reproduction of both transcriptions and shellac discs.
The use of larger styli (radius 4 mils) is advocated by some engineers (Refs. 223, 225)
but it does not appear to be the best all-round compromise.
Needles are made in various sizes. The ordinary steel needles used in mechanical
reproducers are used in the stiff heavy weight pickups whose frequency response
does not exceed about 6000 c/s ; they are also used in some " needle armature "
pickups. For the best high frequency response the " loud tone " or " full tone "
needles should be used, because of their smaller size and weight. " Soft tone " and
" trailing " needles (as used with acetate discs) cause severe treble attenuation when
used with typical pickups. Most light weight pickups with a frequency response
extending to 10 000 c/s or over use miniature needles to enable the armature resonance
to be at a high frequency. A typical example is the H.M.V. Silent Stylus or Columbia
99 (chromium- plated long -playing). Thorn needles are also made of similar size.
Efforts have been made to develop a stylus tip suitable for both 78 r.p.m and fine
groove (see Refs. 279, 281) but a considerable degree of compromise is necessary
and the best results are not obtainable from either type of recording.
(B) Styli for fine groove records
With a few exceptions, all styli for fine groove records are permanent types-usually
jewels. Diamond tips are the only really satisfactory ones for long life, although
expensive to purchase in the first case. Sapphire tips are very common,
but wear rather rapidly-see Sect. 2(vi) for stylus wear. Other materials used are
tungsten carbide, osmium and other metal alloys.
Dimensions of styli for fine groove records
(RTMA Standard REC- 126 -A, Ref. 260, for home phonographs)
Radius of tip 0.001 + 0.0001 -
0.0002 inch
Included angle of tip 40° to 50 °.
Note : With a 90° groove angle, a tip with a radius of 0.001 inch has an effective
radius of 0.0007 at the point of contact.
(C) Colour codes for styli
1. RTMA REC -126 -A for home phonographs (Ref. 260). Needles with a 0.001
inch radius shall be colour coded red.
2. English Gramophone Equipment Panel of the Radio and Electronic Component
Manufacturers' Federation (Jan. 1951). Red -0.0010 in. ; lemon-0.0020 in. ;
green -0.0025 in. ; french blue-0.0030 in. ; orange -0.0035 in. ; violet -universal ;
sky blue-oval tip. Material (marked by band on shaft) black-hard metal ; white
diamond ; no colour-sapphire.
-
(iv) Pinch effect
Owing to the fact that the cutting is done by a stylus having an effectively flat cutting
face, the width of the groove measured at right angles to the groove is narrower at two
places in each cycle (see Sect. 6). As a result of this effect, the stylus tip should rise
and fall twice in each cycle -only a limited number of pickups, however, make ade-
quate provision for this movement.
The flexibility of the stylus, if mounted at an acute angle to the record, together
with that of the record itself, tends to prevent the needle from riding merely the peaks
of the vertical undulations. The pinch effect undoubtedly increases needle wear
and possibly also record wear (Ref. 212).
Various forms of " bent shank " needles have been developed to provide some
vertical compliance. All " bent shank " needles tend to give a drooping high -fre-
quency response the amount of which varies considerably from one make to another ;
i t may, of course, be compensated if desired.
(v) Radius compensation
Radius compensation does not appear to be used with shellac discs, although it
is used with LP discs (R.C.A. Victor) and in transcription discs.
712 (vi) RECORD AND STYLUS WEAR 17.2
the groove. At this point with a pickup having low stylus force, the wear apparently
stops. This wear does not produce any apparent change in quality of the reproduced
sound, although it can be seen as a light streak on the surface of the record. Both
lateral and vertical forces must be considered in producing the wear (Ref. 155). Where -
wear.
ever there is excessive stylus wear, it is always accompanied by excessive record
It is important to distinguish between visible wear and audible wear. Visible
wear does not necessarily cause any audible defect. It is possible that visible wear
actual
may occur earlier with a diamond tip, but audible wear, which is caused by
widening of the groove at corners, will occur far more rapidly with a worn sapphire
or metal tip. The flats on the sides act as scrapers in attempting to negotiate sharp
bends in the record groove. A well polished spherical diamond or sapphire tip is
believed to cause less record wear than a metal alloy tip.
Wear of vinyl records
Wear of vinyl fine groove records is very slight, provided that the stylus force does
not exceed about 7 grams, that a pickup of ample lateral and vertical compliance is
used, that the jewel tip is well polished, that there are no marked mechanical reson-
ances, and that dust is excluded. Dust is the most important cause of record and
stylus wear with vinyl records in the home, and care should be taken to reduce it to
a minimum -see Sect. 2(i). Wear in vinyl records shows itself principally in widen-
ing of the groove thus leading eventually to distortion and rattles (Ref. 155).
Wear of sapphire styli
The wear of sapphire styli is a function of the material of which the pressings are
made, the lateral and vertical compliance of the pickup at the stylus tip, the dynamic
mass of the pickup at the stylus tip, the characteristics of the armature (high frequency)
resonance and the pickup arm (low frequency) resonance, and the stylus force. Ad-
ditional wear may be caused by " skating " (due to the use of a tip radius which is too
small for the groove), by insufficient vertical force to maintain the stylus always in
contact with the groove, and by warped or eccentric records or turntables. When
using a pickup with high lateral and vertical compliance and low dynamic mass at the
stylus tip, an increase of stylus force has only a minor effect in increasing the wear on
the stylus tip. It has been shown (Ref. 282) that with such a pickup (GP20) on a
heavily modulated shellac record, an increase from 7.5 to 14.5 grams in stylus force
causes increased flats on the stylus tip resulting in a drop of only 3 db at 10 000 c/s
on a test record. Such a stylus and pickup may be used for 800 playings of a 12 inch
shellac record with reasonably good fidelity, or 2000 playings with fair fidelity (- 5
db at 10 000 c/s on a test record).
On the other hand a typical light- weight pickup with a stylus force of 1 ounce (28
grams) has been shown to produce wear on sapphire styli on shellac records as follows
(Ref. 59).
Wear just noticeable after 50 playings of 12 inch disc.
1 mil flats after 200 playings 1 The effect of fl
flats on distortion is covered
2 mil flats after 750 playings Jr in Seto flats
2.5 mil flats after 1500 playings
It is obvious that this pickup produces much greater wear than the one with high
lateral and vertical compliance (GP20) described earlier. It is probable that the
greater wear is due principally to the lower lateral and vertical compliance, although
the increased stylus force would also be a contributing factor.
Some excellent photographs of stylus wear are given in Refs. 274 and 290.
The use of a test record at, say, 10 000 c/s to indicate wear of the stylus tip requires
careful interpretation of the results. The level of recording of the test record is lower
than that which may be reached with music, while the effect of wear of the stylus tip
is shown less prominently towards the outer diameter of the record than it would be
with a smaller groove diameter.
It is wise to avoid using sapphire or other permanent stylus tips on records which
have been played previously with steel needles. For a photograph of wear after 20
playings under such conditions see Ref. 274.
714 (vi) RECORD AND STYLUS WEAR 17.2
SECTION 3 : PICKUPS
(i) General survey (ii) Electro- magnetic (moving iron) pickups (iii) Dynamic
(moving coil) pickups (iv) Piezo -electric (crystal) pickups (v) Magnetostriction pickups
(vi) Strain -sensitive pickups (vii) Ribbon pickups (viii) Capacitance pickups (ix)
Eddy -current pickups.
(i) General survey
The performance expected from good quality pickups includes-
1. Wide frequency response with the minimum of equalizing. No sharp peaks
over + 2 db. For most applications a slight droop in the high frequency response
is not undesirable as it may be used to compensate for the pre -emphasis in most
recordings.
2. Low stylus pressure.
3. High lateral compliance.
4. High vertical compliance at stylus tip.
5. Low effective vibrating mass of armature at high frequencies.
6. Fairly high output level.
7. Low distortion.
8. Freedom from major resonances over the useful frequency range (see below).
The low frequency (arm) resonance should be below 30 c/s so that it will have a negli-
gible effect on record wear. The high frequency (armature) resonance should be
as high as possible-certainly over 8000 c/s -and preferably above the useful frequency
range. Even so, the resonance should be effectively damped.
9. Minimum " needle talk " (direct acoustical radiation).
10. High signal to hum ratio.
11. Negligibly small voltage generation from vertical movement of the stylus tip.
12. General ruggedness.
13. Freedom from the effects of excessive humidity. Pickups having a restricted
frequency range (say from 70 to 6000 c /s) should meet the same requirements as
good quality pickups, apart from frequency.
17.3 PICKUPS-(i) GENERAL SURVEY 715
Pickups which have sufficient vertical compliance are sometimes fitted with needle
guards to reduce or prevent damage to the pickup from accidentally dropping it.
Pickup manufacturers sometimes quote the minimum vertical stylus force to pro-
vide perfect tracking. It is generally regarding as sound practice to adopt a stylus
force at least 1.5 times this minimum value, thus making some provision for warped
and eccentric records.
For styli, see Sect. 2(iii).
Single purpose pickups for home use often incorporate a fixed spring counter-
balance. Dual purpose pickups usually apply a higher stylus pressure for 78 r.p.m.
than for fine groove records. In some cases the pickup head (or cartridge) includes
the additional weight for 78 r.p.m., in other cases this is provided by an adjustment
of spring tension or by a counterbalancing weight.
It is desirable in all cases to keep the moment of inertia about both vertical and hori-
zontal pivots to a minimum.
Where a spring counterbalance is used, it should be checked periodically for pres-
sure ; during weighing, the height of the stylus from the baseboard must be exactly
the same as the top of the record.
Up and down movements of the stylus tip, due to the pinch effect and small record
surface irregularities should be absorbed by the vertical compliance of the stylus,
without generating any output noise.
Pickups for 78 r.p.m.
A stylus force not more than 1 ounce (say 30 grams) is desirable, with 2 ounces as
Resonances in pickups
Most pickups have two major resonances. The arm resonance occurs at some
frequency below 100 c /s, while the armature resonance occurs at some frequency
about 3000 c /s. Both resonances have deleterious effects in increasing record wear.
The arm resonance tends to cause the pickup to jump out of the groove when a high
amplitude is recorded at the frequency of resonance -the frequency of resonance
(if undamped) should be well below the lowest frequency for which acceptable track-
ing is required ; say from 15 to 25 c /s.
A recent development is the use of pickup arms with high viscosity oil or other
viscous fluid to damp arm resonance. The damping may be applied as a viscous
film in a ball and socket (hemispherical) joint between the pickup arm and the mount-
ing socket, and gives damping in both horizontal and vertical directions. The arm
is pivoted at a point which is the centre of both ball and socket, and above the centre
of gravity of the arm. The clearance between the two surfaces is about 0.006 inch.
By this means the resonant force is greatly reduced, improved resistance to external
shock is obtained and protection against damage from accidental dropping of the pickup
head is achieved. The amount of mechanical resistance is not a critical value
the upper limit is reached when it interferes with the tracking of records having reason-
-
ably small values of eccentricity or warpage. With the usual values of suspension
compliance and mass, this would occur at several times the amount of resistance neces-
sary to give critical damping of the arm resonance. Even if the damping is con-
siderably below critical, it still has quite a beneficial effect. The viscosity of the fluid
is a function of temperature, but the variation does not seriously affect the performance
over a reasonable temperature range. Ref. 311.
The armature resonance has a serious effect on needle scratch (see Sect. 1).
" During operation the needle is subjected to a continuous shower of blows, and
although highly damped by rubber buffers, it is in a state of perpetual oscillation at its
own resonance frequency " (Ref. 229). In any quality pickup, the armature resonance
with the recommended needle should be over 8000 c/s ; many light -weight pickups
have the resonance at a frequency over 12 000 c /s-in some cases over 18 000 c /s.
Pickups tend to be rich in harmonics at frequencies in the region of half the armature
resonance frequency. In all cases the armature resonance should be effectively
damped. Damping material used to obtain smooth response characteristics may
adversely affect the tracking capabilities of the pickup, and therefore should be in-
vestigated carefully and used judiciously (Ref. 285).
The frequency of the armature resonance is affected by the record material-for
example, a change from a lacquer disc to a shellac pressing causes in one case an
increase in the resonant frequency in the ratio 1: 2 approximately (Ref. 276). A
pickup with an armature resonance well outside the a -f range on shellac discs, shows
a much lower resonance frequency on vinyl -usually inside the a -f range. This is
an unfortunate characteristic of vinyl. The only pickups known to the author with
armature resonances above 15 000 c/s on vinyl fine groove recordings are of the ribbon
armature type.
There is a third resonance caused by the pickup head and arm rotating about the
arm. With heavy pickups this resonance occurs at frequencies between 100 and 400
c /s, but with light -weight pickups the frequency would be higher. This resonance
does not have any effect on the general tonal balance, and in any case should be very
slight with good mechanical design.
Crystal pickups have a fourth resonance which occurs at a high frequency-the
resonance of the crystal itself.
References to resonances in pickups : 10 (Part 2), 17 (second letter), 276, 283, 285,
311, 331.
The testing of pickups
The testing of pickups for frequency response characteristics is covered by R.M.A.
REC -125 -A which states that the test record shall be R.M.A. Frequency Test Record
17.3 PICKUPS-(i) GENERAL SURVEY 717
No. when available. This is recorded on side A at a r.m.s. velocity of 3.16 cm /sec
1
(+ 10 db). Crystal pickups are terminated by a load resistance of 1 megohm (or
5 megohms for ammonium phosphate crystals) shunted by a capacitance of 100 µµF.
Other types of pickups should be terminated as required. The response should be
stated in decibels (0 db = 1 volt).
Pickups should be tested under the same conditions with which they will operate
in normal service. Pickups intended to be used on shellac records should be tested
on shellac discs ; pickups intended for use on vinyl records should be tested on vinyl
discs, while those intended for use on either type of disc should be tested separately
on each. Pickups which are satisfactory on vinyl discs give, in some cases, quite
poor results on shellac discs owing to the increased stiffness of the material and the
reduction in damping by the record. This also applies to pickups with very limited
vertical compliance -while passable on vinyl, they are poor on shellac.
The variable speed turntable is sometimes used as an alternative to the standard
frequency records for the calibration of pickups-see Refs. 273, 295.
In the design and production of pickups it is usual to supply some form of electro-
mechanical calibrator such as in Ref. 284.
For distortion in pickups and the procedure for determining the " tracking "
capabilities of a pickup see Sect. 6(vi).
design is such that magnetic fields cause very little PICKUP 22 .B2Ma
I-300v
hum pickup. 05FF
Mn 1Mn
25
(vi) Strain- sensitive pickups rF
00ñ VOLUME
OSFF
This pickup is based on the principle that the a7Mn
1Mn
4 77Mn
resistance of a conductor changes when the con-
ductor is strained. Direct current is passed 00350F
56,0000 ,5000
through the conductor while in operation. Some
low impedance designs are described in Ref. 234 ; 50,0000
TREBLE
these all require a step -up transformer.
A recent high-impedance commercial type is
the Pfanstiehl which has a resistance of about Fig. 17.6. Pre-amplifier for use
250 000 ohms, giving an output of about 10 to with strain -sensitive pickup (Ref.
15 millivolts and a noise level of about 5 micro- 336)
722 (vii) RIBBON PICKUPS 17.3
volts. Since the pickup is a constant -amplitude device, special methods are required
for frequency compensation. The pre -amplifier of Fig. 17.6 gives the required fre-
quency compensation with continuously -variable controls (see Ref. 336 for settings
for some American records) and an output of about 2 volts with types 6AU6 and 12AU7.
The final stage is a cathode follower. Refs. 96, 303, 336.
provided by R1C,. The capacitance pickup is connected between grid and plate,
and any variation in capacitance will affect the amount of feedback and therefore also
the amplitude of oscillation and the voltage drop across R1. The latter, filtered by
RFC and C,, provides the a -f output which is of the order of 1 volt (Ref. 119).
References to capacitance pickups : 14, 45, 50, 119.
SECTION 4 : TRACKING
(i) General survey of the problem (ii) How to design for minimum distortion (iii)
The influence of stylus friction.
where w = 271
A =
maximum groove amplitude in inches
a =
tracking error expressed in radians (= angle in degrees divided by 57.3)
V =
longitudinal groove velocity in inches per second.
and v =
maximum transverse velocity in inches per second = wA.
Equations (1) and (2) indicate that
(1) Distortion is directly proportional to the maximum transverse velocity.
(2) Distortion is directly proportional to the maximum groove amplitude at any
one frequency.
(3) Distortion is directly proportional to the tracking angle.
(4) Distortion is inversely proportional to the revolutions per minute. For equal
distortion, more careful tracking angle correction is required with 33 -1/3 r.p.m. than
with 78 r.p.m., other conditions being the same
(5) Distortion is constant over the " constant velocity " portion of a recording
characteristic, other conditions being unaltered
(6) The angular tracking error may be increased in the same proportion that the
radius is increased, for the same distortion (roughly 3 1 ratio over a 12 inch record).
:
Example of distortion
Consider the distortion with a 12 inch 78 r.p.m. record having 2° angular tracking
at the innermost groove, recorded with a sine wave having a peak amplitude of 0.002
inch at a frequency of 250 c/s (this being the cross -over frequency), the r.m.s. velocity
being 6.31 cm /sec.
Here a = 2/57.3 radians ; A = 0.002 ; a, = 27r x 250 ; r = 1.875 inches ;
V = 2rrr(78 /60) = 15.3 inches /second.
From equation (1)
2.71. x 250 x 0.002 x 2 x 100
Percentage 2nd harmonic =
57.3 x 15.3
0.72% . -
If the cross -over frequency had been 500 c /s, with the same recorded velocity
(i.e. half the amplitude), the distortion would have been the same. Some frequency
test records are recorded with a velocity less than 6.31 cm /sec, so that the distortion
due to tracking error would be less than the value stated above. The harmonic
distortion as determined for sinusoidal signals gives a fair estimate of the relative
tracking distortion produced by complex signals over the " constant velocity "
portion of the characteristic.
Characteristic for
Constant Velocity Constant Amplitude
Characteristic of Needle Point
0 0
1000 5,000 10000 X000 5,000 10,000
Frequency (c/s) Frequency (c /s)
FIG. 17.9 FIG. 17. 0
Fig. 17.9. Constant velocity recording characteristics showing relative variation of
groove amplitude with frequency.
Fig. 17.10. Frequency characteristic of ideal constant velocity pickup, for constant
amplitude of needle point.
The constant velocity recording characteristic provides a groove amplitude which is
inversely proportional to frequency (Fig. 17.9). The ideal constant velocity pickup
(a high-fidelity electro-magnetic type is a close approximation) provides an output
voltage which is proportional to the frequency, for constant amplitude at the stylus
point. It follows that with tracking distortion (or any other form of distortion caused
by the groove, needle, or pickup) the harmonics are accentuated in proportion to
their frequencies. This effect has been taken into account in the derivation of equa-
tions (1) and (2).
17.4 TRACKING-(i) GENERAL SURVEY 725
It has been proposed (Ref. 228) that a suitable upper limit for tracking distortion
is 2% with 0.001 inch peak amplitude as for transcriptions or 4% with 0.002 inch
peak amplitude as for shellac discs ; it seems that 1% is a preferable limiting value
for good fidelity. On this latter basis the maximum tracking error for 12 inch discs
will be 2.8° for 250 c/s cross -over frequency, or 1.4° for 500 c/s crossover frequency,
both measured at the innermost groove, or three times these values at the outermost
groove.
No additional record wear occurs due to tracking error with permanent needles
having spherical tips. Additional record wear may occur with steel needles or with
worn sapphire tips if the tracking error is serious.
(ii) How to design for minimum distortion
A straight -arm pickup (not offset) is shown in Fig. 17.11 where D represents the
distance from the centre of the record to the pivot of the pickup arm, L represents
the length of the arm measured from the pivot to the needle point while r, and r,
represent the radii of the outermost and innermost grooves. In this example the
pickup is mounted so that the needle point will pass over the centre of the record,
but this is not the position giving the best results. It is obvious that the axis of the
pickup is not a tangent to the groove.
Minimum distortion is always obtained from a straight-arm pickup when it is
" underhung," that is when the needle point comes short of the centre of the record
by a small distance d, the optimum value of which is given below.
FIG. 17.12
These values for d° D, are critical, and should be measured accurately. It is safer
to keep below than to go above the optimum overhang.
The angular tracking error x at any position on the record is the angle between
the axis of the pickup and the tangent to the groove at the needle point. (Figs. 17.11
and 17.12).
With a straight arm, using any value of underhang (Fig. 17.11),
x ti 57.3(2L +)
with an error less than 1°.
(4)
The optimum value of underhang is given by eqn. (3) ; using this value and also
the values of r, and r, as for 12 inch discs, with r = r, (innermost groove) and L = 8
inches, the corresponding value of tracking error is approximately 18.6 °. The dis-
tortion is therefore approximately 6.7% at this point, on the constant velocity charac-
teristic.
Fig. 17.12 shows a pickup with an offset arm for the purpose of reducing dis-
tortion due to tracking error. This differs from the straight arm in that there is an
offset angle ß between the axis of the pickup and the straight line joining the pivot
to the needle point. Note that this angle ß is not equal to the angle of the bend in
the arm. In this case minimum distortion is always obtained when the pickup is
" overhung," that is when the needle point passes beyond the centre of the record
by a distance -d, the optimum value of which is given below, on the assumption
that the optimum value of offset angle is used.
When a pickup is mounted to provide minimum tracking distortion when used
with ordinary needles, the use of bent -shank or trailing type needles will seriously
affect the tracking distortion. In cases where either type of needle may be used, a
compromise may be necessary. When bent -shank needles only will be used, the
increased length of arm should be allowed for.
Offset arm (optimum offset angle) (Fig. 17.12)
tens
Optimum overhang = -d °9, - LL
(rl + r,)' +rira]
(5)
For 10 inch discs -d°,,, = 3.99/L.
For 12 inch discs -d °y, = 4.60/L.
When L = 8 inches :
For 10 inch discs -d°D, = 0.499 inch.
For 12 inch discs -d.,= 0.575 inch.
The optimum offset angle (ß in Fig. 17.12) is given by
sin ß °- rirs(r, + r,)
L[i(ri + rs)' + r,rsl
(6)
For 10 inch discs sin ß °D, = 2.96/L.
For 12 inch discs sin goy, = 3.26/L.
When L = 8 inches :
For 10 inch discs ß °D, = 21 °41'.
For 12 inch discs 1307, = 24 °3'.
The harmonic distortion with optimum overhang and offset angle is given for 12
inch discs approxima',:ly by
5.5
Percentage 2nd harmonic distortion
. 1/L$ - 11.6
If L = 8 inches, 2nd harmonic distortion 0.76%.
FIG. 17.13
The second factor is the peak frequency spectrum characteristic of the music being
recorded. The effect of the decreased amplitudes at high frequencies in speech and
music has been investigated (notably Ref. 42) but there is insufficient information
available concerning the instantaneous distribution of energy within the recorded
spectrum. Curves based on the work of Sivian, Dunn and White indicate that " no
large increase in distortion occurs in reproducing the 16 db pre -emphasized continuous
spectrum . . . yet the addition of a few prominent tones to this spectrum in the
region above 2000 cycles will result in intolerable distortion " (Ref. 42).
Pre -emphasis on shellac discs
If the tracking* difficulty could be overcome, it is possible that the N.A.B. pre -
emphasis characteristic would not be excessive (Ref. 193) but unfortunately it is the
tracking* problem which is the stumbling block. In one test, three records were cut
simultaneously with different values of pre- emphasis and played back on a system with
the maximum de- emphasis. The record with the 6 db pre -emphasis " reproduced
more highs and cleaner highs than the one with the 15, indicating that the overload on
the latter was so bad that it was not being tracked* " (Ref. 193, discussion by J. P.
Maxfield).
The use of the full N.A.B. pre- emphasis (16 db at 10 000 c /s) on standard shellac
discs has come in for much criticism. It is claimed that the cymbals and certain
brass instruments will overload the system at high frequencies, although this could be
taken care of by the use of a limiting amplifier which reduces the gain for the short
period required (Ref. 197). Another writer states that trouble had been experienced
through overmodulation by second and third harmonics of the soprano voice when a
rising characteristic between 1000 and 5000 c/s had been introduced (Ref. 174).
Another writer refers to the " muddiness " and " smearing " in the high level, full -
orchestra passages of many records (Ref. 117). Still another writer states that the
present N.A.B. pre -emphasis curve effectively guarantees excessive distortion and
he refers to the insufficient attention which is paid to the difference between transient
response tests and the steady state (Ref. 257). Another refers to the fact that the
N.A.B. characteristic unduly weights the importance of the signal to noise ratio and
produces the undesirable condition of signals having high velocity with high amplitude
(Ref. 189). Another again states " since the pickup stylus can track* the recorded
high frequencies more clearly without excessive pre -emphasis, the high frequency
reproduction from such records is notable for its clarity " (Ref. 241). Other criticisms
have also been published (Refs. 155, 174).
It is interesting to note that one leading American manufacturer, R. C. A. Victor,
limits the high-frequency pre -emphasis on home records to a maximum value of
12.5 db at 10 000 c/s in place of 16 db as with the N.A.B. characteristic, while the
AES Standard Playback Curve provides for 12 db pre- emphasis at 10 000 c /s.
Pre -emphasis with fine groove recordings
Owing to the lower recording level, smaller radius stylus tip and more elastic
material with fine groove records, a higher value of high frequency pre- emphasis
may be used than with 78 r.p.m. shellac discs, for the same distortion in both cases.
Measurement of recorded velocities
The recorded velocities may be measured either by the light -pattern method (see
(v) below) or by the use of a special high -fidelity pickup. These two methods do
not give the same result, owing partly to the mechanical characteristics of the record
material and partly to the finite needle size which causes tracing. distortion (see Sect. 6).
It is current practice to use the special pickup at low frequencies and the light -pattern
method at high frequencies ; the latter may be checked by the pickup after allowing
for the loss caused by tracing distortion. It has been stated that the mechanical
impedance limits at the reproducing point should be included in any standardization
of frequency characteristic (Ref. 131).
'It is unfortunate that the word " tracking " should be used both here (meaning that the stylus tip
is capable of following the modulations in the groove) and also in " tracking distortion " which is due
to the angle between the axis of the pickup and the tangent to the unmodulated groove (see Sect. 4).
No alternative nomenclature seems to be in current use.
730 (i) RECORDING CHARACTERISTICS 17.5
+5
o
.......
i . _... ,.
n
1C
db
10
s
'
`
Curve E.M.I.
E.M. 1. 7B .p. m.
_ \\
® ,/,' 8 E.M.1. Studios Special Recordings
15 ODecca ff rr
B.B.C. Transcription
20
s r
100 1,000 10,000
Frequency (c /s) FIG. 17.14
Fig. 17.14. Recording characteristics used by all English, Australian and the majority
of European record manufacturers for 78 r p.m., together with B.B.C. transcriptions.
+ 20 +20
Curve I :- Columbia LP micro groove 5,!
5
: -NAB (lateral
í
+1 Curve 2 1949) L2 +15
Curve 3 :-R.C.A. 78,45 It 33 I'S r.p.m.
+1 Curve 4 :-Columbia 78(follows I above 1,000 c /s) 3
+10
Curve 5 :- Ortacoustic
+5
Curve A :- 6db octave (cross over 300c/s) +5
Curve B :- 6db octave (cross over 500c/s)
db
-
-5
-15
I
0
Fri,
0
I.
.11111111111711111.-
its
0 08
10
15
-20
3{
I 20
records. Curve 2 is the N.A.B. (lateral 1949) standard for transcriptions. Curve 3
is that used by R.C.A Victor for 78 r.p.m. shellac discs, 45 r.p.m. and 33 -1/3 r.p.m.
fine groove. Curve 4 is that used by Columbia for 78 r.p.m. shellac discs-the cross-
over frequency is 300 c /s. Curve 5 is that used for Orthacoustic transcription re-
cording-it follows very closely the N.A.B. curve from 60 to 10 000 c/s but is ex-
tended down to 30 c/s (-
20 db) and up to 15 000 c /s.
Standard Playback Curve
For many reasons it has been impossible to achieve a standard recording character-
istic, even in a single country and for a specified cross -over frequency. An entirely
different approach has been made by the Audio Engineering Society of U.S.A. which
has put forward a Standard Playback Curve with the idea of getting this adopted by
all designers of equipment for reproduction from records (Ref. 307). The onus would
then be on the record manufacturers to produce records which sound well when
played with such equipment. The curve (Fig. 17.15A) is based on the frequency
of 1000 c/s as a reference point, and the de- emphasis at 10 000 c/s is 12 db, being
less than the N.A.B. de- emphasis at this frequency. However the AES playback
curve is extended to 15 000 c/s with a de- emphasis of 15.5 db. Both the straight
portions of the curve have slopes of 6 db /octave, and the intersections of the extensions
of these straight portions with the reference axis occur at 400 c/s (the cross -over
frequency) and 2500 c /s.
This playback curve may be duplicated on a flat amplifier with two sections of RC
equalization, as shown in Fig. 17.15B, which is one possible arrangement. Alter-
natively the network of Fig. 17.15C may be used.
+ 20
+ lo NNN.
0
V
W
G
10
20
20 20000
100 1000 10000
It seems that this Standard Playback Curve is what is really needed by the designers
of equipment for reproduction from records. Even if record manufacturers do
not accept it, the error for any cross -over frequency from 325 to 500 c/s is not more
than 2 db, and no problems will be encountered in the reproduction of NAB recording,
all fine groove records and most 78 r.p.m. discs except those with a cross -over fre-
quency of 250 c/s and without any high frequency pre- emphasis. The latter may be
covered by a separate network.
The C.C.I.R, (Geneva, June 1951) proposed a compromise characteristic with a 450
µsec. curve below 1000 c/s and a 50 µsec. curve above 1000 c /s, giving turnover
frequencies of 360 and 2800 c /s, for radio programmes for international exchange.
3 Z L3 C3 R1 R2
(Ohms) (H) (NF) (Ohms) (Ohms)
150 0.545 24.08 12r 30
250 0.910 14.44 204 50
500 1.81 7.22 408 101
FIG. 17.15C
Cross -over frequency 250 300 350 400 500 800 c/s
Bass boost at 70 c/s 11.0 12.6 13.8 15.1 17.1 21.1 db
Bass boost at 50 c/s 14.0 15.6 16 8 18.0 20.0 24.0 db
Bass boost at 30 c/s 18.4 20.0 21.2 22.5 24.5 28.5 db
Cheap pickups usually have a peak in the bass region which reduces the amount
of bass boosting required, and in some cases no bass boosting whatever is provided
by the amplifier.
Some practical recording characteristics have an approach towards a constant
velocity characteristic below some low frequency (e.g. Fig. 17.15 Curves 1, 2 and 5).
The equalizing circuit should be designed to suit the individual recording characteris-
tic. The following treatment, however, is based on the 6 db /octave constant ampli-
tude characteristic as approached by Fig. 17.14 Curve 1 and Fig. 17.15 Curves 3 and 4.
20 / I 66 Octave
19
' s .C Ie
.`
Two ldentical Stages
`
16
B
\
Singk
Stage
14
12
10
I
8
ó
ú
6
\
4 \\\`
y
2 `1 .........
20 25 30
40 50 60
50
100
70
140
100 125 150 200 250
200 2S0 300 400 500
500
Ip00
750 p
Iy500 2,O00
Fequency In c/a FIG. 17.16
Fig. 17.16. Bass equalizer characteristics (A) " ideal " characteristic with slope of
6 db /octave (B) characteristic with single stage r-c equalizer having total boost of 20 db
and max. slope 4.9 db /octave (C) characteristic with two r.c. stages in cascade, each
having total boost 16 db and max. slope 3 db /octave (circuit as Fig. 17.17).
Conventional resistance -capacitance equalizing circuits
The form of bass boosting which is normally used results in a " saturation " shape
of curve as in Fig. 15.4. Maximum slope is obtained at the point of half total boost,
the value being a function of the total boost (Chapter 15 Sect. 2) :
Total boost 20 15 10 6 3 db
Slope at half-boost point 4.9 4.1 3.0 2.0 1.0 db /octave
The desired slope is 6 db /octave, so that there is an appreciable error even with
a total boost of 20 db or more. This is plotted in Fig. 17.16 for a total boost of 20
db with alternative frequency scales for 250 and 500 c/s cross -over frequencies. Witha
-
cross -over frequency of 250 c/s the error is about 1 db at 50 c/s and 2.5 db at 30 c /s, -
which is generally acceptable, even though there is inevitably some additional loss in
734 (iii) INTRODUCTION TO EQUALIZERS 17.5
the extreme bass, due to coupling condensers. In the case of a cross -over frequency
of 500 c /s, however, the error is - 3 db at 50 c/s and - 4 db at 40 c /s, here again
increased by the effect of coupling condensers. Some designers adopt a total boost
of as much as 40 db to provide a satisfactory characteristic for use with high cross-
over frequencies such as 800 c /s. In either case, it is good practice for the amplifier
to include a bass tone control which will permit an adjustment to suit the circumstances.
If no tone control is fitted, it is usually desirable to attenuate the extremely low fre-
quencies ; a value of coupling capacitance may be chosen to give, for example, an
additional 3 db attenuation at 70 c/s and 5 db at 50 c /s.
In the circuit of Fig. 15.3 suitable values of components would be :
V1 = 6J5 (half 6SN7-GT) ; RL = 50 000 ohms ; R1 = 20 000 ohms ; .R5 =
1 megohm ; R, = 2800 ohms ; C = 0.25 µF for very good bass response or
0.05 µF for bass attenuation. Voltage gain at 1000 c/s = 1.5 times = + 3.5 db.
Total boost = 20 db. C2 = 0.2 µF for cross -over 250 c/s or 0 1µF for 500 c /s.
Alternatively the total boost of 20 db could be obtained in two separate stages
each having a total boost of 1.1 db at the cross -over frequency. By this means a
maximum slope of 6 db is obtained at a frequency of 100 or 200 c/s with a cross-over
frequency of 250 or 500 c/s and more accurate equalizing is possible. The frequency
characteristic obtained is shown with a broken line (C) in Fig. 17.16, and the circuit
diagram is given in Fig. 17.17.
Fig. 17.17. Circuit diagram
of two stage bass equalizing
amplifier with bass boosting
in each stage. Pin 4 of the
6SN7 -GT should be used for
the input circuit. Total boost
is 20 db and max. slope 6
db /octave. C = 0.25 µF
for good bass response or 0.1
INPUT
µF for some bass attenuation ;
óó IMn OUTPU1
C, = 0.1µF for cross -over
IMn
250 c/s or 0.05 µF for 500 c/s.
FIG n n
Voltage gain at 1000 c/s = 22
times = 26.8 dbvg.
Instead of having two networks separated by a valve, the complete two- section
equalizer may be incorporated into a single network as in Fig. 17.18. The maximum
slope is 7 db /octave which is adjustable down to 3 db /octave by means of control
which should be tapered ; maximum slope is obtained with maximum value of R,.
R
The plate load resistors may be increased if it is desired to increase the gain. This
circuit provides 25 db bass boosting with a very close approach to the true 6 db octave
slope (when R, is correctly adjusted) and has the additional merit of incorporating
a tone control providing a total control of 13 db in the region 20 to 50 c/s depending
on the cross -over frequency. With an input voltage of about 50 millivolts, the output
will be about 1 volt.
Frequency- selective feedback to provide equalizing
A resistance -capacitance network may be used in the feedback loop of an amplifier
to provide bass boosting which gives a close approximation to the correct degree of
Fig. 17.18. Two section 05NF O.27Mn 0-27Mn OINF OSNF
equalizer to provide for
3 values of cross -over
frequency and adjust-
able slope of characteris-
tic (Ref. 127). V1 -I-
172 = 6SL7 or 6SC7 or
7F7 or any high -mu
triodes.
17.5 (iii) INTRODUCTION TO EQUALIZERS 735
0+300V
LOW FREOUENCY CONTROL
2ßM0
HIGH FREOUENCY CONTROL
O Flot
15,0000
firr
33,000 56M0
01 HF
2.2Mß
51,0000
V +300V =3,qÓ00
18000
15M0 20NF
FIG. 17.19A
8o Z.1
Flo
O
â 6° 70
W
e
W
lt 50 m IO
S 4J
+ W
3
40 I_ o
20
30
2:
ú 20
Z1
2 f 7
u 7 1 i f
100 1,000 4000 1,000 10,000
Frequency Ns) FrequincY (c/0
-_ 6SC7or7F7-
DUAL TRIODE
Fig. 17.20. Circuit diagram of equalizing amplifier for low -level electromagnetic
pickup using negative feedback over the second stage to accomplish equalization (Ref. 144).
A different form of frequency- selective feedback is used in Fig. 17.21 which pro-
vides correct equalizing for English (H.M.V.) records from 25 to 8500 c/s within
2.5 db. V1 is the first pre -amplifier valve having a series tuned circuit L,C1 (with a
minimum impedance below 25 c /s) in its cathode circuit so that at higher frequencies
an increasing proportion of the input signal is fed back. The condenser C, in series
with R, serves to flatten the response above 1000 c/s ; no provision is made for
de- emphasis. The level between 250 and 1000 c/s is flattened by the network L,C,R4
in which L, and C, are tuned to 250 c /s. R4 and L, carry the plate current of the
valve, and the value of R, primarily determines the degree of feedback at 1000 c /s.
FIG. 17.21
OUTPUT
z,
o
FIG. 17.22
Fig. 17.21. Frequency selective negative feedback Fig. 17.22. Bacs equalizer
used to provide equalizing for English records. incorporating L, C and R
(Ref. 199). V, = MH4 (µ = 40, r, = 11 000 (Ref. 188).
ohms), load resistance = 50 000 ohms.
Bass equalizers incorporating L, C and R
A network incorporating L, C and R may be used to provide a very close approach
to the 6 db /octave ideal characteristic, or even higher values if desired. The simpli-
fied circuit is shown in Fig. 17.22 in which L, is tuned by C, to the cross -over fre-
quency, R, controls the shape of the band and 123 controls the slope of the characteris-
tic.
Suggested values for a cross -over frequency of 400 c/s are-L, = 2.9 henrys;
C, = 0.07 µF ; C, = 1.0 µF ; R, = 6800 ohms ; R, = 2200 ohms.
For a slight change in cross -over frequency only C, and R, need be altered (Ref. 188).
References to equalizers (general) : 15, 54, 106, 108, 109, 113, 117, 120, 126, 127,
133, 144, 187, 188, 197, 198, 199, 209, 220, 226, 239, 240.
(B) High -frequency equalizers (de- emphasis)
A shunt capacitance provides for attenuation at the nominal rate of 6 db /octave
(see Chapter 15 Sect. 6). The attenuation is 3 db when the reactance of the shunt
condenser is equal to the resistance of the circuit across which it is connected, that
is when
C = 1 /(2,10R) (1)
17.5 (iii) INTRODUCTION TO EQUALIZERS 737
Limits
of
Attenuation
fo f
17.23. (A) Circuit providing high frequency attenuation with a specified limit to
the attenuation (B) Frequency characteristics.
If the maximum attenuation is 10 db, the maximum slope (which occurs at half
maximum attenuation) is 3 db /octave. This may therefore be used to provide a
3 db /octave de-emphasis characteristic within the limitations of its range.
High frequency equalizers are incorporated in the circuits of Figs. 17.19, 17.20,
17.27.
For general treatment of equalizers see Ref. 258.
(iv) High-frequency attenuation (scratch filter)
In addition to the normal manual tone control it is advisable in good amplifiers
to incorporate a filter which provides very rapid attenuation above a certain frequency,
say 7000 c /s, so that the upper scratch frequencies on old or noisy records may be
rendered inaudible. This is provided by the circuit of Fig. 17.24A (Ref. 15) which
has negligible attenuation at 7000 c /s, -
60 db at 8700 c/s and over 36 db attenuation
at all frequencies above 8000 c /s. The filter is designed for an impedance of 24 000
50,0001 e+
FIG . 17.244 16 P F
*0.30H *0.40H 30,00011
-O.BH
Fig. 17.24A. Amplifier
stage incorporating filter
25N F with 7000 c/s pass -band
and very rapid attenuation
at higher frequencies (Ref.
5 15). V, = MH4, µ =
z 40, rD = 15 000 ohms.
OUTPUT
Tolerance 12
738 (iv) HIGH -FREQUENCY ATTENUATION 17.5
400 02NH r
400NNF NN
1 1
150pH F
NNF
R Fi
15 NF
d
300
yyF
High 2
Output
I
47M11
010F
65C7
, OS
22MR
+F
01NF
G.E. 33A1.0.
Ol 1RM 6C 3.3 Mn 18INA 010F
T FIG,17.24B
Fig. 17.24B. Circuit of pre- amplifier and filter circuit with high attenuation at 10 000
c/s (Ref. 113).
ohms ; the plate resistance of V, is increased
to this value by means of a partially un-
bypassed cathode resistor, while the out-
put end is loaded directly. See reference
for coil winding and other details.
An alternative method is used in the pre-
amplifier circuit of Fig. 17.24B to produce
the response characteristic of Fig. 17.24C lo
This is suitable for use with low level
electro-magnetic pickups. The 10.6 Kc /s
parallel-T network provides a maximum 20
attenuation of 34 db, while the 6 Kc /s
parallel -T network in the feedback loop re-
moves the attenuation which would other-
wise occur at this frequency and thus give 30
mately 100 mH and that of the G.E. pickup 120 mH. The correct value of shunt -
resistance is given by
R (ohms) = 1.2 f, for the Pickering pickup
and R (ohms) = 0.9 f, for the G.E. pickup
where f, is the frequency at which the attenuation is 3 db. Beyond this frequency
the rate of attenuation is approximately 15 db /octave.
The shunt capacitance required for various values of the cut-off frequency is :
Frequency 4000 5000 6000 7000 c/s
Capacitance Pickering 0.02 .013 .009 .0065 pF
G.E.
Fig. 17.25. Equalizer for
use with E.M.I. and Mar-
0.03 .019
goon
.013
---
------__
son
.0095 p.F
Soon
I
LINK I
coniphone Model 12A pFlm 20n
3 4
°
400n I
2
INPUT' I
is 200 ohms ; with link
Ì
r
OOUTPUT
closed, output is 600 ohms. T3NFT16YF 1
TBpFI o
1 I
¡
65,000n 3 65C7 2MO
05NF 05pF Fig. 17.26. Pre -amplifier and equalizer
for use with the G.E. pickup (Ref. 187).
27,0000
INPUT 3M0
01yF
FIG. 17.26
22,00031
a,000n
HIGH FREOUENCY
CONTROL
Mn 20NF 27
INPUT I
.LMn
ONNFC? OF'F'F
R4 .OINF
6SC7 6SL7
INPUT 2 RMn 27/431 C6
3
OUTPUT
GNO.
GAIN CONTROL LOW FREQUENCY
FIG, 17.27A CONTROL
60
50
40
30
z
C' 20
Fig. 17.27. (A) Circuit of
continuously variable equ-
IG
alizing pre-amplifier (B)
Variation in attenuation
02 5 . 5674 5 6 M 6743
in low frequency control 00
7 3
1,000
7 3
10,000
FREQUENCY (Cts)
(C) Variation in attenua- FG. 17,2713
Ñ 30
z
O ISdb
a
a 20
24db
IO
40db
0
2 2 . 7 3 2 2 3 5 6 780 7 3 4 5 67110
100 1000 0,000
FREQUENCY (C/S) FIG, 17,27C
channel. Outputs from the three channels are added in a single valve feedback sum-
ming amplifier ; potentiometers which add flat loss in the auxiliary channels permit
control of the resultant transmission characteristic. The input allows for the con-
nection of two G.E. variable reluctance pick -ups simultaneously. The maximum
output is of the order of 1 volt, and the pre -amplifier may be -used at a moderately
remote location, owing to the cathode follower output.
The frequency characteristics are given in Figs. 17.27B and C (Ref. 246).
17.5 (v) EQUALIZERS FOR ELECTRO- MAGNETIC PICKUPS 741
Mv,,
to an electro- magnetic pickup (Ref. 209). Original
values were V, = EF36, R, = R, = 1 MQ, R = 0.22
MQ, R, = 2.2 MS?, C = 600 µµF, Ebb = 50 volts, C `
unity gain.
Suggested adaptation for 19 db gain and bass boost of 5,Re
26 db is : V, = 6J7 or 1620, R, = 0.1 MQ, R, = 1 MQ, I OUTPUT
RL = 0.47 MQ, R, = 3.0 MD, C = 330 µµF for
500 c /s, or 510 µµF for 300 c /s, or 620 µµF for 250 c /s. INPUT
V, may also be replaced by a high -mu triode (6AT6)
where R, = 0.1 MD, R, = 0.33 MS?, RL = 0.22 MD,
R, omitted, C = 1200 µµF for 500 c /s, 2000 µµF for
300 c/s or 2400 µµF for 250 c /s, giving gain of 8 db and bass boost of 26 db (see
Ref. 287).
A modification of the circuit of Fig. 17.28 is described in Ref. 280 which gives both
bass boosting (with a " very low frequency roll-off ") and high frequency de- emphasis.
Values of components are : R, = 0.33 megohm, R, = 1.0 megohm, RL = 0.22
megohm, C = 600 µµF approx. (adjustable), C, = 25 µµF approx. (adjustable)
shunted across R and V, = EF37.
A different modification of this circuit is given in Fig. 15.13 ; see Chapter 15
Sect. 2(iv)C This modified circuit enables adjustment to be made for any cross-
over frequency by means of a continuously variable potentiometer ; unfortunately
no measured frequency characteristics have been published but this circuit seems to
be close to the ideal for simplicity and to provide a response characteristic as good as
that of any other RC feedback amplifier.
(vi) Equalizers for crystal pickups
A crystal pickup, when working into a high load resistance of the order of 2 to 5
megohms gives a practically level response characteristic over the constant amplitude
section of a gliding -tone frequency test record (A in Fig. 17.30). If the load resistance
is decreased to 0.5 megohm (Curve B) the output at 50 c/s may fall in a typical case
by 6 db, the attenuation ratio at other frequencies being inversely proportional to the
frequency.
High temperatures cause a reduction in bass response ; an increase from 68 °F to
104 °F causing a decrease of about 5 db at 50 c/s in a typical case (Curve B).
The output from a crystal pickup is so high that, if it is not attenuated between the
pickup and the first amplifier stage, it may overload the valve. Unless the equalizer
provides heavy attenuation, it is advisable to place the volume control between the
pickup and the first grid.
Above the cross -over frequency the output voltage from an unequalized high-
fidelity crystal pickup tends to fall at the rate of 6 db /octave. In popular types there
is usually a resonance peak which gives some additional lift to the top end of the charac-
teristic, thus decreasing the amount of equalization required.
742 (vi) EQUALIZERS FOR CRYSTAL PICKUPS 17.5
A simple equalizer for use with crystal pickups is
shown in Fig. 17.29. If R, = 0.5 megohm, R, may
be taken as 10R, or 5 megohms. This gives an attenua-
tion at very low frequencies of 20.8 db. The effect of
variation of C, is shown in curve C Fig. 17.30 ; these
response curves are only satisfactory when considerable
high -frequency de- emphasis is required. Fig. 17.29. Simple equal -
Better equalization is obtained when R, = 40R,, izer circuit for a crystal
thus giving a maximum attenuation of 32 db. Suitable pickup.
values are : R, = 5 megohms, R2 = 0.125 megohm, C, = 100 µµF. The resultant
response curve is shown in curve D.
A more flexible equalizer suitable for use with a high -fidelity crystal pickup is
shown in Fig. 17.31 (Ref. 127). This gives a choice of three load resistors, thus giv-
ing some control of the bass, and three values of capacitance suitable for three cross-
over frequencies.
An improved equalizing circuit suitable for use with the Acos GP12 high -fidelity
pickup is shown in Fig. 17.32 (Ref. 120). This gives correct equalization for H.M.V.
records within + 2.5 db, - 3.5 db from 30 to 8000 c /s.
An equalizer suitable for use with the Brush high -fidelity pickup is shown in Fig.
17.33 (Ref. 06). Position (1) is for quiet shellac records with a 500 c/s cross-over
68°F
10.4
10
,',.1
o,1111LIN,,'!1 .,1"'I
10
oee
SV
oesv
,,
.,'1"1
:IIIIIIIIIIIUI_-_ ,,1.'
_
I.e_70311:m. 111ll
CS
\
OO OCb.O.OSV
0
111
FIG. 17.31
240 NNF
VPF
3
500^
TRANSITION
FREOUENCV
s,
M -F BOOST
we- 005 NF
Volume Control
FIG. 17.32
127).
Fig. 17.31. Flexible equalizer for high-fidelity crystal pickups (Ref.
250 c/s cross -over
Fig. 17.32. Equalizer for Acos GP12 high fidelity crystal pickup, for
frequency (Ref. 120).
Orthacoustic trans-
frequency ; (2) for scratchy and noisy shellac records ; (3) good rumble and
criptions and (4) noisy transcriptions. The low frequency filter reduces
hum.
See also Sect. 5(x).
Switch for
L -F Filter
FIG. 17.33
Fig. 17.33. Equalizer for use with Brush high-fidelity crystal pickup (Ref. 106).
Tocrd
To Cathode
clude -
duction of sound from records may in-
1. A suitable pickup.
2. An equalizer to provide constant
velocity output (only required for elec-
tro-magnetic pickups, and even then
only when not provided as part of the
main equalizer).
3. A first stage pre -amplifier (only
required in the case of a low level pick-
up).
4. A main equalizer, preferably pro-
vided with adjustable cross-over fre-
quencies and variable high-frequency.
de- emphasis.
5. A tone control, preferably com-
prising separate bass and treble controls.
If the main equalizer has flexible adjust-
ments of its characteristics, it may be
used in place of a separate tone control.
6. A rumble filter, preferably with
switch to cut in or out (may be omitted
if response is attenuated below 70 c /s).
(Continued on page 748)
Fig. 17.35A. Circuit of 25 watt ampli-
fier using G.E. variable reluctance or
Pickering 120M pickup (Ref. 240).
(ix) COMPLETE AMPLIFIERS 745
17.5
V1,o
Pickup
to put
Fig. 17.35B.
R80
R81
R82
R83
RC4
R65
0.68 MQ
0.22 MQ
4700 Q
0.22 MI?
20 000 Q
22 000 Q
t7 R44
High -stability
High -stability
High -stability
Composition
Composition
R92
attenuator). Ref. 270. In the original amplifier V1, = V14 = V16 = type E 7.
and V14 may be replaced by type 6J7 or 1620 if R69 = 1MQ, R71
R68
Rey
Value to suit trans-
former
0.1 MD
Type
High-stability carbon
High- stability carbon
carbon
carbon
carbon
Rating
4W
4W
¡W
4W
Output to Tone
Compensation Stages
FIG, 17.358
Williamson pre- amplifier including equalizer and high -pass filter (rumble
Tolerance
20%
20%
20%
20%
10%
20%
10%
R66 0.22 MQ Composition
R87 0.20 MQ* Composition
R88 4.7 MI Composition 5%
R69 0.47 MO Composition 4W 20%
R70 0.22 MI? Composition 4W 20%
R71 2200 Q Composition 20%
R72 2.0 MQ Composition 1 % or
matched
R7, 2.0 MI Composition 1% or
matched
R74 1.0 MI? Composition 1% or
matched
R75 10 M? Composition 5%
R76 47 000 Q Composition 10%
R77 1000 Q Composition 20%
47 000 Q Composition 1W 20%
R76
R79 0.22 MQ Composition 20%
R80 10 000 Q Composition 1W 20%
R81 0.22 M? 4W
R82 0.22 MI 4W
R83 47 000 Q
R84 100 Q
Output
to
Radio Amplifier
Input
FIG. 17.3SC
-61511211M O
A.C.
O
Fig. 17.35C. Williamson tone compensation and filter unit (Ref. 270). In the original
amplifier V9 = 17,0 = VII = type EF37. All may be replaced by type 9002, or by
type 6AU6 with a slight increase in gain, or by type 617 (or 6J5 triode) with a slight
decrease in gain.
Rating Tolerance
RS8 0.25 MS2 log.
R87 47 000 Q 1W
R88 47 000 Q 1W
R89 3300 Q
R10 0.25 MQ log.
R 41 100 000 Q
R42 6800 Q
R43 10 000 Q
R44 0.1 MQ linear
17.5 (ix) COMPLETE AMPLIFIERS 747
Rating
Type (V d.c. Tolerance
working)
C17 50 µF Electrolytic 12
C18 8µF Electrolytic 450
C19 0.25 µF Paper 500 20%
CEO 150 µµF max. Preset
CE1 0.01 µF Paper 250 20%
CEE 0.05 µF Paper 250 20%
C23 1000 µµF Silvered mica 20%
CR4 50 µF Electrolytic 12
C26 0.05 µF Paper 500 20%
C29 100 µµF Silvered mica 5%
CY7 200 µµF Silvered mica 5%
Cae 300 µµF Silvered mica 5%
CY9 500 µµF Silvered mica 5%
C$9 50 µµF Silvered mica 5%
C31 100 µµF Silvered mica 5%
C32 250 µµF Silvered mica 5%
C33 50 µF Electrolytic 12 20%
C34 0.05 µF Paper 500
C95 8µF Electrolytic 450
Cae, 40 75 µµF Silvered mica 1% or
matched
C37,
Cae,
41
42
100
150
µµF
µµF
Silvered mica
Silvered mica
m or
1% or
matched
Cae, 43 200 µµF Silvered mica 1% or
matched
C44 150 µµF Silvered mica I% or
matched
C45 200 µµF Silvered mica
matched
C44 300 µµF Silvered mica 1% or
matched
C47 400 µµF Silvered mica
C48 16 µF Electrolytic 450
C49 16 µF Electrolytic 500
Mains Transformer-
Primary 200- 220 -240 V, 50 c /s.
:
Switches-
S1 Single pole single throw.
S2 Double pole single throw.
S3 Single pole single throw.
S4 5 bank, 5 position selector switch.
7. A high frequency attenuator or filter for worn or noisy records (may be omitted
if a sufficiently flexible tone control is fitted). One very effective form of filter is a
capacitance connected across the secondary of the output transformer-see page 214.
An example of a complete amplifier is given in Fig. 17.35A. This is a 25 watt
amplifier suitable for operation from a G.E. variable reluctance pickup, or Pickering
120M. The total harmonic distortion is under 1%. The pre -amplifier consists of
a triode with proper compensation in the output circuit. Bass boosting is obtained
by the use of frequency -selective negative feedback in the second and third stages.
Hum from the heater is reduced to the vanishing point by making the heaters of
the first two twin triodes part of the cathode resistor of the output stage. The output
stage draws approximately 300 mA, which is the correct current for the heaters of
the 6SL7 valves, and the additional bias required for the output valves is obtained
by a resistor (Ref. 240).
Another example of a complete amplifier is that due to D. T. N. Williamson (Ref.
270) shown in Figs. 17.35 B, C, D, E, F and G. The main amplifier (Fig. 17.35F)
is a new version of the earlier Williamson amplifier of which one adaptation is Fig. 7.44.
The circuit differs in minor details only from Fig. 7.44 ; the balancing adjustment
on V1 and V2 has been omitted as unnecessary, provided that R5R7 R11 and R13 have
the specified values and tolerances, while a transitional phase -shift network R26 C10
has been added to increase the margin of stability at high frequencies.
Output transformer specifications (3.6 ohm secondaries)-Williamson Amplifier :
Core 111 in. stack of 28A Super Silcor laminations (M. and E.A.). The winding
:
consists of two identical interleaved coils each 1+¡ in wide on paxolin formers 11 in.
x 11 in. inside dimensions. On each former is wound :
5 primary sections, each consisting of 440 turns (5 layers, 88 turns per layer)
of 30 S.W.G. enamelled copper wire interleaved with 2 mil. paper,
alternating with
4 secondary sections, each consisting of 84 turns (2 layers, 42 turns per layer)
of 22 S.W.G. enamelled copper wire interleaved with 2 mil. paper.
Each section is insulated from its neighbours by 3 layers of 5 mil. Empire tape. All
connections are brought out on one side of the winding, but the primary sections may
be connected in series when winding, two primary connections only per bobbin being
brought out. Windings to be assembled on core with one bobbin reversed, and with
insulating cheeks and a centre spacer.
Curves showing the loop gain and phase -shift characteristics of the main amplifier
are shown in Fig. 17.35G.
The pre-amplifier Fig. 17.35B has a voltage gain of 250 times from the grid of the
first valve to the voltage across the output terminals. The input grid voltage must
not be less than 0.8 mV ; this minimum voltage can readily be provided by a suitable
transformer from any pickup. Care should be taken to avoid overloading the pre-
amplifier by too high an input voltage. The overall frequency response, when
tested with an ideal " velocity " pickup on an English Decca disc is level within
1 db from 20 to 14 000 c/s ; below 20 c/s there is attenuation at
the rate of 30 db/
octave to eliminate rumble. For the English E.M.I. characteristic, switch S, should
be opened. Other recording characteristics can be handled by the use of the tone
compensation unit.
This overall frequency characteristic, which provides bass boosting and high -
frequency de- emphasis to suit the recording characteristic, together with a rumble
filter, is provided by a careful combination of
(1) bass boosting by V13 together with the feedback network C54 C55 R0,
R64.
(2) Attenuation produced by the combined inter -valve couplings.
17.5 (ix) COMPLETE AMPLIFIERS 749
(3) Feedback over V,4, through the parallel-T network. This provides a peak at
20 c/s which is used to gave a sharp knee to the frequency characteristic instead of
the gradual attenuation produced by the inter -valve couplings. (Note that the
recording characteristic contributes 6 db /octave to the slope below 20 c /s).
The final stage Vi, is merely a cathode follower to permit the use of long leads
between the pre -amplifier and the tone -compensation unit.
If a high-impedance output is permissible, and an attenuation of 8 db at 10 c/s is
sufficient to reduce the rumble, stage Vi, may be used as a single stage pre-amplifier
with a gain of 11 times. Under these circumstances C,, = 0.05 µF, C ,4 = 4000
µµF, R66 = 22 000 ohms, R
= 2.2 MSl, R
is omitted, and the output is taken from
the junction of C,, and R4,.
In the cathode circuit of V,
closing the switch S, reduces the gain to zero in about
1 second for use while changing records.
The noise level, with R
adjusted for minimum hum, is about 3 to 5 µV at Vs,
grid, excluding the noise due to the pickup transformer and auxiliaries. The total
harmonic distortion of the pre-amplifier and tone compensation units combined, is
considerably less than 0.1%.
The tone compensation and filter unit Fig 17.35C provides-
s
o
4
FILTER POSITION
40
to loo ipoo 10,000 70,000
FREQUENCY (c/s) FIG. 17.350
Fig. 17.35D. Measured overall response of low -pass filter (stage Vii in Fig. 17.35C
-
together with network controlled by switch S4). Ref. 270.
+20
tncreodl C Decrease CIo
I
A C
S, Rise S, Rise
R4D Max. R,4M0,.
S, Fall
Rso Max ...1--
ncrease Cal
S. Fall
Rw Max
Decrease C
-2 IOO 1,000
IO
FREQUENCY (c /s) FIG. 17.75E
V9 in Fig.
Fig. 17.35E. Response curves of the tone compensation circuit following
17.35C. Curves are limiting positions for continuously-variable controls (Ref. 270).
750 (ix) COMPLETE AMPLIFIERS 17.5
(1) A universal tone control (switches S,S3 and controls
R40R44) with frequency
characteristics as Fig. 17.35E. The curves may be shifted bodily along the hori-
zontal axis by modifying the capacitance values shown by the arrows in Fig. 17.35E.
(2) A low -pass filter with nominal cut -off values of 5000, 7000, 10 000 and 13 000
c /s, together with a flat position (switch S4) providing response characteristics
as
Fig. 17.35D. This characteristic is provided by a parallel-T network in the feed-
back loop of VII which gives a symmetrical valley at the frequency of resonance.
A capacitor from the grid of VII to earth introduces a lagging phase shift which gives
CH, 450V - CH2
FIG. 17. 35 F
Fig. 17.35F. Williamson main amplifier -new version (Ref. 270). Voltages under-
lined are peak signal voltages at 15 watts output.
R1 1 MO watt ± 20%
R2
R,
33 000 Q
47 000 Q
+
watt
1 f 20%
watt ± 20%
1
R4 470 0 á watt + 10%
R5, R, 22 000 Q 1 watt + 5%
(or matched)
R, 22 000 Q 1 watt ± 20%
R9, Re 0.47 MQ J watt ± 20%
Rio
R11, RIa
390 Q
47 000 Q
J wattf
2 watt ± 5%
10%
(or matched)
1214, R19 0.1 MQ } watt ± 10%
R,S, R2o 1000 Q I watt ± 20%
RIe, RIe 100 Q 1 watt 20%
R17, R2I 100 Q 2 watt wirewound variable
R22 150 Q 3 watt + 20%
Rs5 1200 Q x /speech coil impedance
} watt
R25 4700 Q i watt + 20%
Cl, C2, C6, C, 8 µF 500V wkg.
C3, C4 0.05 µF 350V wkg.
C6, C, 0.25 µF 350V wkg.
Co 8 µF 600V wkg.
CIo 200 µµF 350V wkg.
CHI 30H at 20 mA
CH2 10H at 150 mA
T Power transformer
20
3
Io
1
-
0
a
10
-- -AMPLIF lER WITHOUT FEEDBACK
AMPUF ER WITH FEEDBACK
20
^O
loo
P
JN3S 300
10 102 o pe 0e IOe
FreVeencT We) FIG. 17. 35G
Fig. 17.35G. Loop gain and phase -shift characteristics of Williamson main amplifier
(Ref. 270).
positive feedback below resonance and negative feedback above resonance, thereby
unbalancing the characteristic and giving a sharp knee before the attenuation com-
mences. The addition of an independent RC network controlled by SIA gives
further treble attenuation and leads to the final result in Fig. 17.35D.
(a) Pickups for connection to radio receivers
When a separate pickup and turntable unit is used in conjunction with a radio
receiver, it is often most convenient for the user to operate the device at some distance
(say up to 15 feet) from the receiver. The most desirable arrangement is a Record
Player in the form of a small cabinet enclosing the motor, turntable, pickup, and
(possibly) volume control, pre -amplifier and equalizer, with the top of the cabinet
about 20 inches from the floor so as to be accessible from an easy chair.
The design problem hinges on the use or otherwise of a pre -amplifier. If no pre-
amplifier is permissible, on account of cost or other factors, it is necessary to
employ a pickup delivering a high output voltage or whose output voltage is capable
of being stepped up by a high ratio transformer.
When pickup terminals are fitted to receivers, they usually provide a resistive load
of the order of 0.5 megohm, with some capacitive shunting. The input voltage
required to give full power output is usually about 0.25 volt, although some receivers
require a higher voltage. If a receiver is being designed specially for good perform-
ance with a pickup, full power output should be obtained with an input voltage
not exceeding 0.1 volt. This usually demands a pentode first a -f amplifier, especially
if negative feedback is used.
If bass equalization of 20 db is required for use with an electro- magnetic pickup,
the available voltage must be at least 10 times the input voltage to the amplifier, e.g.
1 volt and 0.1 volt respectively There is a strong temptation to use a popular pickup
which provides at least part of the bass boosting by arm resonance, but this is very
undesirable on account of record wear, distortion and heavy stylus force required.
There are electro- magnetic pickups capable of providing an output of about 1 volt
from the secondary of a transformer, which do not employ arm resonance in the vicinity
of 100 c/s ; these are much to be preferred to those employing arm resonance. How-
ever, the high cost of a good quality transformer, and the hum and poor frequency
response with a cheaper transformer, make this a rather unsatisfactory arrangement.
The necessary bass boosting may be provided by a resistor and capacitor in series,
connected across the secondary of the transformer. A choice of cross -over frequency
may be provided by a tapping switch and several values of capacitance. High -fre-
quency de- emphasis may be provided by one or more values of shunt capacitance,
752 (x) PICKUPS FOR CONNECTION TO RADIO RECEIVERS 17.5
the plane of the record to give a brilliant pattern. The eye should be directed vertic-
ally downwards on to the side of the record nearer to the lamp, and should be as far
above the record as possible.
A pair of dividers may be used to measure the width of the reflected light pattern
for each frequency ; the width of the light pattern is proportional to the voltage
output from a perfect constant -velocity pickup. This method becomes increasingly
less accurate below the cross -over frequency, although it is very satisfactory at higher
frequencies.
The theoretical analysis and calculation of errors is given by Ref. 243.
References 25 (Part 5), 99, 219, 243, 258, 293.
Record fl fo (cross-over) 12
English
Decca ffrr K1802 I4 000 c/s 10 c/s
K1803 14 000 c/s 300 c/s 10 c/s
American
R.C.A. 12 in. (12-5-5) 10 000 c/s 500 c/s ( -1.5 db) 30 c/s
London (album LA-32)
T-4996 12 in. 78 r.p.m. 14 000 c/s (+8 db) 250 c/s ( -1 db) 100 c/s ( -5 db)
T-4997 12 in. 78 r.p.m. 14 000 c/s ( +8 db) 250 c/s ( -1 db) 10 c/s ( -33 db)
Universal D61B Band 1 10 000 c/s (0 db) 500 c/s (0 db)
Band 2 500 c/s (-7 db) 200 c/s (-7 db)
Band 3 200 c/s ( -14 db) 50 c/s ( -14 db)
Columbia XERD281 12
in. 10 000 c/s 500 c/s 50 c/s
fi fo(cross-over) f2
E.M.I. Studios JG449 (1) 20 Kc /s, 18, 16, 14, 12, 10, 8, 6, 4.5, 3.5, 2 Kc /s (0 db) ;
500 c/s ( -1), 160 c/s ( -5.5), 70 c/s ( -12 db).
(2) 19 Kc/s, 17, 15, 13, 11, 9, 7, 5, 4, 3, 1 Kc /s (0 db) ;
250 c/s ( -3), 100 c/s ( -8.5), 50 c/s ( -14 db).
Columbia LOX -650 1000 (0 db), 10 000 ( -1), 9000 ( -1.6), 8000 ( -1.6),
7000 ( -1), 6000 ( -0.8), 5000 ( -0.8), 4000 ( -0.8),
3000 ( -0.8), 2000 ( -0.8), 1500 ( -1), 800 ( -0.8),
500 ( -0.8), 300 ( -3.4), 200 ( -5.2), 150 ( -7.1),
100 ( -9.4), 70 (-11.2), 50 ( -13.6 db).
Decca Z718 50, 70,100, 160, 250, 500, 1000, 2000, c/s
-14 -11 -8 -4 0 0 0 0 db
3000, 3500, 4000, 4500, 5000, 6000 c/s
0 0 0 0 0 0 db
Decca ffrr K1804 14 000 to 400 c/s (constant velocity) 250 to 30 c/s
(constant amplitude) ; in steps. Level +10 db at
1000 c /s.
American
R.M.A. Standard No. 2(A) Constant frequency bands at same frequencies as
REC -128 No. 1 Side A without announcements.
78 r.p.m. No. 2(B) 8000, 6000, 4000, 2000 ( +10 db) ; alternate gliding
and constant frequencies 3000 to 30 c/s (+ 16 to 18
db) ; 8000, 6000, 4000, 2000 (+ 10 db).
R.C.A. (12 -5 -19) 12 in.) Approx. const. vel. 10 000 ; 9000 ; 8000 ; 7000 ;
78 r.p.m. '. 6000 ; 5000 ; 4000 ; 3000 ; 2000, 1000 ; (0 db =
Unfilled vinyl, SF J 8.6 cm /sec. approx.) 800 c/s ; 500 ( -1.5 db) ; Const.
ampl. 300 ; 200 ; 100 ; 50 ; 1000 ( -2 db) ; 10 000.
R.C.A. (12 -5 -25) 12 000 ( -2.5 db) ; 11 000 ( -3) ; 10 000 ( -1.5) ;
(= 460625 -6) 12 in. 33 -1/3 approx. const. vel. down to 700 ( -0.5 db) ; 400 (1.2
r.p.m. Unfilled vinyl, SF db below const. ampl.) ; 300 (0.6 db below const.
ampi.). Also 400 and 4000 c/s tone for intermodula-
tion testing. Groove has bottom radius less than 0.5
mil. Max. stylus radius 3 mils.
R.C.A. (12 -5 -31) 7 in. 45 Approx. const vel. 10 000 ( -0.5 db) ; 9000 ; 8000 ;
r.p.m. 7000 ( -1) ; 6000 ; 5000 ( -1.3) ; 4000 ; 3000
Unfilled vinyl, D.F. ( -1) ; 2000 ; 1000 (0 db = 5.2 cm /sec.) ; 700
(Opposite side of 12 -5 -29) ( -0.5) ; 400 ( -1.5) ; approx. const. ampl. 400 ;
300 ; 200 ; 100 ; 50 ; 1000. Velocity at 1000 c/s
= 5.2 cm /sec.
17.5 (xi) FREQUENCY TEST RECORDS 755
Columbia
TL -1 LP microgroove 10 000 (+12 db) ; 9000 (+12) ; 8000 (+11.5) ; 7000
33-1/3 r.p.m. ( +11) ; ( +11) ; 5000 ( +10) ;
6000 4000 ( +7.5) ;
3000 ( +5.5) ; 2000 ( +3) ; 1500 ( +2) ; 1000 (0 db =
1 cm /sec. r.m.s.) ; 800 ( -1)
; 500 ( -2.5) ; 400 ( -4) ;
300 ( -6);
200 ( -8.5);
150 ( -10); 100 ( -11);
70 ( -12)
; 50 ( -13). This record is cut with the
microgroove recording characteristic including high -
frequency pre- emphasis.
TL-2 LP microgroove 10 000 ( +5.5 db) ; 9000 ( +5.5) ; 8000 ( +5) ; 7000
33-1/3 r.p.m. ( +5.5); 6000 ( +5); 5000 ( +7); 4000 ( +7.5) ; 3000
( +7) ; 2000 ( +4.5) ; 1500 ( +6.5) ; 1000 ( +7.5) ;
800 ( +7.5) ; 500 (+7.5) ; 400 ( +6) ; 300 ( +3.5) ;
200 ( -0.5)
; 150 ( -3.0)
; 100 ( ; 70 ( -6) ; -7.5)
50 ( -10.5) ; 30 ( -12
db). 0 db = 1 cm /sec. r.m.s.
RD -103 LP microgroove 10 000 ( +5 db) ; 9000 ; 8000 ; 7000 ; 6000 ( +5) ;
33 -1/3 r.p.m. 5000 ; 4000 ( +7) ; 3000 ; 2000 ( +5) ; 1500 ; 1000
12 in. Bands ( +7.5) ; 800 ; 500 ( +7.5) ; 400 ; 300 ; 200 (0 db) ;
150 ; 100 (-6)
; 70 ; 50 ; 30 c/s ( ; 0 db = -12)
I cm /sec.Essentially flat over constant -velocity
portion, with cross -over 500 c /s.
RD -103A LP microgroove 10 000 (+12 db) ; 9000 ; 8000 ; 7000 ; 6000 (+10
33 -1/3 r.p.m. db) ; 5000 ; 4000 ; 3000 ( +5) ; 2000 ; 1000 (0 db) ;
12 in. Bands 800; 500( -2.5); 400; 300( -6); 200; 100( -11);
70 ; 50 c/s ( -13 db). 0 db = 1 cm /sec. Levels
arranged to reproduce ± 2 db of " flat " on correctly
equalized LP reproducing system.
(10003 -M) 78 r.p.m. 1000 c/s ; 10 000 ( -7 db) ; 9000 ; 8000 ; 7000 ;
12 in. shellac 6000 5000 ; 4000 ; 3000 ; 2000 ; 1500 ; 1000
;
( -1db) ; 800 ; 500 ; 300 ; 200 ; 150 ; 100 ; 70 ;
50 ( -17 db). Cross -over 300 c /s. Level approx.
4.8 cm /sec. "t 1000 c /s.
(10004-M) 78 r.p.m. As (10003 -M) but with cross -over frequency 500 c /s.
12 in. shellac
Audiotone
78 -1 78 r.p.m. 50 to 250 c/s (peak amplitude 0.0017 inch) ; 250 c/s
upwards (constant velocity).
London Gramophone
T4998 12 in. 78 r.p.m. 14 000 c/s (0 db) ; 13 000 ; 12 000 ; 11 000 ; 10 000 ;
9000 ; 8000 ; 7000 ; 6000 ; 5000 ; 4000 ; 3000 ;
2000 ; 1000 ; 400 (0 db) ; 250 (-2) ; 100 (-6) ;
55 ; 30 c/s ( -16 db).
Clarkstan
(2000S) 78 r.p.m. 50 to 500 (constant amplitude) 500 to 10 000 c/s
(constant velocity) in 17 steps.
(20002S) LP microgroove One side recorded flat, other side with NAB curve.
Cook
Series 10, 10 in. Plastic Constant velocity (9 cm /sec.) above 500 c /s. with 3 db
DF. Side A 78 r.p.m. knee at crossover. Bands 1000 c/s ; 20 000 ; 17 000 ;
15 000 ; 12 000 ; 10 000 ; 9000 ; 8000 ; 7000 ;
6000 ; 5000 ; 4000 ; 3000 ; 2000 ; 1500 ; 1000 ;
700 ; 500 ; 350 ; 250 ; 125 ; 62.5 ; 40 ; 35 ; 1000
c /s.
Side B 33 -1/3 r.p.m. ; Band 1 -LP spot check for
:
R.C.A. Victor
(12 -5 -1) 10 in. 78 r.p.m. Unmodulated grooves at normal recording pitch with
Shellac DF lead out and eccentric. Same both sides.
(12-5 -3) 12 in. 78 r.p.m. Unmodulated grooves at normal recording pitch with
Shellac DF lead out and eccentric. Same both sides.
(12 -5 -7) 12 in. 78 and On 78 r.p.m. bands 2300 ; 1000 c /s. On 33-1/3 r.p.m.
33 -1/3 r.p.m. Shellac 1000 ; 433 c /s. Frequencies constant within 0.2%
DF instantaneous (12 -5 -5 on opposite side).
(12 -5 -9) 12 in. 78 r.p.m. Band 1- silent ; 2-400 c/s at 5.9 cm /sec. ; 3-1000
Shellac SF c/s at 9.6 cm /sec ; 4- silent. Frequencies constant
within 0.2% instantaneous.
(12 -5 -11) 12 in. 78 r.p.m. Landing area with no lead -in spiral ; at least one un-
Shellac DF modulated normal pitch groove ; steep blank spiral ;
normal pitch inside groove ; lead out and eccentric
Same both sides.
(12-5 -13) 10 in. 78 r.p.m. As for (12- 5 -11). Same both sides.
Shellac DF
(12-5 -15) 12 in. 78 r.p.m. Warble frequency bands, sweep rate 5.5 c /s. Band
Unfilled vinyl DF (1) 500 -2500 c/s ; (2) 750 -1250 c/s ; (3) 1250 -1750
c/s ; (4) 1800 -2600 c /s. Same both sides.
(12 -5 -17) 10 in. 78 r.p.m. As (12- 5 -15). Same both sides.
Shellac DF
(12 -5 -21) 12 in. 78 r.p.m. For checking automatic record changers. Three
Unfilled vinyl DF modulated bands with large eccentric groove. Bands
define standard 10 and 12 in. landing areas without
lead in grooves. Inner band 936 c/s (1 minute) for
checking wow.
(12 -5 -23) 12 in. 78 r.p.m. Various modulated bands joined by spiral grooves to
Unfilled vinyl DF indicate limits of standard recording dimensions
(12 -5 -21 on other side).
(12 -5 -29) 7 in. 45 r.p.m. For checking landing and tripping action of changer
Unfilled vinyl DF mechanisms. Bands 1000 ; 400 ; 1000 c /s.
(12-5-35) 7 in. 45 r.p.m. For checking landing and tripping action of changer
Unfilled vinyl DF mechanisms ; for checking pickup sensitivity, turn-
table flutter and rumble.
(12 -5-37) 7 in. 45 r.p.m. Bands of 400 and 4000 c/s signals combined, the
Unfilled vinyl DF (RL- 4000 c/s being 12 db below the 400 c/s level. Peak
419) velocities from 3.8 to 18 cm /sec. in approx. 2 db steps.
0 db = 6 cm /sec. Intermodulation distortion in the
record is less than 4 %. For testing pickup " track-
ing " at various levels and stylus forces (Ref. 285).
(12 -5 -39) 12 in. 78 r.p.m. Bands of 400 and 4000 c/s signals combined, the 4000
Unfilled vinyl DF (RL- c/s being 12 db below the 400 c/s level. Peak re-
420) corded velocities run from 27 to 4.4 cm /sec. in approx.
2 db steps. 0 db = 9.1 cm /sec. Intermodulation
distortion in the record is less than 3 %. Groove has
small bottom radius suitable for testing with 1.0 or
3 0 mil styli. Same use as 12 -5 -37.
17.5 (xi) FREQUENCY TEST RECORDS 757
(12 -5 -41) 12 in. 33.3 or 78 For routine testing of record changer operation with
r.p.m. Unfilled vinyl DF 1.0 or 3.0 mil styli. Standard R.M.A. landing areas
for 10 and 12 inch records are defined by short inter-
rupted tones. Short bands of 400 and 1000 c/s tones
are included for routine pickup sensitivity measure-
ments.
Clarkston
1000 A 12 in. 78 r.p.m. Sweep frequency record for oscillographic observation
vinylite SF of equipment response. Frequency range 70 to
10 000 c /s, flat within ± 1 db. Sweep frequency rate
20 times /sec. Crossover 500 c /s.
1000 D 12 in .78 r.p.m. Sweep frequency record as 1000A but covering range
vinylite SF. from 5000 to 15 000 c /s.
E.M.I. Studios
JH138 12 in. 78 r.p.m. Side 1 400 c/s ( +22.5 db) with approx. 4000 c/s
:
for use with 2.5 mil (+ 10.5 db) superimposed additively for I.M. testing of
radius stylus. pickups. Peak lateral velocity of combined wave is
equal to that of a sine wave at a level +24.5 db. The
succeeding 10 bands have levels of both tones reduced
2 db below those of the foregoing band.
Side 2 : 60 c/s (+8.6 db) with 2000 c/s ( +10.3 db)
superimposed additively. When the 2000 c/s is
reduced in the pickup bass correction equalizer by the
correct amount relative to 60 c/s (i.e. 13.7 db), its
effective level will be -3.4 db, i.e. 12 db below the
60 c/s tone. On a velocity basis, the peak lateral
velocity of the combined wave is equal to that of
a sine wave at a level +15.5 db. The peak combined
amplitude is equivalent to that of a 60 c/s sine wave
having a level of + 10 db. The succeeding 10 bands
each have the level of both tones reduced by 2 db below
those of the foregoing band. 0 db = 1 cm /sec. r.m.s.
(Ref. 310).
768R4215
{1 - IA' +.. } - 73 728R4A7{ - 1 .} + (1)
512R4A5
{1 - ;A$ + . .
- 40960R4A7{1 - }
Fifth harmonic amplitude =
281°215{1 - IA' + }+
25
1 36R'A5{ 1 - 5
+ .. - 73 728/24217{1 - .} . . (3)
49
6144R'Á'
{1 - 2401
- ...} - .. . (4)
where A = 2lra/
R = 2arr /A
a = amplitude of lateral modulation of groove, measured in the plane of
the record
r = stylus radius
and A = wavelength of sinusoidal modulation measured in the direction of an
unmodulated groove.
These formulae only hold when RA does not exceed A/2.
Example Groove velocity 10 inches /sec., lateral groove velocity 2 inches /sec
:
tude).
From (2) : Third harmonic amplitude = 0.0113.
Percentage third harmonic with constant velocity pickup =
-
3 x 0.0113 x 100/0.1839 = 18.4 %.
From (3) Fifth harmonic amplitude = + 0.0035.
:
tracing distortion.
1 45 1.0 4.44 111
However, the deterioration in t 78 30 7.04 1.76
both frequency response and
fidelity is plainly audible to a
critical listener as the groove
radius decreases towards the end
of a record ; this applies to 78 ID ID
`',.
r.p.m., transcriptions and all
types of fine groove recording. R.P.M 33%
The effect of recording char- 3 Mils
acteristics on tracing dis-
tortion
Over the range of constant 41
lie
velocity recording, the tracing
distortion is proportional to the
square of the frequency z
I MII
,+' 78 R.P.M.
3Mi1s
,
With high-frequency pre -em- V r//
y `1 I
C,Q
phasis followed by the correct á \
o
amount of de-emphasis, at the Ñ N_
rate of 6 db /octave above a
hinge frequency " f, (P in Fig
I
MUM
INEmma ó No
17.37B) the third harmonic trac- NIMII11 o III
ing distortion at a recorded fre-
quency f, will be one third of the
'\ I.g
Mó
value calculated by the procedure N
outlined above provided the con-
stants are the same in both cases. o ' 2 4 6 8 I
I --
db
f, f,
rcauency Ns)
FIG. 17 . 37 5
distortion will therefore be four times as great with pre-emphasis : de- emphasis
as at point P. This may be expressed in the general form
113% (constant amplitude) m2
H3% (constant velocity) 3 (5)
where m = 12/11
f2 = recorded frequency
fl = hinge frequency of pre -emphasis
and the pre- emphasis and de-emphasis are both at the rate of 6 db /octave above
If fs = fi f,
m = 1 Ratio of Hs% = 1/3
12 =2f, m =2 Ratio of 1/3% 4/3
fs = 4f1 m = 4 Ratio of H3% = 16/3
Other distortion effects
In the foregoing treatment it is assumed that the pickup stylus and armature are
able to follow ideally all the sharp and sudden changes in the modulation of the
which occur with complex sounds. In many cases this is obviously not the groove
case, so
that additional causes of spurious intermodulation products (mainly inharmonic)
arise. These are reduced by the use of a really good pickup but in most
probably true that the best available pickup will introduce distortion additionalit is
cases
the theoretical tracing distortion. to
The distortion arising from overmodulation at high frequencies due to a high value
of pre -emphasis has already been covered in Sect. 5(i). Similar effects may
occur
at low frequencies with organ music, but may be avoided by the use of a low frequency
limiter in the recording amplifier.
The effect of finite size of needle tip on the reproduction of the fundamental
corded frequency is covered in Sect. 2(i), also translation loss. re-
The pinch effect has already been mentioned. It causes increased combination
products, forced vibration of the armature at its high resonant frequency, increased
noise in the output voltage and needle talk. Fortunately these bad effects
considerably reduced by the introduction of vertical compliance either in themay be
(bent shank or trailer type) or in the pickup itself. needle
References to tracing distortion, intermodulation distortion and pinch
11, 40, 41, 42, 47, 110, 146, 193, 212, 213, 223, 224, 225, 248, 264, 265,
effect :
285.
(ii) Playback loss
Playback loss has been defined as the difference between the recorded and the
re-
produced level at the very same point of a record. It is due to the physical properties
of the record material, being evident as a loss of the higher frequencies. A com-
paratively stiff and hard material such as is used with shellac pressings has very
little
playback loss ; vinyl and lacquer are very much more flexible and exhibit appreciable
playback loss. It is also a function of the pickup used. Reference 227.
channel " limit is ± 0.05% (Ref. 297). In reproduction, most turntables used in
combination sets suffer from an appreciable slowing down on heavily recorded pas-
sages. This effect is reduced by the use of a light-weight pickup with high lateral
compliance. The effect may be noticed, in severe cases, by a movement in the pattern
produced on a stroboscopic disc at the commencement and end of a heavily recorded
passage.
A warped disc or wobbling turntable will produce greater wear and eventually
greater scratch level on the high portions of the record. This is much less with well-
designed light- weight pickups than with older models. However, it is important
that the stylus force should be considerably more than the minimum for correct
tracking on a perfectly flat record. A record with warp in excess of 1/16 inch is
likely to cause trouble.
An eccentric disc or turntable will tend to cause wow ; a very small amount of eccen-
tricity is sufficient to be noticeable in recording (max. ± 0.002 inch N.A.B. ; also
English-Ref. 235). This will produce a tonal pitch variation of 0.1% on reproduc-
tion (10 or 12 inch records).
Flutter is similar to wow but at a high frequency ; the ear is very sensitive to this
form of distortion. This also may occur both in recording and in reproduction.
References to wow etc. in recording : 84, 90, 193, 231.
Reference to tests on wow in phonograph combinations : 267.
See also Chapter 37 Sect. 1 (vi) N for I.R.E. tests on phonograph combinations.
(iv) Distortion due to stylus wear
All theoretical work on disc reproduction assumes that the needle point is perfectly
spherical. A jewel or other form of permanent tip suffers wear on two opposite faces.
The wear is usually measured in terms of the width of the nearly flat portion. The
distortion arising from the worn stylus is independent of amplitude, and consists
principally of odd harmonics. The distortion is a function of d/1
where d = width of flat on needle
and 1 = wavelength of recorded tone.
The following table has been calculated from published data (Ref. 59) for standard
groove constant velocity recording :
at innermost groove
d!1 fundamental H, H,
f= 1000 c/s f = 10 000 c/s
0.16 - 1 db 4.8% 6.6% d = 1.5 mil d = 0.15 mil
0.22 -- 2 db 10% 12% d = 2.4 mil d = 0.24 mil
0.275
0.36
0.5
-
-10
3 db
5 db
db
20%
-
44% --
17% d
d
=
=
3.3 mil
-
4.1 mil
d
d
= 0.33 mil
-
= 0.41 mil
The distortion becomes very severe before there is any appreciable attenuation of
the fundamental caused by the needle wear. Although the harmonics of the higher
frequencies may be outside the frequency range of the equipment, the intermodulation
products will be apparent.
If the recording characteristic is constant amplitude (6 db /octave high frequency
pre -emphasis followed by de-emphasis) the values of distortion in the table should
be divided by 3 and 5 for H, and H, respectively. In this case pre -emphasis : de-
emphasis shows to considerable advantage over constant velocity, in that styli last
longer before requiring replacement, for the same distortion.
It is obvious that, for high fidelity, no observable stylus flat is permissible. This
seriously limits the life of sapphire styli, even when using a good quality pickup with
high lateral and vertical compliance-see Sect. 2(vi).
(v) Noise modulation
The noise level is usually measured as the high -frequency noise developed by an
762 (v) NOISE MODULATION 17.6
unmodulated groove. When a high degree of noise is present and the groove is
modulated by a single -frequency signal, it is found that the signal tends to modulate
the noise. The noise modulation reaches a maximum twice each cycle, and the peak
amplitude may be over ten times the peak amplitude of the measured noise in an
unmodulated groove. The use of pre- emphasis de- emphasis is to reduce the noise
:
itself but to increase noise modulation. This effect is not apparent when the inter -
modulation is less than 4 %.
This effect may be minimized during recording, by special means suggested in the
article.
Reference 145. See also Ref. 193 (proposed standards).
(vi) Pickup distortion
Non -linear distortion in a pickup may vary from practically nil in the best high -
fidelity types to over 5% total harmonics in some types. The total harmonic dis-
tortion with a well designed crystal pickup is well under 2% (Ref. 120).
The usual method to test harmonic distortion is to use a suitable frequency
test record, check it for distortion with a pickup which is known to have very low
distortion, then measure the distortion with a wave analyzer. A simple test is to use
a C.R.O., but this is not capable of indicating less than about 2% or 3% total har-
monic distortion. The distortion is always a function of frequency, so that it should
be checked at all practicable frequencies. " The minimum stylus force for a given
pickup to track a given frequency recorded at a given level can be determined . .
by the observation of waveform on an oscilloscope. The waveform produced by
failure to track has a characteristic ` spiked' appearance, and a very small trace of
this form of distortion is easily detectable " -Ref. 310(b).
Intermodulation distortion is preferable to total harmonic distortion as a general
indication of pickup performance. Recommended test frequencies are 400 and 4000
c /s, with the latter 12 db lower than the former. Test records are available with bands
recorded at different levels, from below normal level, increasing by 2 db steps to
+10 db (e.g. R. C. A. Victor RL -419, RL -420, E.M.I. JH 138).
In most cases the intermodulation distortion of a pickup increases very gradually
up to a certain level, and then rises very rapidly as the level is further increased.
This " knee " should occur at a level of about +6 db, on record RL-420, that is at
about 18.2 cm /sec. A high fidelity pickup, tested on record RL -420 shows about
2.5% intermodulation distortion at a level of +6 db ; most of this distortion is due
to the record (Ref. 291). Other modern lightweight pickups give up to 8% inter-
modulation at 0 db (9.1 cm /sec.).
The " tracking " capability of a pickup may be checked by plotting a curve of inter -
modulation distortion against recorded level, for selected values of vertical stylus
force (for details see Ref. 285).
Intermodulation distortion may be measured either as the r m.s. sum or as the
arithmetical peak sum -see Chapter 14 Sect. 3(ü) and (v). An intermodulation
analyser using the peak sum method is described in Ref. 291.
(vii) Acoustical radiation
Acoustical radiation or needle chatter varies considerably among pickups. Some
pickups have no audible radiation. In other cases the radiation may be reduced,
if desired, by the use of a bent -shank needle or trailer type (Ref. 225).
(viii) Distortion in recording
If the distortion in the cutter head includes, say, 1% third harmonic distortion, this
will result in 4% third harmonic distortion if both the fundamental and the harmonic
frequencies are recorded and reproduced at constant velocity. On the other hand
pre-emphasis followed by de-emphasis will reduce this form of harmonic distortion,
although intermodulation products will still be present (Ref. 189).
See also Sect. 9(i) for distortion on original recordings.
Proposed standards for the measurement of distortion in sound recording :
Ref. 193.
17.7 (i) ANALYSIS OF NOISE 763
1. A very rapid attenuation above a fixed frequency (say 7000 c /s, or a choice be-
tween 4500, 7000 or 10 000 c /s) for use with old or noisy records. A suitable circuit
is Fig. 17.24A.
2. A gradual attenuation characteristic with a hinge point at about 1500 c/s with
variable rate of attenuation up to at least 6 db /octave, in addition to any de-emphasis
as such. Some possible circuits are described in Chapter 15 Sect. 6. See also
Figs. 17.19, 17.23 and 17.27. Crystal pickups used without an equalizer give an
attenuation approaching 6 db /octave.
If it is not practicable to have two independent attenuation characteristics, a single
characteristic may be used provided that it is capable of giving an attenuation of at
least 30 db at 7000 c /s, for use with worn records.
input voltage characteristic is approximately horizontal for an input up to, say, 0.5
volt ; this may be accomplished by means of a network incorporating voltage -delayed
diodes (such as 6H6) or germanium crystal diodes. The distortion products (all
harmonic and most of the intermodulation) are eliminated by splitting the high fre-
quency part of the amplifier into channels each covering only one octave. Some
possible combinations are
1. 0 -3000 c/s ; 3000-6000 c/s (2 channels).
2. 0 -2000 c/s ; 2000-4000 c/s ; 4000 -8000 c/s (3 channels).
3. 0 -1500 c/s ; 1500 -3000 c/s ; 3000 -6000 c/s ; 6000 -12 000 c/s (4 channels).
Filters are used which provide attenuation at the rate of approximately 30 db /octave
at both ends of each channel, except the low frequency end of the first channel. A
filter network for two channels is given in Ref. 115, while one for four channels is
given in Ref. 138.
The only known defects are that signals below the threshold value are lost, and
that there are some intermodulation products.
(vi) Scott dynamic noise suppressor
The Scott dynamic noise suppressor controls the bandwidth of the amplifier by
means of separate high- frequency and low- frequency tone controls which are auto-
matically controlled by the signal. Fig. 17.38 shows one simple application suitable
for home use (Ref. 251 based on Ref. 114). V, is an amplifying stage which also
provides the voltage for application to both high-frequency and low- frequency control
circuits which in turn control the grid voltages of V, and V3. The parallel resonance
between L, and C, provides an attenuation at high frequencies and a high attenuation
above the normal operating range ; it may also be used as a whistle filter. V, is
used as a reactance valve providing variable capacitive reactance which, together with
C, and its companion series condenser, forms a series resonant circuit with L,. The
low- frequency gate including V3 is quite readily understood. Switch S, allows the
suppressor to be opened, providing maximum frequency range when no suppression
is desired. When S, is closed, R3 controls the amount of suppression. For bad
records, switch S2 allows restriction of the maximum frequency range. Switch Se
closes the high -frequency gate and leaves the low- frequency gate open.
The frequency characteristics of an experimental model built in our Applications
Laboratory is given in Fig. 17.39 (Ref. 251). Scott recommends that C, should be
INPUT AMPLIFIER HIGH -FREQUENCY GATE LOW - FREQUENCY GATE
Crystal +
10,00011
Pickup
OI N F 33,O00ß 47M11
47MR IMR 02NF I MA
I MR IMR 002SYF -0SyF
10,000n
10,00011
26wF
220A
22MA
500 001
NF IMA Z2Mn
00INF stni
--i
.005
4.7MR NF
47MÁ TO214F
HF 1 F'l' 2.MAT2NF
FIG. 17. 38 CONTROL FILTER HIGH -FREQUENCY LOW - FREQUENCY
CONTROL CIRCUJT CONTROL CIRCUIT
Fig. 17.38. Simple Scott dynamic noise suppressor (Ref. 251 based on Ref. 114).
765
17.7 (vi) SCOTT DYNAMIC NOISE SUPPRESSOR
C1 and
adjusted for minimum output at 9000 c /s, but it was found preferable to tune
L1 for minimum output at 7500 c/s to avoid too great a rise in output between the two
frequencies of maximum attenuation. This simple circuit cannot provide a response
extending above about 6000 c /s. Ref. 184.
A more recent model with minimum output at 10 000 c/s is described in
Performance data of a more flexible dynamic noise suppressor built into a complete
amplifier are given in Ref. 173.
+10
o MLii%
AINIIIMr.aIMr
ai a
-w-M311111
c9111wr04iri
-
.1111 7411MMIZII11111111111.aNaa
en
lo
IiERAr./
rlVIINI!
Willr,.
qri.
111
E>.aesia
MNswlc7O
INIIIIIIMI1.a&111=111,11111.
MIIIIIIIMS111011t911M1111111
I i' ENE
20 ....
....
.... . --
e. 0.02V
e. 006v
....
.... ----se.0.4V
0IV
-30
...
....
....
....
a 0ß.v
r- Filter Oyer
40 SOO s,000 10,000
30 100 000
Frequency c/s FIG. 17.39
(vii) Amplifier
The amplifier should be capable of an output of at least 10 watts, with low dis-
tortion and good frequency response. The noise and hum level should be at least
768 (viii) PICKUPS FOR LACQUER DISCS 17.8
45 to 50 db below maximum power output (40 db is an extreme limit for poor re-
cordings). The amplifier should preferably have push-pull triodes with negative
feedback, although triodes without feedback or beam power valves with feedback may
be used.
Some 10 inch and 12 inch discs are also used, but these have been covered in Sect. 2.
The pickups to be used with transcription discs should have a stylus force not
exceeding 1 -1/2 ounces (42 grams). Only permanent points should be used.
The total harmonic distortion on an original recording at 1000 c/s or less, in accord-
ance with good practice, would be less than 2% (Ref. 219).
Intermodulation distortion measured on recordings made by three recording heads,
with low frequency peak amplitudes of 2.5 mils and high frequency velocity 12 db
lower, were approximately 1.0, 9.2 and 52% (Ref. 269). The first head is one giving
exceptionally low distortion, while the second is representative of good practice.
The readings were very little affected by the choice of low or high frequency. When
the level was reduced 6 db, the I.M. distortion was reduced to 23% with the third
recording head (say 6% harmonic distortion).
References to transcription recording (general) : 4, 7, 63, (92, 99, 163), (135, 205,
146), 189, 214, 216, 218, 219, 230, 237.
Standards : Refs. 2, 87, 214, 237: Specifications : 185 ; Bibliography : 105.
(ii) Characteristics of record material, wear and noise
Most processed transcription disc are made of vinyl. Vinyl records will reproduce
up to 1000 playings with a suitable light -weight pickup and permanent tip. For the
effects of elastic deformation see Ref. 212.
*Compromise for reproduction of both lateral transcriptions and 78 r.p.m. shellac discs.
tThis is the deflection of a standard volume indicator. Programme peaks up to 21 cm /sec. would
be anticipated.
770 (ii) CHARACTERISTICS OF RECORD MATERIAL 17.9
speed varies from 27 to 13.1 inches per second. As the groove diameter approaches
7 -1/2 inches there is a progressive loss of high frequencies and increase in harmonic
distortion during both recording and play -back. For an analysis of the radius of
curvature see Ref. 146. For tables giving groove /land ratios see Ref. 205.
The B.C.C. have a maximum radius compensation of 10 db, but the frequency at
which it reaches 10 db is decreased as the diameter becomes smaller (minimum 8 -3/4
inches).
There is no doubt that with certain kinds of music it is impossible to apply the
full N.A.B. pre- emphasis of 16 db at 10 000 c/s together with the necessary amount
of radius compensation to provide a nearly level frequency characteristic without
serious over -modulation and distortion. The only answer seems to lie in the use of
fine groove recording.
References 4, 99, 189, 214, 227.
212. Begun, S. J., & T. E. Lynch " The correlation between elastic deformation and vertical forces in
lateral recording " Jour. Accus. Socy. Am. 13.3 (Jan. 1942) 284. Correction 13.4 (April 1942) 392.
213. Pierce, J. A., & F. V. Hunt " On distortion in sound reproduction from phonograph records "
Jour. Acous. Socy. Am. 10.1 (July 1938) 14.
214. Davies, H. " The design of a high -fidelity disc recording equipment " Jour. I.E.E. Vol. 94 Part III
No. 30 (July 1947) 275.
215. Scott, H. H. " Dynamic noise suppressors " Service 17.9 (Sept. 1948) 16.
216. Heaney, K. (book) " The Radio Engineering Handbook " (McGraw -Hill Book Co., New York,
& London, 3rd edit. 1941) Sec. 21.
218. Aldous, D. W. " Manual of Direct Disk Recording " ( Bernards, Publishers, Ltd. London, 1944).
219. " N.B.C. Recording Handbook " National Broadcasting Co. Inc., Engineering Department.
220. O'Brien, E. J. " High fidelity response from phonograph pickups " Elect. 22.3 (March 1949) 118
222. Le Bel, C. J. " Properties of the dulled lacquer cutting stylus " J. Acous. Soc. Am. 13.3 (Jan. 1942)
265.
223. Reid, J. D. " A large radius stylus for the reproduction of lateral cut phonograph records " J. Acous
Soc. Am. 13.3 (Jan. 1942) 274. Review W.W. 49.1 (Jan. 1943) 5.
224. Sepmeyer, L. W. " Tracing distortion in the reproduction of constant amplitude recordings "
J. Acous. Soc. Am. 13.3 (Jan. 1942) 276.
225. Goldsmith F. H. " A noise and wear -reducing phonograph reproducer with controlled response "
J. Acous. Soc. Am. 13.3 (Jan. 1942) 281.
226. Savory, W. A. " Design of audio compensation networks " Tele -Tech 7.4. (April 1948) 34.
227. Kornei, O. " On the playback loss in the reproduction of phonograph records " Jour. S.M.P.E.
37 (Nov. 1941) 569.
228. Baerwald, H. G. " Analytic treatment of tracking error and notes on optimal pick -up design "
Jour. S.M.P.E. 37 (Nov. 1941) 591.
229. McLachlan, N. W. (book) ' The New Acoustics " (Oxford University Press, 1936) p. 88.
230. Dorf, R. H. (book) " Practical Disc Recording " (Radcraft Publications Inc., New York, 1948).
231. Axon, P. E., & H. Davies " A study of frequency fluctuations in sound recording and reproducing
systems " Jour. I.E.E. Part III 96.39 (Jan. 1949) 65.
232. " Columbia LP record specfications " Tele -Tech 7.10 (Oct. 1948) 63.
233. Roys, H. E. " An improved lacquer disc recording head " Audio Eng. 33.2 (Feb. 1949) 21.
234. Bachman, " Phonograph reproducer design " E.E. 65.3 (March 1946) 159.
235. Mittell, B. E. G. " Annexcre to recorded talk on disc- recording standards " delivered to I.R.E.
(Australia) Convention at Sidney, Australia, Nov. 3, 1948.
236. Aldous, O. W. " American microgroove records " W.W. 55.4 (Apr. 1949) 146.
237. " N.A.B. Recording and reproducing standards " Engineering (May 16, 1949)
238. " Stylus ", " Characteristics. of the new 45 r.p.m. record " Audio Eng. 33.3 (March 1949) 6.
239. St. George, P. W., & B. B. Drisko " Versatile phonograph preamplifier " Audio Eng. 33.3 (March
1949) 14.
240. Van Beuren, J. M. " Cost vs quality in of circuits " F.M. & T. 9.2 (Feb. 1949) 31.
241. Scott, H. H. " Audio developments : observations on the quality of records made here and over-
seas " F.M. & T. 9.1 (Jan. 1949) 23.
242. Morse, M. " Sound embossing at the high frequencies " J. Acous. Soc. Am. 19.1 (Jan. 1947) 169.
243. Hombostel, J. " Improved theory of the light pattern method for the modulation measurement in
groove recording " Electronic Eng. 19.1 (Jan. 1947) 165.
244. McClain, E. F. " A distortion reducing stylus for disk reproduction " J. Acous. Soc. Am. 19.2
(March 1947) 326.
245. Duffield, A. W. " Improvements in disc records through constant amplitude recording " Comm.
20.3 (Mar. 1940) 13.
246. Bomberger, D. C. " A continuously variable equalizing pre-amplifier " Audio Eng. 33.4 (Apr. 1949)
14.
247. Maxfield, J. P., & J. K. Hilliard " Notes on pre-equalization for phonograph records " Audio Eng.
32.4 (Apr. 1948) 15.
248. Fleming, L. " Notes on phonograph pickups for lateral-cut- records " J. Acous. Soc. Am. 12.3
(Jan. 1941) 366.
249. Bauer, B. B. " Pickup placement " Elect. 22.6 (June 1949) 87.
250. McProud, C. G. " General purpose 6AS7G amplifier " Audio Eng. 32.6 (June 1948) 24.
251. Aston, R. H. " A dynamic noise suppressor " Radiotronics No. 132 (July /Aug. 1948) 73.
252. Williams, A. L. " Further improvements in light -weight record reproducers Jour. S.M.P.E. 33
(Aug. 1939) 203.
254. Data kindly supplied by Radio Corporation of America, R.C.A. International Division.
255. Data kindly supplied by William S. Bachman, Director of Research, Columbia Records Inc.
256. Data kindly supplied by R. V. Southey, E.M.I. (Australia) Pty. Ltd.
257. Hunt, F. V. (in discussion on " What constitutes high fidelity reproduction 1 ") Audio Eng. 32.12
(Dec. 1948) 8.
258. Frayne, J. G. & H. Wolfe (book) " Elements of Sound Recording " (John Wiley and Sons, New
York Chapman and Hall, Ltd., London 1949).
259. Langford - Smith, F. (Letter) " Recorded supersonic frequencies " W.W. 55.7 (July 1949) 275.
260. R.M.A. (U.S.A.) Standards and other data REC -103 (Oct. 1946) : Dimensional characteristics-
phonograph records for hone use.
REC -105 (June, 1947) : Disc home recording.
REC -125 -A (July 1949) : Phonograph pickups.
REC -126 -A (Oct. 1950) : Playback needles for home phonographs.
REC -128 (May, 1949) : Standard frequency test records.
261. Andrews, D. R. " Importarce of groove fit in lateral recordings " Audio Eng. 33.7 (July 1949) 18.
262. Goldmark, P. C., R. Snepvangers & W. S. Bachman " The Columbia long -playing microgroove
recording system " Proc. I.R.E. 37.8 (Aug. 1949) 923.
263. Carson, B. R, A. D. Burt & H. I. Reiskind " A record changer and record of complementary design "
R.C.A. Rev. 10.2 (June 1949) 173.
264. Corrington, M. S. " Tracing distortion in phonograph records " R.C.A. Rev. 10.2 (June 1949) 241.
265. Roys, H. E. " Analysis by t is. two- frequency intermodulation method of tracing distortion encoun-
-
tered in phonograph reproduction " R.C.A. Rev. 10.2 (June 1949) 254.
267. Institute of Radio Engineer-, U.S.A. "Standards on radio receivers methods of testing amplitude -
modulation broadcast recei. ers " (1948).
268. Sterling, H. T. " Simplifiec preamplifier design " Audio Eng. 33.11 (Nov. 1949) 16.
774 REFERENCES 17.10
269. Yenzer, G. R. " Lateral feedback disc recorder " Audio Eng. 33.9 (Sept. 1949) 22.
270. Williamson, D. T. N. " High quality amplifier -new version " W.W. 55.8 (Aug. 1949)
(Oct. 1949) 365 ; 55.11 (Nov. 1949) 423. 282 ; 55.10
271. " Disc recording system developments -solutions of many problems to bring about a fidelity
of 20 KC " Tele -Tech. 9.1 (Jan. 1950) 14. range
272. Begun, S. J. (book) " Magnetic Recording " (Murray Hill Books Inc. 1949).
273. Haynes, H. E., & H. E. Roys " A variable speed turntable and its use in the calibration
reproducing pickups " Proc. I.R.E. 38.3 (March 1950) 239. of disk
274. Briggs, G. A. (book) " Sound Reproduction " Wharfedale Wireless Works, Bradford,
England, 1949. Yorks.,
275. Scott, H. H. " The design of dynamic noise suppressors " Proceedings of the National
Conference, Chicago, Illinois, Nov. 3-5, 1947, Vol. 3. p. 25. Electronics
276. Brierley, J. H. " Reproduction of Records
and Recordings) Ltd., Liverpool, England.
-1" booklet published by J. H. Brierley (Gramophones
277. Data supplied by J. H. Brierley of J. H. Brierley (Gramophones and Recordings) Ltd., Liverpool,
(England).
278. Fleming, L. " Equalized pre -amplifier using single stage feedback " Audio Eng. (Mar.
See also Ref. 209. 1950) 24.
279. Reid, J. D. " Universal phonograph styli " J.A. S.A. 21.6 (Nov. 1949) 590.
280. Warren, H. G. (letter) " Pre- amplifier circuit " W.W. 46.6 (June 1950) 238.
281. Bauer, B.B. " All- purpose phonograph needles " Elect. 23.6 (June 1950) 74.
282. Wood, G. H. H. ' Record and stylus wear -advantage of compliant stylus mountings " W.W.
56.7 (July 1950) 245.
283. Mallett, E. S. " The determination of gramophone pick -up tracking weights " Electronic
22.267 (May 1950) 196. Eng.
284. Woodward, J. G. " A feedback -controlled calibrator for phonograph pickups " R.C.A. Rev. 11.2
(June 1950) 301.
285. Roys, H. E. " Determining the tracking capabilities of a pickup " Audio Eng. 34.5 (May 1950)
286. " Playback standards " Audio Eng. 34.7 (July 1950) 4. 11.
287. Fleming, L. " Equalized pre -amplifier using single stage feedback " Audio Eng. 34.3 (March 1950)
24.
288. Bachman, W. S. " The Columbia hot stylus recording technique " Audio Eng. 34.6 (June 1950)
289. Worturan, L. A. (Summary of discussion) " Heated stylus recording technique " Audio Eng. 11.
(July 1950) 24. 34.7
290. Marcus, E. J. & M. V. " The diamond as a phonograph stylus material " Audio Eng. 34.7 (July
1950) 25.
291. Fine, R. S. " An intermodulation analyzer for audio systems " Audio Eng. 34.7 (July 1950)
292. Roys, H. E. " Recording and fine- groove technique " Audio Eng. 34.9 (Sept. 1950) 11. 11.
293. Moyer, R. C., D. R. Andrews & H. E. Roys " Methods of calibrating frequency records " Proc.
I.R.E. 38.11 (Nov. 1950) 1306.
294. Godfrey, J. W. " Reproduction of discs and records for broadcasting " B.B.C. Quarterly
(Autumn 1949) 170. 4.3
295. Terry, P. R. " The variable- disc -speed method of measuring the frequency characteristics
pick -ups " B.B.C. Quarterly 4.3 (Autumn 1949) 176. of
296. Anderson, L. J. & C. R. Johnson " New broadcast lightweight pickup and tone arm " Audio
35.3 (March 1951) 18. Eng.
297. Moir, J. " Recorders and reproducers " FM-TV 11.1 (Jan. 1951) 14.
298. Weil, M. " Phono facts " Audio Eng. 35.6 (June 1951) 20.
299. Dutton, G. F. " Gramophone turntable speeds " W.W. 57.6 (June 1951) 22 7.
300. " Frequency test records -Calibration methods discussed by B.S.R.A. " W.W. 57.6 (June
1951) 227.
301. Kelly, S. " Piezo- electric crystal devices " Electronic Eng. 23.278 (April 1951) 134 ; 23.279
1951) 173. (May
302. Shirley, G. (letter) " Diamond vs sapphire " Audio Eng. 35.8 (Aug. 1951) 6.
303. John, R. S. ' A strain- sensitive phonopickup " Radio and TV News 43 (Feb. 1950) 40.
304. Roys, H. E. " Recording and fine groove technique " Broadcast News No. 60 (July-Aug.
305. West, R. L., & S. Kelly " Pickup input circuits " W.W. 56.11 (Nov. 1950) 386. 1950) 9.
306. Goodell, J. D. " Problems in phonograph record reproduction " Radio & TV News 44.5 (Nov.
1950) 39. _.
307. " AES Standard playback curve " Audio Eng. 35.1 (Jan. 1951) 22.
308. " British long-playing records " W.W. 57.1 (Jan. 1951) S7.
309. Pollock, A. M. ' Thorn gramophone needles " W.W. 56.12 (Dec. 1950) 450 ; correspondence
57.3 (March 1951) 121 ; 57.4 (April 1951) 145.
310. Kelly, S. " Intermodulation distortion in gramophone pickups " W.W. 57.7 (July 1951) 256.
Letter L. J. Elliott, 57.9 (Sept. 1951) 370. (b)
311. Bachman, W. S. " The application of damping to phonograph reproducer arms " Proc. I.R.E.
40.2 (Feb. 1952) 133.
312. Discussion on hum in moving coil and ribbon pickups H. J. Leak, W.W. 56.4 (Apr. 1950) 132 ;
:
J. H. Brierley, 56.6 (June 1950) 238 ; 57.2 (Feb. 1951) 81 ; P. J. Baxandall and R. L. West
(Apr. 1951) 146, 147. 57.4
313. Anderson, L. J., & C. R. Johnson " New lightweight pickup and tone arm " Broadcast News No. 64
(May -June 1951) 8.
314. Weil, M. (letter) " Tracking efficiency " Audio Eng. 34.8 (Aug. 1950) 2, also front cover.
315. " Hot stylus technique " W.W. 58.2 (Feb. 1952) 50.
316. Read, O. (book) " The recording and reproduction of sound " (Howard W. Sams &
Indianapolis, Indiana, 2nd ed. 1952). Co. Inc.
317. Parchment, E. D. " Microgroove recording and reproduction " jour. Brit. I.R.E. 12.5 (May 1952)
271.
Additional references will be found in the Supplement commencing on page 1475.
CHAPTER 18
SECTION 1 : MICROPHONES
(i) General survey (ii) Carbon microphones (iii) Condenser microphones (iv)
Crystal and ceramic microphones (v) Moving coil (dynamic) microphones (vi) Pres-
sure ribbon microphones (vii) Velocity ribbon microphones (viii) Throat microphones
(ix) Lapel microphones (x) Lip microphones (xi) The directional characteristics
of microphones (xii) The equalization of microphones (xiii) Microphone transformers
(xiv) Standards for microphones.
775
776 MICROPHONES -(i) GENERAI, SURVEY 18.1
Microphone ratings
Microphones may be rated in terms of either voltage or power, as described in
detail in Chapter 19 Sect. 1(iv). Ratings in common use include
Voltage ratings
A. Open circuit voltage (0 db = 1 V) for sound pressure 1 dyne/cm2, expressed
in
dbv*.
B. Open circuit voltage (0 db = 1 V) for sound pressure 10 dynes /cm', expressed
in
dbv.
C. Volume units (as read by a standard Volume Indicator) for sound pressure 1
dyne/cm2, expressed in vu.
D. High impedance grid circuit voltage (0 db = 1 V, Z = 40 000 ohms) for sound
pressure 1 dyne/cm2, expressed in dbv.
Power ratings
E. Output powert (0 db = 1 mW) for sound pressure 1 dyne/cm2, expressed in
dbm.
F. Output powert (0 db = 1 mW) for sound pressure 10 dynes /cm', expressed in
dbm.
G. Output power (0 db = 6 mW) for sound pressure 1 dyne /cm'. This rating is
now rarely used.
H. Output power (0 db = 6 mW) for sound pressure 10 dynes /cm'. This rating
is now rarely used.
J. Output power (0 db = 1 mW) for sound pressure 0.0002 dyne /cm'. This is
the R.M.A. Microphone System Rating Gm (R.M.A. Standard SE -105 ; see
Chapter 19 Sect. 1(iv)D).
Voltage ratings -To convert from A to B, add + 20 db.
To convert from B to A, add 20 db. -
If a voltage rating is used for a low impedance microphone, the impedance across
which the voltage occurs should be specified.
Volume units-This is not an official microphone rating, but is sometimes used
as a convenient method of measurement. The Volume Indicator is effectively
a
voltmeter in which 0 vu = 0.77 volts r.m.s. across 600 ohms. To convert vu (Rating
C) to voltage (rating A), add -
2 db. When the output from a low impedance micro-
phone is quoted in terms of vu, it may be inferred that this is equivalent to the power
output in milliwatts (Rating E).
Power ratings -The power rating is the power developed in a load of specified
value. Power ratings E, F, G and H may be calculated one from the other as under :
dbm db db db
Carbon (S.B.) -19 -41 -27 -113
Moving coil -45 to -66 -73 to -94 -53 to -74 -139 to -160
Pressure ribbon -57 to -61 -85 to -89 -65 to -69 -151 to -155
Velocity ribbon -54 to -63 -82 to -91 -62 to -71 -148 to -157
in one example), and the second harmonic distortion is considerably less, but the
output is about 10 db lower than that of the single button type.
Carbon microphones require a source of low voltage to pass a current of from 10
to 50 mA through each button. When the circuit is broken by a switch there is a
tendency for the carbon granules to cohere-this may be reduced by connecting two
condensers and three inductors as shown in Fig. 18.1.
All carbon microphones have a tendency to blasting, while their other defects in-
cluding high noise level limit their application. The principal advantage is the high
output level.
References A5, A21, A23, D1, D3, D4, D7, D13, D14.
(iii) Condenser microphones
A condenser microphone consists of a stretched diaphragm and a back plate, with
a polarizing voltage between them. Owing to the high impedance, the pre -amplifier
must be mounted very close to the microphone. A possible circuit arrangement is
shown in Fig. 18.2. The response may be made almost flat from 30 to 10 000 c/s
with an incident angle of 90 °-on the axis there is a high frequency peak of about 8
db as with all pressure microphones. The average response is about
the microphone itself (0 db = 1 volt per dyne /cm2).
-
60 dbv across
References A5, A7, A23, D1, D3, D4, D7, D11, D13, D14.
The high frequency peak which occurs with microphones of standard size may be
reduced by making the diaphragm smaller. A miniature condenser microphone with
a diaphragm having the area of a human ear drum has been introduced with non -
directional pickup characteristics (Refs. A18, A26). The output level from the pre-
amplifier is - 50 dbm in a sound field of 10 dynes /cm2.
FIG.
FIG. I8.2
Fig. 18.1. Circuit diagram of double button carbon microphone with filter (optional)
to reduce cohering (L = 0.0014 H, C = 0.02 ,F, E = 4.5 to 6 V).
Fig. 18.2. Circuit diagram of condenser microphone.
Fig. 18.3. Circuit diagram of crystal microphone.
All crystal microphones using Rochelle salt or similar materials tend to be affected
by excessive humidity and temperatures above 125 °F. Most of the crystals are now
coated to provide considerable protection against humidity, although their use under
tropical conditions appears to be risky.
A crystal microphone is effectively a capacitance, of the order of 0.03 µF for a dia-
phragm type or 0.0004 to 0.015 µF for a sound -cell type. This is effectively in series
18.1 (iv) CRYSTAL AND CERAMIC MICROPHONES 779
with the generated voltage and following grid -resistor. The following grid resistor'
(R in Fig. 18.3) should be from 3 to 5 megohms for a sound cell type, or 1 to 5 megohms
for a diaphragm type. A low resistance causes attenuation of low frequencies.
A long connecting cable will result in loss of output voltage, but will not affect the
frequency response. The loss is given by
db loss = 20 log [1 + (L x CL) /C,u]
where L = length of cable in feet
CL = capacitance of cable in microfarads per foot
and Cm = capacitance of microphone in microfarads.
A few crystal microphones are equipped with step -down transformers for use in
low impedance circuits.
Ceramic piezo -electric microphones have advantages over crystal types as regards
high temperature and humidity. These have substantially similar performance to
that of crystal microphones. One model has a nearly flat response from 30 to 10 000
-
c/s with an output level of 62 dbv (0 db = 1 V per dyne/cm2).
References to crystal and ceramic microphones : A5, A21, A23, D1, D3, D4, D7,
D13, D14.
close to the source of sound, the low frequencies are strongly accentuated. For
this
reason, a good, quiet studio and correct placement of speakers at a reasonable distance
is necessary if a normal velocity microphone is used for speech. The bi- directional
characteristic is useful where there are two speakers, or two groups, arranged on op-
posite sides of the microphone.
One model incorporates a switch and compensating reactor to reduce the accentua-
tion of low frequencies with close speaking (R.C.A. KB-2C, Ref. A27).
A modified form incorporates a large amount of acoustical resistance to give a nearly
flat response at all frequencies when speaking very close to the microphone (1
to 6
inches away), and noise cancellation for sounds originating at a distance. This is
particularly suitable for outdoor public address (R.C.A. KB -3A, Ref. A28).
References A5, A21, A23, A27, A28, D1, D3, D4, D7, D11, D13, D14.
(viii) Throat microphones
A throat microphone is one which is actuated by direct contact of the diaphragm
with the throat. The high frequency response must be accentuated to obtain in-
telligible speech. Both carbon and magnetic types have been used (Refs. A19, D3).
(ix) Lapel microphones
Normal types of carbon, crystal, dynamic and velocity microphones have been used
for this application (Refs. A21, D3).
(x) Lip microphones
The velocity ribbon microphone has been used for this application (Ref. A5).
(xi) The directional characteristics of microphones
Pressure microphones are non-directional (Curve A Fig. 18.4) for low frequencies
but at high frequencies the response is a function of the angle of sound incidence-
this effect becomes smaller as the diaphragm diameter is reduced.
Velocity ribbon microphones are bi-
directional (Curve B) and the sound
source should be on one, other or both
sides.
There are various types of uni- direc-
tional microphones, among which the
cardioid (Curve C) is most popular. This
characteristic may be achieved by a care-
fully designed combination of a velocity
and a pressure microphone. Similar char-
acteristics may be achieved by the use of
acoustical delay systems in combination
with any pressure microphone such as
dynamic or crystal. A reduction of re-
sponse of about 15 db at the back of the
microphone with respect to that at the
front is obtainable with a typical cardioid
microphone. A single-element ribbon
type microphone has been developed which Fig. 18.4. Polar diagrams showing
has a reduction of 20 to 25 db at the back directional characteristics of microphones
of the microphone (Ref. A22). A com- (A) Non -directional ; (B) bi- directional ;
plete survey of directional microphones is (C) cardioid.
given by Olson (Refs. A17, D3).
Polydirectional microphones provide a choice of two or three directional
characteristics or a continuously variable characteristic. A typical example is
R.C.A. 77-D ribbon type. the
The correct use of directional characteristics in microphone placement minimizes
room reflection and reduces acoustical feedback and background noise.
References to directional microphones A17, A21, A23, D3, D13.
18.1 (xii) THE EQUALIZATION OF MICROPHONES 781
This pattern shall be plotted in the form of a polar diagram as a function of the angle
of incidence. The electrical output at any one angle, E9, shall be referred to the output
at the zero position, E0, by the relation 20 loglo (E9 /E0).
The directional characteristic of a bi-directional microphone shall be at least
- 20 db for all frequencies of a specified frequency range for sounds arriving from
any point in a plane perpendicular to the axis of normal usage. For a uni -directional
microphone the angle of minimum response
- .... shall be on an average of at least
10 db, the average to be based on minimum values for a number of frequency bands.
Standards for microphones Refs. A8, A15, A16, A20.
:
SECTION 2 : PRE-AMPLIFIERS
(i) Introduction (ii) Noise (iii) Hum (iv) Microphony (v) Valves for use in
pre- amplifiers (vi) Microphone pre -amplifiers (vii) Pickup pre- amplifiers (viii)
Gain -controlled pre -amplifiers (ix) Standard pre -amplifiers for broadcasting (x)
Standard pre- amplifiers for sound equipment.
(i) Introduction
A pre -amplifier is a voltage amplifier suitable for operation with a low level input
and whose output is intended to be connected to another amplifier operating at a
higher input level. Pre -amplifiers are commonly used with low -level microphones
and pickups. When a pre -amplifier has to feed the a -f amplifier in a typical radio
receiver, the output voltage is usually taken for design purposes as 1 volt r.m.s. across
a high impedance. This allows for a volume control between the pre -amplifier and
the main amplifier. In some cases it is necessary to incorporate the volume control
in the pre -amplifier itself in order to avoid overloading with abnormally high input
levels. In such cases this volume control is usually placed between the first and
second stages of the pre-amplifier--with very high input levels it may be necessary
to incorporate either a volume control or a fixed attenuator prior to the first grid.
A pre -amplifier follows the same general principles of design as any other voltage
amplifier (see Chapter 12) except that particular attention has to be paid to hum, noise
and microphony, all of which tend to be troublesome.
(ii) Noise
(A) The characteristics of random noise
Random noise includes both thermal agitation noise and valve noise. The
energy of such noise is distributed uniformly over the frequency spectrum. The .
distribution of amplitude with time has been covered in Ref. B1- occasional peaks
exceed four times the r.m.s. value.
Random noise voltages must be added in quadrature- -that is to say as the root of
the sum of the squares.
(iii) Hum
The general features of power supply hum filtering and neutralization have been
covered in Chapter 31 Sections 4 and 5. Hum due to conditions within the valves
and hum due to circuit design and layout have been covered in Chapter 31 Sect. 4.
Hum in voltage amplifiers has been covered in Chapter 12 Sect. 10(vi).
In pre -amplifiers the permissible hum voltage on the grid of the first valve depends
on the maximum signal level at this point. If a high ratio transformer is used, with
a reflected secondary resistance of 25 000 ohms or more, the signal level will be high
and an extremely low hum level will not be necessary. On the other hand, if a low
level pickup is coupled directly to the grid circuit without a step -up transformer,
extreme care and ingenuity will be required to make the hum inaudible.
low -level valves is dependent upon the total plate current of the amplifier,
which is liable to vary. Ref. B52.
(b) The heaters may be connected in series, and supplied from a separate full-wave
rectifier and filter, using a thermionic or selenium rectifier-alternatively a
selenium bridge circuit may be used. Refs. B25, B52.
(c) The heaters may be connected in parallel, and supplied from a selenium or other
suitable rectifier, using either a full -wave or bridge circuit. Refs. B25, B52.
All methods employing rectified and filtered heater supply have a ripple component
which may be large enough to cause audible hum.
15. A r -f oscillator with a frequency of 30 or 40 Kc /s may be used as an alternative
to d.c. heater supply (Ref. B25).
16. If a.c. is supplied to the heater of the first stage, the voltage of this stage alone
may be decreased to about 70% to 85% of normal. This is only permissible with a
limited number of valve types and even so may require selection of valves. A cathode
current well under 1 mA is essential (Ref. B20).
17. The cathodes of all pre -amplifier valves should preferably be earthed directly
to the chassis. However, in practice this is sometimes inconvenient and it may be
found that a minimum capacitance of 100 µF shunted across the bias resistor from
cathode to chassis is satisfactory. High resistance grid -leak bias has not been found
satisfactory for low level operation.
18. An electrostatic shield in the power transformer is essential, even where the
transformer concerned supplies only heaters.
19. The pre -amplifier valves may be de- magnetized in a decreasing a.c. field. This
is only useful when the valves have become magnetized by some means.
20. Magnetic shielding of power leads and all conductors carrying a.c. may be ac-
complished by the use of an iron pipe, flexible conductor or -most effective -aper-
malloy wrap similar to that used for continuous loading of submarine cable (Ref. B51).
Additional notes on hum with pentode valves
1. Hum due to lack of sufficient filtering in the plate supply may be neutralized by
some suitable form of neutralizing circuit (e.g. Fig. 12.57).
2. The hum voltage caused by a magnetic field decreases more rapidly than the
gain as the load resistance is decreased. A low load resistance is therefore desirable
if there is a strong magnetic field (Ref. B42).
3. Hum caused by heater -to -plate leakage is also reduced more rapidly than the
gain as the load resistance is decreased.
4. In pentode valves operating with low input levels there is a further source of
hum in that the magnetic field of the heater varies the partition of current between
plate and screen and so introduces hum. This hum is worse with remote cut -off
valves than with those having sharp cut -off characteristics. This form of hum may
be made zero by a suitable adjustment of grid or screen voltage (Ref. B46). See also
Refs. B23 (Jan. 1949) and B50.
(iv) Microphony
There are very large variations between valves with regard to microphony, and
special low-level types are recommended. Individual selection is, however, desirable
in addition if the valves are required to operate at very low levels. In all cases some
improvement may be made by a cushion socket and a thick rubber pad around the
valve.
(v) Valves for use in pre -amplifiers
It is generally desirable to use valve types manufactured and tested for low level
operation (e.g. types 12AY7, 1620 and 5879) but even with any one of these types
there are large variations in hum, noise and microphony. If a particular pre -amplifier
is required to meet stringent test conditions it is advisable to select valves to meet its
requirements, and to use the balance of the valves in other less -stringent pre -amplifiers,
or in later sockets of the same pre -amplifier.
A choice is possible in the American range between pentode types 1620 and 5879.
The former is very much more expensive than the latter, and its use can generally
be avoided by careful design.
18.2 (v) VALVES FOR USE IN PRE -AMPLIFIERS 787
If the designer decides to use ordinary radio valves for the first stage in a pre-
amplifier, it is advisable to test and select the valves in accordance with the following
procedure. A sufficient number of valves should be operated for about 48 hours
under the same electrode voltages as in the pre -amplifier. They should then be
tested for reverse grid current, and only those with reverse grid currents less than 0.1
µA should be used in very critical positions. If a sensitive microammeter is not
available, a 1.0 megohm grid resistor may be cut in and out of circuit and the change
in plate current measured-see Chapter 3 Sect. 3(iv)A. Low reverse grid current
not only reduces the noise, but also permits a higher grid resistor to be used. The
valves with low reverse grid currents should then be tested in the first socket of the
pre -amplifier for noise, hum and microphony.
Some single -ended pentodes have one heater pin adjacent to the plate, and very
high leakage resistance is required to avoid hum (except when this heater pin
is earthed).
Type 6AU6 may be used as a triode with earthed plate and suppressor, using the
screen as the anode, to reduce hum from leakage between heater and plate.
This device is only suitable for the first stage in the pre -amplifier owing to the
limited screen dissipation rating (0.65 watt). The following operating conditions
are suggested with a plate supply voltage of 180 volts (Ref. B48) :
Load resistance 0.05 0.1 0.25 megohm
Cathode bias resistance 450 750 1600 ohms
Plate current 2.4 1.3 0.59 mA
Stage gain* 21 23 21
*With following grid resistor 05 megohm.
Characteristics of some special low-noise valve types
Type 5879 is a 9 -pin miniature low -noise pentode with publishedgm = 1000 µmhos
at a plate current of 1.8 mA. The grid resistor (pentode or triode operation) may
be up to 2.2 megohms under maximum rated conditions, while the voltage gain with
180 volts supply is 87 with RL = 0.22 megohm and following grid resistor 0.47 meg-
ohm. As a pentode, the noise referred to No. 1 grid with a bandwidth of 13 000 c/s
is of the order of 7.2 µV with grid resistor* 0.1 megohm, with only 10% exceeding
16 V. It may also be used as a triode having µ = 21, g,,, = 1530 µmhos and I6 =
5.5 mA at Eb = 250, E, = -
8 volts. As a triode, the noise under the same con-
ditions as for the pentode is of the order of 6 µV with only 10% exceeding 12 V.
Conditions Triode operation
: Hum
1. One side of heater earthed, cathode by-passed (40 SF), zero grid
resistance -median value 9µV
2. One side of heater earthed, cathode unbypassed, zero grid resistance
-median value 100 IN
3. One side of heater earthed, cathode by-passed (40 µF), grid resistor
0.1 mQ-median value 13 µV
4. Centre-tapped resistance across heater supply, returned to fixed bias
point (+ 20 to + 50 volts) will reduce hum on most valves to less
than 20 µV
5. Centre- tapped resistance across heater supply, returned to voltage
to give minimum hum, cathode by- passed (40 µF), grid resistor 0.1
megohm -median value 3.5 µV
-less than 10% exceeding 8 µV
When adjusted for minimum hum, this feeds a small hum bucking signal into the
grid to oppose other minor sources of hum in the amplifier (Ref. B16).
Type 12ÁY7 is a low noise twin triode (Ref. B17) with µ = 40, g,=1750 µmhos
and I b = 3 mA at E b = 250, E , = -4
volts. The following tests were made with
a grid resistor 0.1 megohm and plate load resistor 20 000 ohms. Shot -effect noise
(40 to 13 000 c /s) referred to the grid-median valve 4.7 µV, maximum limit 8 µV.
The noise due to the grid resistor alone is about 3.8 µV.
788 (v) VALVES FOR USE IN PRE-AMPLIFIERS 18.2
Hum referred to the grid -median valve 3.9 µV, maximum limit 12 µV with cathode
resistor by- passed (40 SF).
When used in a typical circuit, placed with the electron stream in a magnetic field*
of 1 gauss, the hum voltage on the grid is about 10 V.
With a grid resistor of 30 000 ohms, approximately 4µV of hum per volt of heater
potential per µµF of capacitance between heater circuit and grid circuit will appear
at the grid. If the heater voltage is 6.3, this will give approximately 25 µV per µµF
of coupling. For applications see Figs. 18.7A, 18.7B.
FIG. 18.5
FIG. 18.6
This circuit has the disadvantage that the cathode follower stage may give rise to
audible hum. It should therefore be restricted to use with a sensitive type of crystal
microphone or to applications where a low hum level is not essential.
(C) A preferred arrangement is a good transformer in the cathode circuit of the 6J5,
stepping down from 20 000 ohms primary impedance to any desired line impedance.
(D) Cathode follower as low -noise input stage
Fig. 18.6A shows the conventional cathode follower, and Fig. 18.6B shows the modi-
fied circuit for use as a low-noise input stage in connection with crystal microphones
dbvg = decibels of voltage gain-see Chapter 19 Sect. 1(ü).
790 (vi) MICROPHONE PRE -AMPLIFIERS 18.2
or pickups. It has been shown (Ref. B53) that the modified circuit will have the
same stability as the conventional circuit if R 4 = (1 -}- G)R,, where G is the voltage
gain of the valve defined as the ratio of the voltage across Rs to that across R1. If
ZL is high, as it will be if coupled to another valve grid, then G will be approximately
constant and a real number. In a properly designed circuit R 4 can be made con-
siderably higher than R1.
80,0000 L zlBn J
X X
3
INPUT'( p 63051
20,0010.
To V2
OE
12,000.0
0t
10,00041 12,0000
HeiF
T16NF 1I6),F Á
640
'
S=
Fig. 18.7. Two stage broadcast station microphone pre -amplifier with gain of 29 db.
18.2 (vi) MICROPHONE PRE -AMPLIFIERS 791
A pre -amplifier gain of 40 db may result in over-loading of the pre -amplifier and
distortion if used with very high microphone input levels -a
device may be incorpor-
ated to give an optional attenuation of 10 db in the pre -amplifier and thus reduce its
total gain to 30 db when there is danger of overloading.
A typical microphone pre -amplifier with a gain of 29 db is shown in Fig. 18.7. Both
valves are connected as triodes, the first stage using the screen as the anode, and the
gain without feedback is 47 db. When operated with an input level of
the signal to noise ratio is 62 db. The maximum power output is 10 mW (+ 10 dbm)
60 dbm, -
at which the distortion is less than 1%-for this level to be reached, the input level
must be -29 dbm. The frequency response is from 30 to 10 000 c/s with less than
0.5 db variation. This circuit may possibly be criticised on two points. The plate
load resistor for the first stage could be reduced to 50 000 ohms without any appreci-
able loss of gain, and the balance .B
of the voltage could be absorbed
by a separate resistor, duly by-
passed. This would reduce the 12AY7
soon Eat).
INPUT OUTPUT
I MP.
SNF
24,00011 IW
S MA
dWNA BONF OB + 250 V.O.C.
Rui,tors 1/2 Watt Unless Specified
FIG. 18.7 B
Fig. 18.7B. Three-stage balanced cross- neutralized pre -amplifier using two 12AY7
valves and 12A U7 (Ref. B17).
792 (vi) MICROPHONE PRE -AMPLIFIERS 18.2
30 to 20 000 c /s, the average noise is 87 db down from the + 24 dbm level and the
distortion is 0.55% at 24 dbm output. The combined hum and noise voltages are
only 7.6 db above the theoretical value on the poorest valves, and only 2.2 db on the
average (Ref. B17).
The essential amplifying and attenuating circuits of the four pre -amplifiers and one
" A " amplifier of a single unit equipment for studio use are shown in Fig. 18.7C.
Each pre -amplifier has a 3- position key. In the pre -amplifiers, type 6AU6 is used as
a triode with earthed plate and suppressor, and negative feedback is applied from
plate circuit to grid. Each pre -amplifier has a separate attenuator normally " hold-
ing " 6 db. The " A " amplifier uses type 6AU6 as a high gain pentode followed by
the master gain control normally " holding " 20 db.
The " B " amplifier of the same equipment is shown in Fig. 18.7D, and has negative
feedback from the secondary of the output transformer to the cathode of V8.
With all controls set at maximum, an input of -
86 dbm gives an output of + 8
dbm. The distortion is less than 1% at 18 dbm output. The noise level through
any one channel is 60 db or more below output level with input -
60 dbm, output
+ 8 dbm.
A pre -amplifier circuit used by the B.B.C. is described in Ref. B41. A special
low -noise microphone amplifier for acoustical measurements is described in Ref. B54.
A pre -amplifier for the Western Electric type 640AA condenser microphone, using
subminiature valves, is described in Ref. B43. Extremely high insulation resistance
is required in the network between the microphone and the first grid, particularly
in the coupling capacitor.
PRE -AMPLIFIER AMPLIFIER
25MIL'
16NF
MIC I
NKEY
5,
160
MI+F
Á
SWITCH
(1.-1) 5rF
M IC 2
MIC3 32, B+
1.16yF
1
KEY
SWITCH
f
'- PRE- AMPLIFIER Not
MASTER GAIN CONTROL
MIC4
Z
TO B.AMPLIFIER
MIC S
I¡
717
KEY J PRE- AMPLIFIER No 3
SWITCH
MIC 6
M 7
II`
KEY
PRE- AMPLIFIER No4
SWITCH FIG. IB.7C
MIC B
Fig. 18.7C. Essen ial amplifying and attenuating circuits of pre -amplifiers and " A "
amplifier of single unit equipment for studio use (Ref. B18). See also Fig. 18.7D.
18.2 (vi) MICROPHONE PRE -AMPLIFIERS 793
References to microphone pre -amplifiers : A16, A24, B19, B28, B41, B43, B48,
B53, B54.
FIG. 18.7 D
Fig. 18.7D. " B " amplifier of same equipment as Fig. 18.7C (Ref. B18).
Step I
Step 2
INPUT R f--O
Sttp P (I)
OUTPUT
Fig. 18.8. Con-
tinuously variable
volume control FIG. 18.9
The values of R, and Re have been tabulated for steps from 1 to 10 db (Table 1
on page 796).
Fig. 18.10. Constant im- Fig. 18.11. Constant im- Fig. 18.12. The L
pedance T section at- pedance II section at- type single section
tenuator. tenuator. attenuator.
The values of resistances given in Table 1 are for Z = 100 ohms. For any other
value of Z, the tabulated values of resistances should be multiplied by (Z/100). See
Figs. 18.10, 18.11 and 18.12.
796 (iii) CONSTANT IMPEDANCE ONE DIRECTION 18.3
The following references will be found helpful in supplying further general in-
formation on attenuators :
Periodicals C6, C12 (the most comprehensive of all), C13, C17 (the basic treatment).
Books D2, D6, D8, Dl 1 (brief treatments are given in several other books).
(iv) Multiple section attenuators
Any convenient number of 17 sections may be connected in tandem, and the effec-
tive resultant is the " ladder " attenuator of Fig. 18.13 in which the two parallel re-
sistors at the end of each section and the commencement of the next are drawn as a
single equivalent resistor (R, /2). The values of R. and R, are calculated as for a
single 17 section (eqn. 2 and Table 1) to provide the desired attenuation per stage.
The value of R. is given by R, Z /(R, -F Z). The minimum attenuation is 3.5 db
on tapping point P1. The input impedance varies considerably as the tapping point
is changed but may be made constant, if desired, by inserting resistors at the points
marked X. The output impedance also varies, although less so than the input im-
pedance. This device may also be used when the impedance of the source differs
from Z.
Loss Series arm Shunt arm Loss Series arm Shunt arm
(db) (each) (db) (each)
INPUT OUTPUT
p y o
'FIG. 16.16
adjustment provided that the input level is not too high. Fixed grid bias must be
used. A universal design curve is given in Fig. 18.18 (Ref. C15).
A circuit using type 6BE6 as a fader -mixer is shown in Fig. 18.18A. This has a
voltage gain of 62 from grid 1 input, and 25 from grid 3 input. The distortion for
798 (v) ELECTRONIC ATTENUATORS 18.3
60
so
40
z
O
30
z
<
20
o
'OOl .01 Fo lo
ama
FIG. 18.16
Fig. 18.18.
Universal design curve for electronic attenuator (Ref. C15).
an output of 5 volts is 0.19% with input to grid 1, and 1.4% with input to grid 3. A
plate load resistor of 50 000 ohms gives lower gain and distortion. For electronic
gain control purposes, it is advisable to apply the signal to grid 3 and the control
voltage to grid 1 (Ref. C20).
OINF
pedance. In some cases the mixing is carried out between the input sources and the
grid of the first amplifier stage, and here the load impedance presented to each source
must be maintained constant.
The simplest type of fader', which does not provide mixing, is shown in Fig. 18.19.
It is only suitable for high level pickups, or the secondaries of step -up transformers,
since the noise from the moving contact is appreciable. The control is a centre -
tapped potentiometer with a total resistance 2R. The value of R should be that
recommended for each pickup.
Fig. 18.20 shows a series network mixer which may be used as a fader, but it has
serious drawbacks. Both sides of input source A are above earth, and any hum picked
up in this channel is fed without appreciable attenuation to the following grid. More-
over stray capacitances to earth of channel A tend to by -pass the high frequency signal
voltages of channel B. This arrangement is not recommended.
A modified series network mixer is shown in Fig. 18.21 in which one side of each
channel may be earthed, but a transformer must be used to couple the outputs to the
grid. This may be used for fading and mixing.
A RI
Fig. 18.19. Simple type Fig. 18.20. Series Fig. 18.21. Modified
of fader for two pickups. network mixer. series network mixer.
The parallel network mixer is shown in Fig. 18.22 where three input sources are
provided for, although any number may be used. The value of R, may be made equal
to or greater than Rt, and similarly with the other channels. If the mixer is con-
nected between two amplifying stages, all resistances may be equal (say 0.5 megohm)
and the maximum insertion loss will be 6 db for 2 channels, 9.5 db for 3 channels or
12 db for 4 channels. The insertion loss varies when the controls are moved -the
maximum variation in insertion loss caused by any one potentiometer is 2.5 db for
two channels but less for more channels. This interaction is less if the source im-
pedance is considerably smaller than the resistances in the mixer, so that triode valves
r+a are preferable to pentodes in the preceding stage.
A The circuit of Fig. 18.22 may also be used when the sources
are microphones or pickups, but R1, R2 and R3 should be the
correct load resistance in each case. The isolating resistors
R4, R5 and RB may be made equal to the highest of the load re-
sistances. This arrangement is only practicable with fairly
high source levels, on account of the insertion loss and its
effect on the signal -to -noise ratio.
c
One application of the parallel network mixer is
given in Fig. 18.23-this general set -up may be modified to
suit any practical case when there is a considerable difference
- FIG. 18.22 in level between the two sources.
Fig. 18.22. Parallel Probably the most popular of all mixing circuits in this
network mixer. class is the method using a common plate load for two or more
valves, When two valves are used with a common
"'The word " fader " is here used in the sense of fading out one input source and fading in another.
The fading of a single source is here called attenuation. Mixing is the combination of two (or more)
sources so that both are amplified simultaneously ; a mixer may however be used also for fading.
800 (vii) NON -CONSTANT IMPEDANCE MIXERS AND FADERS 18.3
68oeo4GOV)
B+
1MA 25,000A
6AU6 05rF or Mghsr
YI 814F v Fig. 18.23. Complete pre- amplifier in-
rsMa corporating parallel network mixer, for
use with high level crystal microphone
CRYSTAL 00n and crystal pickup. R, = 1 to 5MS2 ;
MIC.
SNF R, = 0.5MS2 or higher. Gain in V,
is 30 dbvg.
- CRYSTAL
PICKUP
EQUALIZER
i .5 MA
FIG. 18.23
plate load resistor, the plate resistance of each valve acts as a shunt load on the other,
thereby reducing the gain and the output voltage for a limited distortion. With
triodes, the gain is equal to the normal gain for a single valve multiplied by
(r, -}- RL) /(r, + 2RL). If RL is considerably greater than the gain is slightly
greater than half that for a normal valve. With pentodes, the loss of gain due to shunt-
r
ing is slight, and may be neglected in most calculations. The effect of the shunting
on the maximum output voltage is much more severe than on the gain. The output
voltage for limited distortion is approximately equal to the normal output voltage
multiplied by r, /(r, RL). If RL = 5r, as for a typical triode, then the output
voltage is reduced to one fifth of its normal value. For this reason it is suggested that
RL = 2r, would be more suitable, giving a reduction to one third of the normal value.
Here again, the effect on pentodes is small.
One simple but very effective mixer incorporating
common plate load mixing is Fig. 18.24. V, may be any
twin triode, RL may be determined as outlined above,
while the plate supply voltage may be about 250 volts,
or higher if a high output voltage is required with low
distortion. If desired, a second twin triode with a
further two input channels can be added. Alternatively, e+
two r-f pentodes may be used in place of V, to provide
increased gain. Various combinations may be devised FIG. 18.24
incorporating one or more common plate load mixers Fig. 18.24. Simple
with other types of mixers or faders to meet almost any mixer using twin triode
possible requirements. Fig. 18.25 is one example- with common plate load
V, and V, may be any suitable twin triodes, R, pro- resistor.
vides fading for the two microphones and R, is a combined microphone volume control,
R4 and R, provide fading and mixing for the pickups while R, is a master volume con-
trol. V, has a common plate load resistor.
One of the problems with mixers of the common plate load type is the control of
volume without introducing noise or circuit complications. This problem may be
overcome by the use of 6L7 type valves in which the amplification is controlled by
varying the voltage on the third grid (Fig. 18.26). This mixer is intended for use
S F OSpF
s0,000n
á
Q
C SM11$
j sMn 8YF
O
U_O
O. 0.5,444=
R. FIG. 18.2`.
O
Fig. 18.25. Mixer for two microphones and two pickups. Values of resistors are for
V, = Vg = 6SN7 -GT.
18.3 (vii) NON-CONSTANT IMPEDANCE MIXERS AND FADERS 801
FIG. 15.26
Fig. 18.26. Three channel mixer with electrpnic volume control (Ref. C4).
with three high -level crystal microphones ; the volume controls operate on d.c. and
may therefore be placed some distance from the mixer (Ref. C4).
References to non -constant impedance mixers : C3, C4, D10.
*Constant impedance indicates that both the input impedance and the looking-backwards output
resistance are maintained constant under all conditions.
802 (viii) CONSTANT IMPEDANCE MIXERS AND FADERS 18.3
R1 = R(n - 1) /n R, =O
RL = R(2n -
=
1)/n2 Mixing loss = 10 logro (2n - 1)
n 2 3 4 5 6
Rs =0.50R 0.67R 0.75R
0.80R 0.83R
RL =0.75R 0.56R 0.44R
0.36R 0.31R
Mixing loss =4.77 6.99 8.45
9.54 10.41 db.
If it is desired to maintain all input and output impedances constant and equal to
R, it will be necessary to insert a matching
transformer or to add a matching pad, the FIG. 18.2
n
=
=
n2 /(2n
2
- 1).
3 4 5 6
impedance ratio = 1.33 1.80 2.29 2.78 3.27
loss in pad (N) = 4.77 6.99 8.45 9.54 10.41 db.
(C) Series-parallel type mixers
Fig. 18.29 shows a four channel mixer using T type attenuators in series -parallel.
The input and output are both balanced. The values of R, and Ro are given by
R4 = R(n -
3) /n and R = 4R(2n
where n is the total number of input channels.
3) /n2 -
If n =4 6 8
then R4 =
0.25R 0.5R 0.625R
Ro = 1.25R R 0.75R
and loss in mixer = 7.0 9.5 11.1 db.
With six channels, the output resistance is equal to R and the taper pad will not be
required. In other cases its loss should be added to that in the mixer.
(D) Bridge type mixers
Fig. 18.30 shows a bridge type mixer with four input channels. This can only be
earthed at one point, but is otherwise satisfactory for four input sources. The output
resistance is equal to R, so that no taper pad or matching transformer is required
the mixing loss is 6 db, which is less than that of any other 4 channel resistance mixer.
-
(E) Coil mixing (Ref. Cl8)
Coil mixing refers to the use of a special transformer (Fig. 18.31). The loss is the
theoretical minimum and is given by 10 logo n where n is the number of input sources
-it is 6 db when n = 4. The value of R, is given by
R,= }R(n -1)
and is equal to R when n = 4. The secondary transformer impedance is equal to nR.
18.3 (viii) CONSTANT IMPEDANCE MIXERS AND FADERS 803
FIG. 18.29
Rt - R
Fig. 18.30. Constant A
1. When the impedance of the input circuit is required to be higher than that of
the output circuit, a parallel mixer should be used.
2. When the impedance of the output circuit is required to be higher than that of
the input circuit, a series or coil type mixer should be used.
3. When the input and output circuits are required to be nearly the same im-
pedance, a series -parallel or bridge type mixer should be used.
References to constant impedance mixers :
Cl, C2, C3, C12 (parts 7 and 8), C16, C18, D10,
D11.
(G) Precautions with studio type mixing
systems
Under any possible operating conditions, the
level at any stage should be kept at least 6 db and
preferably 10 db above the critical pre -amplifier
level of - 60 vu.
With mixer controls turned right off, and a
normal level applied to each input in turn, the
ratio of " leakage " at 10 000 c/s to normal pro-
gramme level should be better than 70 db.
Cross -talk between two different circuits may
be reduced by using twisted leads for all speech Fig. 18.31. Constant impedance
circuits with an electrostatic screen around each mixer using coil mixing with
pair. four input sources.
804 REFERENCES 18.4
Where there are long circuits between pre -amplifier outputs and mixer outputs,
it is usually considered wise to use balanced transformer inputs to avoid the possi-
bility of cross -talk or noise affecting the programme.
SECTION 4 : REFERENCES
(A) MICROPHONES AND THEIR APPLICATION, STUDIO EQUIPMENT
A2. Queen, I. " Microphones " Radio Craft (Sept. 1944) 723.
A5. Amos, S. W. and F. C. Brooker " Microphones -a detailed review of their design and
tics " (1) Electronic Eng. 18.218 (April 1946) 109 ; (2) " Sound waves and the physical characteris-
of microphones " 18.219 (May 1946) 136 ; (3) " Pressure operated microphones " 18.220 properties
1946) 190, also 18.221 (July 1946) 221 ; (4) " Ribbon velocity and combined types " 18.222 (June
(Aug.
1946) 255.
A7. Nygren, A. C. " Condenser microphone design-tiny condenser microphone (W.E. 640-AA)
combined with bullet shaped amplifier " F.M. and T. 6.8 (Aug. 1946) 38.
A8. Report of the Standards Committee (1933) of the Institute of Radio Engineers, U.S.A.
A15. American Standards Association " American recommended practice for the calibration of micro-
phones " Z24.4 -1938.
A16. Radio Manufacturers Association (U.S.A.) " Standard audio facilities for broadcasting systems "
TR -105 -B (Nov. 1949).
A17. Olson, H. F. " Gradient microphones " J. Acous. Soc. Am. 17.3 (Jan. 1946) 192.
A18. Hilliard, J. K. " An omnidirectional microphone " (Altec- Lansing miniature condenser type)
Audio Eng. 33.4 (Apr. 1949) 20.
A19. Martin, D. W. " Magnetic throat microphones of high sensitivity " J. Acous. Soc. Am. 19.1
(Jan. 1947) 43.
A20. Radio Manufacturers Association (U.S.A.) " Microphones for sound equipment " SE -105 (Aug.
1949).
A21. Catalogues of microphones.
A22. Olson, H. F., and J. Preston "Single- element unidirectional microphone" Jour. S.M.P.E. 52.3
(March 1949) 293.
A23. Beranek, L. L. (book) " Acoustic Measurements " (John Wiley and Sons Inc. New York ; Chap-
man and Hall Ltd. London, 1949).
A24. Telfer, J. E. " Audio -frequency equipment for broadcasting services " Proc. I.R.E. (Ault.) 11.5
(Mav 1950) 107.
A25. Olson, H. F., and J. Preston " Unobtrusive pressure microphone " Audio Eng. 34.7 (July 1950)
18.
A26. Hilliard, J. K. " Miniature condenser microphone " Jour. S.M.P.E. 54.3 (March 1950) 303.
A27. Anderson, L. J., and L. M. Wigington " The Bantam velocity microphone " Audio Eng. 34.1
(Jan. 1950) 12.
A28. Anderson, L. J., and L. M. Wigington " The KB -3A high- fidelity noise -cancelling microphone "
Audio Eng. 34.4 (April 1950) 16.
A29. Beaverson, W. A., and A. M. Wiggins " A second -order gradient noise cancelling microphone
using a single diaphragm " J. Acous. Soc. Am. 22.5 (Sept. 1950) 592.
A30. Radio Manufacturers Association " Standard amplifiers for sound equipment " SE-101 -A (July
1949).
Additional references will be found in the Supplement commencing on page 1475.
(B) PRE -AMPLIFIERS, NOISE AND HUM
Bl. Landon, V. D. " The distribution of amplitude with time in fluctuation noise " Proc. I.R.E. 29.2
(Feb. 1941) 50.
B2. Bell, D. A. " Measurements of shot and thermal noise " W.E. 18.210 (March 1941) 95 ; (letter)
W. H. Aldous and E. G. James, W.E. 18.214 (July 1941) 278.
B7. Bell, D. A. (letter) " Shot noise and valve equivalent circuits " W.E. 20.242 (Nov. 1943) 538.
B8. Merchant, C. J. " Thermal noise in a parallel r.c. circuit " Elect. 17.7 (July 1944) 143.
B13. Campbell, N. R., and V. J. Francis " A theory of valve and circuit noise " Jour. I.E.E. Part 3 ;
93.21 (Jan. 1946) 45.
B14. Hooke, A. H. " A method of measuring grid primary emission in thermionic valves " Electronic
Eng. 18.217 (March 1946) 75.
B15. Crawford, K. D. E. " H.F. Pentodes in electrometer circuits " Electronic Eng. 20.245 (July 1948)
227.
B16. Heacock, D. P., and R. A. Wissolik " Low-noise miniature pentode for audio amplifier service "
Tele -Tech 10.2 (Feb. 1951) 31. See also Ref. B56.
B17. Knight, C. A. and A. P. Haase " New low noise input tube " Radio and T.V. News 12.3 (March
1949) 15.
B18. Amalgamated Wireless Australasia Ltd., Consolette G52107.
B19. Meyer, R. J. " Open-grid tubes in low -level amplifiers " Elect. 17.10 (Oct. 1944) 126.
B20. Shipton, H. W. " Valve and circuit noise in high gain amplifiers " Electronic Eng. 19.229 (March
1947) 81.
B23. Walter, W. G., H. W. Shipton and W. J. Warren " Demagnetising valves -as a cure for residual
ripple " Electronic Eng. 20.245 (July 1948) 235 ; Correspondence 20.248 (Oct. 1948) 339 ; 20.250
(Dec. 1948) 406 ; 21.251 (Jan. 1949) 30.
B25. Smith, F. W. " Heater supplies for amplifier hum reduction " Audio Eng. 32.8 (Aug. 1948) 26.
B26. Britton, K. G. " Reducing heater hum " W.W. 54.10 (Oct. 1948) 360.
B28. Chinn, H. A. " Audio system design fundamentals " Audio Eng. 32.11 (Nov. 1948) 11.
B40. Pierce, J. R. " Noise in resistances and in electron streams " B.S.T.J. 27.1 (Jan. 1948) 158.
B41. Ellis, H. D. " Studio equipment -a new design " B.B.C. Quarterly (April 1946).
B42. Dickerson, A. F. " Hum reduction " Elect. 21.12 (Dec. 1948) 112.
B43. Le Bel, C. J. " New developments in preamplifiers " Audio Eng. 33.6 (June 1949) 9.
B44. Thompson, B. J., D. O. North and W. A. Harris " Fluctuations in space- charge-limited currents
at moderately high frequencies " R.C.A. Rev. Jan., April, July, Oct 1940, Jan., April, July, 1941.
B45. Williams, F. C. "Thermal fluctuations in complex networks" Jour. I.E.E. Wireless Section 13
(1938) 53.
18.4 REFERENCES 805
B46. Zakarias, I. " Reducing hum in pentodes " Elect. 21.11 (Nov. 1948) 170.
B48. " Grounded plate type 6AÚ6 triode connection for pre-amplifier use " Radiotronics 142 (April
1950) 45.
B49. Data Sheet XIX " Circuit noise due to thermal agitation " Electronic Eng. 14.167 (Jan. 1942) 591.
B50. Zakarias, I. (letter) " Hum in a.c. valves " Electronic Eng. 22.263 (Jan. 1950) 33. See also B23,
B46.
B51. Hopper, F. L. " Noise considerations in sound -recording transmission systems," Jour. S.M.P.E.
54.2 (Feb. 1950) 129.
B52. Hedge, L. B. " D.C. heater supply for low -level amplifiers " Audio Eng. 35.6 (June 1951) 13.
B53. Tucker, M. J. " Cathode followers as low-noise input stages " Electronic Eng. 23.281 (July 1951)
270.
B54. Shorter, D. E. L., and D. G. Beadle " Equipment for acoustic measurements -A portable general
purpose microphone amplifier using miniature valves " Electronic Eng. 23.283 (Sept. 1951) 326.
See also Chapter 31 Refs. 6, 7, 8, 9, 10.
(C) ATTENUATORS AND MIXERS
Cl. Wright, P. B. " Mixer and fader control circuit design " Comm. (1) (Nov. 1943) 44 ; (2) (Dec.
1943) 44.
C2. Cooper, M. F. " Audio frequency mixers " W.E. 21.246 (March 1944) 117.
C3. Crane, R. W. " Audio mixer design " Elect. 18.6 (June 1945) 120.
C4. Patchett, G. N. " Mixing crystal microphones " W.W. 52.2 (Feb. 1946) 57.
C6. Espy, D. Attenuator design " Elect. (Nov. 1941) 51.
C7. Honnell, P. H. " Unsymmetrical Attenuators " Elect. 15.8 (Aug. 1942) 41.
C9. Wright, P. B. " Attenuator design-for amplifier gain controls " Comm. 23.10 (October 1943) 38.
C12.
Cll. Blackwell, R. F., and T. A. Stranglar " Attenuator design' W.E. 21.246 (March 1944) 122.
Wright, P. B. " Resistive attenuator, pad and network, theory and design " Comm. (1) 24.8 (Aug.
1944) 49 ; (2) 24.10 (Oct. 1944) 62 ; (3) 25.1 (Jan. 1945) 50 ; (4) 25.5 (May 1945) 62 ; (5) 25.6
(June 1945) 68 ; (6) 25.7 (July 1945) 50 ; (7) 25.8 (Aug. 1945) 64 ; (8) 25.8 (Sept. 1945) 68 ; (9)
25.10 (Oct. 1945) 72; (10) 25.11 (Nov. 1945) 61.
C13. West, S. S., and E. D. McConnell " Attenuator design " W.W. 46.14 (Dec. 1940) 487.
C14. Baker, W. G. " Notes on the design of attenuating networks " A.W.A. Tec. Rev. 1.2 (June 1935)
19 ; 2.1 (Jan. 1936) 42.
C15. Smith, F. W., and M. C. Thienpont " Electronic attenuators " Comm. 27.5 (May 1947) 20.
C16. Morrical, K. C. " Design and use of mixing networks " Audio Eng. 31.10 (Nov. 1947) 11.
C17. McElroy, P. K. " Designing resistive attenuating networks " Proc. I.R.E. 23.3 (March 1935) 213.
C18. Miller, W. C., and H. R. Kimball " A recording console, associated circuits and constant B equal-
izers " Jour. S.M.P.E. 43 (Sept. 1944) 186.
C19. Leipert, C. J. " Simplified methods for calculation of H and T attenuation pads " T.V. Eng. 1.7
(July 1950) 7.
C20. Circuit Laboratory Report " Audio frequency applications of type 6BE6 " Radiotronics 16.2
(Feb. 1951) 41.
(D) GENERAL REFERENCES-BOOKS
D1. Terman, F. E. " Radio Engineering " (3rd ed. 1947 McGraw -Hill Book Co., New York and
London).
D2. Terman, F. E. " Radio Engineers' Handbook " (1st ed. 1943 McGraw -Hill Book Co., New York
and London).
D3. Olson, H. F. " Elements of Acoustical Engineering " (2nd ed. 1S 47, D. Van Nostrand Company,
New York).
D4. Henney, K. (Editor) " The Radio Engineering Handbook " (4th ed. 1950, McGraw -Hill Book Co.,
New York and London).
D5. Reich, H. J. " Theory and Application of Electron Tubes " (2nd ed. 1944, McGraw-Hill Book Co.,
New York and London).
D6. " Reference Data for Radio Engineers " (3rd ed. Federal Telephone and Radio Corporation, New
York).
D7. Boyce, W. F., and J. J. Roche," Radio Data Book " (Boland and Boyce Inc., Montclair, N.J. U.S.A.).
D8. ' Motion Picture Sound Engineering " (D. Van Nostrand Company Inc. New York, 1938).
D9. Valley, G. E., and H. Wallman (Editors) " Vacuum Tube Amplifiers " (McGraw -Hill Book Com-
pany, New York and London, 1948).
D10. Greenless, A. E. " The amplification and distribution of sound " (Chapman and Hall Ltd. London,
2nd edit. 1948).
D11. Frayne, J. G., and H. Wolfe," Elements of Sound Recording " (John Wiley and Sons, New York;
Chapman and Hall Ltd., London, 1949).
D12. Blackburn, J. F. (Editor) " Components Handbook " (McGraw -Hill Book Company Inc. New
York and London, 1949)
D13. Beranek, L. L. " Acoustic Measurements " (John Wiley and Sons, New York ; Chapman and
Hall, London, 1949).
D14. Kinsler, L. E., and A. R. Frey " Fundamentals of Acoustics " (John Wiley and Sons, New York;
Chapman and Hall, London, 1950).
CHAPTER 19
806
19.1 (i) POWER RELATIONSHIPS EXPRESSED IN BELS AND DECIBELS 807
watt " is misleading unless it is also stated that the original level was 1 watt. A far
more satisfactory way is to state that a rise of 3 db has occurred. This may be cal-
culated quite simply since the gain in decibels is
10 logo (2/1) = 10 log102 = 10 x 0.301 = 3.01 db
or approximately 3 db.
In a similar manner a decrease from 2 watts to 1 watt is a change of approximately
- 3 db.
It has been found that a change in level of 1 db is barely perceptible to the ear,
while an increase of 2 db is only a slight apparent increment. For this reason variable
attenuators are frequently calibrated in steps of 1 db or slightly less. In a similar
manner an increase from 3 watts to 4.75 watts is only a slight audible increment, being
an increase of 2 db.
In order to simplify the understanding of barely perceptible changes the follow-
ing table has been prepared, and it will be seen that a move from one column to the
nearest on left or right is equivalent to a change of 2 db. In this table 0 db is taken
as 3 watts.
db : -10 -8 -6 -4 -2 0 +2 +4 +6 +8 +10 +12
watts : 0.30 0.47 0.75 1.2 1.9 3.0 4.75 7.5 12 19 30 47.5
In addition to the application of decibels to indicate a change in level at one point,
they may also be used to indicate a difference in level between two points such as the
input and output terminals of a device such as an amplifier or attenuator. For ex-
ample, consider an amplifier having an input power of 0.006 watt and an output power
of 6 watts. The power gain is 6/0.006 or 1000 times, and reference to the tables
shows that this is equivalent to 30 db. The amplifier may therefore be described
as having a gain of 30 db, this being irrespective of the input or output impedance.
References 1, 8, 9, 16 (Chap. 32).
widely used at the present time. As an example, a power of 1 watt may be described
as
30 db (reference level 1 mW)
or 30 db (0 db = 1 mW).
or 30 dbm*.
The abbreviation db 6m is sometimes used to indicate a level in decibels with a
6 milliwatt reference level.
To convert from a reference level of 1 mW to 6 mW, add -
7.78 db.
To convert from a reference level of 1 mW to 10 mW, add -
10.00 db.
To convert from a reference level of 1 mW to 12.5 mW, add
To convert from a reference level of 6 mW to 1 mW, add + 7.78 db.
-
10.97 db.
To convert from a reference level of 10 mW to 1 mW, add + 10.00 db.
To convert from a reference level of 12.5 mW to 1 mW, add + 10.97 db.
With any reference level, a power with a positive sign in front of the decibel value
indicates that this is greater than the reference power, and is spoken of as so many
" decibels up." A negative sign indicates less power than the reference power, and
is spoken of as so many " decibels down." 0 db indicates that the power is equal to
the reference power.
A statement of power expressed in decibels is meaningless unless the reference
level is quoted.
References 1, 5, 8, 9, 10.
(B) Voltage
A reference level of 1 volt has been standardized in connection with high impedance
microphones. The abbreviation dbv has been standardized (Ref. 38) to indicate
decibels referred to 1 volt.
(iv) Microphone output expressed in decibels
The output of a microphone may be expressed either in terms of voltage or power.
(A) In terms of output voltage
The response of a microphone at a given frequency may be stated in decibels with
respect to a reference level 0 db = 1 volt (open -circuit) with a sound pressure of
1 dyne per square centimetre (Ref. 36). The abbreviation dbv is used to indicate a
voltage expressed in decibels, with reference level 1 volt (Ref. 38).
For example, the output of a microphone may be stated as - 74 dbv with a sound
pressure of 1 dyne per square centimetre. This is the open -circuit voltage developed
without any loading such as would be provided by the input resistance of the amplifier.
Table 1 [Section 1(x)] may be used to determine the corresponding open-circuit
voltage, which for the example above is approximately 0.0002 volt r.m.s. (Column
1).
If the input resistance of the amplifier is equal to the internal impedance of the micro-
phone (here assumed to be resistive as the worst possible case) the voltage across
input terminals will be only half the generated voltage, giving a loss of 6 db or the
effective input voltage of - 80 dbv.
an
FIG. 19.1
In the general case, the input voltage to the amplifier will be (Fig 19.1) :
Et = ER, /(R + R1)
where E = open -circuit voltage developed by the microphone
Rt = input resistance of amplifier
and R = internal resistance of microphone (here assumed purely resistive).
(B) In terms of output power
Alternatively the output of a microphone may be given in terms of output power
for a stated sound pressure.
dbm indicates a power expressed in decibels with a reference level 1 mW.
19.1 (iv) MICROPHONE OUTPUT EXPRESSED IN DECIBELS 809
Microphone system ratings are most commonly used in a complete sound system
sound, microphone, amplifier, loudspeaker, sound-see (ix) below.
-
See Chapter 18 Sect. 1 for the relationships between various forms of microphone
ratings.
(v) Pickup output expressed in decibels
Although many pickups are rated on the basis of output voltage, some are rated on
a power basis with respect to a specified reference level. The procedure is the same
as for microphones.
Reference 9.
(vi) Amplifier gain expressed in decibels
Much confusion has been caused by the incorrect or careless use of decibels to
indicate the gain of an amplifier -
indicate the gain of a voltage amplifier. Decibels may be used in various ways to
Bridging gain is the ratio, expressed in db, of the power delivered to the bridging
amplifier load to the power in the load across which the input of the amplifier is
bridged (Ref. 33).
ohms,
The commonest case is that of the input to an amplifier having a load of 600 output
with the input terminals of the bridging amplifier connected across it. The
load of the bridging amplifier is most commonly 600 ohms.
(vii) Combined microphone and amplifier gain expressed in decibels
(A) When a microphone, rated in terms of voltage, is connected to a voltage
amplifier which is rated in decibels of voltage gain (dbvg), the output may be
calculated as under :
Output in dbv* = microphone rating in dbv* + coupling factor + amplifier gain
(18)
in decibels of voltage gain (dbvg).
factor = 20 log /(R + R,)] (Fig. 19.1). (19)
The coupling [R1
Typical values of the coupling factor are tabulated below :
Ri/(R + R1) = 0.5 0.56 0.63 0.71 0.79 0.89
Coupling factor = -6 -5 -4 -3 -2 -1 db
Example : Microphone -54 dbv*
Amplifier + 80 dbvg
R1 /(R + R1) = 0.5
Output = -54 -- 6 + 80 = ± 20 dbv* = 10 volts.
It should be noted that the calculated output applies for the rated sound pressure,
voltage
for example 1 dyne per square centimetre. At other sound pressures the
will be proportional to the sound pressure.
(B) When a microphone, rated in terms of effective output level,
is con-
nected to an amplifier having its gain expressed in terms of power in accord-
ance with R.M.A. Standard SE-101 -A :
The amplifier power gain is measured effectively with a constant input
voltage in
as in
series with a resistance equal to the rated source impedance of the amplifier
output
Fig. 19.3. Under these conditions no correction factor is necessary and the output
from the amplifier in dbm is equal to the sum of the microphone effective
will only be
level in dbm and the amplifier power gain in decibels. This output level
attained when the pressure at the microphone is equal to the rated pressure -e.g.
10 dynes /sq. cm.
FIG. 19.3 FIG. 19.4
an amplifier with a power gain of 40 db will give an output level -153 +40 +94 =
-19 dbm. The microphone amplifier and sound pressure in this example are the
same as for (B) above.
(viii) Loudspeaker output expressed in decibels, in terms of acousti-
cal pressure
In accordance with the American R.M.A. Standard SE-103 (Ref. 29) the loud-
speaker pressure rating is the difference between the axial sound pressure level (re-
ferred to a distance of 30 feet) and the available input power level, and is expressed in
db.
It is expressed by the following forms (equations 20, 21, 22, 23) :
Gsp = 10 logro [(Ps /Po)a /(WAS /Wo)] (20)
GSP = 44 + 20 logro ps
GsP = 44 + 20 logro ps
--
10 loam WAS
20 log10 Ec + 10 log R5R
(21)
FIG. 19.5
Fig. 19.5. Loudspeaker
30 ft testing conditions (R.M.A.
Standard SE -103).
The sound pressure (pd) at any distance d feet may be used to compute the pressure
p, at 30 feet by the relation
Pe = (dl30)P d (24)
See also Ref. 34.
Example : If the loudspeaker pressure rating (G SP) is 46 db, what is the axial
sound pressure level at 30 feet, with an available power input of 10 watts, using a
standard test signal ?
GSP = 46 db WAS = 10 watts = 40 dbm
From equation (21),
20 log10 ps = 46 - 44 + 10 log 10 = 12 db.
Therefore ps = 4 dynes /cm' (i.e. 86 db above Po).
The same result may be derived more directly from equation (23), pressure =
46 + 40 = 86 db above Po.
(ix) Sound system rating
The total gain of a system from sound, through microphone, amplifier and loud-
speaker to sound again may be calculated by adding the system ratings of the several
sections and coupling factors (if any).
(A) Using the American R.M.A. system ratings for microphone, amplifier
and loudspeaker we may put
SR = Gm + G + Gsp (25)
where SR = sound system rating (gain in db)
G R.M.A. microphone system rating (equations 10 and 11)
G = amplifier power gain (equation 15)
and Gsp = loudspeaker pressure rating in db (equations 20, 21 and 22).
'See Chapter 20 Sect. 6(a)B for definitions of RSR and RSC
19.1 (ix) SOUND SYSTEM RATING 813
A system rating of 0 db indicates that the sound pressure 30 feet from the loud-
speaker is the same as that at the microphone. Similarly a system gain of x db in-
/
dicates that the sound pressure 30 feet from the loudspeaker is x db greater than that
at the microphone.
(B) Proposed method of rating microphones and loudspeakers for systems
use by Romanow and Hawley (Ref. 11)
This proposed method has not been adopted generally in the precise form ex-
pounded, but the article gives a most valuable analysis of the whole subject of system
gain. This method is also described in Ref. 34.
7.
(a) Tables and charts of decibel relationships
Table 1 Decibels expressed as power and voltage or current ratios
:
Note that the Power Ratio columns give power values in milliwatts when the re-
i
ference level is 1 mW. The Power Ratio columns also give power values in milli -
watts when the centre column represents dbm *.
The Voltage Ratio columns also give values in volts when the centre column re-
presents dbv.t
FIG. 19. 6
lo
=...
Id 1111111111111111111111111E4111
:.....0
0
á
wa 111111111111111111
P'.
.111111111
,.r
Jy.-.....i...
'iiiiiii
1 ,000 1111111111111
./
.11 oEC.
11111111111P_
,0:a:B
loo
10
_ gßi
\
1111111111E111111
11111.11
==8:_=
Zing
IiAMMO=
oi..
111r.1. 11 MIII11111
,rdEM
mum mom
°AIIIIIIII11111111
s
..11111111
!.111111111Ì1Ì111u11111111111
10 +10 +20 +30 +40 +50 +60
DECIBELS
Fig. 19.6. Gain ratio plotted against decibel gain. This chart may also be used for
attenuation by inverting the ratio and making the decibels negative.
Interpolation : If it is required to find the power ratio corresponding to 22.5 db,
or any other value which is not included in the table, the following procedure may be
adopted :-
1. Take the next lowest multiple of 20 db (in this case 20 db), and note the corres-
ponding power ratio (in this case 100).
with 0 db = 1 mW.
'dbm is unit of power expressed in decibels with
tdbv is unit of voltage expressed in decibels 0 db -
1 volt.
814 (x) TABLES AND CHARTS OF DECIBEL RELATIONSHIPS 19.1
2. Take the difference between the specified level and the multiple of 20 db (in
this case 22.5 -
20 = 2.5 db) and note the corresponding power ratio (in this
case 1.778).
3. Multiply the two power ratios so determined (in this case 100 x 1.778 = 177.8).
db db db db
Ratio (Power (Voltage* Ratio (Power (Voltage*
Ratio) Ratio) Ratio) Ratio)
1.0 0 0 5.7 7.559 15.117
1.1 0.414 0.828 5.8 7.634 15.269
1.2 0.792 1.584 5.9 7.709 15.417
1.3 1.139 2.279 6.0 7.782 15.563
1.4 1.461 2.923 6.1 7.853 15.707
1.5 1.761 3.522 6.2 7.924 15.848
1.6 2.041 4.082 6.3 7.993 15.987
1.7 2.304 4.609 6.4 8.062 16.124
1.8 2.553 5.105 6.5 8.129 16.258
1.9 2.788 5.575 6.6 8.195 16.391
2.0 3.010 6.021 6.7 8.261 16.521
2.1 3.222 6.444 6.8 8.325 16.650
2.2 3.424 6.848 6.9 8.388 16.777
2.3 3.617 7.235 7.0 8.451 16.902
2.4 3.802 7.604 7.1 8.513 17.025
2.5 3.979 7.959 7.2 8.573 17.147
2.6 4.150 8.299 7.3 8.633 17.266
2.7 4.314 8.627 7.4 8.692 17.385
2.8 4.472 8.943 7.5 8.751 17.501
2.9 4.624 9.248 7.6 8.808 17.616
3.0 4.771 9.542 7.7 8.865 17.730
3.1 4.914 9.827 7.8 8.921 17.842
3.2 5.051 10.103 7.9 8.976 17.953
3.3 5.185 10.370 8.0 9.031 18.062
3.4 5.315 10.630 8.1 9.085 18.170
3.5 5.441 10.881 8.2 9.138 18.276
3.6 5.563 11.126 8.3 9.191 18.382
3.7 5.682 11.364 8.4 9.243 18.486
3.8 5.798 11.596 8.5 9.294 18.588
3.9 5.911 11.821 8.6 9.345 18.690
4.0 6.021 12.041 8.7 9.395 18.790
4.1 6.128 12.256 8.8 9.445 18.890
4.2 6.232 12.465 8.9 9.494 18.988
4.3 6.335 12.669 9.0 9.542 19.085
4.4 6.435 12.869 9.1 9.590 19.181
4.5 6.532 13.064 9.2 9.638 19.276
4.6 6.628 13.255 9.3 9.685 19.370
4.7 6.721 13.442 9.4 9.731 19.463
4.8 6.812 13.625 9.5 9.777 19.554
4.9 6.902 13.804 9.6 9.823 19.645
5.0 6.990 13.979 9.7 9.868 19.735
5.1 7.076 14.151 9.8 9.912 19.825
5.2 7.160 14.320 9.9 9.956 19.913
5.3 7.243 14.486 10.0 10.000 20.000
5.4 7.324 14.648 100 20 40
5.5 7.404 14.807 1000 30 60
5.6 7.482 14.964 10000 40 80
Ur Current Katio.
and add the decibels of each. For example -
To find the decibels corresponding to ratios above 10, break the ratio into two factors
Voltage ratio = 400 = 4 x 100. Decibels = 12.041 + 40 = 52.041.
818 (x) TABLES AND CHARTS OF DECIBEL RELATIONSHIPS 19.1
db down Level db up
Volts Watts db Volts Watts
1.73 6.00 x 10-' -0-f- 1.73 .006 00
1.54 4.77 x 10-8 1 1.94 .007 55
1.38 3.79 x 10-8 2 2.18 .009 51
1.23 3.01 x 10-' 3 2.45 .012 0
1.09 2.39 x 10-8 4 2.75 .015 1
.974 1.90 x 10-8 5 3.08 .019 0
.868 1.51 x 10-8 6 3.46 .023 9
.774 1.20 X 10-8 7 3.88 .030 1
.690 9.51 x 10-' 8 4.35 .037 9
.615 7.55 x 10-' 9 4.88 .047 7
.548 6.00 x 10-' 10 5.48 .060 0
.488 4.77 x 10-' 11 6.15 .075 5
.435 3.79 x 10-' 12 6.90 .095 1
.388 3.01 x 10-' 13 7.74 .120
.346 2.39 x 10-' 14 8.68 .151
.308 1.90 x 10-' 15 9.74 .190
.275 1.51 x 10-' 16 10.93 .239
.245 1.20 x 10-4 17 12.26 .301
.218 9.51 x 10-5 18 13.76 .379
.194 7.55 x 10-5 19 15.44 .477
.173 6.00 x 10-5 20 17.32 .600
.097 4 1.90 x 10-5 25 30.8 1.90
.054 8 6.00 x 10-5 30 54.8 6.0
.030 8 1.90 X 10-8 35 97.4 19.0
.017 3 6.00 x 10-' 40 173 60.0
.009 74 1.90 x 10-' 45 308 190
.005 48 6.00 x 10-8 50 548 600
.003 08 1.90 x 10-8 55 974 1 900
.001 73 6.00 x 10-8 60 1 730 6 000
.000 974 1.90 X 10-8 65 3 080 19 000
.000 548 6.00 x 10-10 70 5 480 60 000
.000 308 1.90 x 10-10 75 9 740 190 000
.000 173 6.00 x 10-11 80 17 300 600 000
db down Level db up
db down Level db up
Volts Milliwatts dbm Volts Milliwatts
0.774 6 1.000 - 0+ 0.774 6 1.000
0.690 5 .794 3 1 0.869 1 1.259
0.616 7 .631 0 2 0.975 2 1.585
0.548 4 .501 2 3 1.094 1.995
0.488 7 .398 1 4 1.228 2.512
0.435 6 .316 2 5 1.377 3.162
0.388 2 .251 2 6 1.546 3.981
0.346 0 .199 5 7 1.734 5.012
0.308 4 .158 5 8 1.946 6.310
0.274 8 .125 9 9 2.183 7.943
0.244 9 .100 0 10 2.449 10.000
0.218 3 .079 43 11 2.748 12.59
0.194 6 .063 10 12 3.084 15.85
0.173 4 .050 12 13 3.460 19.95
0.154 6 .039 81 14 3.882 25.12
0.137 7 .031 62 15 4.356 31.62
0.122 8 .025 12 16 4.887 39.81
0.109 4 .019 95 17 5.484 50.12
0.097 52 .015 85 18 6.153 63.10
0.086 91 .012 59 19 6.905 79.43
0.077 46 .010 00 20 7.746 100.00
0.043 56 .003 16 25 13.77 316.2
0.024 49 .001 00 30 24.49 1.000W
0.013 77 .000 316 35 43.56 3.162W
0.007 746 .000 100 40 77.46 10.00W
0.004 356 3.16 x 10 -5 45 137.7 31.62W
0.002 449 1.00 x 10 -5 50 244.9 100W
0.001 377 3.16 x 10 -8 55 435.6 316.2W
0.000 774 6 1.00 x 10 -8 60 774.6 1 000W
0.000 435 6 3.16 x 10 -' 65 1 377 3 162W
0.000 244 9 1.00 x 10 -' 70 2 449 10 000W
0.000 137 7 3.16 x 10 -8 75 4 356 31 620W
0.000 077 46 1.00 x 10 -8 80 7 746 100 000W
19.1 (x) TABLES AND CHARTS OF DECIBEL RELATIONSHIPS 821
milli-
Watts dbm volts milli- dbm volts micro- dbm volts
across watts across watts across
5000S2 500052 500052
r- 2. 5
required voltage ratio on the log -log scale
and read the decibels on the C scale. To
obtain good accuracy it is advisable to use the
section of the log -log scale between 3 and 100,
dealing with powers of 10 separately.
For power ratios, set cursor to 10 on the 2
log -log scale, set the C scale to 10 on the
cursor and then proceed as before.
6 8 12 14 18 20
5
-
16 L.
each of the odd numbers of db. Noticing
that 2 db down is equivalent to multiplying 20
by 0.8 we can assume that the error in
taking 1 db down as 0.9 is small, and the
table can then be written in full. Fig. 19.6A.
19.1 (xii) DECIBELS, SLIDE RULES AND MENTAL ARITHMETIC 823
1
true factor
1.00
1.12
1.12
- 1.25
1.26
1.4
1.41
1.6
1.59
1.8
1.78
2
2.00
2.3
2.24
2.5
2.51
2.9
2.82
3.2
3.16
db-
11
factor - 12 13 14 15 16 17 18 19 20
3.6
true factor
3.55
4
- 4.5 5 5.8 6.4
Comparing the multiplying factors computed in this manner with the true factors
it will be seen that the greatest error (at 15 db) is just over 3% which is negligible for
the type of calculation intended to be performed by this method.
When working with " db down," the number of db can be subtracted from 20 and
the multiplying factor from the table can be divided by 10, e.g. 7 db (20-13) down is
equivalent to a multiplying factor of 0.45. A small amount of practice at such mental
conversions soon results in many of the factors being memorized (or else becoming
immediately obvious) without any conscious effort in this direction.
this is for use in measuring instruments etc. where the reading of the meter together
with the reading of an associated attenuator (where necessary) give the actual power
level. With " Scale B " the percentage scale is made more prominent being situated
above the vu scale -this is for use in studio controls etc. where the operator is non-
technical and is not interested in actual levels.
The sensitivity shall not depart from that at 1000 c/s by more than 0.2 db between
35 and 10 000 c/s at an input level of 0 vu, nor more than 0.5 db between 25 and
16 000 c/s (Ref. 19).
The instrument is calibrated by connecting it in shunt with a resistance of 600 ohms
through which is flowing 1 milliwatt of sine -wave power at 1000 c /s, when a reading
shall be 0 vu (or n vu when the calibrating power is n db above 1 milliwatt).
If the instrument is connected across any impedance other than 600 ohms, the
volume indicated must be corrected by adding 10 logic, (600 /Z), where Z is the actual
impedance in ohms.
The total impedance of the volume indicator is usually about 7500 ohms, of which
about 3600 ohms is external to the instrument.
The Volume Indicator is intended to be read as deviations from the reference volume
(0 vu), after making allowance for the sensitivity control (attenuator) which is also
calibrated in vu. The reading is determined by the greatest deflections occurring
in a period of about a minute for programme waves, or a shorter period (e.g. 5 to 10
seconds) for message telephone speech waves, excluding not more than one or two
occasional deflections of unusual amplitude.
References 18, 19, 39.
(i) Nepers
Just as there are two systems of logarithms in general use, so there are two logarith-
mic units for the measurement of difference of power levels. The bel and the decibel
are based on the system of Common Logarithms (to the base 10).
The neper is based on the system of Naperian Logarithms (to the base e).
Nepers are not commonly used in English -speaking countries, but are used by some
European countries.
Equation (1) below should be compared with Section 1, equation (2) above :
N = I loge (Pa /P,) (1)
where N is the ratio of two powers expressed in nepers
P, = reference power
and P2 = power which is referred back to P1.
When the impedances relating to P, and P2 are the same,
N
= loge (E, /Ei) = loge (12/11) (2)
Equation (2) should be compared with equations (4) and (5) in Section 1 above.
826 (i) NEPERS 19.4
SECTION 5 : LOUDNESS
(i) Introduction to loudness (ii) The phon (iii) Loudness units.
Iii
PUONS)
100
80
nlls\:iÍ11I!
II_0,.1110ó17.1/%i1ÌÌN
II
M:í/ÍIMI
MIIIIIISMIE'/.2II01
\`ii \:,II
\\:!\!1111111
N.\\\ilM11III111\iiG111M
so
11.\CIIi1111I1
2 II I\Ils21!1%:ííÍIIIII
-"MOP'
20 00 500 1000 5000 10000
FREQUENCY IN CYCLES PER SECOND
FIG. 19. 7
Fig. 19.7. Contours of equal loudness level (0 db = 10-ís watt /cmz) -Ref. 28, after
Fletcher and Munson (Ref. 40).
100000
10 000
1000
100
10
10 20 30 40 50 60 70 60 90 100
Fig. 19.8. Relation between loudness and loudness level ; reference frequency 1000 c /s-
Ref. 28, after Fletcher and Munson (Ref. 41).
(i) Introduction
Sound includes wanted sound -music or speech -and also unwanted sound
noise. Noise may be measured acoustically, as for example by a sound level meter,
-
or electrically. Acoustical methods of measuring noise and other sounds are de-
scribed in (ii) below. Electrical noise may be measured either with or without a
weighting network. If a weighting network is used, there will be appreciable attenua-
tion of the lower frequencies, including hum frequencies. It is normally assumed in
good engineering practice that the hum components have negligible effect on the
noise reading ; if this is not so, they should be filtered out before the noise voltage is
applied to the measuring equipment.
828 (ii) THE SOUND LEVEL METER 19.0
Small store 52
Theatre (with audience) 42
Hotel 42
Apartment 42
House, large city 40
House, country 30
Average whisper 4 ft. 20
Quiet whisper 5 ft. 10
Rustle of leaves in gentle breeze 10
0
Threshold of hearing
SECTION 7 : REFERENCES
1. Wright, P. B. " Evolution of the db and the vu " Comm. (1) 24.4 (April 1944) 54 ; (2) 24.5 (May
1944) 44 ; erratum and additional data 24.6 (June 1944) 68.
2. Harrison, C. E. " Electrical and acoustical equivalents " Comm. 25.6 (June 1945) 44.
3. Rhita, N. " A decibel nomograph " Radio Craft (Sept. 1945) 769.
4. Hudson, P. K. " Calibration of decibel meters " Comm. 25.7 (July 1945) 58.
5. Scott, F. S. G. " Absolute bels " W.E. 23.272 (May 1946) 132.
6. Miedke, R. C. " Decibel conversion chart " Proc. I.R.E. 34.2 (Feb. 1946) 76W.
7. Haefner, S. J. " Amplifier gain formulas and measurements " Proc. I.R.E. 34.7 (July 1946) 500.
8. Perry, S. V. " The decibel scale" R.M.A. Tech. Bull. U.S.A. (Nov. 1940).
9. Rao, V. V. L. (book) " The Decibel Notation and its Applications to Radio Engineering and Acous-
tics " (Addison and Co. Ltd. Madras, India, 1944).
10. Fowler, N. B. " Measurements in communications " E.H. 66.2 (Feb. 1947) 135.
11. Romanow, F. F., and M. S. Hawley " Proposed method of rating microphones and loudspeakers
for systems use " Proc. I.R.E. 35.9 (Sept. 1947) 953.
12. Morgan R. L. " Noise measurements " Comm. 28.2 (Feb. 1948) 28 ; 28.7 (July 1948) 21.
13. Chinn, H. A. " dbm vs. vu " Audio Eng. 32.3 (March 1948) 28.
14. Shea, T. E. (book) " Transmission Networks and Wave Filters " (D. Van Nostrand Company Inc.,
New York, 1929).
15. Pender, H. and K. McIlwain " Electrical Engineers Handbook -Vol. 5, Electric Communication
and Electronics " (John Wiley and Sons Inc. New York ; Chapman and Hall Ltd., London, 3rd
edit. 1936).
16. Research Council, Academy of Motion Picture Arts and Sciences (book) "Motion Picture Sound
Engineering " (D. Van Nostrand Co. Inc. New York, 1938).
17. Affil, H. A., H. A. Chinn and R. M. Morris " A standard VI and reference level " Comm. 19.4
(April 1939) 10.
18. Chinn, H. A., D. K. Gannett and R. M. Morris " A new standard volume indicator and reference
level." Proc. I.R.E. 28.1 (Jan. 1940) 1.
19. " American recommended practice for volume measurements of electrical speech and program
waves " C16.5 -1942, American Standards Association.
20. McLachlan, N. W. (book) " Noise " (Oxford University Press, 1935).
21. Burrill, C. M. " An evaluation of radio-noise -meter performance in terms of listening experience "
Proc. I.R.E. 30.10 (Oct. 1942) 473.
22. " American standard acoustical terminology " Z24.1 -1942. American Standards Association.
23. " American Standards for sound level meters " Z24.3-1944, American Standards Association.
24. Olson, H. F. (book) " Elements of Acoustical Engineering " (D. Van Nostrand Co. Inc. New York,
2nd edit. 1947).
25. " Methods of Measuring radio noise, " Report of Joint Co-ordination Committee on Radio
Reception of the Edison Electric Institute, National Electrical Manufacturers Association
and Radio Manufacturers Association- R.M.A. Engineering Bulletin No. 32 (Feb. 1940).
26. Burrill, C. M. " Progress in the development of instruments for measuring radio noise," Proc. I.R.E.
29.8 (Aug. 1941) 433.
27. " The characteristics and performance of apparatus for the measurement of radio interference "
B.S.S. No. 727, British Standards Institution, London (1937).
28. " American standard for noise measurement " Z24.2 -1942, American Standards Association.
29. American R.M.A. Standard SE -103 " Speakers for sound equipment " (April 1949).
30. " Reference Data for Engineers " (book) (Federal Telephone and Radio Corporation, New York ;
3rd ed. 1949).
31. Black, W. L., and H. H. Scott, " Audio frequency measurements " Proc. I.R.E. 37.10 (Oct. 1949)
1108.
32. American R.M.A. Standard SE-105 " Microphones for sound equipment " (Aug. 1949).
33. American R.M.A. Standard TR -105 -B " Audio facilities for radio broadcasting systems " (Nov.
1949).
34. Beranek, L. L. (book) " Acoustic Measurements " (John Wiley and Sons Inc. New York ; Chapman
and Hall Ltd. , London, 1949).
35. American R.M.A. Standard SE-101 -A " Amplifiers for sound equipment " (July 1949).
36. American R.M.A. Standards Proposal 198 " Microphones " (adopted Nov. 1938).
37. Di Mattia, A. L., and L. R. Jones " Adding decibel- expressed quantities " Audio Eng. 35.7 (July
1951) 15.
38. I.R.E. ' Standards on abbreviations of radio- electronic terms 1951 " Standard 51 I.R.E. 21 S1 ;
Proc. I.R.E. 39.4 (April 1951) 397.
39. Chinn, H. A. " The measurement of audio volume " Audio Eng. 35.9 (Sept. 1951) 26.
40. Fletcher, H., and W. A. Munson " Loudness, its definition, measurement and calculation " J. Acous.
Soc. Am. 5.2 (Oct. 1933) 82.
41. Fletcher, H., and W. A. Munson " Loudness and masking " J. Acous. Soc. Am. 9.1 (July 1937) 1.
42. King, A. J., R. W. Guelke, C. R. Maguire and R. A. Scott " An objective noise-meter reading in
phons for sustained noises, with special reference to engineering plant " Jour. I.E.E. 88, Part II
(1941) 163.
43. Chinn, H.A. " The measurement of audio volume " Audio Eng. 35.10 (Oct. 1951) 24.
CHAPTER 20.
LOUDSPEAKERS
BY F. LANGFORD- SMITH, B.Sc., B.E.
Section Page
1. Introduction ... ... ... ... 831
2. Characteristics of moving -coil (electro- dynamic) loudspeakers 835
3. Baffles and enclosures for direct-radiator loudspeakers 842
4. Horn loudspeakers ... 851
5. Dual and triple system loudspeakers 860
6. Loudspeakers in operation 861
7. Distortion in loudspeakers 868
8. Summary of acoustical data 871
9. Standards for loudspeakers 874
10. References 876
SECTION 1 : INTRODUCTION
(i) Types of loudspeakers (ii) Direct -radiator loudspeakers (iii) Horn loudspeakers
(iv) Headphones (v) Loudspeaker characteristics (vi) Amplitude of cone movement
(vii) Good qualities of loudspeakers (viii) Loudspeaker grilles.
831
832 (ii) DIRECT RADIATOR LOUDSPEAKERS 20.1
relatively stiff springs to return the armature to the position
of equilibrium, so that the
bass resonance usually occurs at a frequency above 100
c/s.
Refs. 13, 14, 15.
(C) The inductor dynamic type differs from the older electromagnetic
in that the two iron armatures are balanced so that they type
lie at
spring tension. The natural resonance frequence may be aboutrest without a strong
70 c /s, and the am-
plitude of movement is about the same as that of a moving -coil type.
Ref. 15.
(D) The piezo-electric (" crystal ") type generally uses
bimorph elements. It takes two forms, in which the elements Rochelle salt for the
with the voltage applied between the two electrodes. The crystal either has
bend or twist
capacitive impedance, the capacitance being of the order of 0.02 a predominately
The impedance varies considerably with frequency in both scalar F.
angle, so that it is difficult to provide a correct load impedance value and phase
for the output valve.
This is not very important with a triode or if another loudspeaker
is connected in
parallel.
The crystal loudspeaker naturally responds to the higher frequencies,
times used as a " tweeter " in a dual or triple speaker system (Sect. and is some-
5). In this case
the response may be maintained roughly constant over the desired
by a small inductance in series with the primary of the transformer. band of frequencies
speakers are not normally used alone. Crystal loud-
Refs. 13, 14, 24, 60.
(E) The condenser type of loudspeaker usually takes the form of
one large solid
electrode, and a thin movable electrode which is mounted in such
a way that it can
vibrate (usually in sections) without touching the solid electrode.
Either the solid or the movable electrode may be corrugated ; the movable electrode
may be of insulating material with a metal foil coating. An alternative
electrodes flexible (Ref. 22).
form has both
The sensitivity is dependent upon the total area ; with 300 to 500 square
sensitivity may approximate that of an electro-dynamic type. inches the
The Kyle speaker (Ref. 58) has a capacitance of 0.004 µF for an area of 96
inches. The impedance is predominately capacitive. A high polarizing square
required for high efficiency and good performance-500 to 600 volts is usual. voltage is
Condenser loudspeakers are little used at the present time.
Refs. 13, 14, 15, 22, 58.
(F) Throttled air-flow loudspeaker -this consists of a mechanical
ated by an electrical system, which controls a steady air stream, the airvalve, actu-
made proportional to the a -f input current. It is normally limited to very flow being
large sizes.
Ref. 14.
(iii) Horn loudspeakers
Any type of driving unit actuating a cone or diaphragm may be given greater acous-
tical loading by means of a horn, thereby improving the efficiency and power
output.
The horn ceases to have any beneficial effect below a frequency whose value depends
on the law of expansion and the size of the horn. Horn loudspeakers have
considerably
greater efficiency than direct radiators. For further information see Sect. 4.
(iv) Headphones
Headphones (telephone receivers) make use of a diaphragm which is effectively
sealed to the ear by means of a cap with a central opening. The pressure of
the small
quantity of air enclosed between the diaphragm and the ear drum varies in accordance
with the displacement of the diaphragm.
The driving mechanisms of headphones resemble those for loudspeakers.
(A) Magnetic diaphragm : The a -f force caused by the a -f current in the electro-
magnet operates directly upon a steel diaphragm. The bipolar type is the most
popular of all headphones at the present time. A permanent magnet supplies the
steady flux. At frequencies above the second diaphragm resonance the response
falls off rapidly. Refs; 13, 14, 59, 62.
20.1 (iv) HEADPHONES 833
(B) Moving armature type : The principles are the same as for loudspeakers.
Ref. 14.
(C) Moving -coil type : This follows the same principles as the loudspeaker,
and with careful design is capable of greater fidelity and wider frequency range than
the magnetic types. One model has a response from 10 to 9000 c/s ± 2 db except
for a dip of 6 db at 4000 c /s. (Ref. 63). Refs. 13, 14, 63.
(D) Crystal type : The impedance of one model is 80 000 ohms (predominately
capacitive) at 10 000 c /s. A high resistance may be connected in series to raise the
low frequency response relatively to the high frequency response. Refs. 13, 64.
(E) Ribbon type : The principle is the same as for a ribbon microphone. It is
only used for high -fidelity reproduction. Refs. 13, 14, 65.
(F) Inductor type : This has been developed to deliver practically constant
sound pressure to the ear cavity from 50 to 7000 c/s. It has a V shaped diaphragm
driven by a straight conductor located in the bottom of the V. Refs. 13, 14, 66, 67.
Correction circuits for magnetic diaphragm headphones
A typical magnetic diaphragm unit has a pronounced peak slightly below 1000 c /s.
A circuit for attenuating this peak and thereby giving reasonably uniform response
is Fig. 20.1 (Ref. 25).
Directional characteristics
The usual frequency response characteristic is measured with the microphone
on the axis of the loudspeaker, this being the position of maximum high frequency
response. There is a serious loss of higher frequencies (4000 c/s and above) even
with the microphone 30° off the axis.
See Sect. 2(vii) for further information on moving-coil types, and Sect. 4(vii) for
horn types.
Refs. 13, 14, 15, 26 (No. 1), 61, 70, 133.
Non -linear distortion : see Sect. 7(i) and (v).
Efficiency versus frequency characteristic : see Sect. 2(vi).
Transient response : see Sect. 7(iii).
Matching see Chapter 21.
:
DYNAMIC) LOUDSPEAKERS
(i) Ideal " piston" cone in an infinite flat baffle (ii) Practical cones (iii) Special
constructions for wide frequency range (iv) Impedance and phase angle (v) Fre-
quency response (vi) Efficiency (vii) Directional characteristics (viii) Field magnet
(ix) Hum bucking coil (x) Damping.
low frequencies the whole assembly moves together, but at high frequencies the larger
part of the voice coil remains stationary, and the smaller part drives the cone. The
corrugations in the cone are designed to decrease the effective cone area at the higher
frequencies.
Refs. 13A, 14, 57.
(D) Double coil, double cone : This consists of a light coil coupled to a small
cone, the light coil being connected by a compliance to a heavy coil which is firmly
fixed to a large cone. The heavy voice coil is shunted by a capacitance to by -pass
the higher frequencies. At low frequencies both voice coils are operative, and the
whole assembly moves together, but at high frequencies the small cone is driven by
the light coil, and the large cone and heavy coil remain stationary. This is really
equivalent to two separate loudspeakers.
Refs. 13A, 71.
The impedance at 10 000 c/s is always greater than that at 400 c /s, the ratio varying
between 1.1 and 10 times. A high ratio is most undesirable with pentode operation,
with or without feedback, but is of no great consequence with triodes. The im-
pedance at 10 000 c/s is increased by any leakage inductance in the transformer.
The equivalent electrical circuit which gives approximately the same imped-
ance characteristic up to 400 c/s is shown in Fig. 20.3 (Refs. 15, 72). If adjust-
ments are made in the values of Lo, Ro, R, and C1, the circuit may be extended to
higher frequencies. The bass resonant frequency is the parallel resonance of Li and
C1.
838 (iv) IMPEDANCE AND PHASE ANGLE 20.2
The impedance so far discussed is called the " free " impedance ; there is also the
" blocked " impedance which is measured with the voice coil prevented from moving.
The " motional " impedance is found by a vector subtraction of the " blocked "
from the " free " impedance. The difference between the free and blocked im-
pedances is small except in the vicinity of the bass resonant frequency, where the
motional impedance becomes the major portion of the free impedance (Refs 13, 41).
Refs. 13, 13A, 14, 15, 26 (No. 2), 36, 37, 41, 47, 70.
FIG. 20.4
9d
Phase Angle : The phase angle versus
frequency characteristic of a typical mov- 6óm
ing-coil loudspeaker is given in Fig. 20.4.
The impedance is resistive at two fre- s8
quencies only, being capacitive between the
two points and inductive at lower and o°
higher frequencies. The resistive and re-
active components are shown in Fig. 20.5. 1o°
Refs. 13, 13A, 37, 41, 61.
60
6
W
0
N
W
Q
4
á
00
L'
I
J t
8
8 8
n O
ui 6`
FREQUENCY C/S FREQUENCY C/S
Fig. 20.5. (A) Resistive and (B) Reactive components of the impedance of a typical
loudspeaker. Ref. 37.
20.2 (vi) EFFICIENCY 839
(vi) Efficiency
The electro- acoustical efficiency of a loudspeaker
is the ratio of the acoustical output power to the
electrical input power. The efficiency of moving -
coil loud- speakers on flat baffles varies from about
2% to 10% at 400 c /s, depending on the design and
on the flux density, 3% being a typical average.
Maximum efficiency occurs at the bass resonant fre-
quency if the baffle is sufficiently large and the
amplifier output resistance is fairly high, as
with a pentode without feedback, but if the output
resistance is low, the efficiency may be less at this
frequency than at some higher frequency. The
efficiency /frequency characteristic at low frequencies
is affected by the type of baffle -see Sect. 3(iiil and
(iv).
In the electrical equivalent circuit of Fig.
20.7, Lo and Ro are the same as in the circuit of
Fig. 20.3,
Lo = blocked inductance of voice coil
Ra = blocked (d.c.) resistance of voice coil
R, = equivalent radiation resistance
Fig. 20.6. Cross sectional
and R, = equivalent frictional and eddy current diagrammatic view of a mov-
loss resistance. ing-coil loudspeaker.
The maximum possible electro- acoustical efficiency is given by
71max = R, /(Ro + R, + R,) (2)
and the electro-mechanical efficiency is given by
v en = (R2 + R,) /(R° + R2 + R,) (3)
The electro- acoustical efficiency is always less than the electro-mechanical efficiency
owing to the losses represented by R,. It is possible to measure the electro-mechanical
efficiency fairly readily because (R2 + R,) is the motional resistance, and
(R2 + R,) = free resistance - Ra (4)
but the accuracy is very poor except at frequencies in the region of the bass resonant
frequency.
FIG. 20.7
Lo R0
Fig. 20.7. Equivalent circuit for power
considerations.
tional at high frequencies, but the best of the expensive designs are capable of 120°
radiation up to 15 000 c/s (Refs. 36, 61).
General references 13, 14, 15, 54 (Part 2), 70.
:
From Fig. 20.8 it will be seen that critical damping is only possible with a positive
value of RG when Rd is not too high, that is to say with loudspeakers having fairly
high efficiencies. The damping of any loudspeaker may be increased by increasing
the flux density.
There is no advantage whatever in using greater than critical damping.
With enclosed cabinet loudspeakers as a result of the effect on frequency response,
the value of Q should not fall below 1, and the most desirable all-round condition
appears to be with Q from 1.0 to 1.5 -see Sect. 3(iii).
Alternatively, if the impedance is carefully measured and the impedance curve
plotted over the region of the bass resonant frequency, the frequencies of the two
half-power points may be noted, and the Q of the loudspeaker may be computed from
the relation
Q (6)
\2df/ \Z,,, + RG/
where fo = bass resonant frequency
24f = band width at half power point (3 db below maximum impedance)
R. = voice coil resistance
Z,,, = maximum value of the impedance (i.e. at fo)
and RG = effective plate resistance, referred to the secondary.
The measurements of frequency during this test are somewhat critical. The article
by Preisman (Ref. 146) gives further details, although it does not derive eqn. (6) in
this form.
The majority of direct radiator loudspeakers
have values of Q between 8 and 18 with a
high valve plate resistance, although this may
be decreased very considerably with a low
IOR,
LOUD effective plate resistance. Tests have been
SPEAKER carried out to demonstrate the effect of a low
plate resistance on the damping of a loud-
speaker at the bass resonant frequency (Refs
Voltage Proporcional 13A Fig. 6.30, 126, 132).
to Velocity (e /ll)
While there is no doubt that it is desirable
R/10 Scaled to achieve a near approach to critical damping,
Replico of
Loudspeaker
R the results are very much masked by the
,o L/10 Impedance effects of the listening room which behaves as
b a resonant structure. In one case a room
showed 8 resonances below 100 c /s, with Q
FIG. 20.9 values of the order of 12 to 15, and all above
Fig. 20.9. Bridge circuit to give a 8 (Ref. 155).
voltage proportional to the velocity of There are some loudspeakers in which
the cone, to be fed back to the input. the efficiency is too low to permit sufficient
(Ref. 184). damping by reduction of the effective plate
resistance. In these cases improvement
in the electrical damping is possible by the use of positive current feedback combined
with negative voltage feedback which can make the effective plate resistance zero or
negative (Refs. 157, 158, 200, 201).
Alternatively, the e.m.f. generated by the movement of the cone may be fed back
degeneratively to the input of the amplifier. One method of achieving this result is
to arrange a special bridge circuit (Fig. 20.9) to provide the voltage to be fed back.
The amplifier must have good characteristics with very low phase shift. This has
been used satisfactorily over a frequency range from 10 to 1000 c/s, using a standard
12 inch loudspeaker, whose coil velocity was made proportional to input amplitude
and completely independent of frequency (Refs. 184, 186, 148B, 191, 198).
842 (x) DAMPING 20.2
Another method is to wind a separate feedback coil of very fine wire over the
existing voice coil in a conventional loudspeaker. The voltage induced in this coil
by the motion of the voice coil is a pure motional voltage at most frequencies. This
voltage may be used as a feedback voltage to increase the damping and to reduce
distortion arising from non-linearity of the cone suspension and from fringing of the
magnetic field.
At very high audio frequencies the mutual inductance between the driving voice
coil and the feedback coil produces in the feedback voltage a component which is
dependent on the induction between the coils rather than on the motion alone. This
difficulty is overcome by incorporating additional mutual inductance equal in value
but of opposite sign, between the voice coil and feedback circuits at a point external
to the magnetic field (Ref. 148).
baffle and a 10 in. cone, the centre of the cone may be 18 inches from each of two
adjacent sides.
The size of a square baffle, with the loudspeaker mounted off -centre, for desired
bass response is given approximately by the following table for a 10 inch diameter
cone (Ref. 13A) :
Minimum frequency 300 170 80 c/s
Side of square 2 4 8 feet
Below this minimum frequency, and below the loudspeaker bass resonant fre-
quency, the response falls off at the rate of 18 db /octave.
The best flat baffle is provided by mounting the loudspeaker in a hole in the wall
of a room, with the rear of the cone radiating into another room.
In a very large flat baffle, a loudspeaker normally has a peak in the response charac-
teristic at, or somewhat above, the bass resonant frequency ; below the peak frequency
the attenuation is at the rate of 12 db /octave. Refs. 13, 13A, 14, 15 (Part 2), 75, 147,
151, 177.
(ii) Open back cabinets
An open back cabinet, as commonly used with radio receivers of the console, table
or mantel type, has a resonance in the enclosure to the rear of the cone. This resonance
causes a peak in the response characteristic at a frequency which is mainly a function
of the cabinet although also influenced by the loudspeaker characteristics. For
example, a cabinet 2 feet x 2 feet x 8 inches depth gave peak response at 180 c/s
with a loudspeaker having a bass resonant frequency of 20 c /s, while the same cabinet
gave a peak response at 110 c/s with a loudspeaker having a bass resonant frequency
of 100 c/s. (Ref. 13A, Figs. 6.19B and 6.20B). The height of the peak is about
3 to 6 db for shallow cabinets or about 6 to 10 db for deep cabinets.
20.3 (ii) OPEN BACK CABINETS 843
Open back cabinets are undesirable for good fidelity. If unavoidable, they should
be as shallow as possible, with the minimum of acoustical obstruction, particularly
at the back of the cabinet. Open back cabinets should be placed at least 6 inches
\\E
out from the wall.
References to open back cabinets : 13A, 14, 147, 151, 188.
(iii) Enclosed cabinet loudspeakers
An enclosed loudspeaker is one which is totally enclosed so that there can be no
interference between the front and back of the cone. There is no critical value for
the volume of the enclosure but a large volume is desirable because it reduces the rise
25
MMMM
in resonant frequency above that on a large flat baffle. The increase in resonant
frequency for Jensen speakers is shown in Fig. 20.10 and may be taken as fairly
typical. There is practically no non -linear distortion caused by the suspension, since
this only contributes a small part of the total stiffness.
EMMEEEMMEEEE
ME
ME\MEMME
ti:.i MMEMEEEMMEE
Mi!i!i'iSMMM
!i!i!SOi!ìMI
\MMMM
oN n:ém:mwm
iì!ii!:t!i!iílì!i
`'!3:mvx::!ìw
M
MM\ONí:
EWi
`
\:,.!..\..
MM-.
._
:
Vt!::\EEEMME
DMiyV
W . .._
:.\\Ao!i!Ma!ii:i:iMi
MM_
. .
.+. ._e:e_.e:_._.
..\M\.____._
E\ .!::\E\!i'p .'f'YI--IVf'1'A'
s 6!i;i!
EMEie_ei....Mll,... .wr\\-.rs.s
:IA\ì;'!ì;éMM\N.2N!i!Z S!i!i
...-- ....w..a..
i1\ i!i!?i!ii!i!IMM!°C:
\WNSi!íi1iiii.\\aOUKií:ii
i!i i!ì!iOí!i!í!i
ì
MEMEMEEMMlliiMr6dxot*.A.
.
.si.lu
..-__.--.---.-.-
IIIIE
00 5 10 15 20 25 30 35
INCREASE IN RESONANT FREQUENCY
FIG. 20. 10
less than 4 of the wavelength at the lowest working frequency. The enclosure should
have rigid walls -timber at least 4 inch thick, and braced where necessary.
One limitation of the enclosed cabinet loudspeaker is the bass attenuation
which occurs with critical damping, although this only happens with high flux density
and low amplifier output resistance -see below under Equivalent Circuit.
Some loudspeakers have been designed specially for use with enclosed cabinets.
These usually have very low bass resonant frequencies, when measured on a flat
baffle, so that when loaded by the enclosure, the resonant frequency is still sufficiently
low.
The volume of the cabinet is given below for some examples :
Western Electric 755A 8 in. 70 c/s 2 cubic feet
756A 10 in. 65 c/s 24
754A 12 in. 60 c/s 3
Goodmans "Axiom 80" 94 in.* 3} ,>
R.C.A. (Ref. 84) peak at 80 c/s 1.5
Stromberg- Carlson 8 in. 1.7
12 in. 3.9
Equivalent circuit
Fig. 20.11 is a simplified electrical equivalent circuit of the acoustical system of the
loudspeaker (Ref. 135).
Lu represents cone mass + effect of radiation reactance.
Ru represents radiation resistance (which varies with frequency).
Cc represents acoustical capacitance of cabinet volume.
Cu represents equivalent capacitance of cone suspension.
R S represents effect of electrical circuit of loudspeaker and driving amplifier
reflected into acoustical circuit. The mechanical resistance of the cone sus-
pension may be taken as being included with R s.
Es = constant voltage generator.
lu = alternating air current produced by cone, which is proportional to cone
velocity
B2
Now R S oC at low frequencies
+ Ro
RG
where B = flux density in gap,
RG = output impedance of amplifier referred to voice coil circuit,
and Ro = resistance of voice coil.
The acoustical response of the loudspeaker is proportional to luf.
The voltage Lu' 27rfLulucC lufcc acou-
E = the variation
f-i
c U Ru t
stical response. Therefore with fre-
quency of ELU, for any one value of Lu, gives the fre-
quency response of the loudspeaker. Over the
e
LU fre q uenc y ran ge for w h'is h t h o equivalent circuit is
valid, and the wavelength is large compared with the
-L
size of the cone, the loudspeaker can be reduced-so
FIG. 20.11
far as frequency response is concerned -to a half-
section high -pass filter working into open circuit.
Fig. 20.11. Equivalent elec- Small values of R s, resulting from low flux density or
trical circuit of acoustical ele- high amplifier output impedance, give a resonance
ments in an enclosed cabinet peak and bad transient response, while large values of
loudspeaker at low frequencies. R s, corresponding to high flux density and low ampli-
(Ref. 135). fier output impedance, can give a serious loss of bass.
Fig. 20.12 shows curves for three values of R
and hence for three values of Q, for a resonant frequency of 45 c/s. This indicatess
that values of Q less than about 1 result in serious attenuation of low frequencies.
*Resonance 17 c/s on flat baffle.
20.3 (iii) ENCLOSED CABINET LOUDSPEAKERS 845
+15
+ 5
OtOM
db o 0
owo
Q.4
aG E
5 P
SS Ó .G66
10
-15
30 40 50 60 80 roo 200 300 400 500
FREQUENCY (C /S)
FIG. 20.12
Fig. 20.12. Response of enclosed cabinet loudspeaker for various values of the effective
damping resistance R,. (Ref. 135).
References to enclosed cabinets : 13A, 29, 36, 80, 81, 84, 116, 116A, 135, 147,
151, 166, 168, 188.
(iv) Acoustical phase inverter (" vented baffle ")
Also known as a bass reflex bailie.
This has a vent* or duct in the front of the cabinet which augments the direct
radiation from the cone at low frequencies (Fig.
20.13). The box should be at least partially lined
19" 20 -' with sound absorbent material to absorb the
A
INSIDE FACES
4 HOLES 160IÁ.
- EOUISPACED ON
II Y.' P.C. DIA.
Fig. 20.13. Typical corner cabinet with vented baffle, designed for a Goodmans Axiom
12 inch loudspeaker with a bass resonant frequency of 55 c /s, which can be used with
a speaker resonant at 75 c/s by removing the tunnel. Cubic capacity about 8000
cu. ins. = 4.6 cu. ft. (Ref. 182).
*An alternative form uses several tuned resonatorsfrom the inside to the outside of the cabinet (Ref. 18).
846 (iv) ACOUSTICAL PHASE INVERTER 20.3
Mk,
MAW
WM
- 111111M
--
.- L----
m,om ,MMNSIM\M
IM- _ rue
pN. L/S
H.K
7n3 SQNL
MOMI NIMON
-_--\`
M
=
12
.0.1a5
IN.
S K
LA CC
=-
- __-n
M 20.u. 2111-
\\;\om
\\MIWIMMIMMINOMM-IMI
MWM\\MIIMMIMMMIMMMM
MO IMO MIMMIN IIMINIM \\MWIMIMM\_ WMMIM
\\O\SOMIWAMOM
\\\\o7miimm
; \\\\P!!E\'VVMi
ONMERMIN:
,
'so,áme`
\y.l,NE
=____
-``
`\\-
..... -
\SW.da=
==
%I" _--=-
--
--w
_..-.w,--_----
\\._---
=-
--d-
\\011.11\MOMPI.M.,
\\.\M\M!-----
g
i \ \\\
\\\
`\
\N
---I
\,32
----
-.-_L-__
----
--
--- \'=Rili
_-=____=
\Cmili¡----- c
=------_- MO\CM:MOMMEMM
IIIIiii::ii011 1111111111111111
111111111111111
d.... !!!HhIIuIIIIH
.4050
=
=___`=
_ __
-----=
0==b
-__--
=--==
-------
---- m
..--,,,.E-mmim_
II\M1M1\01111MMIIIMIN
.. ft
\rigliS\\OMPIM
\\!ip\\ammimu
:16EltE\1OWAMMO
-a=--.-__=
\an1\i:ii:R:an
ftMMM-
°áo._=
mm-m--m-immm-m
Emilimmm
r----1111
MMINIMMOMI 1 IS IN.
....
L/S
,,,,,,,,1----.1,
.
111111111111111
.. . . . . p
. . .. n ..
TUNNEL LENGTH INCHES
TUNNEL LENGTH -INCHES
FIG. 20.14
Fig. 20.14. Curves showing tunnel length for 10, 12, 15 and 18 inch Goodmans
speakers with bass resonant frequencies from 20 to 100 c/s. (Ref. loud-
182).
20.3 (iv) ACOUSTICAL PHASE INVERTER 847
result of wall vibration under the extremely arduous conditions of operation. Other
materials which have been used satisfactorily include brick and concrete.
Vented baffles are normally designed so that the loudspeaker bass resonant fre-
quency is matched to the acoustical resonant frequency of the cabinet ; unmatched
combinations are, however, also used.
(A) Matched vented baffles
The merits of the vented baffle, when correctly applied, include
(1) improved bass response,
increased power handling capacity,
(2) .
1.5 1 and any subsidiary peaks caused by standing waves should be removed by the
:
addition of damping material (as for the enclosed cabinet above ; see also Fig. 20.17).
Tests for acoustical output should not be made until the cabinet dimensions and
damping have been adjusted to give the correct impedance characteristic. In general,
with a vented baffle the bass range over an octave or more is increased by several
decibels over either a very large baffle or an enclosed cabinet loudspeaker.
The power gain may be greater than 3 db since the air loading of the cone may be
increased appreciably over the lower octave of the frequency range. For about one
third of an octave above and below the resonant frequency of the system, the greater
848 (iv) ACOUSTICAL PHASE INVERTER 20.3
FIG. 20.15
I
8 ) 40
2 .
11
w6 1 30 1 i
X J I
o I I
C '1 1 1
w ' 1
Ì
n
;4 .20á / 1
/I
1
< \
á " i I ,
, I1
2
i \
2 IO /) `
4
on .... ..,. ._ ._
200
FPEOUENCY C/S
Fig. 20.15. Impedance versus frequency characteristics of a 12 inch loudspeaker
(1) in free air (2) on a flat baffle and (3) in a vented baffle. Curve (4) shows
two
" staggered" vented baffle loudspeakers in parallel (measured).
part of the energy is radiated by the port. Phase shift occurs suddenly at the re-
sonant frequency, so that the radiation below this frequency is reduced, being the
vector difference between the two sources.
At the resonant frequency, the air in the vent or tunnel moves vigorously while the
acoustical impedance on the rear of the cone is resistive and reaches a maximum.
At the frequency of the vent resonance, the motion of the speech coil is so small that
no appreciable electrical damping can take place and there is no danger of bass attenua-
tion such as may occur with the enclosed cabinet or flat baffle. The energy dissipated
by mechanical resistances and by the radiation of sound is generally small, so that the
vent resonance on which the maintenance of the bass response depends, is quite lightly
damped. For this reason, when good fidelity is required, the vent resonance fre-
quency should not be greater than 60 c /s. (Ref. 135).
A corner speaker cabinet for 15 inch cones is described in Ref. 100, while one
for 12 inch cones is described in Ref. 105. These both employ a vented baffle arrange-
ment with the vents on both sides of the cabinet, between the cabinet and the walls.
Equivalent circuit (Ref. 135)
The Fig. 20.16 is a simplified electrical equivalent circuit
of the acoustical system and is based on Fig. 20.11 with the
addition of Ry representing the radiation resistance of the vent,
and Ly, the acoustical inductance of the vent plus the effects of
its radiation reactance. Let fv be the frequency of the vent
resonance (i.e. the resonance of Lv and Cc) and assume that
the cone resonant frequency is the same. Below fv, the series
combination LUCu appears as a capacitance and the parallel
combination LvCc. as an inductance, and the overall effect is to
produce a series resonance at some frequency f, in this region. FIG. 20.16
At frequencies above fv, LUCU appears inductive and LvCc Fig. 20.16. Equiva-
capacitive, and a second series resonance appears at some fre- lent electrical circuit
quencyf,. The resonances at f, and f, are responsible for the of vented baffle
characteristic double hump in the electrical impedance fre- loud-speaker at low
quency characteristic of a vented baffle and are subject to the frequencies. (Ref.
circuit damping represented by R s. 135).
20.3 (iv) ACOUSTICAL PHASE INVERTER 849
IU and Iv represent the alternating air current flow in the cone and vent respectively.
!u originates at the back of the cone and I1, must therefore be reversed in phase with
respect to /u if the acoustical output of the vent is to reinforce that from the front of
the cone. At f1 the frequency of the vent resonance, IU and Iv are approximately
in quadrature ; f
above , the desired reinforcing condition is approached, but below
fv the two air currents oppose one another.
The performance of the system is determined by the value of fv, which is normally
made equal to the loudspeaker bass resonant frequency, and also by the ratio Lv/Cc.
With a large cabinet and large vent (Cc large, Lv small) the response can rise to a
peak in the bass, while a small cabinet and a small vent (Cc small, Ly large) will give
much the same effect as a completely closed cabinet, i.e. there may be bass loss.
+15
(a')
IMPEDANCE
+5
o III --.
/ii11(`xb'(°
(b) _'/
+10 5
I IA%IAL FREQUENCY
db
+5
db
io
,l/
i i
i i(a)
RESPONSE
i 15
Yi
The shape of the cabinet, as well as its volume, affect its frequency response and
impedance characteristics. For example, a change from 30 ins. x 15 ins. x 8 ins.
to an equivalent volume 15 ins. x 15 ins. x 16 ins. caused an increase from 45 to
58 c/s in the lower impedance peak and from 75 to 110 c/s in the higher peak (Ref.
117). In the cubic shape of cabinet the addition of an internal partition between
loudspeaker and vent, extending approximately half-way from front to back, causes a
reduction in the frequencies of both impedance peaks, and is claimed to give improved
results on both speech and music (Ref. 117).
Careful attention to damping, along the lines given above for both enclosed cabinets
and matched vented baffles, would be well repaid.
(c) Special types of vented baffle loudspeakers -see Supplement.
Refs. to vented baffles : 4, 9, 12, 13, 13A, 29, 36, 76 (Part 2), 85, 96, 113, 116, 116A,
117, 118, 135, 143, 144, 147, 150, 151, 166, 168, 175, 182, 188, 202, 209, 211, 225,
229, 230.
(v) Acoustical labyrinth loudspeaker
The acoustical labyrinth gives a performance somewhat similar to that of a vented
baffle. The rear of the cone drives a long folded tube, lined FIG. 20.18
with sound absorbing materials, the mouth of which opens in
front of the cabinet (Fig. 20.18). The length of the tube is
approximately 7 feet (measured on the centre -line) for nearly
linear response down to 70 c /s. The loudspeaker bass re-
sonance loaded by the labyrinth is preferably at a frequency at
which the wavelength is four times the length of the tube ; in
the example this is 40 c /s. If this latter condition is not ful-
filled, the frequency response will not be linear. The loud-
speaker resonance frequency in the example was reduced from
50 to 40 c/s by the loading of the labyrinth. This is the only
form of baffle which reduces the bass resonant frequency of a
loudspeaker. The rise of impedance from 400 c/s down to the
bass resonant frequency is reduced considerably by the acous- Fig. 20.18. Sec-
tical labyrinth -in one case the ratio was reduced from 10 :1 to tional view of
4.3 1 (Ref. 27).
:
acoustical labyrinth
References 13, 14, 27, 28, 188, 204, 225 Part 2. loudspeaker.
(vi) The R-J loudspeaker
The R -J loudspeaker has a particularly compact bass unit, to which any desired
tweeter can be added. In one design a 15 inch woofer unit is mounted in a cube with
18 inch sides and fundamental bass reproduction is claimed down to 20 c /s. Only
the forward radiation is used, but both sides of the cone are loaded. The back of the
speaker is completely enclosed within a small stiff cavity. The front of the woofer
works into a carefully designed rectangular duct, and the sound issues from a slot
extending across the base of the enclosure.
Refs. 189, 190, 219, 234.
(vii) Design of exterior of
cabinet
The sharp corners on the usual more-
or -less box shaped cabinet, particularly
those of the side in which the loud- aft I
2 ft 661s
speaker is mounted, produce diffraction
effects causing a sequence of peaks and
valleys up to ± 5 db in the response
characteristic. The best shaped en-
closure is a complete sphere, while the
worst is a cube with the loud- speaker
1
{-Ift->i 4.1 6 ins 1-4- ft -31.4
i
provement over the symmetrical cube, but still far from the ideal. A very close ap-
proach to the ideal is given by a rectangular truncated pyramid mounted on a rect-
angular parallelepiped (Fig. 20.19) Ref. 193.
In all cases it is desirable that the edge of the cone should be flush with the front
of the cabinet.
(i) Introduction
A horn is used in conjunction with a diaphragm or cone loudspeaker for the pur-
poses of increasing the acoustical loading on the diaphragm (over a limited frequency
range) and thereby increasing the efficiency and reducing non-linear distortion.
With a horn, inside the useful frequency limits, the movement of the diaphragm is
much less and the acoustical damping is much greater than on a flat baffle, for the
same acoustical power output. Thus with a horn, a smaller diaphragm can radiate
a given acoustical power.
A horn is essentially a device which transforms acoustical energy at high pressure
and low velocity to energy at low pressure and high velocity.
Horns are of various shapes -conical, parabolic, hypex and exponential, but the
exponential is most widely used.
A well designed and well executed exponential or hypex horn loudspeaker is capable
of giving a flatter frequency output characteristic with less distortion than any other
form of loudspeaker.
References 13A, 14, 15, 59, 82, 142, 177, 188.
Throat
O
F 0 0 0 oN 0 0 0V' c(G co0 0O 0N :y (f) 50 100 150 200 c/s
CO CO C1 gth (a') 22.6 11.3 7.52 5.65 feet
w 0.47 0.93 1.40 1.85 inverse feet
..
o
nouth should preferably not be less than one third of the
th, while the effect of resonances at the low frequencies
increasing the mouth diameter to two thirds of the maxi-
). Taking the smaller mouth diameter, and referring to
ency :
0 0 0 0 0 0NC1
0sNC)N.+ 00
0 0V'Y)fD ace = 1.05.1' = 1200 /f' (3)
plied to any shape of mouth -see table below :
24 12 8 6 feet
;th (a') 22.6 11.3 7.52 5.65 feet
y (f') 50 100 150 200 c/s
moss-section
n is
C) Cl Cl C)V' .) <D
(4)
a
órd I I I I
eet
er in feet.
m = doe"""s
(5)
= do antilog)o (mí/4.605) (6)
= (4.605/m) log10 (d,,, /do) (7)
Ian III.
(4.605/1) logro (d,0 /do) (8)
có equency is taken as f' where f' = 1.2f0, and if the minimum
Ó ° .:0000óo
.r o 000o
n as one third of the wavelengthSth at frequency f)thenm
F ,.0000
...°.
clue of 10.5 /ñ' and
o^o 0o0 f mouth = A' /3 = 380/f' (9)
:500/f') logro (380 /f'd0) (10)
th in feet at frequency f'
1.2 x flare cut -off frequency
at throat in feet
Emensions ó_ _ .) feet.
Horns with square cross-section
These follow the same laws as horns of circular cross -section having diameters
equal to the sides of the square, but the length and mouth dimensions are slightly less
for the same minimum useful frequency f' :-
minimum side of mouth = 300 /f' (11)
minimum length° = (500/f0) logro (300/f'ao) (12)
where a0 = length of side at throat.
For cutting the side of a square horn from a sheet, see Ref. 77.
General references to horn dimensions : 39, 46, 82.
References to theory of exponential horns 13A, 142, 175, 188. :
The cut -off frequency is the same as for an exponential horn having the same
flaring constant -see eqn. (2a). When T = 1, the horn is exponential. When
T = 1 /(mx,) and m is allowed to go to zero, the horn is conical. When T = 0 the
horn is hyperbolic cosine or catenoidal (Refs. 13A, 123, 140, 142, 188). The name
" hypex " is usually applied when T is greater than zero and less than unity.
The value of Tin " hypex " horns is usually between 0.5 and 0.7, and within these
limits the throat resistance of an infinite horn is more nearly constant than that of an
exponential horn, at frequencies slightly above the cut -off frequency. These com-
parisons are for constant throat, mouth and length of horn ; under these conditions
a " hypex " horn with T of the order of 0.6 has improved low frequency characteris-
tics as compared with those of an exponential horn. Consequently, for equivalent
performance, a " hypex " horn may be made more compact than an exponential horn.
A further useful feature of the " hypex " characteristic is that it makes possible a
gradual transition from conical, via " hypex " with varying T, to exponential (Ref. 123).
An analysis of the response peaks in finite hyperbolic horns, and design procedure
for horns to have peaks at pre- determined frequencies, is given in Ref. 139.
(A) (C)
Sound Chamber Sound Chamber
Horn Horn
3
.V//i///////i/ii////i/E
A
FIG. 20.21
Fig. 20.21. Different forms of sound chambers in horn loudspeakers.
For example if a horn with a 40 c/s cut-off is to reproduce a 4000 c/s note, there
will theoretically be 8% second harmonic distortion for an acoustical power of 0.01
watt per square inch of horn throat. The actual distortion appears to be about half
the theoretical value.
Thus for reasons of both efficiency and distortion, a horn should only be designed
to cover a limited frequency range.
(C) Driving units
The simplest driving mechanism is the electro-magnetic type with an iron dia-
phragm as the armature (Fig. 20.21A).
All modern driving units are of the moving-coil type and may have a diaphragm
of aluminium alloy or some form of paper or a cloth base impregnated with synthetic
resin.
Well designed direct- radiator loudspeakers may be used as the driving mechanisms
for horns. The most suitable size is from 8 to 12 inches diameter for medium power
requirements. Some loudspeakers are designed specially for this application.
Where high power output at very low frequencies is required, a 15 inch unit may
be used, as in the Klipsch corner horn described below.
(D) Distortion
In addition to the distortion caused by the throat, there is also distortion due to
the sound chamber. The acoustical capacitance is a function of the position of
the diaphragm, and the effect is most apparent at low frequencies where the amplitude
is large. This distortion may be reduced by the use of a large sound chamber, thus
limiting this unit to low frequencies only. A separate high frequency unit with a small
sound chamber may be used, since it will not be required to handle large amplitudes.
Distortion is also caused by a non-linear suspension ; this is only serious at low
frequencies (Ref. 14, Fig. 8.21). Distortion at low frequencies due to non-
linear suspension may be reduced by
(1) the use of a large dynamic driver,
(2) increased compliance in the suspension,
(3) an enclosed air -chamber at the rear of the diaphragm as in the Klipsch corner
horn.
Another cause of distortion is frequency modulation through the Doppler
Effect -see Sect. 7(ii) -which can be reduced to small proportions by the use of
separate high and low frequency units.
It is impossible to design a horn loudspeaker that covers a wide frequency band
and is simultaneously free from non-linear distortion. Thus two separate units for
low and high frequencies are essential for fidelity. This is a limitation peculiar to
horn loudspeakers.
(E) Efficiency
With horn loudspeakers efficiencies as high as 80% can be achieved over a limited
frequency range. A typical horn of the type used in cinema theatres has an efficiency
from 30% to 45% over its useful range. The Klipsch corner horn has an efficiency
around 50% over its useful range.
856 (v) HORN LOUDSPEAKERS -GENERAL 20.4
DRIVER UNIT
Fig. 20.22. Concentric folded horns (A) three section reflex (B) three section reflex
radial type.
One outstanding design is the Klipsch corner type speaker which includes a folded
low frequency horn.
This is notable mainly on account of the comparatively small space which it occupies
FIG. 20.23 -20 cubic feet in,all- compared with other dual horn systems
having equivalent performance. The efficiency of the low
frequency woofer unit (K-3 -D) is not less than 50% down to
36 c /s, and still fairly high at 32.7 c/s ; it is capable of radiating
a clean fundamental at reduced power down to 27 c/s. Maxi-
mum electrical power input is 15 watts, so that the maximum
acoustical power output is over 7 watts.
The low frequency unit has a 15 inch direct radiator loud-
speaker with an enclosed cabinet baffle at the rear and a folded
horn in front. The recommended driver for ordinary home
and small theatre power levels is the Stephens P- 52 -LX -2
woofer motor ; this is specially treated to increase its com-
pliance before installation. The enclosed cabinet baffle at
the rear is designed to offset the mass reactance of the throat
impedance at low frequencies. The volume of the enclosed
Fig. 20.23. Folded baffle is given
theoretically by
horns for low fre- V = 2.9 AR
quencies actuated (15)
where V = volume in cubic inches
by rear of cone with A = throat area in
area increasing in and R = length of horn square inches
steps.
in inches within which the horn
area doubles.
Fig. 20.24. Folded horns for low frequencies -as part of dual system, with conical
individual sections. (Ref. 52.)
858 (vi) FOLDED HORN LOUDSPEAKERS 20.4
FIG. 20.25
ç
be used with high frequency horns as
an alternative to multi- cellular con-
_ struction, to spread the sound over
FIG. 20.26
a wider angle. In one example the
angle was increased from about 20°
Fig. 20.26. Side and front views of the to over 50° at 8000 c/s by the use of
Klipsch bass reproducer Model K-3. such a diffusing lens. The construction
(Ref. 48). appears cheaper than a multi -cellular
20,4 (vii) HIGH FREQUENCY HORNS 859
FIG. 20.27
nt:
M 00 0W
M Co IOiipL\
Fig. 20.27. Voice coil impedance of two designs of Klipsch bass reproducer (A) Model
K -3-B as Ref. 48 (B) Model K-3 -C (cone with increased compliance). (Ref. 99).
See also the Klipsch high- frequency horn (Fig. 20.28) and page 858.
References to high frequency horns 33, 76 (Part 1), 80, 137, 165.
:
General references to horns 2, 13A, 14, 15, 30, 31, 32, 33, 34, 38 (Parts
: 1 and 2),
39, 45, 46, 47, 48, 51, 52, 59, 69, 73, 76, 77, 82, 99, 115, 119, 128, 130, 137, 142, 147,
165, 175, 177, 206, 209, 211, 217, 232.
(i) Introduction
When a wide frequency range has to be covered there is a choice between
1. A single unit employing some special construction to extend the frequency
range -see Sect. 2(iii),
2. Two or three separate loudspeakers, each covering a limited frequency range,
3. An integral dual or triple system with two or three loudspeakers mounted
(usually co- axially) in one equipment, and
4. Some compromise arrangement -see (iv) below.
If separate loudspeakers are used, they should be mounted as closely together as
possible, with co -axial mounting as the ideal ; this is not so important when the
cross-over frequency is below 500 c /s. In addition, they should be co- planar, with
the plane of horn loudspeakers taken as the plane of the diaphragm. The loud-
speakers should be correctly phased, so as to be additive in the overlap region.
A frequency dividing network is used to split the output between two loudspeakers
so that neither unit is called upon to handle large amplitudes of frequencies beyond
its range. This has the advantage that Doppler Effect distortion (see Sect. 7) is much
reduced. Other advantages are that, on account of limitations in the frequency range
of each unit, the system efficiency is increased, while the directivity characteristic
is improved due to the smaller diaphragm (or horn mouth) for the high frequency
unit. In addition, the transient response is improved, and there is less intermodula-
tion and reduced frequency modulation.
The loudspeakers should preferably have the same efficiency, otherwise one will
have to be attenuated. The directional characteristics of the high frequency unit
should receive careful attention.
Integral dual systems
One excellent arrangement employs two cones, co -axial and co- planar, with the
small high-frequency cone mounted near the apex of the large cone. Both cones
vibrate in unison in the overlap region. (Refs. 61, 116, 116A).
Another arrangement employs a large low- frequency cone and a co -axial high-
frequency horn. This has the disadvantage that the sound sources are not co- planar,
since the diaphragm of the high frequency unit is mounted to the rear of the cone.
Integral triple systems
In one design there is a large low- frequency cone, a mid -frequency horn using the
flared low- frequency cone for its mouth, and a high-frequency horn mounted in
front of the mid- frequency horn.
In a triple system there is no necessity for such a compromise regarding the cross-
over frequency. The " woofer " may handle up to between 300 and 600 c /s, the
middle unit up to between 2000 and 5000 c/s, and the " tweeter " will then look
after the higher frequencies (e.g. Ref. 134).
The design of frequency dividing networks is covered in Chapter 21, Sect. 3
(i) Loudness
Loudness may be measured in loudness units -see page 827 and Fig. 19.8.
IO IO
ACOUSTICAL WATTS OUTPUT
E- I Iói N- ii ii1N-ii N-
10
-IMIM1OIN-MIM\OIIN-11\YIIN i i.i
N1.\1111IC ii ii
111110111111011111116-111111
_N--A111IN--111
-..11111 _ -.'I _-...1111
_ 1111..1I--.\.1I
-..11111 I I I II
MMNomMNImui1111111111mm11MMm111
1111111111=1111111111111IU1111MIINII
_
-..I -...
IIIIIIIIIIIII..IIIIIIIINIIIIII
_-i11111N-\\111-N\\11I
111III./111111111Iglllll
_11111;IIÌÌIIIIIIIIIIIIIIIII
N-\\/ IN-/\L IN-11111-\1\II
N-I.\IIN-\\ N-\\IINN\I/11
III1111mioI1I
1121111111110IMM11111111111111111111 111111I
0.001 111111111111111111111IIAIII M1111111
10? 04 0' 10' 10'
VOLUME IN CUBIC FEET
The " optimum reverberation time " at 1000 c/s depends on the volume of the
room (Ref. 13A) :
Volume 1000 2000 4000
To , 0.7 0.75
20 000 100 000 500 000 cu. ft.
0.82 1.01 1.28 1.62
In a typical room 20 ft. x 15 ft. x 10 ft., with a reverberation time of 0.5 second,
secs.
and with the listener 6 feet from the loudspeaker, the reflected sound adds 3 db to the
direct radiation at 500 c/s under steady state conditions, and 1 db when the sound only
continues for 0.02 second as for speech (Ref. 14, Fig. 13.26). At higher frequencies
the reflected sound becomes rapidly smaller, and may be neglected. At 200 c/s
the reflected sound adds 5.5 db under steady state conditions or 2 db when the sound
continues for 0.02 second. At 100 c/s the reflected sound adds 9.5 db under steady
state conditions, or 6 db when the sound continues for 0.02 second. The effect of
the reflected sound becomes more important as the listener moves further away from
the loudspeaker.
References 13A, 14, 15, 50, 55, 175, 178.
The acoustical power in watts to produce an intensity level of 80 db (0 db =
0.0002 dyne /cm') is plotted in Fig. 20.30 against the volume of the enclosure in cubic
feet (after Olson, Ref. 13A). For living rooms the following information may be
derived-
These values apply to enclosures with optimum reverberation times. Living rooms
may have reverberation times which are lower than the optimum values, and the
acoustical power required will then be greater than indicated above.
The intensity levels for home listening under various conditions are set out in
Chapter 14 Sect. 7(iii).
waves. In a typical case with a room 16 ft. 9 in. x 11 ft. 6 in. x 8 ft. 6 in. the first
nine resonances are 36.8, 51.1, 62.9, 68.6, 73.6, 77.6, 85.5, 102.3, 137.2 (Ref. 53).
The most desirable shape for a listening room is approximately in the ratio height :
width : length = 1 : 1.27: 1.62, so as to distribute the resonances fairly uniformly.
The peaks due to these resonances are of the order of 20 db and completely mask
less pronounced effects. The bad effects may be reduced by
(1) increased acoustical damping at low frequencies, especially near the corners
of the room where it is twice as effective as elsewhere (Ref. 175), and
(2) an open door or window, or preferably both.
In addition to these room resonances, which occur over the whole room, there are
other peaks of response due to focusing or interference between different paths, at
middle and high frequencies, which vary from point to point. Thanks to our two
ears, these sharp peaks and valleys are much less evident than they otherwise would be.
References 13A, 14, 53, 175, 192, 208, 220, 221.
ment varies linearly from a maximum at the centre to zero at either end, the direc-
tionality is improved. The line- source should be tilted slightly forwards so that the
central loudspeaker points directly to the centre of the audience. Better results may
be obtained by using two line- sources, a long one for the low frequencies and a shorter
one (one quarter the length of the long one) for the high frequencies, with a cross-over
network for f. = 1000 c/s (Ref. 197).
In large installations, time delay may be incorporated (Ref. 145, 197).
Multiple loudspeakers
It has been proved by Haas (Ref. 125) that when there are two or more loudspeakers,
the listener only appears to hear from one of the loudspeakers-the nearer one in all
cases -and that a separate echo is not heard until the time delay reaches 50 milli-
seconds, equivalent to a path difference of 56 feet. He also proved that the echo has
negligible disturbing effect on the listener if it is 10 db lower in level than the original.
As a result of the latter effect, listeners in the vicinity of one loudspeaker are not dis-
turbed by the echo from another. It seems reasonable to infer that spacing between
loudspeakers up to about 70 feet would be satisfactory so far as echoes are concerned.
(vii) Open air public address
A typical reflex horn, with an input of 1 watt to the voice coil, gives the following
sound levels on the axis (Ref. 73) :
Distance 10 20 40 80 160 320 feet
Sound level 100 94 88 82 76 70 db
With a single horn, the axis should point directly to the farthest part of the audience.
Horns may be arranged radially to cover a larger audience, allowing 30° for each horn.
For speech reproduction only, a 10 watt amplifier with two horn loudspeakers is
sufficient for a crowd of 5000 people in quiet surroundings. For musical repro-
duction a power of 40 watts with four loudspeakers will be required. It is usual to
allow from 5 to 10 watts for each horn loudspeaker. Alternatively a large number of
any convenient type of loudspeaker may be used, with the spacing between loud-
speakers not greater than 70 feet. When a wider frequency range is required, some
dual horn system is frequently used, particularly with open -air orchestral sound re-
inforcing.
For large crowds, several line -source loudspeakers -see (vi) above -may be used
at the centre of the crowd.
References 13A (pp. 292 -296), 14 (pp. 406 -409), 73, 104, 125, 176.
(viii) Intercommunicating systems
Only speech is to be reproduced, and the usual requirement is merely to have suffici-
ent articulation to be understood. A reduced frequency range is almost universal,
while a very restricted range is used in noisy surroundings (see Chapter 14 Sect. I1).
References 1, 13 (pp. 297, 299), 13A (p. 426).
(ix) Background music in factories
If the noise level is comparatively low, the system may be quite conventional, with
a frequency range from say 100 to at least 6000 c /s. Loudspeakers should be placed
fairly close together to give good coverage of high frequencies. The spacing should
be adjusted so that, with the prevailing noise level, the range of each is slightly over
half the distance between them.
If the noise level is high, the highest noise intensity is often limited in frequency
range at both extremes. In such a case, as an alternative to over -riding the noise,
the full frequency range up to 8000 c/s may be used for the music; which may be at
the same level as the mid- frequency noise. The low .frequency music range may
either be used or attenuated as desired. Careful choice of the source of music is
required to give the full frequency range without distortion. A conventional A-M
receiver is unsuitable, owing to sideband cutting. Many shellac discs, especially
the older ones, are unsuitable on account of limited frequency range.
In all cases some form of volume compressor or limiter is required to reduce the
volume range, and the music should be selected to avoid sudden large changes in
volume.
868 (i) NON -LINEARITY 20.7
deviates from a straight line. The principal causes of non -linearity are
1. Insufficient rigidity in the cone,
-
Non -linearity in a cone occurs when the force versus displacement characteristic
A system has been developed (Ref. 152) for automatically recording the total har-
monic distortion of a loudspeaker over the a -f range. This is particularly helpful
in detecting narrow frequency bands where the distortion is high.
One method of recording the performance of a loudspeaker with regard to non-
linear distortion is to measure on the axis the maximum sound level before a certain
amount of harmonic distortion is produced. This may be plotted as a curve of
maximum undistorted sound pressure as a function of frequency (Ref. 153). Alter -
natively the distortion may be plotted as a function of frequency for constant electrical
input. In all cases the loudspeaker must be tested in the enclosure for which it has
been designed.
References 13A (pp. 163 -173), 20, 21, 23, 36, 61, 106, 108, 109, 133, 152.
39.5 (May 1951) 509; "American Standard of Acoustical Terminology ", A.S.A.
Z24.1 -1951.
(ii) Electrical, mechanical and acoustical equivalents
Electrical Acoustical Acoustical unit
Capacitance acoustical capacitance cm5 /dyne
Inductance inertance grams /cm2 /cm2
Resistance acoustical resistance acoustical ohms
E.M.F. pressure dynes/cm'
Impedance acoustical impedance acoustical ohms
Electrical Mechanical Mechanical unit
Capacitance Compliance cm /dyne
TABLE 1
Frequency ratio
Tone Interval Natural (just) scale Tempered scale
The frequency of all tones is determined by the " pitch." The international
standard pitch is a frequency of 440 c/s for tone A (equivalent to 261.63 for C) whereas
the " physical pitch " is a frequency of 256 for C.
The octaves of C in the two examples above are :
The frequency of any tone may be calculated by multiplying the value of C next
below it by the frequency ratio of the tempered scale as given by the appropriate row
in Table 1.
References 13 (pp. 334 -335), 89, 90, 91, 92, 93, 127.
27. Olney, B. ' A method of eliminating cavity resonance, extending low frequency response,
increasing acoustic damping in cabinet -type loudspeakers " Jour. Acous. Socy. Am. 8.2 (Oct. 1936) and
104. 1936
28. Olney, B. J. " The acoustical labyrinth " Elect. 10.4 (April, 1937) 24.
29. Smith, F. W ` Resonant loudspeaker enclosure design (vented baffle) Comm. 25.8 (Aug.
fflril,
30. Klipsch, P. W. " Design of compact two -horn loudspeaker " Elect. 19.2 (Feb. 1946) 156. 1945) 35.
31. Klipsch, P. W. " A low frequency horn of small dimensions " Jour. Acous. Socy. Am. 13 (Oct.
1941) 137.
32. Klipsch, P. W. " Improved low frequency horn " Jour. Acous. Socy. Am. 14 (Jan.
1943) 179.
20.10 REFERENCES 877
33. Klipsch, P. W. " A high quality loudspeaker of small dimensions " Jour. Acous. Socy. Am. 17.3
(Jan. 1946) 254. With extensive bibliography on horn loudspeakers.
34. Klipsch, P. W. " A note on acoustic horns " Proc. I.R.E. 33.7 (July, 1945) 447.
35. Fletcher, H. " Hearing, the determining factor for high- fidelity transmission," Proc. I.R.E. 30.6
(June, 1942) 266.
36. Olson, H. F., and J. Preston " Wide range loudspeaker developments " R.C.A. Rev. 7.2 (June,
1946) 155.
37. Stanley, A. W. " The output stage- effect of matching on frequency response " W.W. 52.8 (Aug.
1946) 256.
38. Olson, H. F. " Horn loud speakers " R.C.A. Rev. (1) 1.4 (April, 1937) 68 ; (2) Efficiency and dis-
tortion, 2.2 (Oct. 1937) 265.
39. Sanial, A. J. " Graphs for exponential horn design," R.C.A. Rev. 3.1 (July, 1938) 97.
"Or 40. Wait, E. V. " Loudspeaker cones " A.W.A} Tec. Rev. 1.1 (March, 1935) 14.
41. Rudd, J. B. " The impedance of the voice -coil of a loudspeaker " A.W.A. Tec. Rev. 3.3 (Jan.
19381 100.
42. Rudd, J. B. " Equipment for the measurement of loudspeaker response " A.W.A. Tec. Rev. 3.4
(April, 1938) 143.
43. de Boer, J. " The efficiency of loudspeakers " Philips Tec. Rev. 4.10 (Oct. 1939) 301.
44. Shorter, D. E. L. " Loudspeaker transient response," B.B.C. Quarterly 1.3 (Oct. 1946) 121 ; sum-
mary " Loudspeaker transients " W.W. 52.12 (Dec. 1946) 424. See also Ref. 117.
45. Hilliard, J. K. " High fidelity reproduction " F-M & T 6.1 (Jan. 1946) 28.
46. " The design of acoustic exponential horns " Electronic Eng. 19.235 (Sept. 1947) 286.
47. Sanial, A. J. " Loudspeaker design by electro- mechanical analogy " Tele-Tech 6.10 (Oct. 1947) 38.
48. Klipsch, P. W. " The Klipsch sound reproducer -a corner type speaker which uses the walls as
part of the acoustical system," FM & T 7.9 (Sept. 1947) 25.
49. Enhancing sound reproduction " (Review of patent No. 2, 420, 204 by C. M. Sinnett) Audio
Eng. 31.6 (July, 1947) 43.
50. Hopkins, H. F., and N. R. Stryker " A proposed loudness -efficiency rating for loudspeakers and
the determination of system power requirements for enclosures " Proc. I.R.E. 36.3 (March, 1948)
315.
51. Olson, H. F., and R. A. Hackley " Combination horn and direct radiator loud- speaker " Proc.
I.R.E. 24.12 (Dec. 1936) 1557.
52. Massa, F. " Horn -type loud speakers -a quantitative discussion of some fundamental require-
ments in their design," Proc. I.R.E. 26.6 (June, 1938) 720.
53. Moir, J. " Acoustics of small rooms," W.W. 50.11 (Nov. 1944) 322.
54. Roder, H. " Theory of the loudspeaker and of mechanical oscillatory systems " Part 1, Radio
Engineering 16.7 (July 1936) 10 ; Part 2, 16.8 (Aug. 1936) 21 ; Part 3, 16.9 (Sept. 1936) 24 ; Part 4,
16.10 (Oct. 1936) 19.
35. Morison, G. E. " Power and realism-estimating the watts required for a given sound intensity "
W.W. 46.6 (June, 1942) 130.
56. Seabert, J. D. " Electrodynamic speaker design considerations " Proc. I.R.E. 22.6 (June, 1934) 738.
57. Olson, H. F. " A new cone loud speaker for high fidelity sound reproduction " Proc. I.R.E. 22.1
(Jan. 1934) 33.
58. Greaves, V. F., F. W. Kranz and W. D. Crozier " The Kyle condenser loud speaker " Proc. I.R.E.
17.7 (July, 1929) 1142.
59. Terman, F. E. (book) " Radio Engineering " (McGraw -Hill Book Co., New York and London,
3rd edit. 1947).
60. Sawyer, C. B. " The use of rochelle salt crystals for electrical reproducers and microphones "
Proc. I.R.E. 19.11 (Nov., 1931) 2020.
61. Rand, G. E. " An F -M quality speaker " (R.C.A. LC -1A) Broadcast News, No. 46 (Sept. 1947) 24.
62. Kennelly, A. E. " Electrical vibration instruments " (The Macmillan Company, New York).
63. Wente, E. C., and A. L. Thuras " Moving -coil telephone receivers and microphones," Jour. Acosta.
Socy. Am. 3.1 (July, 1931) 44.
64. Williams, A. L. " New piezo-electric devices of interest to the motion picture industry " Jour.
S.M.P.E. 32.5 (1939) 552.
65. Olson, H. F., and F. Massa " A high- quality ribbon receiver," Proc. I.R.E. 21.5 (May, 1933) 673.
66. Olson, H. F., and F. Massa " Performance of telephone receivers as affected by the ear " Jour.
Acous. Socy. Am. 6.4 (1935) 250.
67. Olson, H. F., Jour. S.M.P.E. 27.5 (1936) 537.
68. Fletcher, H. and others " Auditory perspective " ( 6 papers) E.E. 53.1 (Jan. 1934) 9.
69. Massa, F. (Book) " Acoustic design charts " (The Blakiston Company, Philadelphia, 1942).
70. " American recommended practice for loudspeaker testing " C16.4 -1942 (American Standards
Association, Nov. 6, 1942).
71. Olson, H. F. " Multiple coil, multiple cone loudspeakers " Jour. Acous. Socy. Am. 10.4 (April,
1939) 305.
72. Langford -Smith, F. " The relationship between the power output stage and the loudspeaker "
Proceedings World Radio Convention, Australia, April 1938 (reprinted W.W. 44.6 (Feb. 9, 1939)
133 ; 44.7 (Feb. 16, 1939)) 167.
73. Edwards, E. M. " Conversion efficiency of loudspeakers " Service 17.3 (March, 1948) 24.
74. Cook, E. D. " The efficiency of the Rice- Kellogg loudspeaker " G.E. Review 33.9 (Sept. 1930) 505.
75. Massa, F. " Loud speaker design " Elect. 9.2 (Feb. 1936) 20.
76. McProud, C. G. " Two -way speaker system " Audio Eng. (1) 31.11 (Nov. 1947) 18 ; (2) 31.12
(Dec. 1947) 19 ; (3) 32.2 (Feb. 1948) 21.
Logan, G. H. " Design for exponential horns of square cross section " Elect. 12.2 (Feb. 1939) 33.
Reprinted in " Electronics for Engineers " (McGraw -Hill, 1945).
78. Barker, A. C. " Single diaphragm speakers-obtaining wide frequency response and effective
damping " W.W. 54.6 (June, 1948) 217.
79. Lanier, R. S. " What makes a good loudspeaker ? " Western Electric Oscillator No. 8 (July, 1947)
11.
80. Nickel, F. " Quality loudspeakers for every use " Western Electric Oscillator No. 8 (July, 1947) 7.
81. Davies, L. A. " Electro- acoustic coupling -how to get high fidelity from ordinary speakers,"
Australasian Radio World, 11.10 (March, 1947) 7.
82. White, S. J. " Horn type loudspeakers " Audio Eng. 32.5 (May, 1948) 25.
184. Olson, H. F. and Radio Corporation of America " Improvement in folded suspension loudspeaker "
U.S. Patent 545,672 (Div. 5) July 19, 1944.
85. Langford -Smith, F. " The design of a high -fidelity amplifier (1) The power valve and the loud-
speaker " Radiotronics No. 124 (March -April, 1947) 25.
86. ' Stereophonic sound," Elect. 21.8 (August, 1948) 88.
878 REFERENCES 20.10
87. Beard, E. G. (Book) " Philips Manual of radio practice for servicemen " (Philips Electrical In-
dustries of Australia Pty. Ltd., 1947) pp. 257 -259 describing the Philips Miller system.
88. Meyer, E. (Book) " Electro-acoustics " (G. Bell & Sons Ltd. London 1939).
89. Wood, Alexander (Book) " The physics of music " (Methuen & Co. Ltd., London, 3rd edit. 1945).
90. Tillson, B. F. " Musical acoustics " Audio Eng. Part 1, 31.5 (June, 1947) 34 ; Part 2, 31.6 (July,
1947) 25 ; Correction 32.6 (June, 1948) 8.
91. Vermeulen, R. " Octaves and decibels," Philips Tec. Rev. 2.2 (Feb., 1937) 47.
92. Jeans, Sir James (Book) " Science and Music " (MacMillan Company and Cambridge University
Press, 1938).
93. Mills, J. (Book) " A Fugue in cycles and bels " (Chapman & Hall, London, 1936).
94. Soother, H. " Coaxial and separate two -way speaker system design " Comm. 28.7 (July, 1948) 22.
95. " Speakers for sound equipment " American R.M.A. Standard SE -103 (April, 1949).
96. Drisko, B. B. " Getting the most out of a reflex-type speaker " Audio Eng. 32.7 (July, 1948) 24.
97. Angevine, O. L., and R. S. Anderson " The problem of sound distribution " Audio Eng. (1) 32.6
(June, 1948) 18 ; (2) 32.9 (Sept. 1948) 16 ; (3) 32.10 (Oct. 1948) 28.
98. Kendall, E. L. " Electric power required for sound systems " Service 17.9 (Sept. 1948) 12.
99. Klipsch, P. W. " Progress in Klipsch speakers " FM -TV 8.11 (Nov. 1948) 36.
100. McProud, C. G. " A new corner speaker design " Audio Eng. 33.1 (Jan. 1949) 14 ; 33.2 (Feb.
1949) 13.
101. Carnras, M. " A stereophonic magnetic recorder " Proc. I.R.E. 37.4 (April 1949) 442.
102. Seashore, C. E. (Book) " Psychology of music " (McGraw -Hill Book Co., New York and London,
1938).
103. Briggs, G. A. (Book) " Loudspeakers -the why and how of good reproduction " (Wharfedale
Wireless Works, Bradford, Yorkshire, England, 2nd edit. 1948).
104. Greenlees, A. E. (Book) " The amplification and distribution of sound " (Chapman and Hall Ltd.,
London, 2nd edit. 1948).
105. McProud, C. G. " Corner speaker cabinet for 12 inch cones " Audio Eng. 35.5 (May 1949) 14.
106. Cunningham, W. J. " Non -linear distortion in dynamic loudspeakers due to magnetic fields "
Jour. Acous. Socy. Am. 21.3 (May 1949) 202.
107. Hartley, H. A. " New notes in radio " (booklet published by H. A. Hartley Co. Ltd., London).
109. Veneklasen, P. S. " Physical measurements of loudspeaker performance " Jour. S.M.P.E. 52.6
(June 1949) 641.
110. Maxfield, J. P. " Auditorium acoustics " Jour. S.M.P.E. 51.2 (Aug. 1948) 169.
112. Grignon, L. D. " Experiment in stereophonic sound " Jour. S.M.P.E. 52.3 (March 1949) 280.
113. Chapman, C. T. " Vented loudspeaker cabinets-basis of design to match existing loudspeaker
units," W.W. 55.10 (Oct. 1949) 398.
114. American R.M.A. Standard REC-104 " Moving coil loudspeakers for radio receivers " (Jan. 1947).
115. Klipsch, P. W. " New notes on comer speakers " FM-TV 9.8 (Aug. 1949) 25.
116. Olson, H. F., T. Preston and D. H. Cunningham " New 15 -inch duo -cone loudspeaker " Audio
Eng. 33.10 (Oct. 1949) 20.
116A. Olson, H. F., J. Preston and D. H. Cunningham " Duo -cone loudspeaker " R.C.A. Rev. 10.4
(Dec. 1949) 490.
117. Briggs, E. A. (Book) " Sound reproduction " Wharfedale Wireless Works, Bradford, Yorks.
England, 1949.
118. Planer, F. E., and 1. I. Boswell " Vented loudspeaker enclosures " Audio Eng. 32.5 (May 1948) 29.
119. Klipsch, P. W. " Woofer -tweeter crossover network " Elect. 18.11 (Nov. 1945) 144.
120. Harvey, F. K. " Focussing sound with microwave lenses " Bell Lab. Record 27.10 (Oct. 1949) 349.
121. Beranek, L. L. (Book) " Acoustic measurements " (John Wiley & Sons, Inc. New York ; Chapman
& Hall Ltd. London, 1949).
122. U.S. Patents 2,310,243 and 2,373,692.
123. Salmon, V. " A new family of horns " Jour. Acous. Socy. Am. 17.3 (Jan. 1946) 212.
124. Oison, H. F., J. C. Bleazey, J. Preston and R. A. Hackley " High efficiency loud speakers for per-
sonal radio receivers " R.C.A. Rev. 11.1 (Mar. 1950) 80.
125. Haas, H. (report) " The influence of a single echo on the audibility of speech " Translated into
English and available from Director, Building Research Station, Garston, Watford, Herts, England.
126. Moir, J. " Transients and loudspeaker damping " W.W. 56.5 (May 1950) 166.
127. Barbour, J. M. " Musical scales and their classification " J.A.S.A. 21.6 (Nov. 1949) 586.
128. Mawardi, O. K. " Generalized solutions of Webster's horn theory " J.A.S.A. 21.4 (July 1949) 323.
129. Goodfriend, L. W. " Simplified reverberation time calculation " Audio Eng. 34.5 (May, 1950) 20.
130. Gilson, W. E., J. J. Andrea " A symmetrical comer speaker " Audio Eng. 34.3 (March 1950) 17.
131. Corrington, M. S. " Transient testing of loudspeakers " Audio Eng. 34.8 (Aug. 1950) 9.
132. Meir, J. " Developments in England-new loudspeaker measurement technique are being used to
correlate subjective and instrumental performance evaluations." FM -TV 10.9 (Sept. 1950) 23.
133. Olson, H. F. " Sensitivity, directivity and linearity of direct radiator loudspeakers " Audio Eng.
34.10 (Oct. 1950) 15.
134. Plach, D. J. and P. B. Williams " A new loudspeaker of advanced design " Audio Eng. 34.10 (Oct.
1950) 22.
135. Shorter, D. E. L. " Sidelights on loudspeaker cabinet design " W.W. (1) 56.11 (Nov. 1950) 382 ;
(2) 56.12 (Dec. 1950) 436. British Patent Application No. 24528/49.
136. Smith, B. H. " A distributed-source horn " Audio Eng. 35.1 (Jan. 1951) 16.
137. Taylor, P. L. " Ribbon loudspeaker " W.W. 57.1 (Jan. 1951) 7.
138. Ewaskio, C. A., and O. K. Mawardi " Electroacoustic phase shift in loudspeakers " J. Acous. Soc.
Am. 22.4 (July 1950) 444.
139. Molloy, C. T. " Response peaks in finite horns " J. Acous. Soc. Am. 22.5 (Sept. 1950) 551.
140. Thiessen, G. J. " Resonance characteristics of a finite catenoidal horn " J. Acous. Soc. Am. 22.5
(Sept. 1950) 558.
141. Kock, W. E., and F. K. Harvey " Refracting sound waves " J. Acous. Soc. Am. 21.5 (Sept. 1949) 471.
Morse, P. M. (Book) " Vibration and Sound " (McGraw -Hill Book Co. Inc. 2nd ed. 1948).
X142.
143. Worden, D. W. " Design, construction and adjustment of reflexed loudspeaker enclosures " Audio
Eng. 34.12 (Dec. 1950) 15.
144. Smith, B. H. " Resonant loudspeaker enclosures " Audio Eng. 34.12 (Dec. 1950) 22.
145. Parkin, P. H., and W. E. Scholes " Recent-developments in speech reinforcement systems " W.W.
57.2 (Feb. 1951) 44.
146. Preisman, A. " Loudspeaker damping " Audio Eng. 35.3 (March 1951) 22 ; 35.4 (April 1951) 24.
147. Souther, II. T. " Design elements for improved bass response in loudspeaker systems " Audio
Eng. 35.5 (May 195 i) 16.
20.10 REFERENCES 879
148. (A) Tanner, R. L. " Improving loudspeaker response with motional feedback " Elect. 24.3 (Mar.
1951) 142. (B) Letter R B. McGregor 24.6 (June 1951) 298.
149. Gardner, M. F., and J. L. Barnes (Book) " Transients in Linear Systems " (John Wiley & Sons,
New York 1942) Chapter 2. -com-
150. Moir, J. " Loudspeaker performance- achieving optimum reproduction in small rooms 32.
parative advantages of open-cone and horn- loaded speakers " FM-TV 11.4 (Apr. 1951)
151. Plach, D. J., and P. B. Williams " Loudspeaker enclosures " Audio Eng. 35.7 (July 1951) 12.
152. Olson, H. F., and D. F. Pennie " An automatic non-linear distortion analyzer " R.C.A. Rev. 12.1
(Mar. 1951) 35.
153. Hall, H. H., and H. C. Hardy " Measurements for aiding in the evaluation of the quality of loud-
speakers " Abstract J. Acous. Soc. Am. 20.4 (July 1948) 596.
154. Peterson, A. P. G. " Intermodulation distortion " G. R. Exp. 25.10 (March 1951) 1.
155. Moir, J. " Loudspeaker diaphragm control " W.W. 57.7 (July 1951) 252.
156. Salmon, V. " Efficiency of direct radiator loudspeakers " Audio Eng. 35.8 (Aug. 1951) 13.
20.
157. Clements, W. " A new approach to loudspeaker damping " Audio Eng. 35.8 (Aug. 1951)Acous.
158. Hardy, H. C. and H. H. Hall " The transient response of loudspeakers " (summary) J.
Soc. Am. 20.4 (July 1948) 596.
159. Sleeper, M. B. and C. Fowler " The FAS audio system " FM-TV 10.10 (Oct. 1950) 22 ; 10.11
/t". (Nov. 1950) 31 ; 10.12 (Dec. 1950) 24 ; 11.2 (Feb. 1951) 32 ; 11.3 (Mar. 1951) 46 ; 11.5 (May
1951) 34.
Allison, R. F. " Improving the air- coupler " FM-TV (1) 11.10 (Oct. 1951) 26 ; (2) 11.12 (Dec.
1951) 32.
160. Fletcher, H. " The stereophonic sound film systems -general theory " J. Acous. Soc.
Am. 13.2
(Oct. 1941) 89. (including
161. Maxfield, J. P., A. W. Colledge and R. T. Freibus " Pickup for sound motion pictures
stereophonic) " J. Acous. Soc. Am. 30.6 (June 1938) 666.
162. Koenig, W. " Subjective effects in binaural hearing " J. Acous. Soc. 32. Am. 22.1 (Jan. 1950) 61.
163. Goodfriend, L. S. " High fidelity " Audio Eng. 34.11 (Nov. 1950) of dosed rectangular
164. Meeker, W. F., F. H. Slaymaker and L. L. Merrill " The acoustical impedance
loudspeaker housings " J. Acous. Soc. Am. 22.2 (March 1950) 206. Kc " Audio Eng. 34.1 (Jan.
165. mit 3B and W. T. Selsted " A loudspeaker for the range 5 to 20
166. Beranek, L. L. " Enclosures and amplifiers for direct radiator loudspeakers " Paper presented
at
the National Electronics Conference, Chicago, Illinois, Sept. 26, 1950. and T.V. News 44.5
167. Plach, D. J. and P. B. Williams " A 3- channel unitary loudspeaker " Radio
(Nov. 1950) 66. and T.V.
168. Olson, H. F. and A. R. Morgan " A high -quality sound system for the home." Radio
News 44.5 (Nov. 1950) 59.
169. Moir, J. "Stereophonic sound " W.W. 57.3 (March 1951) 84.
170. Pipe, D. W. " Sound reinforcing " W.W. 57.3 (March 1951) 117.
171. Peterson, A. P. G. " Intermodulation distortion " G.E. Exp. 25.10 (Mar. 1951) 1.
1 : 3 (Oct. 1946) 121.
172. Shorter, D. E. L. " Loudspeaker transient response " B.B.C. Quarterly Am. (Jan. 1947) 146.
173. Sordoni, P. G. " Asymmetrical vibrations of cones " J. Acous. Soc. 19.1
174. Powell, T. (letter) " Transients " Audio Eng. 35.9 (Sept. 1951) 12. Sons Inc.,
175. Kinsler, L. E. and A. R. Frey (Book) " Fundamentals of Acoustics " John Wiley &
New York ; Chapman & Hall Ltd., London, 1950. 3rd ed. 1949.
176. " Reference Data for Radio Engineers " Federal Telephone and Radio Corporation,
177. Frayne, J. G. and H. Wolfe (Book) " Elements of Sound Recording " John Wiley & Sons Inc.,
New York ; Chapman & Hall, London, 1949.
178. Sabine, H. J. " A review of the absorption coefficient problem " J. Acous. Soc. Am. 22.3 (May
1950) 387. "
179. Corrington, M. S. and M. C. Kidd " Amplitude and phase measurements on loudspeaker cones
Proc. I.R.E. 39.9 (Sept. 1951) 1021. Group on
180. Beranek, L. L. " Design of loudspeaker grilles " Newsletter of I.R.E. Professional
Audio, Vol. 2, No. 4, July 1951. Reprinted Radiotronics 17.2 (Feb. 1951). 1950) 27.
181. Schneider, A. W. " Sound reinforcing systems " Audio Eng. 34.11 (Nov. 35.9 (Sept. 1951) 18.
182. Youngmark, J. A. " Design data for a bass -reflex cabinet' Audio Eng.
183. Moir, J. " Developments in England -A discussion of the latest trends
1950)
in
26.
the design of English
loudspeakers for high fidelity applications " FM -TV, 10.7 (July
184. Williamson, D. T. N. (letter) " More views on loudspeaker damping " W.W. 53.10 (Oct. 1947) 401.
Replies 53.9
185. Langford -Smith, F. (letter) " Loudspeaker damping " W.W. 53.8 (Aug. 1947) 309 ;
(Sept. 1947) 343 -4 ; 53.10 (Oct. 1947) 401 -2 ; also Ref. 186. 1947) 487.
186. Voigt, P. G. A. H. (letter) " Loudspeaker damping " W.W. 53.12 (Dec.
187. Geppert, D. V. " Loudspeaker damping as a function of the plate resistance of the power
output
tube " Audio Eng. 34.11 (Nov. 1950) 30. Chapter of " Radio Engineering
188. Knowles, H. S. " Loud-speakers and room acoustics " being 16
Handbook" edited by K. Henney, McGraw -Hill Book Co., 4th ed., 1950.
189. Canby, E. T. " Record Revue -The RJ speaker " Audio Eng. 35.10 (Oct. 35.12 1951) 23.
90. Joseph, W., and F. Robbins " The R -J speaker enclosure " Audio Eng. (Dec. 1951) 17.
" Audio Eng. 35.12
191. Wentworth, J. P. " Loudspeaker damping by the use of inverse feedback
(Dec. 1951) 21.
192. Yeich, V. " Listening room design " Audio Eng. 35.11 (Nov. 1951) 28. 35.11 (Nov. 1931) 34.
193. Olson, H. F. " Direct radiator loudspeaker enclosures " Audio Eng.1951) 40.
194. Kramer, K. " A three- channel loudspeaker' FM -TV 11.9 (Sept.
195. Kobrak, H. G. " Auditory perspective -a study of the biological factors related to directional
328.
hearing " J.S.M.P.T.E. 57.4 (Oct. 1951)
system
196. Beranek, L. L., W. H. Radford, J. A. Kessler and J. B. Wiesner " Speech-reinforcement
evaluation " Proc. I.R.E. 39.11 (Nov. 1951) 1401. W.W.
197. Parkin, P. H., and J. H. Taylor " Speech reinforcement in St. Paul's Cathedral "
58.2
(Feb. 1952) 54 ; 58.3 (Mar. 1952) 109.
198. Lowell, H. H. " Motional feedback " Elect. 24.12 (Dec. 1951) 334. 36.2 (Feb.
199. Childs, U. J. " Loudspeaker damping with dynamic negative feedback " Audio Eng.
1952) 11.
200. Clements, W. " It's positive feedback " Audio Eng. 36.5 (May 1952) 20.
201. Childs, U. J. " Further discussion on positive current feedback " Audio Eng. 36.5 (May
1952) 21.
Additional references will be found in the Supplement commencing on page 1475.
CHAPTER 21
(1)
where I9 = signal plate current in amperes, r.m.s.
and RL = load resistance reflected on to the primary of the transformer.
The signal plate current is given by
I, = µE, /(r,, -1- R1) (2)
where is = amplification factor of power output valve
E, = signal grid voltage, r.m.s.
and r, = effective plate resistance of power output valve.
Consequently, at the bass resonant frequency where the resistance R j may rise to
a high value, it is possible to have maximum plate circuit efficiency and yet to have
880
21.1 (ii) OPTIMUM PLATE RESISTANCE 881
less than the maximum electrical power output, because the signal plate current is
reduced by the high load resistance. This effect is most pronounced with low values
of plate resistance, and with triode valves or negative feedback the loudspeaker
acoustical output may actually fall at the bass resonant frequency. With pentodes
without feedback the reverse occurs, since a pentode approximates to a constant
current source, and there is a high peak of output at the bass resonant frequency.
A similar effect occurs at the higher frequencies, the highest level of response
occurring with pentodes, an intermediate value with triodes having RL = 2r9, and
the lowest with constant voltage at all frequencies (a condition approached when a
high value of negative feedback is used).
For any one loudspeaker, there is a value of plate resistance which pro-
vides most nearly constant response at all frequencies.
Reference 5.
Fortunately, owing to the characteristics of music, the input voltage applied to the
grid of the power amplifier at high frequencies will normally be less than the maximum.
This somewhat reduces the distortion at high frequencies.
Pentodes with negative voltage feedback, when operated well below maximum
power output, are less affected by the loudspeaker impedance variations than pentodes
without feedback. If the feedback is large, the output voltage remains nearly con-
stant irrespective of the loudspeaker. There is an optimum value of amplifier output
resistance to give most nearly uniform response at all frequencies with any one loud-
speaker.
When operated at maximum power output with the small degree of feedback usual
in radio receivers, the load resistance is critical. Satisfactory operation with low
distortion can only be obtained by reducing the grid input voltage.
With a large amount of feedback, as for a cathode follower, the impedance at 400
c/s should be the nominal value for the valve. The increase of impedance at other
frequencies will not then cause distortion. See also Chapter 7 Sect. 5(iv).
FIG. 21.1
Alternatively the parallel loudspeakers may be connected across a 500 ohm (or any
other convenient value) line from the secondary of transformer T, as in Fig. 21.2.
Here T, will have a 500 ohm secondary and will reflect an impedance R L into the
primary. T, and T, will have nominal impedances each 1000 ohms. If there are
N loudspeakers in parallel, each will have a nominal impedance of N x 500 ohms.
21.2 (i) MULTIPLE LOUDSPEAKERS- GENERAL 883
An alternative arrangement, which has much to commend it, is the use of an ex-
tension speaker with an impedance of about twice that of the speaker in the set. This
means that the extension speaker will operate at a lower sound level than the one in
the set, but the mis- matching will be less severe and the maximum volume obtain-
able from the set will not be seriously affected.
If negative feedback is used in the receiver, the effect of the connection of the
external speaker on the volume level of the loudspeaker in the set will not be very
noticeable at low or medium levels, although the same problem arises in regard to
the overload level.
FIG. 21.5
Fig. 21.4. Extension loudspeaker with provision for switching either loudspeaker alone.
Fig. 21.5. Extension loudspeaker connected to the voice coil circuit.
It is possible to modify the previous arrangement by means of switches which can
open -circuit the voice coil of the first speaker and at the same time close a switch in
the primary circuit of the second speaker. This is illustrated in Fig. 21.4 from which
it will be seen that only one speaker will be operating at the one time and there is
therefore no problem with correct matching. With this arrangement, the extension
speaker should have the same impedance as the one in the set and the power input
to both speakers will then be equal. The two switches S, and S2 could, of course,
be combined into a single wafer switch.
The previous arrangements have all adopted an extension from the primary of the
loud speaker transformer and therefore at a high impedance. There are advantages
to be gained in using the voice -coil circuit for the extension, as illustrated in Fig. 21.5.
This avoids the necessity for a step -down transformer on the extension speaker and
for a blocking condenser. The set manufacturer may fit two terminals on the loud-
speaker housing, connected to the voice coil, as an alternative to Fig. 21.3. To obtain
the same power from both speakers, it is necessary for the voice -coil impedances to be
equal. If it is required to have one speaker operating at a higher level of sound than
the other, the impedance of the second (lower output) speaker should be made higher
than that of the first speaker. In such a case it is possible to obtain correct matching
by calculating the impedance of both voice -coils in parallel and selecting a step -down
transformer to suit. Very few voice-coils have an impedance less than about 2 ohms,
so that it is possible to use quite ordinary wiring in the connections of the extension
speaker. If the second speaker is to be situated more than say 10 feet from the first
speaker, it may be desirable to use heavy wire, such as power flex (twin plastic is very
convenient). In some cases it may be desired to operate the extension speaker at a
rather lower level than the first speaker, in which case losses in the extension line may
be desirable.
A single-pole double -throw switch may be used to change over from one to the
other voice -coil as shown in Fig. 21.6 (A). Here switch S is used to open- circuit
the first voice -coil and at the same time to close the circuit to the second voice -coil.
The further refinement of a series volume control R is shown in the extension speaker
circuit, so that the volume may be reduced below that of the speaker in the receiver.
The resistance R should have a maximum value about 20 times that of the voice-coil
impedance, but even so this arrangement cannot be used to reduce the volume to zero.
In order to have a complete control over the volume from the extension loudspeaker,
the arrangement of (B) may be used in which R is a potentiometer with the moving
contact taken to the extension speaker.
21.2 (iii) EXTENSION LOUDSPEAKERS 885
The series resistor volume control shown in Fig. 21.6 (A) increases the effective
impedance of the extension loudspeaker circuit at low volumes, but this is not a
serious detriment since the volume will be low and the mis-matching of only minor
importance, particularly if negative feedback is used. The potentiometer method of
volume control in Fig. 21.6 (B) has to be a compromise, and is incapable of giving
satisfactory matching under a wide range of conditions. A reasonable compromise
for the resistance R would be about five times the impedance of the voice -coil, but
this will result in appreciable loss of power even at maximum volume. For perfect
matching, resistance R should be taken into account, but for many purposes the effect
(B)
FIG. 21.6
may be neglected provided that R is not less than five times the voice -coil impedance.
At low levels this arrangement has a high impedance but here again the effect will
not be serious.
Another form of volume control is a tapped secondary winding, with the extension
speaker connected through a rotary switch to a selected tap. Taps may be made
with each one 70% of the turns of the one above it, giving roughly 3 db steps, or
80% for 2 db steps. This method causes serious mis- matching, and is not advisable
unless there are several main loudspeakers and only one " extension " with volume
control.
The best method of volume control is a L type level control (Fig. 21.11) which
provides constant input resistance and may be continuously variable (see Chapter 18).
All the preceding arrangements are limited to the use of either of two speakers or
have given a choice between one and both. The ideal arrangement is to permit
the use of one or other or both. This is particularly helpful when the extension
speaker is used in a different room and one may wish to operate the extension speaker
alone. Tuning-in may be done by switching over to both speakers, adjusting the
volume level to suit the extension speaker, and then switching over so that only the
extension is in operation. Then, if at any time the speaker in the receiver is required
to operate, this may be done simply by moving the switch without causing any inter-
ference to the extension speaker.
eterence Z11.
Although 500 ohms is a popular value, any lower or higher value may be used pro-
vided that the resistance of the line is not too great, and that the capacitance
across
the line does not seriously affect the high frequency response.
A very popular line impedance is 600 ohms.
21.3 LOUDSPEAKER DIVIDER NETWORKS 887
The number of turns may be calculated from the formulae of Chapter 10 Sect.
2(iv)A or alternatively from curves (Ref. 11 for 17 A.W.G. D.C.E.).
Iron cores introduce distortion, and are therefore undesirable. However if iron -
cored inductors are used, an air gap of 0.008 in. or 0.010 in. should be provided
(Ref. 6 p. 101). Ferrite cores present possibilities for this application.
In the filter type arrangement which is also popular, the two inductors and the
two capacitors have different values, and the nominal attenuation is 12 db per octave
(Fig. 21.8). Taking the design constant m = 0.6, we have
Lo = Ro/(271 ,) Co = 1 /(2nf,Ro)
L, = Ro /(3.27fr) C, = 0.8 /(nf,Ro)
where L is in henrys and C is in farads (Refs. 6, 19).
In all cases the series connection is preferable to the parallel connection. Resistance
in the inductors has a slight effect on the attenuation at the cross -over frequency, while
it also introduces insertion loss (0.3 to 1 db). Care should be taken to keep the re-
sistance as low as practicable. The usual position is between a single power amplifier
valve (or two in push -pull or parallel) and the loudspeakers. The insertion loss is
therefore a loss of maximum output power. The divider network may be connected
either on the primary or secondary side of the output transformer, the latter being
more usual- in this case the transformer must be suitable for the total frequency
range of both units.
888 LOUDSPEAKER DIVIDER NETWORKS 21.3
If the dividing network is placed on the primary side, each of the two output trans-
formers is only called upon to handle a limited frequency range, and may be of cheaper
construction. One interesting application is Fig. 21.9 (Refs. 9, 16). For a plate -
to -plate load of 5000 ohms and a cross-over frequency of 400 c/s the component
values are :
L, = 2.0 henrys (series aiding) ; inductance T1 primary = 50 henrys min. ;
L, = 1.0 henry (with air gap) ; T, primary inductance = 2.0 henrys (with air
gap) ; C1 = 0.16 µF : C, = 0.04 µF : leakage inductance of T1 not over 1
henry : leakage inductance of T2 not over 0.05 henry.
FIG. 21.9
It has been shown that the arrangement of Fig. 21.9 produces less distortion than
the conventional circuit : this is due to two features. Firstly, the low -pass filter
Li, 2C, attenuates high frequencies propagating in either direction ; as a result
the harmonic components of magnetizing current in T1 have less effect on the high-
pass channel. Secondly, the low -pass filter attenuates high frequencies on their way
to T, so that the transformer is only called upon to handle a limited range of fre-
quencies, thereby reducing the distortion (Ref. 9).
In the case of separate amplifiers for low and high audio frequencies, the divider
network preferably precedes the amplifiers. This permits separate attenuators to
be used for each channel ; these may be used for balancing the two units, or for tone
control.
If the high frequency unit is more sensitive than the low frequency one, a simple
form of fixed attenuator may be incorporated (Fig. 21.10). Here
R1 + R, = Ro and R, /R, = (El /E,)
where R, = R2 R, /(R,
-
1
R0).
For example, to give (E1 /E,) = 2 (i.e. an attenuation of 6 db) :
R1 = ¡R, and R, = R,.
If it is desired to use this as a variable attenuator to give tone control, it should pre-
ferably be of the constant input impedance type such as the L pad [Fig. 21.11, also
see Chapter 18 Sect. 3(iii)]. If a simpler type must be used, it may be designed as
for Fig. 21.10 to give the correct input impedance at the normal operating position,
but with a sliding contact. It will then have an incorrect impedance at any other
21.3 LOUDSPEAKER DIVIDER NETWORKS 889
setting ; the effect is only slight if the sensitivity of this unit is very much greater than
the other, as happens with horn and direct- radiator assemblies.
If one loudspeaker unit has a lower impedance than the other, an auto -transformer
may be used to provide correct matching. This may not be necessary if an attenuator
is used on one unit.
FIG. 21.11
It is important to check for the phasing of the two loudspeakers ; in the vicinity of
the cross -over frequency both units should be in phase (aiding one another). This
may be checked by the use of a single dry cell connected in series with the secondary
of the transformer ; in the case of 2 ohm voice coils there may be a 5 ohm limiting
resistor connected in series with the cell.
References 1, 2, 6, 9, 11, 12, 16, 19.
SECTION 4 : REFERENCES
1. Sowerby, J. McG. " Radio Data Charta 10-loudspeaker dividing networks " W.W. 49.8 (Aug.
1943) 238.
2. Sieder, E. N. " Design of crossover networks for loudspeaker units " Q.S.T. 28.12 (Dec. 1944) 35.
4. " Matching transformers for loudspeakers " Philips Tec. Corn. 78 (Nov.-Dec. 1940) and 79 (Jan. -
Feb. 1941).
5. Stanley, A. W. " The output stage -effect of matching on frequency response " W.W. 52.8 (Aug.
1946) 256.
6. Terman, F. E. " Radio Engineers' Handbook " (McGraw-Hill Book Company, New York and
London, First edition 1943) pp. 249-251.
7. Langford -Smith, F. " The relationship between the power -outout stage and the loudspeaker "
A.W.A. Tec. Rev. 4.4 (Feb. 1940) 199. Originally published in Proceedings of World Radio Con-
vention, Sydney, April 1938 (Institution of Radio Engineers, Australia). See also Ref. 7a.
7a. Langford- Smith, F. " The output stage and the loudspeaker " W.W. 44.6 (Feb. 9, 1939) 133 ;
44.7 (Feb. 16, 1939) 167.
8. R.C.A. " Application Note on receiver design-output transformer " (variation of impedance with
current) No. 75 (May 28, 1937).
9. Klipsch, P. W. " Low distortion cross -over network " Elect. 21.11 (Nov. 1948) 98.
10. Langford- Smith, F. " The design of a high fidelity amplifier -(1) The power valve and the loud-
speaker," Radiotronics 124 (March -April 1947) 25. (2) " Negative feedback beam power amplifiers
and the loudspeaker," Radiotronics 125 (May -June 1947) 53.
11. McProud, C. G. " Design and construction of practical dividing networks " Audio Eng. 31.5 (June
1947) 15.
12. Schuler, E. R. " Design of loudspeaker dividing networks " Elect. 21.2 (Feb. 1948) 124.
13. McProud, C. G. " Two-way speaker system " Part 3 Audio Eng. 32.2 (Feb. 1948) 21.
14. Angevine, O. L. " Impedance matching " Audio Eng. 31.11 (Dec. 1947) 20.
15. Jonker, J. L. H. " Pentode and tetrode output valves " Philips Tec. Corn. 75 (July 1940) 1.
16. Klipsch, P. W. " Woofer -tweeter crossover network " Elect. 18.11 (Nov. 1945) 144.
18. Amos, S. W. "Feedback and the loudspeaker" W.W. 50.12 (Dec. 1944) 354.
19. Hilliard, J. K. " Loudspeaker dividing networks " Elect. 14.1 (Jan. 1941) 26.
20. " Impedance matching and power distribution in loud speaker systems " Technical Monograph
No. 2, Jensen Radio Mfg. Co.
21. Smith, B. H. " Constant-resistance dividing networks " Audio Eng. 35.8 (Aug. 1951) 18.
22. White, S. " Design of crossover networ.s" FM-TV 12.1 (Jan. 1952) 42.
Additional references will be found in the Supplement commencing on page 1475.
CHAPTER 22
AERIALS AND TRANSMISSION LINES
BY W. N. CHRISTIANSEN, M.SC.
Section Page
1. Introduction 890
2. The transmission line ... 890
3. Aerials and power transfer 892
4. Characteristics of aerials ... ... ... 893
5. Effects of the earth on the performance of an aerial 894
6. Effects of the ionosphere on the reception of radio signals 901
7. The impedance of an aerial ... 901
8. Dummy aerials ... 906
9. Types of aerial used for broadcast reception 907
10. References ... 911
SECTION 1 : INTRODUCTION
Aerials and transmission lines differ from simple electrical networks in that their
inductance, capacitance and resistance are not lumped but are distributed over dis-
tances such that the time required for electrical energy to travel from one part to
another has to be taken into account. In a single chapter it is, of course, impossible
to attempt to give a theoretical treatment of these devices. What will be done will
be to present some useful formulae, a few physical pictures of the processes which
occur and some results of practical experience with aerials and transmission lines.
(i) Introduction
A transmission line consists of an arrangement of electrical conductors by means
of which electromagnetic energy is conveyed, over distances comparable with the
wavelength of the electromagnetic waves, from one place to another. The theory
of transmission lines provides a link between circuit theory and the field theory of
electromagnetic waves inasmuch as the properties of such lines may be determined
either from the picture of a transmission line as a filter network with an infinite number
of elements, or from the picture of electromagnetic waves guided between (usually)
a pair of conducting surfaces.
Some of the properties of transmission lines will now be given.
890
THE TRANSMISSION LINE 891
long. The importance of this characteristic impedance lies in the fact that if any
length of line is terminated in an impedance of this value, then all the energy flowing
along the line is absorbed at the termination and none is reflected back along the line.
A result of this is that the input impedance of any length of transmission line terminated
in its characteristic impedance is equal to the characteristic impedance. At radio
frequencies, the characteristic impedance of all normally-used types of transmission
line is almost purely resistive.
The value of the characteristic impedance for a !ow -loss line is
Zo = VL /C (1)
where L and C are the distributed inductance and capacitance per unit length of the
line. The velocity of propagation of electromagnetic waves along such a line is
y = 1 /VLC (2)
With air as the dielectric, y = 3 x 108 metres per second, which is the velocity of
light*. For other dielectrics
v = 3 x 108 /\/K metres per second (3)
where K = dielectric constant.
In cables with polythene dielectric, for example, K = 2.2 and y is therefore 0.67
of the velocity of light. This means that the wavelength of the waves in the cable
is only 67% of the wavelength in air.
For a low -loss transmission -line the characteristic impedance may be found if C
and L are determined. Actually it is not necessary to determine both-one is suffici-
ent.
If we combine (1) and (2) we obtain
Zo = 1 /v C = vL. i4)
For a transmission line consisting of a pair of parallel conductors
276 d
Zo logro ohms (5)
VK
where d is the distance between the centres of the conductors and r is the radius of
each conductor.
For a co -axial line
138
Zo
where rt is
=
K rtlogro
the inside radius of the outer conductor and r, is the outside radius of the
(6)
inner conductor.
(i) Introduction
The function of a receiving aerial is to collect electromagnetic energy which is pass-
ing through the space surrounding the aerial, and to pass this energy into a radio re-
ceiver.
at a receiver when the frequency of the emitted wave is (a) 600 Kc /s and (b) 150 Mc /s.
We assume that the energy radiated in each case is 1000 watts and the distance be-
tween transmitter and receiver is 20 kilometres.
On applying the formula (13) we find that in the medium -frequency case, the power
available at the receiver is 8.8 milliwatts, while in the v -h-f case the power available
is only 0.142 microwatts.
This means that if both signals are to be amplified to produce, say 1 watt at the de-
tector then the amplification required in the first case is approximately 20 decibels
while in the second, it is nearly 70 decibels.
These figures partly explain why a crystal receiver is very effective for long -wave
reception, whereas a receiver required for v -h -f work usually has a large number of
stages of amplification.
It will be seen from the table that the effective area for capture of electromagnetic
energy by a short dipole is approximately equal to an area bounded by a circle which is
at a radial distance of X /1.6n from the dipole. The physical significance of this is
that such a circle very roughly forms the boundary between the region in which the
local induction field of the aerial predominates and the region in which the radiation
894 (i) EFFECTIVE AREA OF A RECEIVING AERIAL 22.4
field is the major component. Hence we can picture the aerial as capturing the energy
which falls within the region in which the induction field of the aerial is of significant
magnitude.
A broadside array of dipoles with a reflecting curtain can absorb all the energy that
falls on it. Without the reflectors, it absorbs only half the incident energy and radiates
one quarter back towards the transmitter and one quarter in the opposite direction.
A similar effect occurs with a dipole. If it absorbed all the energy available to it
and reradiated none, then its effective area would be double the value given in Table 1.
A reflector is required to achieve this.
OF AN AERIAL
(i) Introduction (ii) A perfectly- conducting earth (iii) An imperfectly- conducting
earth (iv) The attenuation of radio waves in the presence of an imperfectly -reflecting
earth.
(ï) Introduction
In the previous section the effect of reflecting surfaces has been disregarded. In
practice, these effeVts must be taken into account, except in rare cases. The two
" surfaces " to be considered are the earth and the ionosphere. We deal here with
the effect of the earth.
In Fig. 22.1 are shown the effects of the earth on the field pattern of a short vertical
aerial and of a short horizontal dipole. It will be seen that the earth has no effect on
the field pattern of the vertical aerial apart from removing the lower half of the pattern.
With the horizontal aerial, however, there is complete cancellation of the waves in
the horizontal plane and the resultant effect is zero.
The conclusion may be drawn from the diagram that where waves are arriving at
the aerial in directions parallel to the plane of the earth, then one would expect a
vertical aerial to be very effective while a horizontal aerial would be ineffective. (This
applies to a transmitting as well as to a receiving aerial). It will be seen later that this
conclusion does not fit the facts, in all cases, because the earth is not a perfect con-
ductor. At medium and low radio -frequencies, however, it is in agreement with the
facts-- horizontal aerials are of no use in communicating between two points both close
to the earth.
When waves are arriving at the aerial in directions inclined to the horizontal (as,
for example, in the case of short -wave communication via the ionosphere) then either
horizontal or vertical aerials may be used. If the distance between transmitter and
receiver is not great, however, then the waves will be arriving at nearly vertical in-
cidence and the horizontal aerial will be superior, provided that it is placed at a suitable
height above the earth. The height at which the aerial should be placed above the
earth depends on the angle of elevation of the waves which are to be received. The
height is different for horizontal and vertical aerials.
In Fig. 22.1, for an angle of elevation B, the distance d between the path lengths of
waves to the aerial and to its image should be equal to an even number of half-waves
in the case of a vertical aerial, and an odd number of half-waves in the case of a hori-
zontal aerial. Simple trigonometry shows that this corresponds to a height h given by
h = n0/(4 sin 0) (20)
where n is an even integer for a vertical aerial and an odd integer for a horizontal aerial.
(A)
Field Intensity
(o)
Angla of
direction of roy
FIG.22.I
(A) and
Fig. 22.1. Effects of the earth on the field pattern of a short vertical aerial and in (B)
of a short horizontal dipole (B). Both (A) and (B) are cross- sectional views,indicates
-
the sign + indicates direction of current into the paper while the sign short horizontal
the
opposite direction. Polar diagrams of a short vertical aerial and a
dipole, both close to earth, are shown in (C) and (D) respectively.
896 (iii) AN IMPERFECTLY -CONDUCTING EARTH
22.5
(iii) An imperfectly-conducting earth
The propagation of radio-waves close to the surface of a finitely conducting earth
has been studied theoretically for many years. Zenneck first produced a solution
to the problem, and later Sommerfeld gave a more accurate analysis. Unfortunately,
the incorrectness of Zenneck's analysis combined with a small error in Sommerfeld's,
created some confusion that has persisted until the present day. As a result
of
errors, it appeared that a special type of wave was propagated in the vicinity of the
ground- plane. This wave was called a " surface- wave." It is known now, that the
Zenneck " surface-wave " does not exist, but the name still appears in the literature this
on wave propagation. It is applied to the waves that travel between transmitting and
receiving aerials when both are close (in terms of a wavelength) to the surface
of the
earth, as for example, in medium-wave broadcasting. When the aerials are raised
several wave -lengths above earth, then the waves travelling between them
are
" space-waves." These terms are artificial, but they serve some purpose in thatcalled they
indicate that the conditions of propagation between two aerials placed close to the earth,
are markedly different from those between two aerials elevated several wave
-lengths
above the earth's surface (as in v -h -f broadcasting).
In the conditions of propagation to which the term " surface-wave " is applied,
vertical aerials are superior to horizontal aerials. Thus vertical aerials are always
used in medium- frequency broadcasting. (The horizontal portion of a medium
frequency receiving or transmitting aerial plays no useful part, except to increase -
the
efficiency of coupling to the receiver or transmitter).
When the aerials are raised several wave -lengths above the earth, as is usually the
case in v -h -f communication, then horizontal and vertical aerials have roughly
equal
effectiveness.
When waves are not arriving at the aerial in a horizontal direction, then the differ-
ence between an imperfectly conducting earth and a perfectly conducting
one is not
so marked. Hence for short -wave communication one can treat the earth as
a perfect
conductor without introducing serious error in the calculations.
' S'`41111111111111111I1111111111I1111111111111111111111111111111111111111111111I11111111
2 11 h.''! I!:; :!!!IIIIIIIIIIIIIIIIIIIIIIIIII11111111111111111111111111111111111111111
16e 1111111\:_
I II
I I I I II I I I I I I I I II I I 111111I I I
I I I I I I II I I I II I
I I I I I I I I! lö
I I I I II I I I I I I II I I I II I I I
s 11111111111tic= z :S!l111111111111111111111111111111111111111111111111111111111111! s
2 n11I111I1uIUIhi:.. ;;ull,lll1u1111II11I1uIUIUIIIIIIIII000III111I1I11
Ia
2
IIIIIIIIIIIIIIIIIIIIIII111II1i;
-111111111111111111111111111111111111111iVM§RV!1llllwn ,p
`ìiu
.IIIIIIIIIIIIIIIIIIIIIIIIi'`: :!1: `!ì;lIIIIIIi1111II111111111111111I1I1IlI1II1111111
III'IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
11111111111111111
2
IOt
I IInII111IIlIIIIII11nII1IIIIIIIIIIIII1I11111111111IIIIIIIIIìiliamllimIII:'
5,8á 11I11111111111111111111111111111111111111111I111111111111111I1I1:1ill;lllh!: 2
V 51111111111111111II111111I111111111111I1I11I11111I1II11111I11Il1lllIli.qlli;.
10
á 1111111111111111 I 1111111111111111111111111111111111111111111111111111110i:
10 11111111111111111111111111111111111111111111111111111111111111111111111111IIi\
n11111111111111 I1I111I11111111I1111II1II111111111111111IIlIIl11I111I1111111111; , vi! S
21111111=IIIIIIII I 1111111IIIIIIIII1111111111111111IIIIIIII111111II111111111111111i, i
,Ie 1111111 2
s II1111 IIIIIII IIIIIIIIIIIIIIIIIIIIIIIII11111111 1111IIIIIIIIIIII1111111111II1111II11
IÓe
16' ÌIIÌIÌ
0
i11111111Ì1II111111ÌI ÚIIIIIÌIÌ1IÌIII I1IIÌÌIIÌÌIIII11111ÌI11IWIÌIÌÌIIII
40 50 708090N70 30 60
2 3
-- 5 6 7 8 9 10
22.
Distance
20
Miles FIG 2
s 11111.!11II11IIIII111u':9111111III111111111m111In1111RI1II1111I1111I II 1111111! s
2 :!111111ilIIIIIIIIIIIIIIIIIIii4 9II111I111111111111IIIIIIIIIIIIIIIINI II 1111111
111;1111111IiVIII1111n111II1I11II11i::111111111111I111IIIIIIIII111IIII I11111111
2.
IIIIì.1I1III119111I1IlI1II1III111111I1ì:!1n111II11111II1111111I11ll11I 111111
2 Ic111I1ì!IIIIIIIII lilIIIIII IIIII111111I111111ìi:!! IIIIIIIIIIIIIII IIIIIIIIIÌ;l11111 o'
Io11I1.91111111i 1I11111111i!1u1111111111111I111111111I1di:!11 III1111 IIIN
5.51111:!11111111G:!11111111!1111I1111111111111111111111111111é; III !
°°'
-
Ia.,111`i1!lI1IIIIIIIII11Ih,,1111II1111111111IIIIIIIII IIii I
111,111111;i!III11111ñ!IIIIIIIIIIIIIIIIII1111111I111IIIIIIIHIIÌÌIIII:.._ 1
11llllai:!IIIIIGii!IIIIInì:iIlllllll!IIIIIII 111111I11111111 m1u11111m1
11111111ÌiNii!lIII11'!IIIIINiil111111G111 HIIIIIIIIIilI11IÌÌIÌIIIIIlIl111I1111
I11I11111111n;!11;10Ìlìì:lIIIIIIIì11111I IIIIIIIIIIIIIIIIIIIIIIIII111111111
IIIIll111111111IIGRì:1111111'111111:111III; 1111111111III111111I111111111111111!
>r 2 C111111111111111111111i:li!!IIÌIÌììy11111i11111,111111111111111111111111111111111111
-. IIL,1111NU11I1I11111ui1u11I.a1N1`n1i!1nI1111111I1uu111u1II1111
° Ilil1i11111111111111111111111i:!Ii111B'1II1, 11111i!1111111111111111IIIIIII1111111
; 21111111fiN1111IIIIIIIIIIIIIIIIIIIINÌ!IIIISt1111 IIIIii!IIIIIIIIIIIIIIIIIIIIIIIIII
1111111111111:111111111111N1111111n111i!111. 1lli!111111h:11111111111I1II11111111
II
4 l
S
1111111111111111.1111111NIIIIIIIIlIV;1:1Iip5lllll0I11III0I111111I1II1I1111II! 6
1;. =11111111111111111111`2NIIUIINIII1111NI1Cl'91Ì0I11111ú!11111I11111111111111111
b5'11I1111111111111III11N1:P111YIII1111fl11IIloollllìi!IIIIIIì1111G!IIIIIIIIIIIIIII
1 2
11111y111111:111111:1111111111111
2 Recririn, Aerial °abort Earth ImIiI111111111al911l11110IIIIII111111111 IÓ
1I111111111mI11811fì1111111111M
a
IO
5 flñllln11II1n11I1IIU.;11111111111I111i:1iIII,1II1Ii1111111111!
ñIM.9111111111IIIIIIIN4011111091111:!111111\1111111 s
111111111111111111111NIIIIIIIIIINIIIÌ111ì1nllllllllllllnllì:Illllì;!11111111
11111111N1111111111111111111111111111111111111IIIIII11111111111111PIIIIh111111
pro 2 3 4 5 6 7 6 910 20 30 40 SO 60 7O60901Ó0
Distance -Miles FIG. 22.3
Fig. 22.3. Effect of raising the receiving aerial above the earth (1 KW at 45 Mc /s
radiated over earth with conductivity o = 10 -18 e.m.u. and transmitting aerial 30 feet
above the earth). The same curves also apply when the transmitting and receiving
aerials are transposed.
Fig. 22.3 shows the effect of raising the aerial above the earth* for a frequency of
45 Mc /s, and indicates that unless the aerials are close to the earth, horizontal and
vertical polarizations have practically equal effectiveness.
Fig. 22.4 and 22.5 for frequencies of 75 and 150 Mc /s were derived by T. L. Eckers-
ley for vertically polarized waves, but may be used also for horizontally polarized
.
waves with little error when the aerials are at heights above earth corresponding to
normal practice.
10
3 ìmIIIIIInllllifll INIIIIIIIIIUluI111111111111111nII111nI111I11Imlllllllu1111
2
n11
11lI1ì1..!IIIIIIIII IIIII IIIII III I111111II1111I I IIII IIIIIIII II IIIII IIIII lil /-794o I
I I
iÓ 1IIIIIIIUUIIi!lIIIIIIIIIIIIIIIIIIU11111111111111111111111I1111I11uuuu11I11111
s nI111IIIIIIIIInII11Ili::111111n11,1n11111111111111111I111111111111111II11u11n1111.
IIIIIIIIIIIIIIIIIIIIIIIIIIIIiiin.IIIIIIllllllllIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
2
IÓ I/Illllllll 111111t11111111111III111u:l11I11111111111111111111111111111111111111111
mnuulll1I1111uuu11lllul
IIIIIIIIIIIIIIIIIIIIIU111111111U1I11puM¡n
2
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIi;.. IIIIIIIIIIIIIIIIIIII!
1IIIIIIIUIIIInllñ5ul1111111I1111IIIII11111111I111I111u,.. 11111111111111111111
IÓ
s 11111II000UUUII;:IIIIIIIIII1111111111111111111111111111111ì::!nn11;1111111u!
IIIII1111I111111IIIIIIIÌi!1111111111111111111111111111111111111111111iii11!111111
1
E
á 1IIINIIlI111111111111uI1111ii!I11111111111111111111IIIIIIIIIIIIIIIIIIIIIIIIIIì::l1
51NIIIIIIIIIIIIIi;!11111111111111111111111111111111111111111111111111111111111111111
IIIIIIIIIIIIIIIIIIIiìP111111111111111H111111111111111111111111111111111111111111111!
211111111101I11I11IIIH:lII11IIIIi1111i!111111111111IIIIIIIIUIIIII1I1111111111111 y
E '5 u111111111111ì!Illllllllllì:lllll ullllll . 1111111111111111111111111111111111111! _
2IIIIIIIIIIIIIIII10UlIIIIIIIÌi llllll111111 IIIIIIIIIIIII111111IIIIIlu11111111111 oIIIIIIIIIIIIIIIIIIIIIglIIIIIIIII:111111111111:11111111111111111111111111111111111112=
iO IIIIII111111IIIIIIIIIIIIIÌilII1111111:II111111111111111111III11111111IIIIIIIIIIIIIiI
1111111111IIIIIIIIIIIIIIIIIIIÌi !I 111111ill111I1ìlIIIIIIIIIIIIIIIIIIIIIIII111111I1
111
''' Wi
11111111111111IIIIIIIIIIIIIIIIIIII 211111l:1: 111111II:lIIIIIIIIIIIIIIIIIIIII11111111
in Feet
11_ I
11!' Height of Tran,mitring Aerial ::ÍII1111111111111111;11111111111111III1I11111111.
1
I._II1,111.1111111111111.111111111111111111111111
Il,rrommuuurmmumudlllllill, III11:1111111111I11111111111111I11111111 2
11111111111111111111111111II II IIII 1111111111111111:!II IIII1I111;1I I I II I I III I1111111
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII IIIIIIII111IIIIIIIIilIIIIIIIIII!IIIIIIIIIIIIIIIII!
I
.
1111111111111111111111111111111111 II IIIUlI111l:IIIIIIIIh:1111111G11111111111111111
111111111111111111111111111111111111111111111111101IIIIIIIIli:11111111IIII111111111l.
1111111111111111111111111111111111111111111111111111ó1111111111111111111111111111111
IIIIIIiIIUnInulu1u11I1uu1111111u11u1I111116n11n111au1n11:u111111111
2
442 2 3 3 6 7 1 9.10 20 30 40 50 60 70 e0 90
Distance -Mks FIG. 22-4 A,
10e
4 5 6 7 9 910 20 30 40 60 708090100
Height of Receiving Aerial °bore E°rth , in Feet FIG. 22.4
IÓ
5 ì:!1111A' II IIIIIII I11111111 II II IIII III IIIIII IIIIIi111111111 I I I I IIIII ll Il11ll
11 I
1111ìe,.11I11IIIIIIIIIIIII1111IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII111lIlIIIIIIf-I
2
I
I I II I II II I IIm ::!!11 1 11I 11111111 I 111111111111111111111I1111 I I 1 I I I I 1111111111111i IÓ
lo=
IlllluuuuuIIl
5 :I"IllI11WIIllnuun I
Imlllnml 5
11111III1II IIIII111!lIIÌIIÌIIi
b
2
l
1
li:!1
11111u11"111 11111111
I
I!lIII
III11111111111111
r
1111111111 111111111 2
5 IIIIIIIIII011111 IIIIIIIIII111111111111110.1 w Illlllllll1111 I II II 1 1 1111
IIII:IIIIIIIIIIIil1IIIIIIIIIIIIIIIIIIIIIIIIIIIII11iE IIIIIIIIIIIIIIIIIIIlió
11 IIIIIIIIIIIIIIIIi!IIIII,IIIIIIII IIIIIIIIIIIIIIIl111.tF 1111111111111111,11
IQ
5
'
11I11111111111111II1111i!111111111111111111111111111111111111111ì:!IIIIIIIIIII111111! 3
IIII IIIIIIIIIIIIIIIIIIIIIII11il1II I IIIIII IIIIIIIIIIIIIIIIII ì l!i-111111111
5
ö
IIIIIIIIlIIIIlI11111111111111111iu111111::!11
1111111111111111111111111111iiIIIII
c 2
IIIIIIII I gnmulmul;!lnuullllW.
1111I1I111111illIIIIIIIIIIIIInIIIIIIIIIIíIIIIIIIIIlIIIIIIIIIII IIIII IIII IIII Illl
11111111111111111111111111111111111111! 5t
11IIIIIIIIIIIIIIIi lIIIIIIIIIIIIIIi1 IIIII1:11111111111111111111111111111111111111
'°s 2LL
IIIIIIIIIIIIIIIIIIIIIi!11111111111111!1111 11111111111111111111111111111111111111111
2 11111111IIIIIIIIIIIIIIl1i;l1IIII11111111l1 I11ì;1IlIIIIIIIIIIIIIIIIIIIII1111111111!Io
` IIIIIIIIIIIIIIIIIIIIIIIIIIIG!lIllllllull
a
..ri..111_:.rilt. 11IU1umu11ummmnmllllun
5 IP 1
1l11111ì.1II1I11111111111111111111111! s
Hei9bt af TraasmRtin9 Aerial
2111 n f:et I!1111111111.
' 1lYuuummummnnummlllil 511111111C111111111111i!11111111111111111111111
Illlllllli. '1I1111111II1II1111111111111
iIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIlII,
1IIIIIì!IIIIIIIIIII11111111111111111! Io
11111111I111IIIIIIII 11u11111111I11I1111111111111111111 III 1 t 111111111 11
5 IIIlII11ulIIlI1111I1uI111111I1111I1I1I111111I.1I111116Illl 111 IIIIIIIIIlII11! 5
2111111II II IIIIII IIII II III I I III I II I111111111111111ì!1I IIII IIlI IIII I IÌIi! IIII1111111II
010
0 2 3 4 5 6 7 8 9 IO
Distance - 20
Milt,
30 40 SO 60 7080 901ÓÖ0
FIG. 22.5 A
100
4 5 6 7 8 9 10 20 30 40 50 60 708090100
Hei9bt of Receiving Aerial abaft Earth,m Fnct
FIG. 22.5 5
(i) Introduction
In the previous sections, the energy ideally available at the receiver terminals has
been calculated. This energy is available, however, only if the receiver is matched
to the aerial, and if there are no resistive losses associated with the aerial or aerial -
earth system.
If the input circuit of a receiver is to be designed to extract the maximum energy
from the radio -waves that are passing the aerial, then a knowledge of the component
parts of the impedance of the aerial is required.
902 IMPEDANCE OF AN AERIAL-(i) INTRODUCTION 22.7
.....
S i1II111111,
tion. l
4
If the impedance is measured A':...
3
at the base of an ideal vertical E C:2... .. ...
aerial, then the resistive compon-
ent is called the Base Radiation 2 11111II1I1n1rP.Iir.1I1111r.1111IIII1I111I1I1111H
In1I1N n111' r/ Ny11111 v
Resistance. Values of this ó 11111111111P/11 r d11111 A111111111111111111111111
111111111111/I11i1H111i111111111111111111111H111
for the types of aerial com- nunuurnriuHlduwnuuuunH 1
monly used in medium -fre-
r 1111111111M41111121111111111111111111111
:I: ..g1
quency reception are given in .s:.
Fi 22.6.
Fig. cos '::::::'
If the r -f resistance at the base O 6 uunnu.un.iu.
$
.Ilr.n'. .
:':C :::.............Ci:l::::
Therefore X, _ -
j 603 cot 29.3° = -j 1070 ohms.
Next, to obtain RR we use Fig. 22.6. For this aerial the ratio of the horizontal
portion of the aerial to the total length is 0.625. By interpolation between plotted
values we find that RR is approximately 1.0 ohm. The series loss resistance is
given
as 20 ohms, hence the total resistance at the base is 21 ohms.
The base impedance of the aerial is then
Z{ = 21 -
j 1070 ohms.
(b) Calculation of voltage at receiver terminals
From Fig. 22.2 we see that at the given frequency and distance the power flux
unit area 0 is 2.0 x 10 -8 watts /m2 for a radiated power of 1 KW. Hence 0per is
1.0 x 10 -8 for 500 watts radiated. The effective area of the aerial is given
in Table 1
as 3á2/161r. Hence the power ideally available is
PR = 1.0 x 10 -8 x 3 x 3002/16w
= 0.535 x 10 -' watts.
From (25A) we find that the e.m.f. induced in the aerial (referred to the base radia-
tion resistance) is e = 2A/PARR = 1.46 x 10 -2 volts or 14.6 millivolts.
21 ohms 1070 ohms
FIG. 22.8
The voltage appearing at the aerial terminals will be greater than the induced e.m.f.
because of the highly reactive impedance of the aerial. The equivalent circuit
the aerial and input circuit of the receiver are shown in Fig. 22.8. We see that of
aerial current is the
i = 1.46 x 10 -2/42 and the voltage at the receiver terminals is approximately
V = i 1070 = 0.373 volt.
(vi) Dipoles
The second type of aerial, the impedance of which we may wish to calculate is the
dipole aerial. Dipole aerials well removed from the earth may be treated in a manner
similar to that of a vertical aerial, because, at least approximately, a vertical aerial
with its image forms a dipole aerial. With the vertical aerial we measure the imped-
ance with respect to the neutral point (earth) of the equivalent dipole, so the impedance
is only half that of the equivalent dipole. Hence in calculating the impedance
of a
dipole we can use the results obtained previously, but multiplied by a factor of two.
Because a dipole aerial (remote from the earth) has no earth loss, it is in general more
efficient than the equivalent vertical aerial with earth return.
The most commonly used variety of dipole aerial is the resonant, or half-wave
variety. When remote from earth it has twice the impedance of a quarter-wave
aerial, i.e. its impedance is resistive and has a magnitude of 73 ohms. When in
the
vicinity of the earth, the impedance depends on the height above earth. The variation
is shown in Fig. 22.9.
Reflectors are sometimes used with half-wave aerials at very high frequencies.
22.7 (vi) DIPOLES 905
In Fig. 22.10 is shown the effect, on the radiation resistance of a half-wave dipole,
of the spacing between aerial and reflector, when the latter is tuned to provide maxi-
mum forward radiation. (For further information on the effects of reflectors the
reader is referred to Ref. 14).
00 2 4 6 8 1.0 M2 b4 I6
Aerial Height Above Earth (In Wavelengths)
FIG. 22.9
Fig. 22.9. Radiation resistance of half-wave dipole aerial plotted against height of
aerial above the earth.
906 (vii) LOOP AERIALS 22.7
0 02 03
Spac'ng in Wavelengths
FIG. 22.10
Fig. 22.10. Radiation resistance of half-wave dipole aerial plotted against spacing
between aerial and reflector (tuned to provide maximum forward radiation).
X`
-
21.íC
(31)
where C is the total capacitance of the aerial.
Hence, for such an aerial, the dummy consists of a fixed resistor in series with a
fixed capacitor. The impedance of a longer aerial may be represented moderately
well over a range of frequencies by a resistor, inductor and capacitor in series, all
shunted by a capacitor. For such networks we refer the reader to sources listed in
the bibliography.
For standard dummy aerials to be used in receiver testing, see Chapter 37 Sections
1 and 2.
DUMMY AERIALS 907
22,8
use a dummy
With a receiver designed to operate with a loop aerial, it is not usual toto the receiver
aerial when testing the receiver. Instead the aerial is left connected a second
and a known e.m.f. is induced in the loop. This is effected by connectingfrom each
to signal generator, and placing the two loops at a suitable distance
loop a
addition to
other. The e.m.f. induced in the receiving loop can be calculated if, in current in
loops, the number of turns, and the
the distance, the dimensions of the
the transmitting loop are known.
Very small aerials of this type are sometimes used indoors. Such aerials are use-
ful only when the required signals are of large intensity, because these indoor aerials
do not fulfil any of the three conditions listed above, except when close to broadcast
stations. Besides being inefficient because of their small size, they are also partially
shielded by earthed conductors in the house- wiring. They are also very liable to
pick up r-f noise carried along the power mains.
Loop aerials are sometimes used in broadcast receivers. When used indoors
they have the same drawbacks as the indoor capacitance type aerial, except that if
shielded, or balanced with respect to earth, they are less sensitive to inductive inter-
ference from nearby sources of r -f noise. The reason for this is that a balanced or
shielded loop responds to the magnetic component of the field of an aerial, whereas
the capacitance type aerial responds to the electric component. In the radiation
field, these two components are equal and are mutually dependent. In the im-
mediate vicinity of an aerial they are not equal, and the electric component increases
more rapidly than does the magnetic field as one approaches the aerial. Hence in
the vicinity of a radiation source of r -f noise, a loop aerial will pick up less energy than
will a vertical aerial.
A loop aerial is more sensitive to waves arriving at steep angles to the plane of the
earth than is a vertical aerial. Hence it should not be used where interference be-
tween ground and skywaves is experienced. In such conditions a straight vertical
aerial is normally the best type of aerial that can be used. In special circumstances,
such as occur sometimes in country areas, it is possible to use another type of aerial,
called a wave -antenna, to reduce ground -wave - sky -wave interference.
The wave antenna consists of a horizontal wire several wave lengths long suspended
a few feet from the ground and directed towards the required broadcast station.
It functions because of an effect that has not been mentioned previously, called
wave -tilt. This is the production of a radial component of the electromagnetic field,
when the waves are passing over imperfectly conducting ground. This component
is picked up by the horizontal wire which, when a wavelength or more long, has
maximum directivity along its axis.
(iii) Short-wave receiving aerials
Since short-wave reception usually is concerned with signals of low field intensity,
efficient aerials placed well away from sources of interference are required. In an
earlier section it was shown that the height of the aerial above ground is also important.
Indoor aerials are most unsuitable for short-wave reception. A half-wave dipole
placed at the correct height above earth and connected to the receiver by a trans-
mission line provides an efficient receiving system over a restricted range of fre-
quencies. In short-wave reception, however, one is concerned with a range of ap-
proximately 3 to 1 in frequency. If the aerial is to provide a reasonably good match
to the transmission line or receiver over such a frequency range, then a more elaborate
aerial than the simple half -wave type is required.
One way of doing this is to arrange a number of half-wave dipole aerials in the form
of a fan and connect them in parallel. Each dipole is tuned to one of the frequency
bands allotted for short-wave broadcasting. At resonance the half-wave dipole
matches a 70 ohm transmission line. The dipoles that are off resonance provide high
and predominately reactive impedances of both signs in parallel with the 70 ohms
of the resonant dipole, and have small overall effect.
Such an aerial is connected to the receiver by a balanced transmission line. The
receiver must be provided with an input circuit that is also balanced with respect to
earth. Alternatively an aperiodic balance /unbalance transformer must be connected
between the transmission line and the receiver.
An aerial that has similar characteristics to the one described above, but is not of
the balanced type, is shown in Fig. 22.11. It is essentially half of the fan arrange-
ment of dipoles. A common type of co -axial transmission line, having a characteristic
impedance of 50 ohms, is suitable for use with this aerial. A few buried radial wires
roughly half-wave (at the mid- frequency) in length provide a good earthing system
for this aerial.
22.9 (iii) SHORT -WAVE RECEIVING AERIALS 909
75 ohm Lint
(e) (b)
(c)
(e)
()
(d)
FIG. 22.12
FIG. 22 .I24)
22.9 (iv) V -H -F AERIALS 911
(f) Is a modification of (e). The aerial conductors are made in the form of a conical
cage to increase the frequency -range over which the aerial is effective. In this aerial
another method of providing the transformation from the unbalanced transmission
line to the balanced dipole is shown. It consists of splitting the outer conductor of
the co -axial line into halves, for a distance of one quarter-wave. At the top of this
" split " the halves of the dipole are connected directly to the outer conductor, while
the inner conductor of the transmission line is connected to one of them.
One of the most easily constructed aerials is the folded half-wave dipole, which is
shown in Fig. 22.12 (g). Whilst the radiation characteristic is the same as that of a
conventional half -wave dipole, the folding produces an impedance transformation.
When the conductors are of the same diameter throughout, the input impedance of
the aerial is approximately 300 ohms. This enables the aerial to be matched directly
to a 300 ohms moulded transmission line, which is commercially available. The
spacing between the parallel conductors is not critical but, as in the case of a trans-
mission line, it must be very small compared with a wavelength.
SECTION 10 : REFERENCES
1. Southworth, " Certain factors effecting the gain of directive antennas " Proc. I.R.E. 18.9 (Sept.
1930) 1502.
2. Burgess, R. E. " Aerial characteristics -relation between transmission and reception," W.E. 21.247
(April 1944) 154.
3. Schelkunoff, S. A. " Theory of antennas of arbitrary shape and size " Proc. I.R.E. 29 (1941) 493.
4. Pistolkors, A. " The radiation resistance of beam antennas " Proc. I.R.E. 17 (1929) 562.
5. Colebrook, F. M. " An experimental and analytical investigation of earthed receiving aerials "
Jour. I.E.E. 71 (1932) 235.
6. Sturley, K. R. ' Receiver aerial coupling circuits " W.E. 18.211 (April 1941) 137.
7. Burgess, R. E. " Receiver input circuits-design considerations for optimum signal/noise ratio "
W.E. 20.233 (Feb. 1943) 66.
8. Wells, N. " Shortwave dipole aerials " W.E. 20.236 (May 1943) 219, also letter J. S. McPetrie
20.237 (June 1943) 303.
9. Stewart, H. E. " Notes on transmission lines-use of general equations in determining line pro-
perties " Q.S.T. (August 1943) 25.
10. Lapon, E. A. " Open wire radio frequency transmission lines " Proc. I.R.E. 31.6 (June 1943) 271.
11. Wheeler, H. A., and V. E. Whitman, " Design of doublet antenna systems " Proc. I.R.E. 24.10
(Oct. 1936) 1257.
12. Carlson, W. L., and V. D. Landon, " New antenna kit design " R.C.A. Rev. 2.1 (July 1937) 60.
13. Carter, P. S. " Circuit relations in radiating systems " Proc. I.R.E. 20.6 (June 1932) 1004.
14. Brown, G. H. " Directional antennas " Proc. I.R.E. 25.1 (Jan. 1937) 78.
15. Data Sheet 53 " Aerial characteristics II- radiation resistance and polar characteristics " Electronic
Eng. 15.188 (Oct. 1943) 197.
16. King, R. " Coupled antennas and transmission lines " Proc. I.R.E. 31.11 (Nov. 1943) 626.
17. King, R., and C. W. Harrison " The receiving antenna," Proc. I.R.E. 32.1 (Jan. 1944) 18.
18. Burgess, R. E. " Reactance and effective height of screened loop aerials " W.E. 21.248 (May 1944)
210.
19. Burgess, R. E. " The screened loop aerial " W.E. 16.193 (Oct. 1939) 492.
20. Colebrook, F. M. " The application of transmission line theory to closed aerials " Jour. I.E.E. 83
(1938) 403.
21. Burgess, R. E. " Aerial characteristics " W.E. 21 (1944) 154.
22. Kronenberg, M. H. " A multiple antenna coupling system -a method of operating several receivers
on a single antenna" Q.S.T. (Aug. 1944) 9.
23. " Loops and small doublet antennas for F.M. receivers " A.R.T.S. and P. Bulletin No. 136 (9th
Oct. 1944).
24. Carter, P. S. " Simple television antennas " R.C.A. Rev. 4.2 (Oct. 1939) 168.
25. Sherman, J. B. " Circular loop antennas at u- h-f's," Proc. I.R.E. 32.9 (Sept. 1944) 534.
26. Kendell, E. M. " Receiver loop antenna design factors " Comm. 25.11 (Nov. 1945) 62.
27. Moody, W. " F-M antennas " Service 14.1 (Jan. 1945) 24.
28. Burgess, R. E. " Aerial-to -line couplings- cathode follower and constant -resistance network"
W.E. 23.275 (Aug. 1946) 217.
29. Vladimir, L. O. ' Low impedance loop antenna for broadcast receivers " Elect. 19.6 (Sept. 1946)
100.
30. Sturley, K. R. (book) ' Radio Receiver Design " Part 1 Chapter 3 (Chapman and Hall, London
1943).
31. ' The A.R.R.L. Antenna Book " (American Radio Relay League, 1949).
32. McPherson, W. L. " Electrical properties of aerials for medium and long -wave broadcasting "
Elect. Comm. 17.1 (July 1938) 44.
33. Bruce, E. " Developments in s.w. directive antennas " Proc. I.R.E. 19.8 (Aug. 1931) 1406.
34. " Capacity aerials for mains receivers " Philips Tec. Com. 3 (March 1947), 24, based on C.A.L.M.
Report No. 38.
35. Design Data (11) " Aerial coupling in wide -band amplifiers " W.W. 53.2 (Feb. 1947) 50.
36. Browder, J. E., and V. J. Young " Design values for loop antenna input circuits " Proc. I.R.E. 35.5
(May 1947) 519.
37. Kobilaky, M. J. " A note on coupling transformers for loop antennas " Proc. I.R.E. 35.9 (Sept
1947) 969.
38. Scroggie, M. G. " Is a big aerial worth while ? " W.W. 53.9 (Sept. 1947) 314.
Additional references will be found in the Supplement commencing on page 1475.
CHAPTER 23
RADIO FREQUENCY AMPLIFIERS
BY B. SANDEL, A.S.T.C.
Section Page
1. Introduction 912
2. Aerial stages ... 915
3. R -F amplifiers ... ... 922
4. Image rejection ... ... ... 925
5. Effects of valve input admittance 927
6. Valve and circuit noise 935
7. Instability in r-f amplifiers 942
8. Distortion ... ... 944
9. Bibliography 945
SECTION 1 : INTRODUCTION
(i) Aerial coupling (ii) Tuning methods (iii) R -F amplifiers (iv) Design con-
siderations.
912
23.1 (i) AERIAL COUPLING 913
and the inductance of the metal parts making up the capacitor. With the ganged
capacitor arrangement considerable trouble is experienced due to coupling from one
circuit to another, because it is not an easy matter to earth the shaft effectively be-
tween sections. The earthing wipers or leads have appreciable reactance and resist-
ance at v-h -f and these factors are no longer negligible in comparison with the tuned
circuit impedances. Insulated rotors and stators are of considerable assistance in
overcoming difficulties due to common shaft coupling.
ponents (including valves) and because the circuit layout is increasingly important
if satisfactory results are to be obtained. Electron transit time effects in valves be-
come of greater importance as the frequency is raised because of the changes which
occur in the valve input and output admittances *. The valve input admittance must
be considered when designing an aerial or r -f circuit as it will affect the dynamic im-
pedance of the circuit, and consequently the selectivity and gain. Valve sockets
and other components will add further damping to the input circuits. The effects
of valve noise will also be governed by the dynamic impedances of the input circuits
and so the various valve effects must be considered in conjunction with the external
circuits. This becomes of greater importance as the frequency increases.
A design difficulty with aerial and r-f stages is in obtaining satisfactory perform-
ance over a large tuning range. The three to one frequency coverage normally em-
ployed on dual wave receivers for the medium (540-1600 Kc /s) and shortwave
(e.g. 6 -18 Mc /s) bands presents some problems in regard to tracking, and constancy
of gain and selectivity. The L/C ratios obtainable with a variable capacitor giving
an incremental capacitance range of at least 9 1 are satisfactory on the medium wave
:
range but are rather poor on the shortwave band. For these reasons it is common
practice to limit the coverage on shortwave bands, in the better class of receiver, to
enable improved all round performance to be obtained ; a frequency ratio of about
1.3 : 1 is usual in the region of 10 to 20 Mc /s, with larger ratios at the lower frequen-
cies. An obvious advantage of multi -band receivers is the greatly improved ease of
tuning, apart altogether from other considerations of better performance.
To obtain the reduced frequency coverage with variable ganged capacitors having
a capacitance range of about 10-400 µµF, various arrangements of series and parallel
fixed capacitors are used.. If the circuits are arranged to use a combination of series
and parallel capacitors for bandspreading, it is possible to make the tuning follow
practically any desired law. Further, if the tuning range is small, the oscillator fre-
quency in a superheterodyne receiver can be above or below the signal frequency
and three point tracking can still be obtained using series padders in both the signal
and oscillator circuits. These matters will receive further consideration in Chapter 25
Sect. 3 on superheterodyne tracking.
Methods for measuring the value of k, the coefficient of coupling, are given in
Chapter 26 Sect. 4(ii)E and (iii)B, C. Methods for measuring the primary resonant
frequency of an aerial or r-f transformer are covered in Chapter 37 Sect. 4(ii).
2 to C2 \/R2R Ai (7a)
RD
RA1
(7b)
where RD is the dynamic impedance of the secondary circuit.
From this it can be shown, with M and C2 adjusted for maximum secondary current,
that the selectivity of the tuned circuit is half the value it would have if considered in
the absence of the primary. This condition of optimum coupling is not usual in
broadcast receivers, except where matching to a transmission line, since the coupling
between primary and secondary is made loose to prevent serious tracking errors due
to the variable reactance " reflected " into the secondary circuit over a band of fre-
quencies. The expression most generally required is that of eqn. (2) and the calcu-
lation of voltage gain need only be made at, say, two or three points across the tuning
range.
It should be noted, however, that in some cases loose coupling can lead to a con-
siderable reduction in signal -to -noise ratio if site noise is not predominant, e.g. at
the higher frequencies on the short -wave band it is advantageous to use optimum
coupling, or slightly greater, to improve the signal-to -noise ratio (see Ref. A2 Chapter
26).
The effects of valve input admittance can be included by considering an additional
resistance and a capacitance shunted across C2. The value of the resistance com-
ponent may be negative or positive depending on the type of valve and the operating
conditions, e.g. some types of converter valves have a negative input resistance.
Having obtained the stage gain it is next important to find the selectivity of the
circuit. As an approximation for ordinary types of aerial coils it is usually con-
venient to determine the selectivity of the secondary as for a single tuned circuit.
The simplest approach is to consider the resultant Q of the secondary circuit at the
resonant frequency when all the secondary resistance, including that due to reflection
from the primary and valve loading, is included. The value of Q is given by
wL eg /R eg, where Leg is the resultant secondary inductance. The selectivity is then
found as for a single tuned circuit, resonant at the frequency under consideration,
using the universal curve given in Chapter 9 (Fig. 9.17) in conjunction with the methods
detailed in Chapter 26 Sect. 4(iv)A. This method is satisfactory provided that the
amount of mistuning is not too large a proportion of the central reference frequency,
and that the aerial coil follows the usual practice with the primary resonated above or
below the tuning range or is designed for connection to a transmission line (or aerial)
of which the characteristic impedance is largely resistive ; otherwise the variations
in Ref¡ and L would have to be considered. Further, in a number of practical
cases it is permissible to calculate Reg and use L2 when finding the value for Q, since
918 (iii) MUTUAL INDUCTANCE COUPLING 23.2
selec-
the reflected reactance may not be very large. It is often sufficient to calculate
tivity at, say, the mid -frequency of the tuning range, although it should be noted that
selectivity is a function of frequency and dependent on the circuit operating conditions.
It is now worth while considering three special cases for the aerial coupling trans-
former :
(1) Where the aerial can be represented by a capacitive reactance in series with a
resistance.
(2) Where the aerial behaves like an inductive reactance in series with a resistance.
a pure
(3) Where the aerial is coupled to a transmission line which " looks " like
resistance at the aerial terminals ; or the aerial itself, coupled to the receiver, looks like
a pure resistance.
Any aerial can be made to appear resistive if the reactance components are tuned
out by a suitable arrangement of the aerial coupling network ; this is a common pro-
cedure with fixed frequency installations.
In case (1) it is usual to make the primary circuit resonate below the lowest tuning
frequency, since this offers a good compromise with regard to constancy of gain,
aerial
tracking, etc. This arrangement is also advantageous because the effect of the
is merely to move the primary resonant frequency still further away from
the tuning
range, thereby minimizing the effect of the aerial on the secondary without losing too to
much gain. In general the high impedance primary winding is resonated at 0.6
of the lowest tuning frequency, either with the aerial capacitance alone, or with
0.8 to the
an added capacitor. In either case the resonant frequency must not be close band
intermediate frequency in a superheterodyne receiver, and normally a broadcast
primary is resonated at a lower frequency than the i -f so that no aerial which is likely
to be used will resonate the aerial primary at the intermediate frequency.
Since the gain obtained from an aerial transformer with its primary resonant
out-
side the low frequency end of the tuning range decreases towards the
high frequency
end of the range, a small capacitance is often added across the top of the transformer.
This capacitance increases the gain at higher frequencies and allows a substantially desired
flat gain and noise characteristic to be obtained from the transformer over the
likely to be
range. This type of aerial transformer is fairly representative of those
encountered on the medium and long wavebands. Practical values for the coefficient
factors
of coupling between primary and secondary are a compromise between various 0.15 to
such as gain, tracking error and signal -to-noise ratios ; usual values are from top
0.3, with most aerial coils using values of about 0.2 (this should include any added
capacitance coupling). The value for primary Q is of the order of 50, and the second-
ary Q is about 90 to 130 in typical aerial coils using the range of coupling coefficients
stated, although Q values greater than 200 can readily be obtained by using specialk
iron cores. Values of secondary Q above 100 generally call for smaller values of
than 0.2 if good tracking is to be achieved.
If for some particular reason it is necessary to use a higher coefficient of coupling resonate
than say 0.4, then the primary inductance should be small and arranged to
with the aerial at about V/2 of the highest signal frequency. This improves the
signal -to -noise ratio but may give rise to tracking difficulties.
Case (2) normally applies with untuned loop aerials but the condition can occur
with some types of aerials on the shortwave bands. In this case it is usual to make
rule is
the coefficient of coupling fairly high, of the order of 0.5 or more, and a usefulloop (or
0.4 of the
to make the value of primary inductance approximately equal to
aerial) inductance (see Refs. B17 and A2, p. 44).
standard
For use with domestic type receivers operating on the short wave band, thereactance
dummy antenna is taken as being 400 ohms resistive, even though the aerial operating
will not be zero across the tuning range, as this gives an indication
of average
The aerial coupling transformer for this type of receiver can be designed
conditions.
using the procedure given in Case (3) below.
The untuned loop antenna (which should be balanced to ground for best results)
A-M broad-
finds application with portable receivers, and in some cases with ordinary
Loops have also been used in F -M receivers operating on the 40-50
cast receivers.
23.2 (iii) MUTUAL INDUCTANCE COUPLING 919
Mc /s band, but not in receivers for the 88-108 Mc/s band because of the ease with
which a half-wave dipole can be arranged inside the receiver cabinet. With loops it
is necessary to keep stray capacitances to a minimum so as to avoid resonance effects
within or close to the band of frequencies to be received ; this also assists in achieving
good tracking because a loop that behaves like a small fixed inductance can be com-
pensated for quite readily in the aerial coil secondary.
The number of turns used in untuned loop aerials is not very critical, since a loss
in effective height by using fewer turns can be offset by the increased voltage step -up
possible with the coupling transformer. To keep stray capacitances low the fewer
turns used the better. The best compromise for receivers operating on the medium
or short -wave bands is a single turn, loop, or perhaps a few turns (say not more than
four) of wire wrapped around the carrying case for medium wave portable receivers
(this wire can be litz or copper braid, and even rubber covered hook -up wire is often
satisfactory). The best type of single turn loop is one made out of copper tubing (or
any other good conductor such as aluminium or one of its alloys) with the tube dia-
meter not less than about 0.25 in. and preferably considerably larger than this ; the
loop area should always be as large as possible. If the loop is to be used on the long-
wave band it is preferable to use a few turns (about ten) of litz wire rather than a single
turn loop. This is necessary because of the drop in Q which occurs with the single
turn loop at the lower frequencies. However, satisfactory results have been achieved
with a 0.25 in. diameter copper tube loop at a frequency of 500 Kc /s.
In many cases a two winding transformer is not used and the loop is made part
of the tuned circuit (being in series with the main tuning inductance and earthed at
one side). An alternative arrangement is to tap the loop across part of the main tuning
inductance ; this can be made to give very satisfactory results.
Because the inductance of loops (of the type being considered) is usually very small,
it is often difficult to measure the Q at the working frequencies. This difficulty
arises with standard types of Q meters because the maximum capacitance is limited
to about 400 µµF. A useful method (suggested by J. B. Rudd) is as follows :
Couple the loop fairly tightly to an inductance which is sufficiently large to be tuned
to the working frequency. Measure the Q of the inductance with the loop open cir-
cuited (call this (2,) and then the Q (call this Q') with the loop short circuited at its
terminals. The Q of the loop (QL) is given by
.vQ1Q'
QL
where Q,
-Q1(1 - y) -Q'
= magnification factor for the primary inductance with the loop open
circuited.
Q' = magnification factor for the primary inductance with the loop terminals
short circuited.
y = 1 -
(L'ILi)
L' = inductance of primary with loop short circuited.
and L1 = inductance of primary with loop open circuited.
See (vi) below for some discussion of directly tuned loops.
Case (3) permits of a fairly simple solution. Normally k is about 0.2 or less, and
the primary winding reactance (at the geometrical mean frequency of the tuning range)
is made equal to the characteristic impedance of the transmission line or aerial. A
particularly simple solution has been made by Rudd (Ref. B12) to give conditions
for two point matching in the tuning range. The procedure is as follows :
wy/ w,
Cl
=
+
j'Q/f,
(8)
920 (iii) MUTUAL INDUCTANCE COUPLING 23.2
and Q, = secondary circuit Q in the absence of the primary. The value for Q,
should include the effect of valve loading, and so
QR (9)
Q' QwoLa -i- R
where Q is for the unloaded secondary
and R is the valve input resistance.
A useful approximate formula for estimating the aerial coil gain under this condition
is
Ea 1 Q,wLa (10)
E, 2 Zo
This expression gives the maximum possible value of gain and corresponds to eqn. (7a)
given previously.
Although elaborate methods are available for calculating optimum signal -to -noise
ratios etc., the details above should give a fairly close approximation to the required
practical conditions where the aerial coupling circuit is a compromise between the
various conflicting factors discussed previously. These factors of gain, selectivity
etc. should be carefully considered as in nearly every case the performance of a re-
ceiver is largely governed by the aerial coupling circuit.
na QawLa
n, Zo
where na =
total number of turns
n, =
number of turns across which aerial is connected
=
2Tr x operating frequency
Q, =
magnification factor for complete circuit
and Zo =
impedance of the aerial (or the characteristic impedance of a trans-
mission line).
The gain is found from equation (10),
E,
= 1 /QawLa
E, 2,v Zo
The selectivity is calculated using the same methods as before, from the equivalent
circuit. Whether this equivalent circuit takes the form of a series or parallel tuned
circuit is usually not important, as the results obtained are substantially the same,
and provided the circuit Q exceeds about 10 the Universal Selectivity Curves can be
applied directly in either case.
It should be noted that the treatment is only approximate, as the exact tapping point
is dependent on the coupling between the two sections of the aerial coil winding.
However, for a wide variation in coupling, the correct tapping point is not very differ-
ent from that obtained by the above procedure, which assumes unity coupling, and
the loss in secondary voltage as compared with optimum coupling is quite small.
23.2 (iv) TAPPED INDUCTANCE 921
FIG. 23.4
This shows a common disadvantage with the circuit of Fig. 23.3 in that the variation
in gain across the tuning range is quite appreciable. The treatment is quite straight-
forward and is left to the reader ; a fairly complete analysis is given in the references.
(vi) General summary
From the considerations detailed above, it can be seen that the coupled circuit
arrangement of Fig. 23.2 is generally most satisfactory since it readily lends itself,
with minor modifications, to applications using balanced or unbalanced aerial systems.
Under some conditions of receiver operation other methods are used because of the
practical consideration of ease of adjustment for best operation, or for use with a
particular type of aerial system.
Fixed tuned aerial stages are sometimes used with receivers covering a limited range
of frequencies. A typical case would be in the range of 88 -108 Mc /s where the loaded
Q of the aerial coil may be less than 20. In such a case a simple arrangement is for
the secondary to be resonated near the centre frequency in the band ; this introduces
gain variations of about 2 1 across the tuning range. The gain variations can be
:
offset by resonating the secondary at a point in the band such that the gain variations
offset those in r -f or mixer stages. The overall gain variations for the band under
consideration can be limited to less than 1.5 : 1 with a superheterodyne receiver using
a tuned r -f stage, and as the loss in gain and signal -to -noise ratio is not very serious,
the saving in the cost of a gang section is often well worth while when weighted against
the other factors. If the aerial circuit loaded Q is above about 20, this method is not
recommended because of the deterioration in circuit performance. In some cases it is
preferable to tune the aerial circuit and use an untuned r-f stage (when this is prac-
ticable).
One type of coupling circuit which has not been mentioned previously is the directly
tuned loop. In this case the loop provides the tuned circuit inductance as well as the
direct signal pick -up. As for any other loop, the area and the Q should be as large as
possible. This loop will probably be used, in most cases, with portable receivers.
If the loop can be placed in such a position that its Q is not materially affected by the
presence of the receiver and batteries, then its performance will usually be superior
to that of the single turn loop (an improvement of about 6 db can be expected). How-
ever, this arrangement is seldom convenient and in many cases better results are
possible with the single turn loop (or one with a few turns as discussed previously)
because any effect on its Q can be offset to a large extent by the use of an aerial coupling
transformer having a high Q winding in the tuned circuit. The improvement can
be seen most readily by considering a low Q loop, having a small inductance, connected
directly in series with an aerial coil which has a high Q and whose inductance is
practically the whole of the tuned circuit inductance.
In some cases it is required to use an additional external aerial with the loop to
increase signal pick -up. There are many possible arrangements but amongst the
simplest is the use of a small capacitance (say 10 to 20 µµF) connected between the
922 (vi) GENERAL SUMMARY 23.2
external aerial and the top of the tuned circuit (i.e. to the grid connection) or alter-
natively a tap on the aerial coil secondary a few turns up from the earthy end. For
a directly tuned loop the aerial is tapped into the loop in the same way as for the other
coil arrangement ; alternatively the series capacitor arrangement is satisfactory. The
main disadvantage of both these simple arrangements is the large change in gain which
occurs across the tuning range, but for medium -wave commercial portable receivers
additional circuit complications are seldom justified.
The subject of aerial-to- receiver coupling is a large one and only a few of the more
important design factors have been considered. For more complete information
on particular systems it is necessary to consult the text books and references.
If the winding is of the high impedance type, then it is usual to resonate the primary
at approximately 0.6 to 0.8 of the lowest tuning frequency, but the resonant frequency
should not be close to the intermediate frequency in superheterodyne receivers.
Under this condition the valve has a capacitive load which causes degeneration by
introducing a positive resistance component plus a capacitance into the r-f valve input
circuit due to the " Miller Effect." It is not usual to resonate the primary winding
within the tuning range because of the very large changes in gain and selectivity which
would be introduced. In addition, the problem of tracking the various tuned circuits
becomes almost hopeless.
When the aerial coil primary is of the low impedance type it is usual, although not
essential, for the r -f stage to be of this type also and vice -versa. This simplifies
tracking problems.
The methods to be used for calculating the gain and selectivity of this arrangement
are set out in detail in Chapter 9 Sect. 6(ii) under the heading " Coupled Circuits
Tuned Secondary." For the usual practical arrangement the stage gain is given
-
approximately by g,,,Q2 wM,
where g,,, = mutual conductance of r -f amplifier valve,
w = 277 x operating frequency,
M = kV/L,L, = mutual inductance,
and Q2 = secondary circuit Q including effect of valve input resistance, but not
" reflected " resistance from the primary.
The selectivity can also be calculated, approximately, by the method suggested for
aerial coils.
When designing a stage of this type it is necessary, as in the case of the aerial coil,
to arrive at a compromise between gain, selectivity and tracking errors. This usually
leads to a value for the coefficient of coupling of approximately 0.15 to 0.3, with the
lower values preferred for good tracking between aerial, r-f, and, in the case of the
superhet., the oscillator stage.
As with aerial coils, the primary Q is about 50 (or perhaps less) and the secondary
Q from say 90 to 130 in typical coils using this range of k values. Higher Q generally
calls for the lower values of k.
In the case of receivers working at say 50-400 Kc/s it is often inconvenient to re-
sonate the primary winding in the manner suggested. Under these circumstances
it is sometimes necessary to damp the primary winding heavily so that no pronounced
resonance occurs. It is then permissible to resonate the winding within the tuning
range. This procedure can lead to a severe loss in gain but is sometimes justified
under practical conditions where the loss of gain is outweighed by other factors.
The addition of top capacitance coupling, i.e. a small capacitance from the plate
connection to the grid connection of the following stage, is often employed to equalize
gain variations across the tuning range. The value of the capacitance is usually of
the order of a few micro -micro -farads and the exact value is best determined experi-
mentally. It should be noted that the presence of this capacitance will alter the co-
efficient of coupling, stage gain, selectivity and the primary resonant frequency, as
well as circuit tracking. There is always some capacitance coupling present with any
practical transformer and this largely determines the relative physical arrangement of
the primary and secondary windings, which are connected (in nearly all practical
cases) so that the capacitance coupling adds to the mutual inductance coupling.
(iii) Parallel tuned circuit
A simple parallel tuned circuit is sometimes used in the r -f stage, and can be made
to give higher gains than the transformer coupled arrangement. The difficulties
encountered are that
(1) large gain variations occur across a band of frequencies
(2) the skirt selectivity is rather poor
(3) tracking with conventional aerial circuits is difficult
(4) the tuning capacitor has to be isolated from B-}-.
For these and other reasons this circuit is not in common use in broadcast receivers,
but it might be incorporated in a receiver working at high frequencies covering a
924 (iii) PARALLEL TUNED CIRCUIT 23.3
restricted tuning range with some possible advantage. The gain of the stage is given
by g,Q,f1wL
where g,,, = mutual conductance of r -f amplifier yalve,
Q eff = Q of r -f coil when loaded by plate resistance of r-f amplifier and input
resistance of following stage,
w = 27r X frequency,
and L = inductance of tuned circuit.
Selectivity is calculated as for any other single tuned stage, using the methods of
Chapter 9 or Chapter 26, Sect. 4(iv)A.
(iv) Choke- capacitance coupling
Fig. 23.6 shows a common r -f coupling circuit for use at v-h -f such as on the 88 -108
Mc /s F-M broadcast band. The usual arrangement takes the form of a r -f choke,
resonated well below the lowest tuning frequency, coupled to a tuned circuit L,C,
by means of a capacitor C. The choice of a suitable value for C will allow some re-
duction in variations of stage gain across the tuning range as it can be considered as
part of a voltage divider formed with the tuned grid circuit. A further useful function
of the arrangement is that when a suitable resonance frequency has been chosen for
the choke, the total stray capacitance across the tuned circuit is effectively reduced,
thereby reducing limitations on the tuning range ; a small value for C assists
in this latter regard. However, the application of the circuit is largely one of practical
convenience and, as always, is a compromise.
With a choke resonated at 1/-% of the
MIXER lowest tuning frequency in a receiver with
GRID a low frequency limit of 87.5 Mc/s and an
Lp
upper limit of 108.5 Mc /s, the values
obtained for the additional capacitance
;,hunted across the tuned circuit were 3.6
µµF at 88 Mc /s and 4 at 108 Mc /s,
excluding the input capacitance of the fol -
A c
V.
Bt
lowing stage. The value of C was 10 µµF.
BIAS. Because of the low value of dynamic im-
FIG.25.6 CAPACITANCE COUPLED R -F STAGE.
pedance for this circuit at v-h-f, the effect
of the plate resistance of the r -f valve can usually be neglected (but see Ref. A8).
In this case the selectivity for the circuit can readily be calculated as for a single tuned
circuit, including the effects of the input resistance of the following stage, without
introducing very large errors. The gain may be calculated with sufficient acctracy
as for the case of the simple tuned circuit acting as the anode load and multiplying
the result by a suitable factor to allow for the voltage division due to the capacitive
reactances at the frequency being considered. This factor is readily found from a
knowledge of the equivalent capacitance due to the choke circuit, at the working
frequency, and the value selected for C.
Voltage gain - r2
RL
+ RL
(µ + 1)
rD RL
Resistance loading across input circuit ti µ
+
+ 1
RL = rn1/l + R,g,,
Noise factor = N = / 2 + R7 + Rz
11/// RDAs
2
RDA, Aa 2
where RL = load impedance connected between plate and B +
r7 = valve plate resistance
µ = valve amplification factor
g,,, = valve mutual conductance
R, = valve input resistance at working frequency
R, = equivalent noise resistance of first valve
Rz = equivalent noise resistance of second valve
Ra = generator impedance e.g. aerial resistance
Al = voltage gain of input circuit
and AZ = voltage gain of first stage, excluding the voltage gain of the input
circuit.
It should be noted that it is always advisable to operate the heater and cathode of
the valve at the same r -f potential. This is readily achieved by using suitable chokes
in the heater leads. For more detailed information on grounded -grid amplifiers
see References B8, B18, A10 and A4.
When one of these frequencies is the desired one (it may be either the lower or the
higher), the other is referred to as its image frequency. Clearly the desired and un-
desired signal frequencies are separated by twice the intermediate frequency. Putting
this statement in symbolic form :
fimave = f,,, + fi -1 when f,,, > fe,,
fima,e = fnec -fi -1 when f,,, < feie
where f,,, = desired signal frequency.
As an example : On the 540 -1600 Kc /s A -M broadcast band the oscillator fre-
quency is normally above the signal frequency. If the reasons for this are not obvious,
consideration of the oscillator tuning range and the values of variable capacitance
required will make it so. The usual i-f is 455 Kc /s and if a signal frequency of 1000
Kc /s is taken,
then fi m ,, e = (1000 + 455) + 455 = 1910 Kc /s.
Since the converter cannot discriminate between the two signals of 1000 Kc/s and 1910
Kc /s, it is necessary for the aerial circuit, and r -f circuit if this is used, to make the
magnitude of the 1000 Kc /s voltage much greater than the magnitude of the 1910
Kc /s voltage. Otherwise severe interference can result when signals are being trans-
mitted at both these frequencies.
When making calculations for image rejection the worst conditions are usually
found at the high frequency end of the tuning range, provided that the selectivity of
the tuned circuits does not vary appreciably with change in frequency. This is
because the separation between the two frequencies is a smaller proportion of the fre-
quency to which the receiver is tuned. It should also be clear that higher values
of i -f will materially assist in reducing image effects because of the wider frequency
separation between the desired and undesired signals ; it is for this reason that some
v-h -f receivers use intermediate frequencies of the order of 10.7 Mc /s.
coil is halved, under some conditions of operation, by using two point matching to
an aerial transmission line, the greater part of the image protection may here again
be due to the r-f stage. These considerations are particularly important at the higher
frequencies. On the long and medium wave bands sufficient image protection can
often be obtained from the discrimination afforded by the aerial coil alone, but on the
shortwave bands deterioration in image rejection is serious (and rapid) unless the inter-
mediate frequency is increased above the usual value of about 455 Kc /s.
It is important to note that a good image rejection ratio is desirable since this also
indicates the selectivity of the signal circuits and the degree of rejection against spurious
frequency combinations. A high degree of selectivity preceding the first r -f amplifier
valve has obvious advantages in reducing cross -modulation effects.
ably accurate results are possible using a high frequency Q meter to determine the
capacitive and resistive loading effects. The usual precautions as to length of leads,
earthing etc. must be carefully observed when making these measurements, if the
results are to be of any practical value. Something of this nature is usually required
when considering the effects of the converter valve on its input circuit, since the pub-
lished information is rather meagre. In this regard it is well to remember that some
types of converter valves, such as those using inner grid injection, e.g. types 6A8,
6K8, 6SA7, 6BE6 etc., give negative loading whilst other types such as X61M, X79,
6J8 -G, 6L7, etc. (which use outer grid oscillator injection) give positive loading. In-
creasing the negative bias on the signal grid will reduce the loading effect with all types,
but at very high values of bias the negative loading may reverse its polarity and be-
come positive. [The reasons for these effects are discussed later in this section in
connection with input loading of receiving valves.] To reduce input conductance
some valve types have more than one cathode connection (see Ref. A8), and typical
examples are types 6AK5 and 6AG5, each having two cathode terminals. Even when
the cathode terminals are directly connected to ground the length of the cathode lead
is still sufficient to provide appreciable inductive reactance at frequencies of the order
of 100 Mc /s. For this reason all plate and screen by -pass capacitors should be re-
turned to one cathode lead and the grid returns to the other. The alternating volt-
ages developed across the cathode lead inductance due to currents from the plate and
screen circuits are in this way prevented from being directly impressed in series with
the grid circuit, since it is the grid to cathode voltages which are of major importance.
Direct current divides between the two available cathode paths, but this is not import-
ant since the path taken by this current normally does not affect the grid input admit-
tance. To obtain grid bias the " grid " cathode lead employs the usual resistor and
capacitor combination. The r-f by -pass capacitors from the plate and screen circuits
connect directly to the other cathode terminal which is not directly connected to
ground in this arrangement. A typical circuit is shown in Fig. 23.8.
In pentodes working at v-h -f it is advis-
able to connect the suppressor grid terminal
directly to ground, when a separate terminal
is available for this electrode, rather than
to the cathode, because the suppressor lead
inductance would then be connected in
series with any external cathode lead induc-
tance. If this precaution is not taken the
coupling between control grid and plate is B'
increased, because of the capacitance from FIG.234 R-F AMPLIFIER WITH TWO CATHODE
CONNECTIONS.
control grid to suppressor and from sup-
pressor to plate, the junction of these two
capacitances having an impedance to ground depending on the total lead inductance.
Similar considerations also apply to screen grid circuits, particularly when the valve
has a single cathode lead, and the shielding action of this grid can be seriously affected
if proper precautions are not taken, such as directly earthing by -pass capacitors.
These circuit arrangements are not very important at low frequencies where degenera-
tive effects may be more serious, necessitating the connection of the suppressor grid
and by -pass capacitors directly to the cathode.
6A8
6J7
Pentagrid
Pentode
250
250
100
100
-3
-3
-
0
0.3
0.3
-0.05
0.05
-3
6K7
6K8
Pentode
Triode -hexode
250
250
100
100 -3
-3
---
0 0.3
0.3
0.05
-0.08t
6L7
6SA7t11
Mixer
Pentagrid
6SA72f Pentagrid
250
250
250
100
100
100
0
-2
-- 0.3
0.3
0.3
0.15$
-0.03§
-0.03§
6SJ7f Pentode 250 100 -3 0 0.3 0.05
6SK7f Pentode 250 100 -3 0 0.3 0.05
954 Pentode 250 100 -3 0 0.3 0.005
1851 Pentode 250 150 -2 0 0.3 0.13
6AC7/
1852f Pentode 250 150 -2 0 0.3 0.13
6AB7/
1853¡ Pentode 250 200 -3 0 0.3 0.065
practically constant for all operating conditions. Also, when the transconductance
of a valve is changed by a change in signal -grid bias, k,, varies directly with trans -
conductance over a wide range. In the case of converter types, the value of k,, de-
pends on oscillator -grid bias and oscillator voltage amplitude. In converter and
mixer types, k,, is practically independent of oscillator frequency.
In eqn. (1), the term k,f is a conductance which exists when the cathode current is
zero. The term k,, f 2 is the additional conductance which exists when cathode current
flows. These two terms can be explained by a simple analysis of the input circuit
of a valve.
(B) Cold input conductance
The input impedance of a valve when there is no cathode current is referred to as
the cold input impedance. The principal components of this cold impedance are a
resistance due to dielectric hysteresis, and a reactance due to input capacitance and
cathode -lead inductance. Because these components are in a parallel combination,
it is convenient to use the terms admittance, the reciprocal of impedance, and sus-
930 (ü) INPUT LOADING OF RECEIVING VALVES 23.5
ceptance, the reciprocal of reactance. For most purposes, the effect of cathode -lead
inductance is negligible when cathode current is very low. The cold input admittance
is, therefore, a conductance in parallel with a capacitive susceptance. The conductance
due to dielectric hysteresis increases linearly with frequency. Hence, the cold input
conductance can be written as kcf, where k, is proportional to the power factor of
grid insulation and is the k, of eqn. (1).
The amplitude of Q, is proportional to the amplitude of the grid voltage ; the grid
current, which is the time rate of change Q is thus proportional to the time rate of
change of grid voltage. For a sinusoidal grid voltage, e, = E, sin wt, the time rate
of change of grid voltage is wE, cos wt. Therefore, for a given valve type and opera-
ting point, the amplitude of grid current is
I, = KE, w
and the absolute value of grid -cathode admittance due to induced charge on the grid is
Y, I, /E, = Kw (3)
The conductive component (g,) of this admittance is
g, = Y, sin 9 = Y,9 = KwO (for small values of 9).
Because B = CT, this conductance becomes, for a given operating point,
g, = Kw'r. (4)
Thus, the conductance due to electron transit time also varies with the square of the
frequency. This conductance and the input conductance, g,,,w'LCh, due to cathode -
lead inductance, are the principal components of the term k,,f' of eqn. (1).
This explanation of input admittance due to induced grid charge is based on a
space -charge -limited valve, and shows how a positive input admittance can result
from the induced charge. The input admittance due to induced grid charge is nega-
tive in a valve which operates as a temperature- limited valve, that is, as a valve where
cathode emission does not increase when the potential of other electrodes in the valve
is increased. The emission of a valve operating with reduced filament voltage is
temperature limited ; a valve with a screen interposed between cathode and grid
acts as a temperature -limited valve when the screen potential is reasonably high. The
existence of a negative input admittance in such a valve can be explained with the aid
of Fig. 23.12.
When the value of Ef, in Fig. 23.12 is sufficiently high, the current drawn from the
cathode divides between g, and plate ; any change in one branch of this current is
accompanied by an opposite change in the other. As a first approximation, there-
fore, it is assumed that the current entering the space between g, and g, is constant
932 (ii) INPUT LOADING OF RECEIVING VALVES 23.5
and equal to pv, where p is the density of electrons and y is their velocity. g2 may
now be considered as the source of all electrons passing to subsequent electrodes.
Suppose now, that a small alternating voltage is connected in series with grid g3,
as shown in Fig. 23.12. During the part of the cycle when e, is increasing, the elec-
trons in the space between g2 and g3 are accelerated and their velocities are increased.
Because the current pv is a constant, the density of electrons (p) must decrease. In
this case, therefore, the charge at g2 is 180 degrees out of phase with the grid voltage,
as shown at a and b of Fig. 23.13. This diminution in charge propagates toward
the plate with finite velocity and induces a decreasing charge on the grid. Because
of the finite velocity of propagation, the maximum decrease in grid charge occurs at
a time later than that corresponding to the maximum positive value of e as shown in
Fig. 23.13(c). The current, which is the derivative of Q, with respect to time, is
shown in Fig. 23.13(d). If there were no phase displacement (B = 0), this current
would correspond to a negative capacitance ; the existence of a transit angle B, there-
fore, corresponds to a negative conductance. By reasoning similar to that used in
the derivation of eqns. 3 and 4, it can be shown that the absolute value of negative
admittance due to induced grid charge is proportional to w, and that the negative
conductance is proportional to w2. These relations are the same as those shown in
eqns. 3 and 4 for the positive admittance and positive conductance of the space- charge-
limited case.
A negative value of input conductance due to transit time signifies that the input
circuit is receiving energy from the " B " supply. This negative value may increase
the gain and selectivity of a preceding stage. If this negative value becomes too large,
it can cause oscillation. A positive value of input conductance due to transit time
signifies that the signal source is supplying energy to the grid. This energy is used
in accelerating electrons toward the plate and manifests itself as additional heating
of the plate. A positive input conductance can decrease the gain and selectivity of a
preceding stage.
It should be noted that, in this discussion of admittance due to induced grid charge,
no mention has been made of input admittance due to electrons between grid and plate.
The effect of these electrons is similar to that of electrons between grid and cathode.
The admittance due to electrons between grid and plate, therefore, can be considered
as being included in eqn. (3).
greater than gmg,L. Therefore, in a space- charge -limited valve, where K is positive,
the hot input capacitance is greater than the cold input capacitance. In a temperature -
limited valve, where K is negative, the hot input capacitance is less than the cold
input capacitance. In both valves K changes with change in transconductance.
Because of this change, the input capacitance changes somewhat with change in a.v.c.
bias. In many receivers, this change in input capacitance is negligible because it is
small compared with the tuning capacitances connected in the grid circuits of the high -
frequency stages. However, in high- frequency stages where the tuning capacitance
is small, and the resonance peak of the tuned circuit is sharp, change in a.v.c. bias
can cause appreciable detuning effect.
(E) Reduction of detuning effect
The difference between the hot and the cold input admittances of a space- charge-
limited valve can be reduced by means of an unbypassed cathode resistor, Rk in Fig.
1651 , 6AC7/ IB52
TYPICAL CHARACTERISTICS
Ef = 6.3 VOLTS
PLATE VOLTS 250
SUPPR ESSOR VOLTS .0
SCREEN VOLTS 150
GRID VOLTS VARIED
22
FREOUENCY
UNBYPASSED
40 Mc%
CATHODE RESISTOR 4
R (OHMS)
200
giAll
175
150
Wilt a
25
a40f
125
ICE WAU
ll 511 IM 2-0
LL
;;c" 15V
V 75
á z
IF
O
5
I
á
a
a
i ill
z V
25
o
,MIE
A Ai UU
WM.
--
IN60..
GC
.
---IIIIIII
Ili
30 III 05
O 5
z
z
ú
X
HO
0 2 4 6 B 10
PLATE MILLIAMPERES FIG. 23.15
934 (ii) INPUT LOADING OF RECEIVING VALVES 23.5
23.14. The total hot input admittance of this circuit is made up of a conductance
and a capacitive susceptance C,'. Analysis of Fig. 23.14 shows that, if cathode -lead
inductance is neglected, the total hot input capacitance, C;', is approximately
C C, +K/C,
I
C,' { 1
1 + gmRk (6)
Inspection of this equation shows that if K is positive and varies in proportion with
g,,,, the use of the proper value of RI, will make C; independent of gm. In a space -
charge-limited valve, K is positive and is found by experiment to be approximately
proportional to gm. It follows that the proper value of Rk will minimize the detuning
effect of a.v.c. in a space- charge -limited valve. Eqn. (6) is useful for illustrating the
effect of RI, but is not sufficiently precise for computation of the proper value of Rk
6AB7 /1853 to use in practice. This
TYPICAL CHARACTERISTICS value can be determined
E ¡6.3
VOLTS
by experiment. It will
PLATE VOLTS 250 be found that this value,
SUPPRESSOR VOLTS O in addition to minimizing
SCREEN VOLTS 200
capacitance change, also
GRID VOLTS VARIED
FREOUENCY 40 Mc/t
reduces the change in in-
UNBYPASSED CATHODE RESISTOR Rk (OHMS)
put conductance caused
12
by change in a.v.c. bias.
60
The effect of unbypassed
cathode resistance on the
00 ....... ar.0
2.0 change in input capacit-
ance and input conduct-
ance of types 6AC7/1852
90
G n 15 W and 6AB7/1853 is shown
Y. in Figs. 23.15 and 23.16.
u These curves were taken
5
,,,..................__. á at a frequency of 40 mega -
9?
° cycles. The curves for
3
Aill ....."*"..°....-
the 6AC7/1852 also hold
, ...
The hot input conductance of a valve with an unbypassed cathode resistor can be de-
termined by modification of the values of k,, in the table. The value of k,, in the table
should be multiplied by gm /(1 gmRk). The resultant value of k,,, when sub-
stituted in eqn. (1), with k, from the table, gives the input conductance of a valve
with an unbypassed cathode resistor. In the factor (1 + gmRk), gm is the grid -
cathode transconductance when Rk is by- passed.
When an unbypassed cathode resistor is used, circuit parts should be so arranged
that grid- cathode and plate -cathode capacitances are as small as possible. These
capacitances form a feedback path between plate and grid when there is appreciable
impedance between cathode and ground. To minimize plate-cathode capacitance,
the suppressor and the screen by -pass condenser should be connected to ground rather
than to cathode.
23.6 (i) THERMAL AGITATION NOISE 935
Maximum receiver sensitivity is not, in most cases, determined by the gain of the
particular receiver but by the magnitude of the input circuit noise, which is generated
by the antenna, the tuned input circuit, and the first tube. This is true of A-M, F -M
and television except that in F -M and television the random noise effect assumes
a far greater degree of importance than in the standard broadcast band. The reason
for this is twofold :
(1) At the frequencies where these two services operate, 50 to 250 Mc /s, the re-
lative values of the several different noise sources assume entirely new proportions
and the heretofore unimportant and little known induced grid noise becomes one
of the predominant components of the total.
(2) Most random input and tube noise is proportional to the square root of the band-
width. Both television, with a 4 Mc /s band, and F-M, with a 200 Kc /s band, occupy
much wider sections of the frequency spectrum than anything previously encountered
by the commercial receiver engineer.
Fig. 23.17(b) is a simple parallel -tuned circuit where the noise generating resistance
is equal to the tuned circuit impedance. Again let us assume the bandwidth to be
five Kc per second.
R = Q(wL) = 100 x 1900 = 190 000 ohms
e 2 = 1.6 x 10 -20iFR
e _ -/1.6 x 10 -20 x 5 000 x 190 000
e = 3.9 microvolts.
/
Thermal agitation noise voltage may be calculated easily with eqn. (2) but by using
the graph shown in Fig. 23.18 the room temperature values may be found directly.
...
10
s
M -/
II riMil 1/210
°°..
% /.N O,iQ'.os
.MP.
2
° ..i
/Eii
.,.,,.
.,,.., ,..
.onsI
mom. FO/ A
....M..
oF
°
F.'`
oF:
.,°°
,,,a
F.\+`ÓcQt
bF_
P,
, .Q s
j
T.20°CENT
Mi'H %..%..
5 2 5 2 5 2 5
100 IQoo IC;G00 IGQoOO MEG 1 10MEG
RESISTANCE (ohms) FIG. 23.18
Fig. 23.18. Thermal agitation noise voltage versus resistance and band -width (Ref B33).
(ii) Shot noise
Another important component of the total receiver noise is shot noise. This noise
is generated inside the vacuum tube and is due to the random fluctuations in the plate
current of the tube, or, to state it in another manner, random variations in the rate of
23.6 (ü) SHOT NOISE 937
arrival of electrons at the plate. 'When amplified, this noise sounds as if the plate
were being bombarded with pebbles or as if a shower of shot were falling upon a metal
surface, hence the name shot noise.
Although generated essentially in the plate circuit of the tube, which is not a con-
venient reference point for sensitivity or signal -to-noise ratio calculations, the shot
noise is nearly always referred to as a noise voltage in series with the grid. Since the
following equation is true,
e, = i,, /g, (3)
where e, = a.c. grid voltage
i, = a.c. plate current,
and g,,, = transconductance,
by simply dividing the noise current in the plate circuit by the transconductance of
the tube, the shot noise may be referred to the grid and expressed in terms of grid
voltage.
Another step is taken, however, to simplify the noise nomenclature. Suppose a
given tube has a shot noise equal to e
microvolts in series with its grid. It is per-
fectly valid to imagine that this voltage could be replaced by a resistance whose thermal
agitation noise is equal to e, (the shot noise) and to consider the tube to be free of noise.
This imaginary resistance, which when placed in the grid of the tube generates a
voltage equal to the shot noise of the tube, is known as the shot noise equivalent
resistance or just as the equivalent noise of the tube. The advantage of this termino-
logy is that when the equivalent noise resistance of the particular tube is known, the
noise voltage may be calculated directly for any given bandwidth by substituting values
e 'R
in the following formula :
= 4KT4FR
= equivalent noise resistance,
(4)
'
where
or at room temperature
e = 1.6 x 1C-204FR (5)
If the noise were expressed as a voltage or current its value would be correct only for
one particular bandwidth.
2_S any
By knowing the
two
R of
given tubes their
TRIODE AMPLIFIER REO ° GM
relative shot noise merit is
also known regardless of
IB /2_5 2 ,)
PE NTODE AMPLIFIER REO -
GM
1.
what bandwidth they are
IB +Ic G¡,E
to operate at, while if the
noise voltages were given
GC-
TRIODE MIXER REQ° G2/ 4 alone the operating band-
width at which the calcu-
IB / 4 201' lation was made would
PEN TODE MIXER +
also have to be noted if
REO
Gc
150
2,300
1,200
5,200
8,800
2,100
2,660
6BA 6 PENTODE AMPLIFIER 260 100 ,400 3,520
6886 PENTODE MIXER 250 100 1,100 14,080
68E6
6C4
CONVERTER
TRIODE AMPLIFIER
250
100 --
100 475
3,100
190000
810
6C4
6C5
TRIODE MIXER
TRIODE AMPLIFIER
100
250 -- 770
2,000
3,240
1,250
6C5
645
TRIODE MIXER
TRIODE AMPLIFIER
250
250 -- 500
2,600
5,000
960
645
+6,16
+646
TRIODE MIXER
TRIODE AMPLIFIER
TRIODE MIXER
250
IOO
100
-- 650
5,300
1,320
3,840
470
1,880
668 CONVERTER 250 IOO 350° 290,000
65A7 CONVERTER 250 100 450° 240,000
6587-Y
6507
6567
CONVERTER
TRIODE AMPLIFIER
PENTODE AMPLIFIER
2S0
250
250
-
100
125
950°
1,325
,700
62,000
1,890
3,100
6567 PENTODE MIXER 250 125 1,180 12,400
65.17 PENTODE AMPLIFIER 250 100 1,650 6,100
6567
651.7
6507
PENTODE AMPLIFIER
TRIODE
TRIODE
AMPLIFIER
AMPLIFIER
250
250
250
-
100 2,000
1,600
1,100
11,000
1,560
2.300
(4) VALUES OF PLATE VOLTAGE AND CURRENT ANO SCREEN
VOLTAGE AND CURRENT ARE FOR TYPICAL OPERATING
CONDITIONS.
( °) CONVERSION TRANSCONDUCTANCE - MICROMHOS
FIG. 23. 20
After the equivalent noise resistance is known the value of r.m.s. noise voltage at
the grid of this tube can be calculated by applying the same expression that is used for
thermal agitation noise,
e 2 = 1.6 x 10 -20dFR
or, by using the graph of Fig. 23.18.
Fig. 23.20 presents calculated equivalent noise resistance values for a number of
commonly used tubes acting as various types of circuit elements. These are, of
course, approximate figures.
It can be seen from Figs. 23.19 and 23.20 that the noise resistance or voltage is at
a minimum for a triode, increasing for the pentode and the multigrid tube, following
in that order.
Shot noise is unique among the noise sources in the sense that the shot-noise voltage
should be considered to exist in series with the grid inside the tube. The reason for
this is that nothing can be done to the external grid circuit that will alter the magnitude
of this component. Even though the shot noise must lie tolerated, its effect can be
minimized by designing the input circuit for maximum signal at the grid. This
does not reduce the magnitude of the noise but does improve the signal -to- noise-ratio
of the receiver.
23.6 (iii) INDUCED GRID NOISE 939
-
duced grid noise in vacuum -
An expression for induced -grid- noise* for tubes tube circuit (Ref. B33).
with control grid adjacent to the
cathode follows :
where
9.
:
=
T5
1.4 x 4KTkJFR ,;rß;(6)
= cathode tempera-
_;"
ture (degrees Kel-
vin)
G el a r, = electronic (transit
time) component of
input conductance
z electronic compon-
and R et e e
mob,
ent of input resist-
ance.
From eqn. (6) it can be seen that
u
2QIX 11Mtl
aw
the induced grid noise is propor-
tional to the electronic or transit
time component of the input resist -
\
ance. Measurement of the total
m
input resistance is a comparatively
simple matter with the use of a high
soo
frequency Q meter, but the separa-
I tion of the electronic and the cathode
< inductance components, which are
2 00 essentially two resistances in parallel
between the grid and ground, is a
very difficult matter. Since most
Ii0 20
FREGVIENC5 (megacycle) 200 high- frequency tubes are con-
FIG' 23'22 structed with either two cathode
Fig. 23.22. Approximate electronic input leads or one very short lead, assum-
resistance versus frequency (Ref. B33). ing the total measured input resist-
ance to be electronic would not introduce too great an error. Another factor in favour
of this approximation is that it would be the case for maximum induced grid noise and
any error introduced would more than likely be on the safe side.
O. North " Fluctuations induced in vacuum -tube grids at high
frequencies " Proc. I.R.E. Feb.
D.
1941.
940 (iii) INDUCED GRID NOISE 23.6
Cathode temperature in most receiving tubes, which almost exclusively use oxide -
coated cathodes, is approximately 3.6 times the normal room temperature in degrees
K. Eqn. (6) can be rewritten therefore as
e'{,. = 5 x 4KT4FRPteee (7)
where T = room temperature (degrees Kelvin),
or, when T = 300 degrees Kelvin,
e2 j.,. = 8 x 10- aozFRetct (8)
In circuit calculations this noise is essentially in series with a resistance equal to
R.tect located between the grid and ground -Fig. 23.21.
The approximate input resistance for a number of common receiving tubes in the
frequency range of F-M and television is given in Fig. 23.22. This chart can be used
to find approximate input resistance values for induced grid -noise calculations.
(iv) Total noise calculations
Calculations of total input noise are made by using the grid of the input tube as a
reference point. There are many sources of noise and each must be calculated and
referred to the grid reference point before a summation is made. Since noise is a
random effect and calculated on a power basis, the separate components cannot be
added directly but as the square root of the sum of the squares.
Total Noise = 1/e1' -{- e,' +- e,' + etc. (9)
The various noise voltages that must be referred to the first grid are :
(1) Thermal agitation noise of the antenna radiation resistance.
(2) Thermal agitation noise of the tuned grid circuit.
(3) Shot noise of the input tube.
(4) Induced grid noise of the input tube.
(5) Grid circuit noise of the following stages referred back to the first grid.
In Fig. 23.23(a) appears a diagram of a practical input circuit and the location of all
the circuit parameters and noise voltages. Fig. 23.23(b) is essentially the same except
that the antenna circuit is reflected through the transformer and considered to exist
at the grid. This is the diagram that is most useful in calculating the total input
circuit noise.
FIG. 23.23
The steps necessary to find specific
values for each of these factors are shown in
Fig. 23.24. Antenna radiation resistance
varies widely with the type of antenna
chosen, but for F -M and television work
it is generally in the order of 75 to 300
I+N
(a) ohms. When the noise is known in terms
Rant Antenna Radiation Resistance of an equivalent resistance, as is the case
coot Antenna Resistance Thermal Noise here for the antenna, tuned circuit, and
Signal voltage
esig shot noise, the equivalent voltage can be
RcktTuned Circuit Impedance
ecktTuned Circuit Thermal Noise
either calculated or obtained directly
RekctEkctronlc Input Resistance from Fig. 23.18.
ei Induced Grid Noue
Reg Equivalent Shot Noise Resistance
Valve Shot Noise
In order to add the antenna, tuned cir-
eshet
N. Turns Ratio of Coupling Transformer cuit, and induced grid noise to the shot
noise the effective voltage of these three
components at the grid, or between the
points A and B, must be known. Each
must go through what is essentially a re-
sistive divider and may be calculated as
shown in Fig. 23.25.
FIG. 23,24
Fig. 23.24. Procedure for calculating various noise voltages (Ref. B33).
t
etoeae = 1/(eskot)2 + (e , at A -B)2 + (e;_, at A -B)2 + (eeke at A-B)2 (10)
One other factor may affect this total, however. If the total noise of the following
stages, which is calculated similarly, ignoring the antenna of course, is appreciable,
it must be added to the constants of Fig. 23.25. In reflecting it to the first grid the
second stage noise should be divided by the gain of the first tube. When the gain is
about ten or more this factor may usually be neglected.
Effective signal voltage across A -B is calculated in the same way as the antenna
noise in Fig. 23.25. The signal-to -noise ratio is now also known.
Since the signal -to -noise ratio is determined by the signal strength and the total
noise at the grid of the input tube, for a receiver that has a mixer, such as 6SK7, for
the input tube, the signal -to-noise ratio may be considerably improved by the addition
of an r -f tube, such as a 6SG7, which has considerably less total noise. By adding
additional r -f tubes (6SG7's), however, since the total noise and signal at the grid
will be the same, the signal -to -noise ratio will not be improved.
Fig. 23.25. Circuit for reflecting various voltages to the grid (Ref. B33). To find
the effective voltage of the antenna, the tuned circuit, and the induced grid noise at
the grid of the tube let R, equal one of the above noise resistances and el its generated
voltage. If R, and R3 equal the other two noise resistances the effective voltage at
the grid is
e, 122123
eid -B - Ri+ R R3
<
R2 + R3
R2 +R3
This calculation must be performed for the three components in turn.
(y) Sample circuit calculations
For a sample problem let us calculate the total noise at the grid of an F -M receiver
r -f amplifier stage, assuming the circuit in Fig. 23.26(a) to be under consideration.
As a simplification of procedure the steps in the calculations will be numbered.
(1) N2R, t
= 1200 ohms (calculated)
(2) Ree,ct = 1200 ohms (Figure 23.22)
(3) 12222 Q c,L = 8000 ohms (calculated)
=
(4) R =
3100 ohms (Figure 23.20)
At this point it will be convenient to redraw the circuit as shown in Fig. 23.26(b).
(5) Ne, t
= 2 microvolts (Fig. 23.18)
(6) et,,. = V'8 x 10 -E0 x 200 x 103 x 1200
= 4.4 microvolts (equation (8))
942 (v) SAMPLE CIRCUIT CALCULATIONS 23.6
(vi) Conclusions
Selection of an input tube for a television
or F -M receiver is dependent upon many
varying circuit conditions and individual
requirements. The choice of using bal-
anced or unbalanced input, permeability or
capacitor tuning, noisy pentodes or quiet
triodes that possibly require neutralization,
among others, lies entirely with the design
engineer. Considering these reasons and
various engineering and economic compro-
mises no particular tube can be chosen and
defined as the input tube. Complete
noise information about the circuits in-
volved is necessary, however, as this is one N ilgnol B
(b) FIG. 23.26
of the determining factors for good sensi-
Fig. 23.26. Typical F -M receiver
tivity and signal -to -noise ratio input circuit (Ref. B33).
Obviously, even for values of C,n less than those obtained from the above expres-
sion, there would be a marked effect on the gain and selectivity of the circuits before
oscillation actually started.
(iii) Summary
Possible sources of feedback giving rise to instability can often be predicted from
the circuit and component layout diagrams of a receiver. An estimate can be made
of the possible magnitudes of many of the undesired voltages involved.
It should be evident that it is far better to avoid possible feedback and instability
by good electrical and mechanical design rather than spend many fruitless hours
tracking down an oscillation which could have been avoided.
A most helpful discussion of instability problems is given in E. E. Zepler's " Tech-
nique of Radio Design " (in particular Chapter 9) and the reader is recommended to
consult this book as an excellent practical guide.
Some further considerations will be given to circuit instability in Chapter 26, in
connection with i -f amplifiers. These circuits, being fixed tuned, are usually more
amenable to calculation of possible instability than r-f stages and the results obtained
more closely approximate to the practical set -up.
It is necessary to mention that at very high frequencies (say above 50 Mc /s or so)
the inductance of the screen-grid lead in screen -grid tetrodes and pentodes can cause
an apparent change in plate to control -grid capacitance which is often sufficient to
cause instability. If the screen is earthed by means of a capacitor, it is often possible
to select a capacitance value which will be series resonant with the screen lead in-
ductance, at the working frequency, and so form a low impedance path to ground.
This arrangement is often sufficient to prevent instability from this cause, even over
a range of frequencies. For further discussion see Ref. A8.
It is also important to note that at the higher frequencies a capacitor does not behave
as a pure capacitance, and its effective inductance and resistance become increasingly
important as the frequency increases. With many types of capacitors it is possible
that the inductive reactance will exceed the capacitive reactance even at frequencies
as low as 30 Mc /s (with electrolytics an additional r-f by -pass should always be used).
This effect is sometimes used to make the capacitor series resonant at the working
frequency. Even if the capacitor behaves as a small inductance it will be appreciated
that in some cases the reactance can be very low and effective by- passing is still pos-
sible. In tuned circuits the Q can be materially affected by the increase in r-f resist-
ance of the capacitors and both this and the previous effects should be checked.
The effectiveness of by -pass capacitors and their Q at the working frequency can
be readily checked with the aid of a Q meter by placing the capacitor in series with
a coil of known inductance and Q. See Ref. B45. Finally, r-f chokes do not always
behave as such, and their resonant frequencies should always be checked (see also
Chapter 11 Sect. 6).
SECTION 8 : DISTORTION
(i) Modulation envelope distortion (ii) Cross modulation distortion.
distortion because of the selectivity of the tuned circuits. Third order terms (those
containing cubes) do introduce distortion of the modulation envelope.
R -F amplifier valves are usually designed so that the third order curvature of their
characteristics is minimized as far as possible. This is achieved by using a variable
pitch for the control grid winding and results in the well known variable -mu (or re-
mote cut -off) characteristic. The variable -mu (or more exactly " variable g,,, "
characteristic) and the resultant shape of the eo -
i,, curve, have a large bearing on
cross modulation distortion as will be discussed presently.
SECTION 9 : BIBLIOGRAPHY
(A) BOOKS DEALING WITH RADIO FREQUENCY AMPLIFIERS, AERIAL COUPLING,
NOISE, ETC.
Al. Sturley, K. R. " Radio Receiver Design " Parts 1 and 2 (Chapman and Hall, London, Part 1, 1943 ;
Part 2, 1945).
A2. Zepler, E. E. " The Technique of Radio Design " (Chapman and Hall, London, 1943, John Wiley
and Sons, New York, 1943).
A3. Everitt, W. L. " Communication Engineering " (2nd edit., McGraw -Hill, New York and London,
1937).
A4. Terman, F. E. " Radio Engineers' Handbook " (1st edit. McGraw-Hill, New York and London,
1943).
A5. Terman, F. E. " Radio Engineering " (2nd edit. McGraw -Hill, New York and London, 1937).
A6. Glasgow, R. S. " Principles of Radio Engineering " (McGraw -Hill, New York and London, 1936).
A7. Moullin, E. B. " Spontaneous Fluctuations of Voltage " (Oxford University Press, London, 1938).
A8. Harvey, A. F. " High Frequency Thermionic Tubes " (Chapter 2) John Wiley and Sons, New
York, 1943.
A9. Kiver, M. S. " F-M Simplified " (Chapter 12 -R.F. Tuners for F-M Receivers). D. Van Nos -
trand Co. Inc., Toronto, New York, London, 1947.
A10. Valley, G. E., and H. Waltman, " Vacuum Tube Amplifiers " (McGraw -Hill, New York, Toronto
and London, 1940).
946 REFERENCES 23.9
All. Thomas, H. A., and R. E. Burgess " Survey of existing information and data on radio noise over
the frequency range 1 -30 MC (His Majesty's Stationery Office, London, 1947).
Al2. Moxon, L. A. " Recent Advances in Radio Receivers " (Cambridge University Press, London,
1949).
A13. Goldman, A. " Frequency Analysis, Modulation and Noise " (McGraw -Hill Book Company,
1948).
A14. " Valve and Circuit Noise -survey of existing knowledge and outstanding problems " Radio
Research Special Report No. 20 (1951) -Dept. of Scientific and Industrial Research, London.
SECTION 1 : INTRODUCTION
In this chapter it is proposed to consider, briefly, some of the fundamental types of
oscillator circuits (used in conjunction with valves) which are commonly employed
in radio receivers. Of course there are numerous variations of the fundamental
circuits, but generally these changes are only a practical convenience for obtaining
some required special condition of operation from the basic circuit.
The types of circuits to be considered are :
(a) The tuned -plate oscillator
(b) The tuned -grid oscillator
(c) The Hartley oscillator
(d) The Colpitts oscillator
(e) The electron -coupled oscillator
(f) The negative transconductance oscillator.
For those interested in the general theory of oscillation, and in the many special cir-
cuits available, a list of suitable references is given at the end of the chapter which will
serve as a starting point, at least, for a more complete survey of this field of knowledge.
Crystal oscillator circuits are not discussed here but a number of useful circuits are
given in Refs. 27 (page 164), 26 (page 97) and 31 ; the latter giving a discussion of
overtone (harmonic mode) crystal oscillators.
Before proceeding to a discussion of particular types of oscillator circuits, a few of
the fundamental principles will be briefly reviewed.
The simplest form of electrical oscillator consists of a combination of inductance
(L) and capacitance (C) connected together (as shown in Fig. 24.1), to which has been
added, initially, electric or magnetic energy. Suppose that the
capacitor C has been charged by some means. The energy
stored in the capacitor is then ICE', where E is the maximum
potential difference between the plates. At this instant, when
E has its maximum value, the current in the circuit is zero.
FIG.24.1 The presence of the inductor will allow the energy stored
947
948 INTRODUCTION 24.1
in the electric field of the capacitor to be transferred, and to form a magnetic field
around the inductor. The capacitor discharges until finally E becomes zero and the
current I becomes a maximum. At the instant at which I is a maximum the energy
in the magnetic field is ¡LI2, all the available energy is stored in the magnetic field
and there is no electric field. The process now reverses, the magnetic field collapses
and energy is transferred back to the electric field of the capacitor. This process
repeats itself indefinitely if there is no loss of energy in the circuit (radiation is not
considered here).
Since the total energy which is stored in each field in turn, is the same, it is per-
missible to write
ICE2 = ILI2
(1)
where E and I have their maximum values.
Also E = 1 /wC (2)
and so 1 /w2C = L (3)
or f = 1 /27rA/LC (4)
f
where is natural resonant frequency of the oscillations occurring in the circuit.
Since there is always some resistance (R) present with practical circuit elements,
-
the amplitude of each successive oscillation will decrease until eventually all of the
energy is dissipated mainly in the form of heat in the resistance -and the oscilla-
tions will cease. (This is discussed further in Chapter 9, Sect. 1). The addition of
extra energy to the circuit from some external source, such that the added energy
equals that being lost, would allow the oscillations to continue indefinitely. In the
circuits to be considered the power supply is the external source of energy, and the
valve is the device controlling the energy which is added to the L C R circuit, in the
correct phase and amplitude, to maintain oscillations.
With any valve oscillator an exact analysis of the method of operation is very difficult,
if not impossible, and it is usual to treat the circuits as being linear (at least for simple
design procedure) although they depend on conditions of non -linearity for their opera-
tion. This simplification is valuable because the mathematical analysis which can
be carried out yields a great deal of useful information concerning the behaviour of
the circuits. That the circuit operation is non -linear can be readily appreciated by
considering the fact that the amplitude of the oscillations, once started, does not con-
tinue to build up indefinitely. The energy gain of the system reaches a certain am-
plitude and then progressively falls until equilibrium is established. The limits are
usually set by the valve-plate current cut-off occurs beyond some value of the nega-
tive grid voltage swing, and plate current saturation or grid current damping will
limit the amplitude of the grid swing in the positive direction. For a discussion of
the factors governing oscillation amplitude, the reader should consult Refs. 1 and 2.
In the sections to follow, typical circuits applicable to radio receivers will be dis-
cussed. It is not proposed to give a mathematical analysis or detailed physical ex-
planations of the operation of the circuits, as this has been more or less adequately
done in many text books and periodicals. Suitable references are 1, 3, 5, 7, 10, 11,
24.
At the outset it may be mentioned that, since high oscillator efficiency is not usually
as important a factor as some other requirements in radio receivers, the design is
generally a combination of empirical and experimental techniques. A large number
of the circuit component values are pre-determined by such considerations as tracking
the tuned circuit with signal circuits over a band of frequencies and maintaining
constant oscillator amplitude over the tuning range. Other factors will be discussed
in Sect. 9 of this chapter.
Design procedures such as those used for class " C " power oscillators in trans-
mitters are not carried out in detail for receiver oscillators, although obviously the
basic principles are the same in both cases ; this will receive further consideration in
Sect 9.
24.2 (i) TUNED -PLATE OSCILLATOR 949
(1) The rotor plates of the tuning capacitor can be directly earthed without using
additional series blocking capacitors directly in the tuned circuit.
(2) The resistor R1 can often be selected to assist in maintaining a constant ampli-
tude of oscillation over the tuning range, since it will give greater damping of the
tuned circuit as the frequency increases.
Either arrangement of the grid-leak resistor (Re) shown in circuits (B) and (C) is
satisfactory, but in circuit (C) it gives increased damping on the feedback winding
(II) which may be advantageous in maintaining constant grid voltage amplitude over
a range of frequencies.
6 Bi
FIG.24.2 TUNED - PLATE OSCILLATOR
Tight coupling and low values of mutual inductance (M) are helpful in maintaining
frequency stability with mains voltage variations, as these factors tend to make the
circuit relatively independent of valve constants other than interelectrode capacitances
(see Ref. 21). L1 is made as small as possible, to reduce M, and also to keep its
natural resonant frequency well above the tuning range ; this latter factor means
that stray capacitance across the feedback winding must be kept as small as possible.
The high natural resonant frequency for the feedback circuit is helpful in reducing
variations in the tuned circuit (due to reflected impedance), this being particularly
important since it reduces tracking error and limitations on the maximum possible
tuning range.
The resonant frequency (f) of the tuned -plate oscillator is given approximately by
f= 2a
11
LoG
/ 1
v r, =fo /
+ R
1 R
r,
I
(5)
The important point to observe is that other components, including the valve,
affect the frequency of oscillation. As these components are capable of variation with
voltage or temperature fluctuations they will affect oscillator stability, and their effects
should be minimized as far as possible. Some improvement in stability is possible
by making the grid current small, which agrees with the condition of a low value for
Lf, but this is also governed by the permissible value of the grid resistor (Ro).
The condition for the maintenance of oscillation is given by
M = -[µD +
-
where L,, G and R have the same meanings as previously,
(7)
Lo =
tuned circuit inductance
r9 =
plate resistance of valve
Lf =
feedback winding inductance
R =
series r -f resistance (total) in tuned circuit
and G =
total tuning capacitance.
The minimum value of M required to maintain oscillation is given approximately by
L,A GRl
M [ µ(1 + A) + gm J (9)
24.2 (ii) TUNED -GRID OSCILLATOR 951
(A) (e)
PIG. 24.4 HARTLE OSCILLATOR
The angular frequency of oscillation for the Hartley circuit of Fig. 24.4(A) (see
Ref. 5) is given approximately by
w = wo1/1 + R, /ra (10)
where wo = l /1/LoG
wo = 2+r x fo (fo is natural resonant frequency of tuned circuit)
R, = r-f resistance of L,
r, = plate resistance of valve
Lo = oscillator coil inductance (= L, + L, + 2M)
and G = tuning capacitance.
For the maintenance of oscillation
L, + M Gr,(R, + R,)L.
(11)
L, +M +(L, +MXL, +M)
where µ = amplification factor of valve.
L, and L,, are the two sections of Lo coupled by mutual inductance
M and R, and R, are the r-f resistances of L, and L2 respectively.
The frequency equation shows that the oscillation frequency f is higher than fo,
and to reduce this difference r, should be high and R, small. Also, for the circuit
to oscillate readily, the mutual conductance of the valve should be high.
A Hartley oscillator offers advantages over the tuned -plate and tuned -grid circuits
at frequencies in excess of about 40 Mc /s, the greatest advantage being that the feed-
952 (iii) HARTLEY OSCILLATOR 24.2
back winding, being a part of the tuned circuit, does not offer the same difficulties
as the other cases, in which the natural resonant frequency approaches the operating
frequency.
In the case of the other two types of oscillators mentioned, if the feedback winding
has greater inductance than the tuned winding, the oscillator can easily change over
from one type to the other (i.e. tuned -plate becomes tuned-grid and vice- versa) and
satisfactory tuning is obviously impossible. This trouble is not as unlikely as may be
thought since it is often very difficult to obtain sufficient amplitude of oscillation as
the frequency increases.
The Hartley circuit can be made to give satisfactory operation to frequencies at
least as high as 150 Mc /s, even with multi-grid converter valves (depending, of course,
on the valve type used), but its chief disadvantages are
(1) Its liability to parasitic oscillations as the frequency increases,
(2) The possibility of the circuit acting like a modified Colpitts oscillator because
of stray and valve interelectrode capacitances,
(3) That it is rather awkward to find the optimum tapping point for the best con-
ditions of oscillation, particularly when the frequency becomes fairly high.
This is particularly so with valve types such as the 6BE6 on the F-M broadcast
band since the tapping point on the oscillator coil has a large effect on conversion
gain.
(1) and (2) generally manifest themselves as sudden changes in oscillation frequency
when tuning over the working range, or as " dead " spots, or through the oscillator
stopping altogether when the gang capacitance becomes less than a certain value.
It may be mentioned finally that the Hartley circuit has been extensively used on
all broadcast bands including the F -M 88-108 Mc /s band ; its application in re-
ceivers has been largely confined to use with pentagrid converters such as types 6SA7,
6BE6, 6SB7 -Y and 6BA7.
R -r
CHOKE
e+
- (A)
fIG.24S COLPITTS OSCILLATOR
The circuit of Fig. 24.5(B) has been successfully used with type 6BE6 converters
tuning the 88-108 Mc /s F -M broadcast band and avoids the necessity for tapping the
oscillator coil as in the Hartley circuit.
The angular frequency of oscillation of the Colpitts oscillator, from the solution
of the circuit of Fig. 24.5(A), is given by
R C,
w- coo,
° 1 i r, C, + Ca /
(12)
where wD = 27r X fD (f;, is the natural resonant frequency of the tuned circuit alone)
R = r -f resistance of LD
rD = plate resistance of valve
µ = amplification factor of valve
Lo = oscillator coil inductance
and C, and C2 tune Lo ; they also form a capacitive voltage divider across Lo ; and
the excitation voltage on the grid is proportional to C, /(C, + C2).
This circuit is most convenient for use at v-h -f, it is very easy to make oscillate,
and is not so liable to parasitic oscillations as the Hartley circuit. Its use on the
medium and short-wave bands has been rather limited because of difficulties in circuit
arrangement when covering frequency bands of about 3 : 1, and also because of the
very simple and satisfactory manner in which the tuned -grid and tuned -plate circuits
can be made to operate over these ranges. The Colpitts circuit readily lends itself
to inductance tuning and with this arrangment has been used satisfactorily on all
of the frequency ranges encountered in normal broadcast receiver design.
(v) Electron-coupled oscillator
Electron-coupled oscillators (Refs. 1, 22) can use a large variety of fundamental
circuits, such as the Hartley or Colpitts, for generating oscillations. The circuit
usually employs a tetrode or pentode valve, or, most often in receivers, is used with a
pentagrid converter. Valves having suppressor grids will generally give better fre-
quency stability.
The fundamental principle is that the actual oscillating circuit is connected to the
load circuit only by means of the electron stream within the valve. In this way changes
in load conditions and high tension voltages have a reduced effect on the actual fre-
quency of oscillation. The improvement in frequency stability with voltage varia-
tions appears to be closely bound up with compensating effects due to simultaneous
voltage changes on the plate and screen. Variation in load conditions would, ideally,
leave the oscillation frequency unchanged since the two circuits are only connected
unilaterally through the electron stream. Actually, of course, interelectrode capacit-
ances and stray coupling prevent this condition from being completely fulfilled.
I -F Those pentagrid converters which com-
monly use the Hartley oscillator circuit
arrangement are typical examples of the
SIGNAL application of electron- coupling. One such
arrangement is shown in Fig. 24.6. It is
often advantageous in this type of circuit to
Et*
connect the screen to the rectifier output
through R as this reduces " flutter " ; the
FIG.24.6 PENTAGRID CONVERTER USING ELECTRON - high tension for the plate is obtained from
COUPLED HARTLEY OSCILLATOR the power supply in the usual manner.
With this latter arrangement C2 generally consists of an electrolytic capacitor and a
mica capacitor (to act as a r -f by -pass) connected in parallel.
Since this oscillator is a two terminal type it is often a very convenient arrangement ;
one particular example is its use as a beat -frequency oscillator in a radio receiver. A
particular form of the negative transconductance oscillator (Ref. 13)-which employs,
also, the principle of electron coupling -using a pentagrid converter valve (e.g. type
6A8, but not 6SA7, 6BE6 etc.) is shown in Fig. 24.7. The negative transconductance
is brought about as follows. Electrons moving towards the plate are turned back to
the inner screen (G,) and the oscillator anode (G,) when the control grid (G,) has a
more negative voltage applied to it. The net effect of an increase of negative voltage
on the signal grid is to increase the current to the oscillator-anode and to grid G,.
OUTPUT
50.000(1
2040011 .
8+ 250V
F IC. 24.7 NEGATIVE TRANSCONDUCTANCE OSCILLATOR
(i) General
Frequency drift in the oscillator section of superheterodyne receivers (F -M or
A-M) is important for a number of reasons. The most obvious is that as a result
of such drift the actual intermediate frequency output from the frequency changer
will not be that to which the i -f amplifier is tuned, giving a loss in amplification and
the added possibility of considerable frequency distortion in A-M receivers, and non-
linear distortion in F-M receivers. Other special effects will be discussed separately.
Oscillator frequency variations which can be offset in the original design may be
due to changes in
(a) Supply voltage
(b) Temperature
(c) Humidity
(d) A.V.C. bias on the signal grid of multielectrode converter valves. This
will be discussed in Chapter 25 Sect. 2.
Also, it should be noted that a high percentage of oscillator harmonics can lead to
oscillator instability. Mechanical stability of the circuit is a prime requirement.
Detailed discussion regarding oscillator frequency stability can be found in Refs.
1, 4, 5, 10, 11, 15, 19, 20, 21 and 28.
This is minimized by the use of high quality dielectrics such as special grades of por-
celain.
Inductance changes due to temperature rise also cause frequency variation, and
low -loss dielectric formers are essential. If sufficient mechanical stability is possible
air cored coils are often preferable, but not always.
Good circuit layout is an essential requirement and all sources of heat should be
kept away from tuning capacitors and coils.
When all the requirements as to dielectrics, circuit layout etc. have been met as
far as is possible, final temperature compensation is made by using capacitors having
negative temperature coefficients. These capacitors are available commercially in a
wide range of capacitance values, and with various negative temperature co- efficients.
The manufacture of this type of capacitor is possible because of the availability of
ceramic materials having dielectric constants which decrease with an increase in
temperature *. It should be realized that exact compensation by simple capacitance
adjustment is only feasible, in general, at one point in a given tuning range, but a very
appreciable improvement over a band of frequencies is possible without exact com-
pensation at any one point. A method for obtaining exact compensation at two points
(although here again a compromise adjustment may be preferable) in a given tuning
range, using the principles of superheterodyne tracking, has been given by Bushby
(Ref. 19) ; it is also shown that a compromise adjustment is possible, although exact
compensation is not practicable, at three points in the tuning range. Methods for
carrying out long period frequency-drift compensation, due to temperature changes,
are outlined in Ref. 19.
Changes in oscillator frequency with humidity can be minimized by using coils,
capacitors and insulating materials which are properly baked to remove moisture
and then impregnated (with suitable waxes or varnishes) to prevent the absorption of
moisture.t
It may be worth noting here that carbon resistors have negative temperature co-
efficients. Wire -wound resistors have positive or negative temperature co- efficients
depending on the type of resistance wire used.$
this frequency should not coincide with the bass resonant frequency of the loud speaker.
Often it is necessary to mount the capacitor on rubber and provide flexible leads to
the dial, etc.
(C) Microphonics caused by heater-cathode capacitance variations. This effect
is minimized by connecting one side of the heater directly to cathode, where this is
possible, and using a r-f choke in the other heater lead to avoid shunting the tuned
circuit with the heater circuit, e.g. see Fig. 24.6 where the result of omitting a r-f
choke would be to short out part of the tuning inductance.
Microphony in superhet. oscillators is covered in Ref. 32.
However, almost any oscillator circuit is suitable. Variation of the beat -frequency
is best accomplished by variation of the beat -oscillator tuning (say a range of ± 3
Kc /s) since it should be clear that the best conditions for reception normally occur
when the receiver is tuned exactly to the required carrier frequency.
The a.v.c. should not be operated by the beat oscillator as this would cause a serious
loss in receiver sensitivity. For this reason a.v.c. is often disconnected when hetero-
dyne reception of this type is being carried out.
Some receivers have the beat -oscillator voltage injected directly into the diode
circuit, but it does not necessarily follow that this is the most suitable point. At the
detector the incoming carrier voltage has its largest amplitude, and so the beat oscilla-
tor voltage also has to be correspondingly high. A simple solution which is helpful
in some cases is to inject the oscillator output voltage into the grid circuit of the last
i -f amplifier valve, which can be operated with fixed bias to provide constant gain.
Harmonic operation of the beat oscillator is sometimes helpful in obtaining im-
proved stability and in reducing undesired interference effects.
Considerations as to stability of operation are of prime importance, and the factors
previously discussed should be kept in mind when designing this circuit.
Useful data on beat oscillator circuits are given in Ref. 8 (page 128), and also in
Refs. 26 and 27.
SECTION 10 : BIBLIOGRAPHY
(1) Thomas, H. A. (book) " Theory and Design of Valve Oscillators " (Bibliography of 172 references)
Chapman and Hall, London, 2nd. ed. 1951.
(2) Pol van der, B. " The nonlinear theory of electric oscillations " Proc. I.R.E. 22.9 (Sept. 1934) 1051.
(Includes a bibliography of 87 items).
(3) Rider, J. F. (book) " The Oscillator at Work " John F. Rider Publisher Inc., New York, 1940.
(4) Colebrook, F. M. " Valve Oscillators of Stable Frequency -A critical survey of present knowledge "
H.M. Stationery Office, London (1934).
(5) Sturley, K. R. (book) " Radio Receiver Design "-Part 1 (Chapter 6) Chapman and Hall, London ;
Wiley and Sons, New York (1943).
Schlesinger, K. Cathode- follower circuits " Proc. I.R.E. 33.12 (Dec. 1945) 843 (Oscillating
(6)
H.
. " Theory and Applications of Electron Tubes " (Chapter 10) McGraw -Hill
(7) Reich, J. (book)
Book Co. Inc., New York and London (2nd edit. 1944).
(8) Zepler, E. E. (book) " The Technique of Radio Design " (Chapter 4) Chapman and Hall, London ;
Wiley and Sons, New York (1943).
(9) Beard, E. G. " Some notes on oscillating valve circuits " Philips Tec. Com. No. 9 (Oct. 1947) 6.
(10) Terman, F. E. (book) " Radio Engineers' Handbook " McGraw-Hill Book Co. Inc., New York
and London (1st edit. 1943).
(11) Terman, F. E. (book) "Radio Engineering" McGraw -Hill Book Co. Inc., New York and London
(2nd edit. 1937).
(12) Brunetti, C. " The Transitron oscillator " Proc. I.R.E. 27.2 (Feb. 1939) 88.
(13) " Negative transconductance oscillator -a useful circuit " Radiotronics No. 114 (July 1941) P. 48.
(14) Dedman, E. A. " Transitron oscillators " (letter) W.W. 49.5 (May 1943) 152 (gives circuit using
6A7, freq. limit 30 -40 Mc /s.).
(15) White, S. Y. " V -H-F receiver oscillator design " Elect. 16.7 (July 1943) 96.
(16) Tucker, D. G. " The synchronisation of oscillators " Electronic Eng. (Part 1) 15.181 (March 1943)
412.
(17) Butler, F. " Cathode coupled oscillators " W.E. 21.254 (Nov. 1944) 521. Letter, M. Felix 22.256
(Jan. 1945) 14.
(18) Adler, R. ' A study of locking phenomena in oscillators " Proc. I.R.E. 34.6 (June 1946) 351.
(19) Bushby, T. R. W. " Thermal frequency drift compensation " Proc. I.R.E. (Dec. 1942) 546. [Dis-
cussion 31.7 (July 1943) 385]. Also, A.W.A. Tec. Rev. 6.3 (1943) 14.
(20) ' Effect of Temperature on Frequency of 6J5 Oscillator " R.C.A. Application Note No. 108 (Nov.
13th, 1940). Reprinted Radiotronics No. 110, p. 15.
(21) Llewellyn, F. B. " Constant-frequency oscillators " Proc. I.R.E. 19.12 (Dec. 1931) 2063.
(22) Dow, J. B. " A recent development in vacuum tube oscillator circuits " Proc. I.R.E. 19.12 (Dec.
1931) 2095.
(23) Herold, E. W. " Negative resistance " (gives a bibliography of 55 items) Proc. I.R.E. 23.10 (Oct.
1935) 1201.
(24) Everitt, W. L. (book) " Communication Engineering " (Chapters 17 and 18) 2nd edit. McGraw -
Hill Book Co. Inc., New York and London (1937).
(25) Ratcliffe, J. A. (book) " The Physical Principles of Wireless " (5th edition) Methuen and Co. Ltd.,
London, 1941.
(26) ` The Radio Amateur's Handbook " 24th edit. A.R.R.L., Connecticut, 1947, and later editions.
(27) " The Radio Handbook " 10th edit. Editors and Engineers, Los Angeles, 1946.
(28) Miller, J. M. " Thermal drift in superheterodyne receivers " Elect. 10.11 (Nov. 1937) 24.
(29) Keen, A. W. " Negative resistance characteristics -graphical analysis " W.E. 27.321 (June 1950)
175.
(30) and M. McKenna with negative resistance load-measurement of
Tomes, D. M. .32. 990) Amplifier
(31) " Overtone crystal oscillator design " Elect. 23.11 (Nov. 1950) 88.
Additional references will be found in the Supplement commencing on page 1475.
CHAPTER 25
FREQUENCY CONVERSION AND TRACKING
Section Page
1. The operation of frequency converters and mixers ... 962
2. Converter applications ... ... ... 987
3. Superheterodyne tracking ... 1002
4. References ... ... 1017
SECTION 1 : THE OPERATION OF FREQUENCY CONVERTERS
AND MIXERS
(i) Introduction (ii) General analysis of operation common to all types (iii) The
oscillator section of converter tubes (iv) The detailed operation of the modulator or
mixer section of the converter stage (v) Conclusion (vi) Appendix.
Reprinted by special permission from the article by E. W. Herold entitled " The
operation of frequency converters and mixers for superheterodyne reception " in the
Proceedings of the I.R.E. Vol. 30 No. 2 (February 1942) page 84.
(i) Introduction
The better modern radio receivers are almost universally designed to use the super-
heterodyne circuit. In such a circuit, the received signal frequency is heterodyned
with the frequency of a local oscillator to produce a difference frequency known as the
intermediate frequency. The resultant signal is amplified by a selective, fixed -
tuned amplifier before detection. Since the heterodyne action is usually accomplished
by means of a suitable vacuum tube, it is the purpose of this paper to discuss the chief
similarities and differences among the tubes which might be used, as well as to explain
their behaviour.
The combination of signal and local- oscillator frequencies to produce an inter-
mediate frequency is a process of modulation in which one of the applied frequencies
causes the amplitude of the other to vary. Although this process was originally
called heterodyne detection and, later, first detection, it is now called frequency con-
version. The portion of the radio receiver which produces conversion may, therefore,
be identified as the converter. If conversion is accomplished in a single vacuum tube
which combines the functions of oscillator and modulator, this tube may logically be
termed a converter tube. When separate tubes are used for the oscillator and the
modulator portions of the converter, respectively, the tube for the latter purpose is
conveniently called a modulator or mixer tube. This terminology will be used in this
paper.
Although in some of the earliest superheterodynes, frequency conversion was
accomplished by a triode oscillator and triode modulator (Ref. 1) other circuits used
a single triode which served as both modulator and oscillator (Ref. 2). A triode used
in the latter way could, therefore, be called a converter tube. The introduction of
two-grid tubes (i.e. tetrodes) permitted a wide variety of modulator and converter
arrangements which frequently gave superior performance to that possible with
triodes (Ref. 3 -7).
When indirectly heated cathodes became more common, conversion circuits in
which the oscillator voltage was injected in the cathode circuit were used. These
962
25.1 FREQUENCY CONVERTERS-(i) INTRODUCTION 963
circuits reduced considerably the interaction between oscillator and signal circuits
which would otherwise be present. (Ref. 8). When tetrodes and pentodes became
available, the use of the triode was dropped except as the local oscillator. It was not
long, however, before the desirability of more complete separation of oscillator and
signal circuits became evident. Multigrid converter tubes were, therefore, devised
to permit this separation in a satisfactory manner, at least for the frequencies then
in common use. (Refs. 9 -14). In some of these it was also possible to control the
conversion gain by an automatic- volume-control voltage, a decided advantage. The
most satisfactory of the earlier multigrid tubes was known as the pentagrid converter,
a type still widely used. A similar tube having an additional suppressor grid is used
in Europe and is known as the octode.
When it became desirable to add high -frequency bands to superheterodyne receivers
which also had to cover the low broadcast frequencies, the converter problem became
more difficult. The highest practicable intermediate frequency appeared to be about
450 to 460 kilocycles, a value which was only about 2 per cent of the highest frequency
to be received. Its use meant that the oscillator frequency was separated from the
signal frequency by only 2 per cent and the signal circuit, therefore, offered appreci-
able impedance at the oscillator frequency. A phenomenon known as " space -
charge coupling," found in the pentagrid converter, indicated that signal and oscillator
circuits were not separated as completely as would be desirable. (Ref. 15). In ad-
dition, the permissible frequency variations of the oscillator had to be held to less
than the intermediate- frequency bandwidth, namely, 5 to 10 kilocycles ; at the highest
frequency to be received, the oscillator frequency was required therefore to remain
stable within 0.05 per cent. In the pentagrid converter, the most serious change in
oscillator frequency occurred when the automatic-volume -control voltage was changed,
and was sometimes as much as 50 kilocycles. Economic considerations have led to
the use of at least a three -to-one frequency coverage for each band in the receiver.
With capacitance tuning, the circuit impedance is very low at the low-frequency end
of the high- frequency band so that failure to oscillate was occasionally observed in
the pentagrid converter.
In Europe, where converter problems were similar, a tube known as the triode -
hexode (Ref. 16) was developed to overcome some of the disadvantages of the pentagrid
converter. In the pentagrid tube, the oscillator voltage is generated by, and therefore
applied to, the electrodes of the assembly closest to the cathode (i.e., the inner elec-
trodes). In the European form of triode -hexode, the oscillator voltage is generated by
a separate small triode section mounted on a cathode common to a hexode -modulator
section. The triode grid is connected internally to the third grid of the hexode
section. In this way, by the application of the oscillator voltage to an outer grid and
the signal to the inner grad of the modulator, space-charge coupling was greatly re-
duced and automatic -volume -control voltage could be applied to the modulator section
of the tube without seriously changing the oscillator frequency. In some European
types, a suppressor grid has been added so that such tubes should be called triode -
heptodes.
The first American commercial development to provide improved performance
over that of the pentagrid converter also utilized oscillator voltage injection on an
outer grid but required a separate tube for oscillator (Ref. 17). This development,
therefore, resulted in a modulator or mixer tube rather than a converter. There were
many advantages accompanying the use of separate oscillator tube so that such a
solution of the problem appeared to be reasonably satisfactory.
The demand arose shortly, however, for a one -tube converter system with better
performance than the original pentagrid type for use in the standard all -wave receiver.
A tube, the 6K8, in which one side of a rectangular cathode was used for the oscillator
and the other side was used for the mixer section, was developed and made available.
(Ref. 18). This tube used inner -grid oscillator injection, as with the pentagrid con-
verter, but had greatly improved oscillator stability. Another solution, also intro-
duced in the United States, was a triode -heptode which is an adaptation of the Euro-
pean triode -hexode. This type used outer -grid injection of the oscillator voltage
964 FREQUENCY CONVERTERS -(i) INTRODUCTION 25.1
where wo is the angular frequency of the local oscillator. Use of the cosine series
implies that the transconductance is a single-valued function of the oscillator elec-
trode voltage which varies as cos wot. When a small signal, e, sin w,t, is applied
to the tube, the resulting alternating plate current to the first order in e, may be
written
i, = g,e, sin w,t
co
= aoe, sin w,t -F e, a sin w,t cos nwot
n = 1
co
= aoe, sin w,t + ,¡e, X a sin (w, + nwo)t
n = 1
-
If a circuit tuned to the frequency (w, n wo) is inserted in the plate, the modulator
tube converts the incoming signal frequency w, to a useful output at an angular fre-
quency (w, - nwo) which is called the intermediate frequency. Since n is an in-
teger, it is evident that the intermediate frequency, in general, may be chosen to be
the difference between the signal frequency and any integral multiple of the local-
oscillator frequency ; this is true even though a pure sine -wave local oscillation is
applied to the tube. The harmonics of the local -oscillator frequency need only be
present in the time variation of the signal- electrode transconductance. The ordinary
conversion transconductance is simply a special case when n = 1. The conversion
transconductance at the nth harmonic of the local oscillator is give.. .y
(ii) GENERAL ANALYSIS OF OPERATION 965
25.1
iWs - WO a
ge = 2
Substituting the value of the Fourier coefficient a it is found that
27r
gefl = 27r
g, cos n word( wot).
o
previously
When n is set equal to unity, this expression becomes identical with the one
derived. (Ref. 17).
of
Thus, the conversion transconductance is obtained by a simple Fourier analysis time.
the signal- electrode -to- output -electrode transconductance as a function of
B
Fig. 25.1. Signal - electrode
transconductance versus oscil-
lator- electrode voltage for a
typical mixer tube. The r-
applied oscillator voltage is
shown at A and B is the
resulting time variation of OSCILLATOR ELECTRODE VOLTS
transconductance.
FIG. 25.1
Such an analysis is readily made from the tube characteristics directly by examination
of the curve of signal -electrode transconductance versus oscillator -electrode voltage.
The calculation of the conversion transconductance at the nth harmonic of the os-
cillator is made from this curve by assuming an applied oscillator voltage and making
a Fourier analysis of the resulting curve of transconductance versus time
for its nth
harmonic component. The analysis is exactly similar to the one made of power
output tubes, except that, in the latter case, the plate- current- versus-control -electrode-
voltage curve is used. Fig. 25.1 shows a curve of signal -grid transconductance
versus oscillator -electrode voltage for a typical modulator or mixer tube. In
the
usual case, the oscillator voltage is applied from a tuned circuit and so is closely
sinusoidal in shape as at A in the figure. The resulting curve of transconductance used
versus time is shown at B. Any of the usual Fourier analysis methods may be
to determine the desired component of curve B. Half of this value is the conversion
transconductance at the harmonic considered. Convenient formulas of sufficient
oscillator
accuracy for many purposes follow. Referring to Fig. 25.2a, a sine -wave
voltage is assumed and a seven -point analysis is made (i.e., 30- degree intervals).
The conversion transconductances ge are
g,,1 =
1
12 [(87
- g2) + 1.73 (g, - g,)1
- g1) + (g,
as in-
The values g1, g2, etc., are chosen from the transconductance characteristic
dicated in Fig. 25.2a. The values computed from the above formulas are, of course,
the
most accurate for g,1 and of less accuracy for gee while a value computed from
formula for g,, is a very rough approximation.
966 (ii) GENERAL ANALYSIS OF OPERATION 25.1
on the electrode currents is so small that usually it may be neglected. Thus, the
conversion transconductance of these converter tubes may be found exactly as if the
tube were a modulator or mixer, only.
With the circuit of Fig. 25.3 (Refs. 19 -21) a Hartley oscillator arrangement is used
and oscillator-frequency voltage is present on the cathode. The effect of such a
voltage is also to demodulate the electron stream through the action of the alternating
cathode potential on the screen-to- cathode and signal-grid -to- cathode voltages.
When a relatively high-transconductance signal grid is present, as in the figure, this
demodulation is considerably greater than in the normal cathode -at- ground circuit.
In order to determine the conversion
transconductance of a tube to be used in
SIGNAL
__r I.F. OUTPUT
'
under amplifier conditions is readily derived from theory (Ref. 24), the latter pro-
cedure is convenient. Thus, if i, is the mean- squared noise current in the output
of the converter or mixer tube considered as an amplifier (i.e. steady direct voltages
applied) the mean- squared intermediate- frequency noise is
1 r2n
2
ft -te 2a iv 'd(cot)
'
are usually proportional to these quantities. Thus, the converter-stage output noise,
which is the average of i, over the oscillator cycle, is usually proportional to the aver-
age electrode currents and average transconductance when the oscillator is applied.
Specific examples will be given in following sections of this paper treating typical
modes of converter operation.
Tube noise is conveniently treated by use of an equivalent grid- noise -resistance
concept whereby the tube noise is referred to the signal grid. The equivalent noise
resistance of a converter or mixer tube is
ii-f2
R°0
(4kTRdf)gc8
968 (ii) GENERAL ANALYSIS OF OPERATION 25.1
where k = 1.37 x 10-23, TR is room temperature in degrees Kelvin, and df
is the
effective over-all bandwidth for noise purposes. Since df is invariably
associated
with Ì;_,2, the bandwidth cancels in the determination of R eQ which
is one of the
advantages of the equivalent- resistance concept. For TR = 20 degrees
centigrade,
1 1;_2
R e, --= 0.625 x 1040
gent df
A summary of values of R e, for common types of converter will be
found in a pre-
ceding paper. (Ref 23).
The equivalent noise resistance R e alone does not tell the entire story as
regards signal -to-noise ratio, particularly at high frequencies. For example, if the
converter stage is the first stage of a receiver, and bandwidth is not a consideration,
the signal energy which must be supplied by the antenna to drive it will be inversely
proportional to the converter-stage input resistance. On the other hand, the noise
energy of the converter or mixer tube is proportional to its equivalent noise resistance.
The signal -to-noise ratio therefore, will vary with the ratio of input resistance to
equivalent noise resistance, and this quantity should be as high as possible. When
bandwidth is important, the input resistance should be replaced by the reciprocal
of the input capacitance if it is desired to compare various converter systems for signal-
to-noise ratio.
There are three methods of operation of mixer or modulator tubes. The oscillator
voltage may be put on the same grid as the signal voltage, it may be put on the inner
grid (the signal applied to an outer grid), or it may be impressed on an outer grid
(with the signal on the inner grid). Each of these modes of operation has charac-
teristics which depend on the mode rather than on the tube used in it. Tubes which
may be used in any one mode differ from one another mainly in the degree in which
they affect these characteristics. The treatment to follow, therefore, will not neces-
sarily deal with specific tube types : instead, the phenomenon encountered will be
illustrated by the use of data taken on one or more typical tubes for each of the modes
of operation.
A. Tubes with oscillator and signal voltages applied to same grid
Typical tubes used for this type of operation are triodes and pentodes. The
oscillator voltage may be introduced in series with the signal voltage, coupled to the
signal input circuit inductively, capacitively, and /or conductively, or 't may be
coupled into the cathode circuit. In all but the last case, by operating below the grid -
current point, the oscillator circuit is not loaded directly by the mixer tube. When
cathode injection is used, however, an effective load equal to the mean cathode con-
ductance (slightly greater than the mean transconductance) is imposed on the oscillator
circuit. The cathode injection circuit has the advantage that oscillator-frequency
voltage between the signal input circuit and ground is minimized, thus reducing
radiation when the converter stage is also the first stage of the receiver.
A typical transconductance -versus-bias curve
for a variable -g radio -frequency pentode is shown
in Fig. 25.4. The use of the Fourier analysis for
conversion transconductance at oscillator funda-
mental indicates that a value of approximately a
quarter of the peak transconductance can be at-
tained. Because of the tailing off of the lower
end of the curve, highest conversion transconduct-
ance requires a large oscillator swing. Very nearly
the maximum value is obtained, however, at an op-
erating bias shown by the dotted line, with an os-
cillator peak amplitude approximately equal to
the bias. With lower oscillator amplitudes, and
the same fixed bias, the fundamental conversion -is -Io -s o
transconductance drops in approximate propor- CONTROL GRID BIAS VOLTS
FIG.25.4
tion to the oscillator amplitude Fig. 25.4. Transconductance
Strictly speaking, when the cathode injec- characteristic of a typical variable-
tion type of operation is used the effect of µ, radio frequency pentode.
the oscillator voltage which is impressed between screen and cathode, and plate
and cathode should be considered. Practically, however, there is little difference
over the simpler circuit in which the oscillator voltage is impressed on the signal grid
only. It is for this reason that the cathode- injection circuit is placed in the same
category as those in which the oscillator voltage is actually impressed on the same
electrode as the signal.
In a practical circuit the effective oscillator voltage is, of course, the oscillator voltage
actually existent between grid and cathode of the tube. When the oscillator voltage
is impressed in series with the signal circuit or on the cathode, this effective voltage
is different from the applied oscillator voltage by the drop across the signal circuit.
In the usual case, with the oscillator frequency higher than the signal frequency, the
signal circuit appears capacitive at oscillator frequency. This capacitance and the
grid -to- cathode capacitance, being in series, form a capacitance divider and reduce
the effective oscillator voltage. The reduction would not be a serious matter if it
remained a constant quantity ; but in receivers which must be tuned over an appreci-
able frequency range this is not the case. The result is a variation in conversion
gain over the band. A number of neutralizing circuits have been described in the
970 (iv) MIXER SECTION OF CONVERTER STAGE 25.1
patent literature which are designed to reduce the oscillator- frequency voltage across
the signal circuit and thus minimize the variations. (Refs. 27, 28).
Coupling of the oscillator voltage into, or across, the signal circuit is also accom-
panied by changes in effective oscillator voltage when the tuning is varied. These
changes are not so great with pure inductive coupling as with pure capacitance coup-
ling. In many practical cases, both couplings are present.
A method of reducing the variation of conversion gain with effective oscillator
voltage in tubes in which oscillator voltage and signal areplaced on the same grid,
employs automatic bias. Automatic bias may be obtained either by a cathode self-
bias resistor (by- passed to radio frequency) or by a high -resistance grid leak, or both.
An illustration of the improvement which may be obtained in this way is shown in
Fig. 25.5. Three curves of conversion transconductance, at oscillator fundamental,
against effective peak oscillator volts are shown for the typical variable -µ pentode of
Fig. 25.4 used as a mixer. For the curve a, a fixed bias was used at approximately
an optimum point. The curve is stopped at the grid-current point because operation
beyond this point is not practicable in a receiver. Curve b shows the same tube
operated with a cathode self-bias resistor. This curve is also stopped at the grid-
current point. Curve c shows operation with a high -resistance grid leak. It is
evident that, above an oscillator voltage of about 3, curve b is somewhat flatter, and c
is considerably flatter than the fixed -bias curve a. The high-resistance grid lead
used for c may be made a part of the automatic-volume -control filter but care must
be taken that its value is considerably higher than the resistance in the automatic -
volume-control circuit which is common to other tubes in the receiver. If this is
not done, all the tubes will be biased down with large oscillator swings. When a
high-resistance leak is used, the automatic -volume-control action does not begin
in the mixer tube until the automatic -volume -control bias has exceeded the peak
oscillator voltage. Because of the high resistance of the leak, the signal circuit is
not loaded appreciably by the mixer tube. In a practical case, precautions must be
taken that a pentode in the converter stage is not operated at excessive currents when
accidental failure of the oscillator reduces the bias. A series dropping resistor in the
screen -grid supply will prevent such overload. When a series screen resistor is
used, the curve of conversion transconductance versus oscillator voltage is even flatter
than the best of the curves shown in Fig. 25.5. Series screen operation, therefore,
is highly desirable (Ref. 23).
One of the effects of feedback through interelectrode capacitance in vacuum tubes
is a severe loading of the input circuit when an inductance is present in the cathode
circuit. Thus, in mixers using cathode injection, the signal circuit is frequently
heavily damped since the oscillator circuit is inductive at signal frequency in the
usual case. The feedback occurs through the grid -to- cathode capacitance and can
be neutralized to some extent by a split cathode coil with a neutralizing capacitance.
(Ref. 28). Such neutralization also minimizes the voltage drop of oscillator frequency
across the signal circuit.
Loading of the signal circuit by feedback from the plate circuit of modulators or
mixers may also be serious when the signal- grid -to-plate capacitance is appreciable.
This is especially true when a low capacitance intermediate- frequency circuit, which
presents a comparatively high capacitive reactance at signal frequency, is used, as in
wide -band intermediate-frequency circuits. The grid -plate capacitance of radio -
frequency pentodes is usually small enough so that the effect is negligible in these
tubes. In triodes, however, feedback from the intermediate -frequency circuit may
be serious and the grid -plate capacitance should be minimized in tube and circuit
design. Although neutralization is a possible solution to the plate feedback, a more
promising solution is the use of a specially designed intermediate-frequency circuit
which offers a low impedance at signal frequency by the equivalent of series tuning
and yet causes little or no sacrifice in intermediate-frequency performance.
At high frequencies, the converter stage exhibits phenomena not usually observable
at low frequencies. One group of phenomena is caused not by the high operating
frequency, per se, but rather by a high ratio of operating frequency to intermediate
971
25.1 (iv) MIXER SECTION OF CONVERTER STAGE
óS
12000 300
Z
b
F IS00
200
Ó H
U
o I000
Z
io
Io
a
I- S00
z a
o Z
a
w
8 10 12 -IS -10 -S 00
2 4 6
Z VOLTS CONTROL GRID BIAS VOLTS
PEAK OSCILLATOR
ú FIG. 25.5 FIG. 25.6
gee 1 + a
where gm is the average control-grid -to -plate transconductance (averaged over an
oscillator cycle), I,. is the average screen current, g, is the conversion transconductance,
and a is the ratio of the screen current to plate current. Valuable additions to the
above relation are given by formulas which enable a simple calculation of noise re-
sistance from amplifier data found in any tube handbook. These additional relations
are approximations derived from typical curve shapes and are based on the maximum
peak cathode current lo and the maximum peak cathode transconductance go. The
data are given in Table 1. It has been assumed that oscillator excitation is approxi-
mately optimum. In this table, E,0 is the control-grid voltage needed to cut off the
plate current of the tube with the plate and screen voltages applied, and a is the ratio
of screen to plate current.
TABLE 1
Mixe[ Noise of Triodes and Pentodes
(Oscillator and Signal both Applied to Control Grid
As an example of the use of the table, suppose it is desired to find the equivalent
noise resistance of a particular triode operated as a converter at the oscillator second
harmonic. The local oscillator can be permitted to swing the triode mixer grid to
zero bias. With a plate voltage of 180 volts and zero bias, the tube data sheet shows
a transconductance, go = 2.6 x 10-8 mho. Thus the equivalent noise resistance is
31 /go or 12 000 ohms and the conversion transconductance at second harmonic is
0.13 go, or 340 micromhos. Since, with this plate voltage the tube cuts off at about
8 volts, a peak oscillator voltage of around 12 volts will be required.
The above table may also be used to obtain a rough estimate of the input loading
of pentode or triode mixers, since the high -frequency input conductance is roughly
proportional to the average transconductance g,,, and to the square of the frequency.
Thus, if the loading at any transconductance and frequency is known, the loading as
a mixer under the conditions of the table may quickly be computed.
2,000
o
-20 -10 o +10
Not GRID BIAS VOLTS
FIG.25. B
direction and which cuts off this transconductance over slightly less than half the
cycle, in the other. Curves of conversion transconductance against peak oscillator
voltage are shown in Fig. 25.9. Curve a is for fixed -bias operation of the oscillator
grid, curve b is with a high -resistance (i.e. several megohms) grid leak and condenser
for bias, and curve c is with the recommended value of grid leak (50 000 ohms) for
this type of tube. It is seen that best operation is obtained with the lower resistance
value of grid leak. With this value, the negative bias produced by rectification in the
grid circuit is reduced enough to allow the oscillator grid to swing appreciably positive
over part of the cycle. An incidental advantage to the use of the low-resistance leak
when the tube is self-oscillating (i.e., a converter) is that undesirable relaxation os-
cillations are minimized.
i
600
In mixers or converters in which the oscillator voltage is present on both the cathode
and the oscillator grid in the same phase (e.g. Fig. 25.3) it is usually necessary to
utilize a relatively sharp cut off in the design of the oscillator grid so as to cut off the
cathode current when the signal grid is positive (Ref. 19). By this means, the signal
grid is prevented from drawing current. At the same time, however, the high currents
needed for a high peak value of signal-grid transconductance cannot be obtained
without a greater positive swing of the oscillator grid than with a more open oscillator
grid structure. Thus, it is clear that it is desirable to have a negative bias on the
oscillator electrode which is considerably smaller than the peak oscillator voltage.
For this reason, optimum results are obtained on these tubes with very low values of
oscillator grid leak (e.g. 10 000 to 20 000 ohms).
The effects of feedback through the interelectrode capacitance are small in well-
designed multigrid mixers and converters of the kind covered in this section. The
signal- grid -to-plate capacitance is usually small enough to play no part in the opera-
tion ; even with a high L -to -C ratio in the intermediate-frequency transformer, the
capacitive reactance of the intermediate-frequency circuit at signal frequency is only
a very small fraction of the feedback reactance. The other interelectrode capacitance
which plays some part in determining circuit performance (excluding, of course,
the input and output capacitances) is the capacitance from the oscillator electrode or
electrodes to the signal grid. This capacitance is a source of coupling between these
two circuits. In well -designed converter or modulator tubes of the type discussed
in this section, however, the coupling through the capacitance may be made small
compared with another form of internal coupling known as " space -charge coupling"
which will be treated later in this discussion.
Coupling between oscillator and signal circuits is of no great consequence except
when an appreciable voltage of oscillator frequency is built up across the signal-grid
circuit. This is not usually possible unless the signal circuit is nearly in tune with
the oscillator as it is when a low ratio of intermediate frequency to signal frequency
is used. The effect of oscillator- frequency voltage induced across the signal circuit
depends on its phase; the effect is usually either to increase or to decrease the relative
modulation of the plate current at oscillator frequency and so to change the con-
version transconductance. This action is a disadvantage, particularly when the
amount of induced voltage changes when the tuning is varied, as usually occurs.
25.1 (iv) MIXER SECTION OF CONVERTER STAGE 975
In some cases, another effect is a flow of grid current to the signal grid ; this may
happen when the oscillator-frequency voltage across the signal -grid circuit exceeds
the bias. Grid current caused by this effect can usually be distinguished from grid
current due to other causes. By- passing or short -circuiting the signal -grid circuit
reduces the oscillator-frequency voltage across the signal-grid circuit to zero. Any
remaining grid current must, therefore, be due to other causes.
Current to a negative signal grid of a tube operated with inner -grid oscillator in-
jection is sometimes observed at high frequencies (e.g. over 20 megacycles) even when
no impedance is present in the signal -grid circuit. This current is caused by elec-
trons whose effective initial velocity has been increased by their finite transit time
in the high-frequency alternating field around the oscillator grid. These electrons
are then able to strike a signal grid which is several volts negative. The magnitude
of the signal -grid current is not usually as great as with tubes applying the oscillator
voltage to an outer grid* although it may prevent the use of an automatic- volume-
control voltage on the tube.
An investigation of coupling effects in the pentagrid converter showed that the
coupling was much larger than could be explained by interelectrode capacitance.
It was furthermore discovered that the apparent coupling induced a voltage on the
signal circuit in opposite phase to that induced by a capacitance from oscillator to
signal grid. (Ref. 15). The coupling which occurred was due to variations in space
charge in front of the signal grid at oscillator frequency. A qualitative explanation
for the observed behaviour is that, when the oscillator-grid voltage is increased, the
electron charge density adjacent to the signal grid is increased and electrons are re-
pelled from the signal grid. A capacitance between the oscillator grid and the signal
grid would have the opposite effect. The coupling, therefore, may be said to be
approximately equivalent to a negative capacitance from the oscillator grid to the signal
grid. The effect is not reversible because an increase of potential on the signal grid
does not increase the electron charge density around the oscillator grid. If anything,
it decreases the charge density. The equivalence to a negative capacitance must be
restricted to a one -way negative capacitance and, as will be shown later, is restricted
also to low-frequency operation.
In general, the use of an equivalent impedance from oscillator grid to signal grid
to explain the behaviour of " space- charge coupling " is somewhat artificial. A
better point of view is simply that a current is induced in the signal grid which de-
pends on the oscillator-grid voltage. Thus, a transadmittance exists between the
two electrodes analogous to the transconductance of an ordinary amplifier tube.
Indeed, the effect has been used for amplification in a very similar manner to the use
of the transconductance of the conventional tube. (Ref. 32, 33).
It is found that the transadmittance from the oscillator to the signal electrode
Y,,,o_, is of the form
Ymo -a = kt w2 + jk, w.
At low frequencies (i.e., lei w2 <
k,w) the transadmittance is mainly a transuscept-
ance but, as the frequency rises, the transconductance component ktw2 becomes of
more and more importance, eventually exceeding the transusceptance in magnitude.
The early work on " space -charge coupling " indicated that the effect was opposite
to that of a capacitance connected from oscillator to signal grid and could be cancelled
by the connection of such a capacitance of the correct value (Refs. 15, 34). The effect
of such cancellation could be only partial, however, since only the transusceptance
was balanced out by this arrangement. For complete cancellation it is also necessary
to connect a conductance, the required value of which increases as the square of the
frequency, between the oscillator grid and the signal grid so that the transconductance
term is also balanced out. (Ref. 35, 36).
The cancellation of " space -charge coupling " may be viewed in another way.
A well -known method of measuring the transadmittance of a vacuum tube is to con-
nect an admittance from control grid to output electrode and to vary this admittance
until no alternating -current output is found with a signal applied to the control grid.
The next part of this section contains a more detailed discussion of signal-grid current in outer -grid
oscillator injection tubes.
976 (iv) MIXER SECTION OF CONVERTER STAGE 25.1
(Ref. 37). The external admittance is then equal to the transadmittance. In exactly
the same way, the transadmittance which results from the space- charge coupling may
be measured. As a step further, if an admittance can be found which substantially
equals the transadmittance at all frequencies or over the band of frequencies to be
used, this admittance may be permanently connected so as to cancel the effects of
space -charge coupling. As has been previously stated, the admittance which is re-
quired is a capacitance and a conductance whose value varies as the square of the
frequency. Such an admittance is given to a first approximation by the series con-
nection of a capacitance C and a resistance R. Up to an angular frequency c, _
0.3 /CR the admittance of this combination is substantially as desired. At higher
values of frequency, the conductance and susceptance fail to rise rapidly enough and
the cancellation is less complete. Other circuits are a better approximation to the
desired admittance. For example, the connection of a small inductance, having the
value L = 1 /2CR2, gives a good approximation up to an angular frequency w _
0.6 /CR. The latter circuit is, therefore, effective to a frequency twice as high as the
simple series arrangement of capacitance and resistance. Inasmuch as in some cases
the value of inductance needed is only a fraction of a microhenry, the inductance may
conveniently be derived from proper proportioning and configuration of the circuit
leads.
+10
FIG. 25.10
Fig. 25.10. Signal grid (grid No. 4) admittance of a typical mixer for inner-
grid injection of oscillator at 31.5 megacycles. Curves taken with nodesigned
oscillator voltage
applied. Data represents electronic admittance only (i.e. " cold" values were subtracted
from measured values before plotting.)
It is of interest to note the order of magnitude of the transadmittance which is
measured in the usual converter and mixer tubes. (Refs. 26, 35, 36, 38). In the
formula for Y,,0_, given above, k, is in the neighborhood of 10 -E1 and k, is
around
10- 12. Cancellation is effected by a capacitance of the order of one or two micro
-
microfarads and a series resistance of 500 to 1000 ohms.
The correct value of the cancelling admittance may be found experimentally by
adjustment so that no oscillator voltage is present across the signal -grid circuit when
the latter is tuned to the oscillator frequency. Another method which may be used
is to observe either the mixer or converter plate current or the oscillator grid
current
25.1 (iv) MIXER SECTION OF CONVERTER STAGE 977
as the tuning of the signal is varied through the oscillator frequency. With proper
adjustment of the cancelling admittance there will be no reaction of the signal- circuit
tuning on either of these currents.
There are two disadvantages which accompany the cancellation of space- charge
coupling as outlined. In the first place, the signal -grid input admittance is increased
by the cancelling admittance. This point will be brought up again after discussing
the input admittance. The second disadvantage is that the oscillator frequency shift
with voltage changes in converter tubes may be somewhat increased by the use of
this cancellating admittance. When separate oscillator and mixer tubes are used,
the latter effect may be made less serious.
The next point to be considered is the input admittance of the signal grid. Signal-
grid admittance curves of a typical modulator designed for use with the oscillator
voltage impressed on the first grid are shown under direct-current conditions (i.e.,
as a function of oscillator -grid bias for several values of signal -grid bias) in Fig. 25.10.
The admittance is separated into conductive and susceptive components, the latter
being plotted in terms of equivalent capacitance. The admittance components of
the " cold " tube (no electrons present) have been subtracted from the measured value
so that the plotted results represent the admittance due to the presence of electrons
only. The data shown were taken at 31.5 megacycles with a measuring signal which
did not exceed 1.0 volt peak at any tithe. A modified Boonton Q meter was used
to take the data. It should be noted that the presence of a marked conductive com-
ponent of admittance is to be expected at frequencies as high as those used.
-1000
o
X
100
Fig. 25.11. Signal -grid (grid No. 4)
conductance of a typical mixer
o designed for inner -grid injection of
oscillator. Lines are drawn with
slope of 2. Curve a taken with
E,, = 0, E,, = -3 volts. Curve
10
b taken with E, = -6, E,Q =
-6 volts.
10 100
FREQUENCY (MEGACYCLES)
FIG. 25.11
The most striking feature of the data of Fig. 25.10 is that both susceptive and con-
ductive components are negative over a large portion of the characteristic. The
Appendix discusses this feature in somewhat more detail. The measurements show
that the susceptive component is analogous to a capacitance. The capacitance curves
given are independent of frequency up to the highest frequency used (approximately
50 megacycles). The conductive component, on the other hand, increases as the
square of the frequency also up to this frequency. The conductance is, therefore,
negative even at very low frequencies although its magnitude is then very small.
Thus, the conductance curves of Fig. 25.11 are valid for any frequency by multi-
plication of the conductance axis by the square of the ratio of the frequency considered,
to the frequency used for the data (i.e., 31.5 megacycles). Data taken at various
frequencies for two particular values of grid bias voltage E,, are plotted in Fig. 25.11.
The square -law relation is shown to check very closely.
Fig. 25.10 should be considered remembering that the oscillator voltage is applied
along the axis of abscissas. Considering an applied oscillator voltage, the admittance
978 (iv) MIXER SECTION OF CONVERTER STAGE 25.1
curves must be integrated over the oscillator cycle to find the admittance to the signal
frequency. The operation is just as if the tube were an amplifier whose input ad-
mittance is periodically varied over the curve of Fig. 25.10 which corresponds to the
signal -grid bias which is used. Curves of the modulator input conductance at 31.5
megacycles for various applied oscillator voltages are shown in Fig. 25.12. The
oscillator -grid bias is obtained by means of the recommended value of grid leak for
the tube (50 000 ohms). Curves are shown for two values of signal-grid bias volt-
age E,,. As before, data for other frequencies are obtained by multiplying the con-
ductance by the square of the frequency ratio.
The practical effect of the negative input admittance in a circuit is due to the con-
ductive portion only, inasmuch as the total input capacitance remains positive in
general *. An improved image ratio, and somewhat greater gain to the converter
signal grid over other types of modulator is to be expected when this type of oscillator
injection is used. At high frequencies, when a comparatively low intermediate fre-
quency is used, it is usually desirable to cancel the space -charge coupling of the tube
in the manner previously discussed. When this cancellation is made reasonably
complete by the use of a condenser and resistor combination connected from the
oscillator grid to the signal grid, the losses in this admittance at signal frequency are
usually sufficient to wipe out the negative input admittance. The net positive input
conductance however is often less than that found with other types of mixer.
The change in signal -grid input capacitance with automatic-volume -control is
small in this type of modulator, particularly with the larger values of oscillator swing
because of the integrating action of the oscillator voltage.
The fluctuation noise which is found in the output of inner -grid oscillator -injection
mixers and converters is not readily evaluated quantitatively. The fluctuation noise
is primarily due to current-distribution fluctuations but is complicated by the possi-
bility of a virtual cathode ahead of the signal grid. Data have been taken, however,
which indicate some degree of proportionality between the mean-square noise current
and the plate current. The signal -to -noise ratio for this type of modulator is, there-
fore, approximately proportional to the ratio of conversion transconductance to the
square root of the plate current. It is considerably less than for the pentode modulator
with both signal and oscillator voltages on the control grid.
The noise of the converter or mixer with oscillator on an inner grid may be ex-
pressed in terms of an equivalent grid resistance as
20I,
RR F E
ge 2
where 7, is the operating plate current, g, is the conversion transconductance, and FE
is a factor which is about 0.5 for tubes with suppressor grids and at full gain. For
tubes without suppressor or for tubes whose gain is reduced by signal -grid bias, FE
is somewhat larger and approaches unity as a maximum. With this mode of opera-
tion there is not so much value in expressions for R
based on maximum trans-
conductance and maximum plate current because these quantities are neither available
nor are they easily measured. For operation at second or third harmonics of the
oscillator (assuming optimum oscillator excitation) the plate current I6 and the con-
version transconductance g, are roughly + or 1/3, respectively, of their values with
fundamental operation so that the equivalent noise resistance for second- harmonic
and third- harmonic operation is around two and three times, respectively, of its
value for fundamental operation.
periodically alters the current distribution between anode and screen grid. The
connections of signal and oscillator voltages to this type of modulator are just the
reverse, therefore, of the mixer treated in the preceding section. The behaviours
of the two types are also quite different although they both include internal separation
of signal and oscillator electrodes through a shielding screen grid.
W ó 200-
s tamed by a 50 000 -ohm grid leak and con-
denser. The latter connection is most
=ó
o widely used, however, because of its con-
H - venience. A compromise using fixed bias
° - . together with a grid leak is most satisfac-
s lo IS 20 tory of all.
`
PEAK OSCILLATOR VOLTS
(Ref. 40). When this com-
FIG. 25.14 bination is used, the curve of conversion
Fig. 25.14. Conversion transconduct- transconductance follows curve a of Fig.
ance of a typical mixer designed for 25.14 to the intersection with curve b and
outer grid injection of oscillator. Signal- then follows along the flat top of curve b.
grid bias, E = -3 volts. Curve a In a well -designed mixer with the signal
voltage on the grid adjacent to the cathode
corresponds to fixed No. 3 grid bias, and the oscillator voltage on an outer grid,
E,3 = -
8 volts. Curve b corresponds effects due to feedback through the inter-
to bias obtained through a 50 000 -ohm electrode capacitance may usually be ne-
grid leak. glected. The only effect which might be of
980 (iv) MIXER SECTION OF CONVERTER STAGE 25.1
importance in some cases is coupling of the oscillator to the signal circuit through the
signal-grid -to- oscillator-grid capacitance. In many tubes a small amount of space -
charge coupling between these grids is also present and adds to the capacitance coup-
ling (contrary to the space -charge coupling discussed in Section B which opposes
the capacitance coupling in that case). Measurements of the magnitude of the space -
charge coupling for this type of modulator show that it is of the order of 1/5 to 1 /10
of that present in inner -grid- injection modulators. Coupling between oscillator and
signal circuits causes a voltage of oscillator frequency to be built up across the signal
input circuit. This oscillator-frequency voltage, depending on its phase, aids or
opposes the effect of the normal oscillator -grid alternating voltage. The action is
additive when the signal circuit has capacitative reactance to the oscillator frequency,
as in the usual case. When the oscillator-frequency voltage across the signal input
circuit exceeds the bias, grid current is drawn to the signal grid, an undesirable oc-
currence. This grid current may be distinguished from signal -grid current due to
other causes by short -circuiting the signal -input circuit and noting the change in grid
current. With the majority of tubes, another cause of signal-grid current far exceeds
this one in importance. This other cause will now be discussed.
The most prominent high- frequency effect which was observed in mixers of the
kind under discussion, was a direct current to the negative signal grid even when no
impedance was present in this grid circuit. This effect was investigated and found
to be due to the finite time of transit of the electrons which pass through the signal
grid and are repelled at the oscillator grid, returning to pass near the signal grid again.
(Refs. 17, 41, 42). When the oscillator frequency is high, the oscillator -grid potential
varies an appreciable amount during the time that such electrons are in the space
between screen grid and oscillator grid. These electrons may, therefore, be acceler-
ated in their return path more than they were decelerated in their forward path. Thus,
they may arrive at the signal grid with an additional velocity sufficient to allow them
to strike a slightly negative electrode. Some electrons may make many such trips
before being collected ; moreover, in each trip their velocity is increased so that
they may receive a total increase in velocity equivalent to several volts. A rough
estimate of the grid current to be expected from a given tube is given by the semi -
empirical equation
ICl = AIkE0.eW72-9EBE°l
Where A and B depend on electrode voltages and configuration, In is the signal-grid
current, E,1 is the signal -grid bias, Ik is the cathode current, E. is the impressed
oscillator voltage on the oscillator grid, w is the angular frequency of the oscillator,
and 72_3 is the electron transit time in the space between screen grid and oscillator
grid.
Data on the signal -grid current of a typical mixer at 20 megacycles are shown in
Fig. 25.15 where a semi -logarithmic plot is used to indicate the origin of the above
equation.
The reduction of signal -grid current by operation at more negative signal-grid
bias values is an obvious remedy. When this is done, in order to prevent a reduction
in conversion transconductance, the screen voltage must be raised. A better method
of reducing the undesired grid current lies in a change of tube design. It will be
shown in a later part of this discussion that the constant A and /or the transit time
72_3 of the above formula may be reduced considerably by proper electrode con-
figuration.
Another high -frequency phenomenon which is particularly noticed in outer -grid-
injection mixers is the high input conductance due to transit -time effects. The cause
for this was first made evident when the change of signal -grid admittance with os-
cillator -grid potential was observed. Fig. 25.16 gives data on the susceptive and
conductive components of the signal-grid admittance of this type of modulator as a
function of oscillator -grid bias (no oscillator voltage applied). The data were taken
at 31.5 megacycles and, as in the other input admittance curves, show the admittance
components due to the presence of electrons only. It is seen that when the No. 3
grid is made sufficiently negative the input admittance is greatly increased. This
25.1 (iv) MIXER SECTION OF CONVERTER STAGE 981
2.0
0^
I- \
Z 1.0
¢
uW
¢
uu
-15
No 3
-10
GRID
-5
BIAS VOLTS
0 +5
-415
OZó
OB.
03 INCREMENT DUE
t
Q:
TO ELECTRONS RETURNED
BY No3 GRID
_s
010 +5
-6 -4 -2 -IS -10 -5 0
No I GRID BIAS VOLTS No 3 GRID BIAS VOLTS
FIG. 25.15 FIG.25.16
Fig. 25.15. Signal grid (grid No. 1) current in a typical mixer with a 20-megacycle
oscillator voltage applied to grid No. 3. E, = -10 volts, E,$ 100 volts, Eb
and 4 =
250 volts.
Fig. 25.16. Signal-grid (grid No. 1) admittance of typical mixer designed for
outer grid injection of oscillator. Data taken at 31.5 megacycles with no os-
cillator voltage applied. E, = -3
volts, E,E and = 100 volts, E5 = 250 volts.
behaviour coincides, of course, with plate -current cut off. It seems clear that the
electrons which are turned back at the No. 3 grid and which again reach the signal
grid are the cause of the increased admittance. Calculations based on this explana-
tion have been published by M. J. O. Strutt (Ref. 43) and show reasonable quantita-
tive agreement with experiment. As in the other cases above, the upper curve of
Fig. 25.16 is approximately independent of frequency while the lower one may be
converted to any other frequency by multiplying the ordinates by the square of the
frequency ratio.
When an oscillator voltage is applied, the No. 3 grid bias is periodically varied at
oscillator frequency. The net input admittance is then the average value over the
oscillator cycle. Such net values of the conductance component are shown in Fig.
25.17. The frequency for these curves is 31.5 megacycles. Values for other fre-
quencies are obtained by multiplying the ordinates by the square of the frequency
ratio. Curve a coincides with the fixed bias condition of curve a of Fig. 25.14 while
curve b corresponds to the grid- leak -and -condenser bias as in b of Fig. 25.14. The
conductance is approximately twice as high when the tube is used as a mixer as when
it is used as an amplifier. This is a serious disadvantage, particularly at very high
frequencies.
ti
200 Fig. 25.17. Signal-grid (grid No. 1) con-
ductance of typical mixer designed for outer -grid
W injection of oscillator. Frequency, 31.5 mega -
.. cycles, signal-grid bias, E, = -3 volts.
Curve a corresponds to fixed No. 3 grid bias,
E,. = -8 volts. Curve b corresponds to bias
6 Do s ID 15 20 3 through a 50 000 -ohm grid leak.
PEAK OSCILLATOR VOLTS
FIG. 25.17
0
\
-
áa,0 3)9ci
"` \\
' /
/
p
\
Irj
N
ká
-"'7`.
--
\
41]
/
\
d
/
/
b
\
\
(öDà
N.
/
\
ci
/ -
b 6
s
//
d
200
-IS -10 -5 0 +5
Noi GRID BIAS VOLTS
FIG. 25.19
It should be noted that the use of the oscillator-grid support rods in the centre of the
electron streams as shown in Fig. 25.18 (d) was found to improve the performance.
No change in signal -grid conductance with oscillator-grid potential could be observed
with this construction.t The conductance of the tube as a modulator, therefore, was
reduced to less than half of that of construction (a). At the same time, a check of
signal -grid current with a high-frequency oscillator applied to the No. 3 grid showed
that this current was reduced to 1/20 of that of the original construction (a). The
change in construction may be looked upon as dividing the constant A in the grid -
current formula previously given, by a factor of more than 20.
*The same principles have now been applied to inner -grid -injection mixers and converters. See
references 19 and 45.
possible to construct tubes in which the signal-grid conductance
tit should be mentioned that it is alsonegative
decreases somewhat with increasingly No. 3 grid bias. This effect is caused by the inductance
of the inner screen -grid lead which causes a negative conductance in the input circuit when the inner
screen current is high, as at negative No. 3 grid bias values. This negative conductance cancels part
of the positive conductance of the signal grid.
25.1
/ ----\
(iv) MIXER SECTION OF CONVERTER STAGE
\
p
1
\
bQ
\ _Q i
q
//
p
\
._ /q
(9) / (h)
FOUR SIDEROD LARGE Nol GRID FINS ON No2 CLOSE- SPACED
Nol GRID SIDERODS GRID CHANNELS Noi GRID
Fig. 25.20. Cross -sectional views of improved mixers designed for outer grid injection
of oscillator. The views show only the portions of the tube inside of and including the
oscillator injection grid.
Another method of reducing the effect of electrons returned by the oscillator grid
is to reduce the effect of electron transit time in the tube. This may be done by
reducing the spacings, particularly the screen-grid -to- oscillator-grid spacing. This
method of improving modulator performance has two disadvantages compared with
the one discussed in connection with Fig. 25.18. The reduction in spacing is accom-
panied by a more sloping (i .e. less steep) signal -grid transconductance versus oscillator -
grid voltage curve. This change in construction requires an increase in applied
oscillator voltage to attain the same conversion transconductance. The second dis-
advantage is that such a method reduces the transit time and hence, the undesirable
high -frequency effects only by an amount bearing some relation to the reduction
in spacing. Since this reduction is limited in a given size of tube, the method whereby
electron paths are changed is much more effective. The method of reducing spacing,
on the other hand, is extremely simple to adopt. A combination of both methods
may be most desirable from the point of view of best performance with least com-
plexity in the tube structure.
B -6 -4 -2 ú0
Nol GRID BIAS VOLTS
FIG. 25.21
is included and is drawn as a. All four of the constructions of Fig. 25.20 were satis-
factory as regards signal -grid conductance ; in every case the change in conductance
as the oscillator grid was made negative was a negligible factor. Construction (h)
required approximately 20 per cent more oscillator voltage than (e), (f), or (g) because
of the reduction in slope of the transconductance versus No. 3 grid voltage curve
which accompanied the reduced spacing between the screen and the No. 3 grid.
Outer -grid-injection mixers have the same or slightly greater signal -grid capacitance
changes with automatic volume control as are found in amplifier tubes. In this
respect they are inferior to inner -grid- injection converters or mixers. The use of a
small unbypassed cathode resistance (Ref. 29, 30) is a help, however.
In closing this section, the subject of fluctuation noise will be considered. Experi-
mental evidence indicates that the major portion of the noise in mixers with oscillator
voltage on an outer grid is due to current- distribution fluctuations. (Ref. 24). The
oscillator voltage changes the current distribution from plate to screen so that the
mixer noise is given by the average of the distribution fluctuations over the oscillator
cycle. In terms of the equivalent noise resistance the average has been found to be
(Ref. 23),
-- [rb I62
Ra, =
g.,Y
where I n is the average (i.e. the operating) plate current and I O is the average of the
square of the plate current over an oscillator cycle. I. is the cathode current of the
mixer section and is substantially constant over the oscillator cycle. This relation
is not very useful in the form given. It is usually sufficiently accurate for most pur-
poses to use an expression identical with that which applies to tubes with inner -grid
oscillator injection, namely,
20/6
R = go
Y F,
Y
where FY is about 0.5 for tubes with suppressor grids and somewhat higher for others.
By assuming a typical tube characteristic, the noise resistance may be expressed in
terms of the cathode current I. of the mixer section and the maximum signal -grid-
to-plate transconductance g,,,,x as
I,
ReQ=120
for operation at oscillator fundamental. For operation at second or third harmonic
of the oscillator, the noise resistance will be approximately doubled, or tripled, respec-
tively.
(y) Conclusion
It has been shown that the principle of frequency conversion in all types of tubes
and with all methods of operation may be considered as the same (i.e. as a small -
percentage amplitude modulation). The differences in other characteristics between
various tubes and methods of operation are so marked, however, that each application
must be considered as a separate problem. The type of tube and method of operation
must be intelligently chosen to meet the most important needs of the application.
In making such a choice, it is frequently of assistance to prepare a table comparing
types of tubes and methods of operation on the basis of performance data. An
attempt has been made to draw such a comparison in a qualitative way for general
cases and for a few of the important characteristics. Table 2 is the result. It must
be understood, of course, that the appraisals are largely a matter of opinion based on
experience and the present state of knowledge. Furthermore, in particular circuits
and with particular tubes, the relative standings may sometimes be quite different.
A study of the fundamentals brought out in the previous sections of this paper should
help in evaluating such exceptions.
25.1 (v) CONCLUSION 985
TABLE 2
Approximate Comparative Appraisals of Methods of Frequency Conversion
Oscillator
Voltage on
Oscillator and Signal Oscillator Voltage on No. 1 Grid
Voltages on No. 3 Grid, Signal on Signal on
Desirable No. 1 Grid No. 1 Grid No. 3 Grid
Characteristic
Hexode Hexode
Triode Pentode Pentode or or
Heptode Heptode
High conversion trans -
conductance Good Good Fair Fair Fair
High plate resistance Poor Good Poor Good Good
High signal -to-noise
ratio Good Good Poor Poor Poor
Low oscillator-signal
circuit interaction
and radiation Poor Poor Good Good Fair
Low input conductance
at high frequencies Poor' Fair Poor Poor' Good
Low signal-grid current
at high frequencies Good Good Poor Poor' Fair
Low cost of complete
converter system Good Fair Fair Poor Good
(vi) Appendix
Discussion of negative admittance of current -limited grids
In Figs. 25.10, 11 and 12 it was seen that the electronic signal -grid (i.e. input)
admittance components (i.e. the admittance due to the presence of electrons) of a
mixer designed for No. 1 grid injection of the oscillator are negative over a considerable
portion of the normal operating range. Figs. 25.10 and 11, however, were taken with
static voltages applied and so indicate that the phenomenon is not caused by an alter-
nating oscillator voltage but is associated with the characteristics of the tube itself.
The input admittance of negative grids in vacuum tubes is the sum of three factors :
(1) the "cold" admittance, or the admittance of the tube with the electron current
cut off ; (2) the admittance due to feedback from other electrodes through tube and
external capacitance, etc. ; and (3) the admittance due to the presence of the electrons
in the tube. The first two factors have been well known for many years although
certain aspects of the second have only recently received attention. (Refs. 29, 30
and 48). The third factor, however is not so well understood although the excellent
work done during the last ten years has paved the way for a complete understanding
of the subject. (Ref. 49). The present discussion is concerned only with this last
point, namely the admittance of negative grids due to the presence of electrons in the
tube.
Early work on transit -time effects in diodes and negative -grid triodes had indicated
that, at very high frequencies, the conductance became negative in certain discrete
bands (i.e., at large transit angles). It was not, at first, appreciated that conditions
were possible with negative -grid triodes in which the input conductance could be-
come negative even at low frequencies (i.e., at small transit angles). Data taken
on the input (No. 4 grid) conductance of pentagrid converters by W. R. Ferris of this
986 (vi) APPENDIX 25.1
laboratory during 1934 showed that these tubes had a negative input conductance
which varied as the square of the frequency and which remained negative at low fre-
quencies. The conductance appeared, therefore, to behave in the same way as the
positive input conductance of ordinary negative-grid tubes, except for a reversal in
sign. The data on the pentagrid were taken with an external oscillator voltage applied
to the No. 1 grid. The work of Bakker and de Vries (Ref. 50) disclosed the possi-
bility of a negative input conductance at small transit angles in a triode operated under
current-limited conditions. They gave an experimental confirmation for a triode
operated at reduced filament temperature. Data taken by the writer during 1936
on a pentagrid converter showed that the negative conductance was present in this
tube even when direct voltages, only, were applied and that it was accompanied by a
reduction in capacitance. A fairly complete theory of the effect was developed in
unpublished work by Bernard Salzberg, formerly of this laboratory, who extended
the theory of Bakker and de Vries to the more general case of multigrid tubes with
negative control grids in a current- limited region. Other experimental work was
done on the effect during 1936 by J. M. Miller and during the first half of 1937 by
the writer. In the meantime, the papers of H. Rothe (Ref. 51), I. Runge (Refs. 52, 53),
and L. C. Peterson (Ref. 54) showed that independent experimental and theoretical
work had been done on the negative-admittance effect in other laboratories.
In a rough way, the negative admittance found under current-limited conditions
may be explained as follows : The electron current in a tube is equal to the product
of the charge density and the electron velocity. If this current is held constant, a rise
in effective potential of the control electrode raises the velocity and so lowers the
charge density. A reduction in charge density with increase in potential, however,
results in a reduction in capacitance, provided no electrons are caught by the grid.
Thus, the susceptive component of the part of the admittance due to the current
through the grid, is negative. Because of the time lag due to the finite time of transit
of the electrons, there is an additional component of admittance lagging the negative
susceptance by 90 degrees, i.e., a negative conductance. The value of the negative
conductance will be proportional to both the transit angle and to the value of the
susceptance. Since both of these quantities are proportional to frequency, the
negative conductance is proportional to the square of the frequency.
The general shape of the curves of Fig. 25.10 may be explained as follows : At
a No. 1 grid bias of about -20 volts, the cathode current is cut off and the electronic
admittance is zero. At slightly less negative values of No. 1 grid bias, the electron
current is too small to build up an appreciable space charge ahead of the signal grid
(No. 4 grid). The latter grid, although it exhibits some control of the plate current
does not control the major portion of the current reaching it and is thus in a substanti-
ally current- limited region. Its susceptance and conductance are, therefore, negative.
Higher currents increase the negative admittance until, at some value of No. 1 grid
bias, the electron current is increased to the point at which a virtual cathode is formed
in front of some parts of the signal grid. At these parts, the current which reaches
the grid is no longer independent of this grid potential, and as a result, a positive
susceptance and conductance begin to counteract the negative admittance of other
portions of the grid. The admittance curves reach a minimum and for still higher
currents approach and attain a positive value. The current necessary to attain the
minimum admittance point is less when the signal -grid bias is made more negative
so that the minima for increasingly negative No. 4 grid -bias values occur at increasingly
negative No. 1 grid -bias values.
It may be noted that the signal- grid -to -plate transconductance is at a maximum
in the region just to the right of the admittance minima of Fig. 25.10 (compare Fig.
25.8). The admittance of such a tube used as an amplifier remains negative, there-
fore, at the maximum amplification point.
25.2 (i) BROADCAST FREQUENCIES 987
The grid-circuit loading is thus directly proportional to the grid -plate capacitance
of the converter, and it is only in types in which this is high that input loading be-
comes appreciable. For example the 6A8-G has a plate to signal grid capacitance of
0.26 µµF and with an i -f transformer tuned by 85 µµF to 455 Kc /s the resistive com-
ponent of the loading due to feedback across the valve is about 0.25 megohm at 600
Kc /s rising to 0.45 megohm at 1000 Kc /s and 0.55 megohm at 1400 Kc /s.
The resistive component of the loading is also proportional to the reactance of the
capacitor tuning the plate circuit of the converter. This capacitor should be given
as large a value as possible if grid loading is the main consideration.
The reactive component of the input loading, which is always capacitive, appears
in parallel with the tuning capacitor and can usually be ignored.
Some converters have much lower plate -grid capacitance, e.g. the 6J8 -G has a
maximum of 0.01 µµF, and in such cases the signal -grid circuit loading due to feed-
back from the plate circuit is negligible.
External coupling between oscillator and signal frequency circuits can cause another
effect which may be mistaken for input circuit loading. This effect occurs in triode -
hexode and similar types of converters when a tuned-plate oscillator circuit is used,
the loss of gain being most noticeably at the high-frequency end of the broadcast
band. In such a case, conversion sensitivity over the broadcast band is reasonably
flat, but aerial (or r -f grid) sensitivity shows poor gain at the high- frequency end even
although the coil, checked separately, gives constant gain over the band.
The coupling (capacitive or inductive) between the oscillator plate and the signal -
grid causes an oscillator voltage to be applied to the signal grid which opposes the
effects of the correctly injected oscillator voltage-see Sect. 1(iv)A of this chapter.
In severe cases the coupling may even be sufficient to give an oscillator voltage on the
signal grid in excess of the bias, leading to severe damping of the input circuit. Tuned-
plate oscillator circuits are most likely to produce this effect because the required
oscillator grid voltage is a constant whether plate tuning or grid tuning is used and
with a tuned plate circuit this voltage must be developed across the smaller untuned
grid winding of the oscillator coil. At the same time the oscillator plate voltage,
instead of being appreciably smaller than the oscillator grid voltage as in a tuned-
grid circuit, is appreciably greater.
A mechanical re-arrangement of the layout to isolate the oscillator -plate circuit
from the signal-grid circuit will usually cure the trouble. In a difficult case it may be
possible to arrange the wiring so that an opposite coupling between oscillator-grid
and signal -grid will produce the required result.
2. Signal-circuit regeneration
It is possible to use controlled regeneration in a converter stage to give improved
gain, image ratio and signal -to-noise ratio. Various possibilities are given in Ref. 65.
FIG. 25.22
amount of control for a given a.v.c. voltage. Such a circuit will also minimize cross-
modulation. However in converters without a suppressor grid the screen voltage
must not approach the plate voltage because secondary emission from the plate will
cause a large reduction in plate resistance and lead to damping of the first i -f trans-
former, so that some bleed from screen to ground may be needed. The best com-
promise can readily be determined with resistor boxes.
If maximum a.v.c. control is required from a triode-hexode converter, for example
when it is followed by a reflexed amplifier, and play-through must be reduced to a
minimum, a large screen bleed may be needed.
Maximum a.v.c. control can be obtained in pentagrid converters by reducing the
oscillator plate voltage to a minimum consistent with satisfactory oscillator perform-
ance. For example with 20 volts bias on the signal grid, the conversion transcon-
ductance of a 6A8 -G is reduced to less than -one half of the original figure when the
oscillator plate voltage is changed from 200 to 100 volts, although at minimum signal-
grid bias the oscillator plate voltage has little effect on the conversion transconductance.
Final operating conditions should not be decided upon until the tests outlined above
have been repeated with a number of valves, some of which should preferably have
characteristics near the upper and lower acceptance limits for the type.
of the band is due to confusion between the correct response and the image. This
results from the fact that the correct signal appears to differ depending on whether the
signal generator is tuned or the receiver is tuned. For example, if a receiver with
a 450 Kc /s intermediate frequency and with its oscillator on the high- frequency side
of the signal frequency is correctly tuned to an 18 Mc /s signal the receiver oscillator
is set to 18.45 Mc/s. The generator can then be tuned to a frequency 450 Kc /s
higher than the oscillator i.e. 18.9 Mc/s and the image response will be found, or
alternatively with the generator still set to 18 Mc /s the receiver can be tuned to the
lower frequency of 17.1 Mc/s when the local oscillator will be at 17.55 Mc/s, 450
Kc/s from the signal, and an image will be heard again.
(ii) SHORT WAVES 991
25.2
Whenever there is any doubt as to which is the correct response, the frequencies
should be worked out -conditions will be reversed from the example above if the
oscillator is at a lower frequency than the signal -but a simple rule is that if the re-
ceiver is being tuned and the oscillator is on the high frequency side of the signal,
then the higher frequency signal is the correct one.
A similar effect to an image response can also occur during signal -frequency trim-
mer adjustment. With the oscillator on the high-frequency side of the signal it may
be found that as the signal-frequency trimming capacitance is increased, the output,
with rocking, rises to a maximum, falls and rises to a second maximum, which is the
correct one. On the other hand if the receiver is left aligned at the first maximum it
is found that the adjustment is approximately correct for the image frequency.
This effect has no direct connection with the image response, but is due to the
decrease in sensitivity which can occur when the signal -grid circuit is tuned to the
oscillator frequency -owing to grid damping in a severe case, or out -of-phase os-
cillator injection in other cases. Thus as the signal -circuit tuning initially approaches
the correct setting the gain rises, decreases again when the separation from the os-
cillator frequency is small, and then rises to the correct peak. In a sensitive receiver
the illusion of correct alignment on the wrong peak is heightened by an increase in
the level of the background noise due to the increased receiver sensitivity near the
image frequency.
B. Operating conditions
Converters can be set up for operation on the short-wave band in a manner similar
to that described for the broadcast band. However additional steps are also worth-
while. Particularly when harmonic mixing -see C(4) below -is being used,
it is advisable to investigate the effects of different values of oscillator -grid resistor.
To avoid varying two quantities at the same time the oscillator voltage should be kept
as constant as possible while the value of the grid leak is altered, and readings should be
taken at least at each end of the short-wave band. After the optimum grid leak
value has been chosen, previously selected operating potentials should be rechecked.
When the converter is a type with inner -grid oscillator voltage injection, improved
sensitivity may be obtained with the oscillator operating at a lower frequency than
the signal instead of under the more usual higher frequency condition.
In Sect. 1 of this chapter it was explained that the space charge in front of the signal
grid develops a voltage at oscillator frequency on the signal grid and that this voltage,
depending upon its phase, increases or decreases the oscillator -frequency modulation
of the electron stream. In a particular case with the oscillator at a higher frequency
than the signal, the tuned circuit connected to the signal grid would present capacitive
reactance to the oscillator voltage. A positive increment of oscillator -grid voltage
increases the current flow and thus increases the space charge in front of the signal
grid, i.e. the space charge becomes more negative. Since there is capacitive reactance
between the space charge and signal grid and between signal grid and ground, a
negative increment of space charge potential results in a (smaller) negative increment
of signal -grid potential. Thus a positive change in potential on the oscillator grid
results in a negative change in potential on the signal grid, so that the effective os-
cillator modulation is reduced with a consequent reduction in sensitivity.
However when the oscillator is operated at a lower frequency than that of the signal,
the signal -grid circuit presents an inductive reactance to the oscillator voltage and the
phase of the oscillator voltage on the signal grid is reversed. Thus the effective
modulation of the electron stream is increased and sensitivity is improved.
In practice, if neutralizing is not used, an improvement in sensitivity of two or three
times is sometimes obtainable by changing from high- frequency to low-frequency
oscillator operation, with an increase of the image ratio in the same proportion. The
improvement in sensitivity and image ratio decreases towards the low-frequency end
of the short-wave band as the separation between signal and oscillator circuits (i.e
the intermediate frequency) becomes a larger fraction of the oscillator frequency
For correct tracking with the oscillator on the low- frequency side of the signal,
the padder is moved from the oscillator to the signal- frequency circuit(s) but this
992 (ü) SHORT WAVES 25.2
does not introduce serious complications. However, a large band coverage is more
difficult with the oscillator on the low side because the oscillator circuit covers a greater
frequency ratio than the signal circuit, whereas on the high side it covers a smaller
ratio.
It is possible for external coupling between oscillator and signal circuits to exceed
that which occurs due to internal coupling in the converter itself and it is advisable
to check that the external coupling is negligible before investigating the effects of
low-side and high -side oscillator operation. A valve voltmeter can be used across
the circuit and the variation of the indication as the signal circuit is tuned through the
oscillator frequency, with the circuit connected to and disconnected from the signal
grid, shows the relative amount of coupling due to internal and external sources.
A further desirable step in the investigation of short -wave performance is the de-
termination of the best sensitivity that can be obtained by adjustment of the magnitude
and phase of the oscillator voltage on the signal grid. This can be done by using a
very small capacitor, variable in order to adjust the amplitude of the oscillator voltage,
and connected between either signal -grid and oscillator grid or signal -grid and os-
cillator plate to vary the phase of the oscillator voltage. In the absence of appreciable
coupling within the receiver, some increase in sensitivity is usually obtained with
a particular value of neutralizing capacitor, because maximum sensitivity is obtained
not when the oscillator voltage on the signal grid is a minimum but when it is the
correctly-phased maximum that can be present without signal -grid current flowing.
Thus, when using a plate tuned triode -hexode converter, it may be found experi-
mentally that sensitivity is improved at 18 Mc /s with a 0.5 µµF capacitor connected
between oscillator grid and signal grid. In such a case it is usually possible to obtain
a capacitance of approximately the correct value without using a separate component
by a suitable arrangement of the wave -change switch wiring.
However, when neutralizing is used, a measurement should be made of the effect
of the neutralizing capacitance on the frequency stability of the receiver at 18 Mc /s.
In general, sensitivity improvements due to oscillator voltage on the signal-grid must
be offset against decreased frequency stability. The frequency shift in pentagrid
converters of the 6SA7 type increases so rapidly with increased capacitance between
signal -grid and oscillator grids that a neutralizing capacitance cannot be used in a
normal short-wave receiver if a.v.c. is applied to the control grid.
Unless special precautions are taken, grid current variations over the short-wave
band are greater than desired. The increase at the high -frequency end of the band
occurs because, although the coupling between primary and secondary of the oscillator
coil which is required to maintain oscillation is reduced, the actual coupling remains
unchanged. A small resistor (say 25 ohm watt) wired between the tuned circuit
and the oscillator grid is the usual method of obtaining reasonably uniform grid current
over the band. The effective parallel circuit damping of the resistor in series with
the capacitance of the oscillator grid is inversely proportional to the square of the
frequency so that a large reduction in grid current can be obtained at the high-fre-
quency end of the band without noticeably affecting the grid current at the low -
frequency end.
Such a resistor is sometimes also required to prevent squegging of the local oscillator .
when the amplitude of oscillation is high and the time constant of the oscillator grid
circuit is much greater than the time required for one cycle of the oscillator frequency.
For further information on squegging see Chapter 35 Sect. 3(vi)C and Ref. 73.
C. Frequency stability
All the problems of frequency stability experienced with a separate oscillator (see
Chapter 24) occur when the oscillator section of a converter is used as the local os-
cillator in a superheterodyne receiver and in addition there are further causes of
instability due to interaction between the oscillator and mixer sections of the valve.
1. Frequency variations due to the local oscillator
Local oscillators are always self-biased triodes so that the only potential applied to
the circuit from external sources is the plate-supply voltage. Two common types
25.2 (ii) SHORT WAVES 993
of frequency instability are due to this link with the rest of the receiver. The first
is " flutter " due to fluctuations in the B supply voltage as a result of the output -valve
current varying at audio frequencies. These fluctuations vary the oscillator frequency
so that the signal is continually detuned, giving a -f signals which make the " flutter "
self-sustaining. The mechanism of " flutter " is more fully described in Chapter 35
Sect. 3(vi)B and methods of overcoming it are given.
The second detuning effect for which the B supply is responsible is due to the ap-
plication of a.v.c. to the controlled stages of a receiver. This reduces the B current,
causes the B voltage to rise, and thus produces detuning which will vary, for instance,
with the instantaneous level of a fading station.
Although in a particular case the detuning may not be sufficient to produce the
symptoms described, it may still give unpleasant tuning at the high- frequency end
of the short -wave band, a typical effect being that the receiver does not tune smoothly
but jumps from one side of a signal to the other.
When testing for these effects, modulated and unmodulated signals of magnitudes
varying from maximum to minimum should be used at the highest tuning frequency
of the receiver and the volume control setting should also be varied. If equipment
for measuring frequency shift is not available a useful test is to tune the receiver as
accurately as possible to a large signal (perhaps 0.1 volt) with the volume control
suitably retarded and then reduce the input to the smallest usable signal and turn
up the volume control. If the signal is still tuned, frequency shift may be considered
satisfactory. The test should be repeated with different signal inputs in case positive
and negative frequency shifts should cancel over the range of inputs first selected.
If flutter occurs on a large unmodulated signal there are circumstances under which
it can be ignored. When the modulation is switched on it may be found that the
output valve is severely overloaded, and that when the volume control is turned down
to give no more than full a -f output the flutter may not occur. In such a case the
flutter would not be noticed in normal use of the receiver and if a cure cannot be
effected cheaply it may not be warranted.
In practice it is found that frequency stability can often be improved by the simple
expedient of increasing the oscillator grid current. In one particular triode-hexode
the 18 Mc/s frequency shift for a given change in control grid bias was reduced to one
tenth by increasing the oscillator grid current from 200 to 600 A. Frequency shift due
to other causes, and with other types of converters can be improved in the same way,
although in varying degrees. The improvement may of course have to be offset
against a decrease in sensitivity and an increase in noise.
A third type of oscillator frequency instability due to the B supply is experienced
in F -M receivers. If there is a hum voltage superimposed on the B supply to the
oscillator plate this will produce frequency modulation of the local oscillator, and the
F-M detector will convert the resultant hum modulation of the signal carrier into
a -f hum.
2. Frequency variations due to the mixer
The space charge adjacent to oscillator electrodes in a converter valve can be varied
by altering the potentials on mixer electrodes. This, of course, gives rise to oscillator -
frequency variations. The mixer screen voltage is particularly important because one
of the two grids comprising the screen is adjacent to an oscillator electrode, so that it
is desirable to keep this voltage as constant as possible. This frequently necessitates
a separate decoupling resistor and capacitor for the converter screen because with a
common supply to converter and i -f amplifier screens the application of a.v.c to the
i-f amplifier (even if the converter stage is not controlled) alters the mixer screen
voltage and thus causes frequency shift. On rare occasions this shift may be used to
offset another in the opposite direction due to a different effect.
When frequency shift is troublesome it is desirable to isolate each of the causes
and determine its direction, as cancellation is sometimes possible even if a complete
cure is not.
The effects upon oscillator frequency of varying a.v.c. voltages applied to a signal -
grid electrode are more complex. With either outer-grid or inner -grid oscillator
994 (ii) SHORT WAVES 25.2
injection a space charge adjacent to the oscillator grid can be varied by altering the
direct potential applied to the signal grid. That this is not always the main source
of frequency drift can be seen in some receivers with this type of instability by measur-
ing the frequency shift for a given change in converter a.v.c. voltage with the signal
generator connected firstly to the aerial terminal and then to the converter grid. It
may be found that the frequency shift in the former case is many times that with the
signal generator connected to the converter -grid, and it may even be in the opposite
direction.
This type of instability is due to coupling between oscillator and control -grid
circuits. In Sect. 1(iv)B of this chapter it is explained that at high frequencies it is
usual for oscillator voltage to appear at the signal -grid of the converter. Coupling
between signal- and oscillator -grid circuits results in some of this voltage being
returned to the oscillator circuit and, in a manner similar to the operation of a re-
actance-tube modulator, this results in a modification of the effective capacitance at
the oscillator grid. A constant variation of this capacitance would be of no signific-
ance, but any alteration of converter operating conditions which resulted in a variation
of oscillator voltage on the signal grid, e.g. the application of a.v.c. voltages, would
alter the effective reflected capacitance at the oscillator grid and thus cause oscillator
frequency shift. (Ref. 87).
With a given set of conditions the simplest method of curing this type of fre-
quency shift is usually to minimize the coupling between oscillator and signal grid
circuits. In general, coupling is capacitive, partly within the valve and partly ex-
ternal to it. In most modern valves the internal capacitance is not troublesome and
to effect a cure it may be necessary to separate short-wave aerial and oscillator coils
from each other, to separate the wiring from the two grids or from the two sections of
the gang condenser, and to use wave -change switch contacts on the opposite side of
the wafer. In the particular case of the 6BE6, it is desirable to reduce the external
capacitance between the two grids of the valve to 1 µµF or less.
Another possible cure for this type of frequency shift is the use of harmonic mixing,
in which the frequency difference between signal and oscillator circuits is so great
that no significant amount of oscillator frequency voltage appears on the signal grid.
3. Tuned -plate operation
An analysis of tuned -grid and tuned -plate oscillators (Chapter 24 Sect. 2) shows
the differences between the two types of operation. However, it is mainly because
of the structure of some types of converters that tuned -plate operation is preferable.
When the oscillator section of a converter has a common electron stream with the
mixer section, the oscillator -plate electrode is designed to modulate the electron
stream as little as possible since modulation by the oscillator plate cancels that from
the oscillator grid. It is also desirable in equipment design to keep the oscillator
plate alternating voltage as small as possible. With a tuned -grid oscillator, the plate
voltage is smaller than the grid voltage and the reverse is the case with a tuned-plate
oscillator. Accordingly, tuned -grid oscillators are always used for converters having
a common electron stream for the oscillator and mixer sections.
Triode -hexode and triode -heptode converters have a separate triode for use as a
local oscillator, and oscillator injection into the mixer section is carried out by a special
injector grid which is in most cases internally connected to the oscillator grid. The
capacitance and coupling effects of the injector grid are thus introduced into the
oscillator circuit, and if a tuned-grid oscillator is used they appear directly across the
tuned circuit.
However with a tuned plate oscillator these effects are coupled to the tuned circuit
from the oscillator primary winding and their effect is thus minimized. An im-
provement in oscillator frequency stability, with respect to the effect of a.v.c. applica-
tion to the signal grid, of five times has been measured with a typical triode -hexode
on converting from a tuned -grid to a tuned -plate oscillator, while frequency shift
for a given change in B supply voltage was halved under the same conditions.
4. Harmonic mixing
Many of the difficulties experienced in high- frequency applications of converters
are due to the small percentage of frequency difference between the oscillator and
25.2 (ii) SHORT WAVES 995
signal- frequency circuits. It was shown in Sect. 1 of this chapter that a sinusoidal
oscillator voltage applied to the injector grid of the mixer section can also give mixing
at harmonics of the oscillator frequency. By taking advantage of this inherent con-
verter characteristic it is possible to separate widely the resonant frequencies of the
oscillator and signal frequency tuned circuits and thus eliminate many difficulties
in high- frequency converter applications.
There are disadvantages to harmonic mixing but in some cases they are outweighed
by the advantages to be obtained. Firstly there is some loss in sensitivity over a part,
at least, of the tuning range and secondly there are spurious responses due to incoming
signals mixing with the oscillator fundamental.
With careful design the loss in sensitivity is small. In the normal 6 to 18 Mc /s
short -wave range it will probably not be greater and may be less than 2 db from 10
to 18 Mc /s. The maximum loss is usually at 6 Mc/s, and should not exceed 3 db,
about 2 db or less being normal. Even when the converter is the first valve in the
receiver no serious decrease in signal -to -noise ratio of the complete receiver should
be experienced.
Because of a number of secondary effects which occur with harmonic mixing, this
performance is better than would be expected from an investigation of maximum
second harmonic conversion transconductance alone, although it presupposes that
the best conditions for harmonic mixing are used. It will be found that higher grid
current is necessary, particularly at the low-frequency end of the band, and a lower
value of grid leak than normal is often useful in obtaining maximum sensitivity. In
general, it is desirable to investigate the converter conditions carefully, as described
previously, and for satisfactory performance inner -grid mixers should operate with
the second harmonic of the oscillator on the low-frequency side of the signal.
The reason for this is as follows. The modulation of the electron stream by grid 1
gives a transconductance-time curve similar to Fig. 25.1B, and the space charge in
front of the signal grid fluctuates at the same rate. This fluctuation, although of the
same frequency as the local oscillator, has large harmonic components and since the
space charge is coupled capacitively to the signal -grid circuit and this is tuned ap-
proximately to the second harmonic frequency, most of the voltage induced into the
signal-grid circuit is at the frequency of the second harmonic of the oscillator. This
second harmonic voltage remodulates the electron stream with a phase which depends
on whether the signal frequency circuit has capacitive or inductive reactance at the
second harmonic frequency so that the re-modulation aids or opposes, depending on
its phase, the original second-harmonic modulation by grid 1. For inner -grid in-
jection converters, the oscillator second -harmonic frequency should be lower than the
signal frequency for aiding re-modulation.
It will be found that with harmonic mixing the same padder can be used as that
required for the same frequency coverage with fundamental mixing.
Although the spurious responses which arise due to fundamental mixing are a
disadvantage of harmonic operation, a tuned r-f stage ahead of the converter will
reduce them to negligible proportions, not greater than -60 db in a typical case.
Moreover, even when fundamental mixing is used there are responses, although
smaller ones, due to signals mixing with the oscillator harmonics. Fundamental
spurious responses 30 db below the signal could be taken as representative of receivers,
without a r-f stage, using second-harmonic mixing so that these responses cause
appreciably less interference than the normal image.
The benefits to be obtained from harmonic mixing are evident when operating,
for example, the 1R5 converter on the 6 to 18 Mc /s short-wave band. Because the
1R5 has no separate oscillator -plate electrode, an oscillator circuit is often used in
which the screen is connected to the oscillator primary winding to operate as the
oscillator plate. A neutralizing capacitor is then needed between signal grid and
oscillator grid and the determining of a suitable capacitance and its tolerances, the
minimizing of coupling between signal and oscillator circuits and the stabilizing of the
oscillator with reference to supply -voltage variations involve a considerable amount
of work, as does the duplicating of the results in production receivers.
996 (ii) SHORT WAVES 25.2
is shown in Fig. 25.23. This converter is very stable in operation which makes it
useful in dual -wave receivers in spite of its low conversion transconductance. It
has a high plate resistance (4 megohms) under recommended operation conditions
so that the use of a high impedance first i -f transformer is more effective in increasing
conversion gain than in the case of other converter types, for example the 6A8 -G.
25.2 (iii) TYPES OF CONVERTERS 997
It is desirable not to apply a.v.c. to the 6J8-G on short waves if this is possible,
but if a.v.c. is used a tuned -plate oscillator circuit will greatly improve the frequency
stability.
(2) Types 6AE8, X79 and X61M : The electrode arrangements of these triode -
hexodes is similar to that of Fig. 25.23 but there is no suppressor grid adjacent to the
plate. Because of this it is necessary in circuit design to avoid conditions which may
lead to secondary emission from the hexode plate to the screen.
Frequency stability with respect to B voltage variations is good, although it is de-
sirable to use plate tuning of the oscillator if a.v.c. is applied to the hexode section
on short waves. The conversion transconductances are high (approximately 750
micromhos with 2 volts bias and a screen voltage of 85) and few special precautions
are required to obtain non -critical operation and good performance on the broadcast
and short-wave bands.
OSC.ANODE--I SCREEN
SIG.GRID I
o
o o E o b o óo o O 0
0 o oo o
o
O
o
0
ó ó o o O
o o o o o o O
o o o o o
oo o o
i o
oo
o
o oo
o o
oo
oo
o
o
o
o
PLATE oo o
o
o
o
o o oo
PLATE o o o o o
qq
The operating characteristics of most outer -grid injection triode -hexodes are similar
to those of the above types.
B. Inner-grid oscillator injection
Converters using inner -grid oscillator injection, with the exception of the 6K8(G),
have combined oscillator and mixer sections using a common cathode stream. In
general the oscillator frequency stability is poorer than that of outer -grid injection
types and coupling between control and signal grids is greater, although modern
valves are much improved in these respects. Under maximum gain conditions the
input loading of the signal grid is negative but becomes less negative and ultimately
positive as the signal grid bias is increased.
To obtain maximum short-wave sensitivity without neutralizing, the oscillator
should operate on the low- frequency side of the signal.
(1) Type 6A8 -G : The structure of the pentagrid (or heptode) type 6A8-G is
shown in Fig. 25.25. It will be noticed that the oscillator plate consists only of two
side rods without a normal grid winding, and that there is no suppressor grid, which
leads to a low plate resistance (0.36 megohm under typical operating conditions).
Nevertheless the conversion transconductance of 550 micromhos is sufficient to give
reasonable conversion gain.
When used with high impedance i -f transformers the 6A8 -G may introduce some
grid loading at the low- frequency end of the broadcast band due to feedback (Miller
998 (iii) TYPES OF CONVERTERS 25.2
Effect) from the plate circuit to the control grid through the relatively high (0.26 µµF)
plate -to-grid capacitance. This effect is normally only just noticeable but may be
aggravated by locating the 6A8 -G against the back of the signal- frequency section
of the tuning condenser so that appreciable capacitance is present between the plate
of the valve and the stator assembly of the gang. The effect shows up as apparently
poor aerial coil gain so that the cause may not be suspected. It is not of course peculiar
to the 6A8 -G but the already high plate -to -grid capacitance, the size of the valve and
the lack of internal or external shielding make the effect more likely to occur with this
type. The effect may be minimized by external shielding of the valve.
Part of the oscillator plate (grid 2) current is due to electrons which have passed
from the cathode through grids 1 and 3 and then been repelled by grid 4 back through
grid 3 to grid 2 again. The oscillator characteristics and thus the oscillator frequency
stability are therefore very dependent on the potentials applied to grids 2, 3 and 4 and,
for short wave operation, a.v.c. should not be applied to the valve. In addition it
may be necessary to decouple the oscillator -plate voltage supply with an electrolytic
capacitor and even to obtain it directly from the rectifier output-suitably decoupled
to eliminate hum -in order to avoid feeding a -f variations from the output valve back
to the oscillator plate and thus causing flutter.
To obtain consistent performance it is desirable to keep the oscillator -plate voltage
higher than the screen voltage.
OSC.GRID -t r SCREEN
SIG.GRI
OSC.G
,T2
ID
rc,5 SCREEN
! llob
oo
o
o
o o
o o o
o o o
o o o o
o o o o
o o o o
o o o o
o o o o
o
o
o
i o
o
o
o
o
o
o
o 0
SUPPRESSOR PLATE
OSC.PLATE PLATE CATHODE
CATHODE SIG. GRID
/ms's
o
I
..i
I I
r
SHIELD
FIG. 25.75 FIG. 25. 27
Fig. 25.26. Electrode ar- Fig. 25.27. Electrode ar-
rangement of the triode- rangement of the pentagrid
hexodc type 6K8(G). type 6SA7(GT).
(2) Type 6K8-G : The structure of the inner -grid injection triode-hexode type
6K8 -G is shown in Fig 25.26. It was designed to provide more stable short -wave
performance than the 6A8 -G and has a higher oscillator transconductance and operates
with lower oscillator excitation. Its lower conversion transconductance (350 mic-
romhos) gives a lower conversion gain than that obtainable from the 6A8 -G although
the plate resistance (0.6 megohm) is higher.
A.V.C. can be applied to the 6K8-G on short -waves but the frequency stability
with respect to changes in other electrode voltages is only fair.
(3) Types 6BE6 and 6SA7(GT) : The main point of interest in the structure
of the pentagrid types 6BE6 and 6SA7(GT), Figs. 25.27, and 25.29 is that there is no
electrode which functions only as an oscillator plate.
By the omission of this electrode, it is possible to obtain a relatively high oscillator
transconductance, without greatly increasing the total cathode current.
The oscillator circuits employed with these types have certain unconventional
features and Fig. 25.28 may be taken as typical. The lack of a separate oscillator
25.2 (iii) TYPES OF CONVERTERS 999
I -F
OUTPUT
and the oscillating voltage (E,) on the No. 1 grid are at their peak positive values.
For maximum conversion transconductance this peak value of plate current should
be as large as possible. The effect on plate current of the positive voltage on the
cathode is approximately the same as would be produced by an equal voltage of
negative sign applied to the signal grid. Hence the amplitude of oscillator voltage
on the cathode limits the peak plate current. This amplitude should therefore be
small.
PL ATE
FIG. 25.29
During the negative portion of an oscillation cycle the cathode may swing more
negative than the signal grid. If this occurs, positive signal -grid current will flow
unless the oscillator grid is sufficiently negative to cut -off the cathode current. This
signal -grid current flowing through the signal -grid circuit resistance will develop a
negative bias on the signal grid and may also cause a negative bias to be applied to the
i -f and r -f stages through the a.v.c. system. As a result, sensitivity will be decreased.
In order to prevent signal -grid current, the d.c. bias developed by the oscillator grid
should be not less than its cut -off value.
1000 (iii) TYPES OF CONVERTERS 25.2
Because the peak plate current depends on how far positive the oscillator grid
swings with respect to the cathode, it is desirable that this positive swing be as large
as possible. It follows that the oscillator grid -leak resistance should be
low, but
not so low as to cause excessive damping of the tank circuit. It has been found,
for operation in frequency bands lower than approximately 6 Mc /s, that all these
requirements are generally best satisfied when the oscillator circuit is adjusted
to
provide, with recommended values of plate and screen voltage, a value of Ek of ap-
proximately 2 volts peak, and a d.c. oscillator-grid current of 0.5 mA through a grid -
leak resistance of 20 000 ohms. This will give a peak positive voltage of the oscillator
grid with respect to cathode of about 4 volts
On the normal short-wave band of 6 to 18 Mc /s, minimum grid current occurs
at the low-frequency end of the band and the design procedure consists in adjusting
the oscillator circuit so that sufficient grid current (200 µA minimum) is obtained at
6 Mc /s without developing excessive cathode voltage (approximately 2.5 volts
r.m.s.
maximum) at 18 Mc/s. The oscillator -grid bias is then somewhat less than cut-off
at 6 Mc /s, but the signal -grid current should not be so high as to cause trouble.
Oscillation at the high-frequency end of the band however, may be greater than
optimum unless a grid stopper is used, but over- excitation will improve frequency
stability.
If, for manufacturing reasons, the use of a tapped coil is not desirable, a normal
primary winding can be used for the cathode connection, although it will be necessary
to reverse the connections from those for a plate -tickler oscillator. This coil arrange-
ment allows one side of the padder to be grounded.
Another method of connection which is satisfactory for the broadcast band is to use
a plate-tickler oscillator circuit with the screen electrode as the oscillator plate and the
cathode grounded. This connection can also be used on the short-wave band but
the interaction between signal and oscillator circuits causes severe pulling and other
-
associated troubles. A neutralizing capacitor between signal and oscillator grids
will minimize these effects neutralizing is not recommended with a cathode -coupled
oscillator-and an alternative system is the use of harmonic mixing.
Whenever the screen is connected to an oscillator primary winding the oscillator
voltage on this electrode must be kept to a minimum, i.e. primaries must have as few
turns as possible with maximum coupling to secondaries. In addition, it is often
desirable to use a small carbon resistor in series with the oscillator grid electrode to
prevent the oscillator voltage on the screen from rising excessively at the high-fre-
quency end of the band. The reasons are, firstly, that modulation by the oscillator
voltage on the screen is out of phase with the modulation due to the oscillator grid,
which decreases conversion sensitivity and, secondly, that as the signal grid is between
the two electrodes forming the screen a large oscillator voltage on the screen results,
through capacitive coupling, in a comparatively large oscillator voltage on the signal
grid, particularly on short waves. This voltage can be neutralized, but normal
neutralization is effective for only one frequency and set of operating conditions and
when a large amount of neutralization is used and a balance between two large equal
and opposite voltages is obtained, any small change in conditions results in a large
oscillator voltage reappearing on the signal grid.
The arrangement of shields on grid 2 and of the siderods of grid 3 make the oscillator
section of these types much more stable with respect to electrode voltage variations
than the 6A8 -G, and it can be used satisfactorily with a.v.c. applied on short waves.
Nevertheless in any short-wave application care must be taken to reduce to a mini-
mum any coupling between signal and oscillator grids if frequency stability is to be
satisfactory. In the case of the 6BE6, the external capacitive coupling between
oscillator and signal grids should be limited to 1 µµF for satisfactory short wave per-
formance.
25.2 (iii) TYPES OF CONVERTERS 1001
A+ a+ 8+ Cn Neutralising Capacity
FIG. 25.31 FIG. 25.32 B+
Fig. 25.31. Self- excited 1R5 converter Fig. 25.32. Circuit for short -wave oper-
circuit. ation of 1R5 converter.
(4) Type 1R5 The electrode structure of the 1R5 is similar to that of the 6BE6
:
and 6SA7(GT) without the shields on grid 2, and with a filament instead of the heater
and cathode. Circuits for use with the 1R5 are complicated by the lack of a separate
cathode electrode but this may be overcome by using a choke as shown in Fig. 25.31
Alternatively the choke may be omitted and a filament lead returned to A+ through
a separate winding of the same number of turns as, and wound over the tapped section
of, the oscillator coil.
However, neither of these circuits is satisfactory on short waves if the full range -
6 to 18 Mc /s -is to be covered, and even on the broadcast band there may be diffi-
culty in obtaining sufficient grid current. The circuit of Fig. 25.32, with or without
the padder feedback shown, is more suitable.
Neutralizing is desirable on short waves if conversion is carried out with the oscillator
fundamental. The neutralizing capacitor should be connected across the short -wave
coils only, because a different value is normally required on the broadcast band, and
oscillator voltage on grid 3 is usually less with no neutralizing than with the capacitance
(from 2+ to 5 µµF) required on short waves.
The purpose of the neutralizing capacitor in a circuit such as Fig. 25.32 is not so
much the balancing out of oscillator voltage due to the space charge in front of grid 3,
but the neutralizing of a voltage due to capacitive coupling between grid 3 and grids
2 and 4. Therefore, as previously mentioned, it is essential to reduce the oscillator
voltage on the screen to the lowest practicable value.
If there is a tendency for the 1R5 to squeg at the high -frequency end of the short-
wave band it is often possible to save the small carbon resistor in the oscillator grid
lead, which is the usual cure, by reducing the value of oscillator grid leak to say 30 000
ohms or slightly lower. This frequently increases sensitivity, other conditions being
unchanged, and it is worth trying even if there are no signs of squegging. In most
receivers the oscillator grid circuit damping caused by the low value of oscillator
grid leak or by the grid stopper is desirable to reduce the oscillator voltage on the
screen of the converter at the high- frequency end of the band with a consequent im-
provement in sensitivity.
A very satisfactory way to operate the 1R5 is to use harmonic mixing as described
previously. The increased freedom in layout, the increased stability in all respects
which simplifies production alignment -and the removal of the need for neutralizing
-
more than compensate for the slight disadvantages which in any case are of little
importance if the receiver has a r -f stage.
1002 TRACKING -(i) GENERAL 25.3
(i) General
The problem of tracking (whether it be in a straight or a superheterodyne receiver)
is to set, simultaneously, to some desired resonant frequency, each of a series of tuned
circuits which are mechanically coupled together and operated from a single control.
In a superheterodyne receiver the problem is one of maintaining a constant fre-
quency difference (equal to the intermediate frequency) between the signal circuits
(such as the aerial and r-f stages) and the oscillator circuit. It is a relatively simple
matter to make the difference between the signal and oscillator frequency equal to
the i -f at two points in the tuning range. This condition is called two point tracking,
and is applied whenever the error in frequency difference between the circuits does not
become a large percentage of the total pass band. Where the tuning error is likely
to become excessive, it can be reduced by the addition of another component, in the
form of a capacitor or inductor, into the oscillator or signal circuits. In this case it is
possible to proportion the circuit components so that zero frequency error exists
between the intermediate frequency and the difference between the signal and os-
cillator circuits at three frequencies in the tuning range, instead of two ; furthermore
the error at frequencies between the tracking points is appreciably reduced.
The most usual application of three point tracking has been in receivers covering
the standard long, medium and short wave bands, and where the tuning element is a
variable capacitor. For this reason attention will be confined to this system of tuning
and a suitable design method will be detailed. Those interested in tracking per-
meability tuned circuits are referred to the articles of Refs. 107 and 111. These
give a method of three point tracking using identical variable inductors in the signal
and oscillator circuits.
SIGNAL OSCILLATOR It is necessary to point out
that the theoretical solutions,
so far published, of the three
point tracking problem are all
idealized in so far that they
ignore in part, or completely,
the effects of primary wind-
FIG. 25.33 J ings on the signal and oscilla-
tor coils. The best approach
Fig. 25.33. Circuits used for tracking analysis. under these conditions ap-
pears to be to select a method of determining values for circuit components which
(a) Does not involve an excessive number of arithmetical operations and will allow
the use of a slide rule (or four figure logs) for all calculations, except in special cases.
(b) Gives values for the components that fall within a few per cent of those actually
required in the circuits.
(c) Allows the change required in other circuit components to be rapidly estimated,
when the value of one component in the circuit is changed by an amount which falls
within a previously determined range of values.
(d) Lends itself to graphical methods of estimating the component values.
After the component values have been calculated it is usually advisable to build
a pilot model receiver in which the values of all the elements (padder, coils and trim-
mers) can be varied over a small range (say ± 10 %). In this way it is a fairly simple
matter to secure tracking at the three points required, and at the same time to de-
termine what tolerances are permissible in the component values before mistracking
becomes excessive. Fixed padders and coils can then be used for further models
of the same receiver ; although production difficulties can be reduced by using
`` slug " inductance variation in the oscillator coils as well as the usual variable parallel
25.3 TRACKING-(i) GENERAL 1003
capacitance trimmers in the oscillator and signal circuits (many receiver manufacturers
also use " slug " tuning of the signal circuits).
A few points which may be of interest are :
(1) The tracking error on the broadcast bands need not exceed a few Kc /s e.g.
the error between tracking points need not exceed about 3 Kc /s on the 540-1600 Kc /s
medium wave band using tracking frequencies of 600, 1000 and 1400 Kc /s. This
error is negligible, in most cases, since the oscillator tuning takes charge and the lack
of alignment affects the aerial and r-f circuits which are relatively unselective.
A typical curve of tracking errors is shown in Fig. 2534.
(2) High impedance primaries, on the aerial and r-f coils, are practically always
used in modern receivers covering the medium waveband. The coefficient of coup-
ling in aerial and r-f coils, and the location of the primary resonant frequencies have
very important effects on tracking. Suitable values for these factors are discussed
in Chapter 23, Sects. 2 and 3.
(3) When a series capacitance (padder) is used in the oscillator circuit only, the
operating frequency of the oscillator must be higher than the signal frequency if
three point tracking is to be secured.
Three point tracking is obtainable if both the signal and oscillator circuits use
series padders, whether the oscillator frequency is higher or lower than the signal
frequency provided, of course, that the component values are correctly proportioned.
If only the signal circuits are padded the oscillator frequency must be lower than
the signal frequency to obtain three
point tracking.
A typical example of padding of
.+4
N
Q
cd
tom;
the signal and oscillator circuits oc-
curs in band -spread receivers, where
the short wave band is covered in a iu
.
\°
Nhi
number of steps (e.g. 6 bands may be -5 cp
used to tune from 6 -18 Mc /s), and it
is required to have approximately ,A4J
straight line frequency tuning with a ° to 900 1100 1300 1500
500 700
standard variable capacitor
p ( i.e. one SIGNAL FREQUENCY KC/5
whose capacitance versus rotation FIG.25.34 (B)
quencies, and make the third point either the arithmetical-mean or geometrical -mean
of the outer frequencies. This leads to fairly simple design calculations, and gives
values for the components which are reasonably close to the optimum. The small
differences in component values required to set to the optimum tracking frequencies
(or whatever the designer chooses) can then be made experimentally. It should be
observed that it has been usual to align receivers to tracking points on various bands
which have been established by long practice as giving sufficiently satisfactory results
e.g. on the medium wave band 600, 1000 and 1400 Kc /s are in common use, but
these are a compromise, due partly to a lack of more exact knowledge, and also because
of the frequency allocations of the main broadcasting stations. If a station were to be
located at, say, 540 Kc /s then the loss in receiver sensitivity, due to mistracking,
would be serious if 600 Kc /s were to be taken as the low frequency tracking point.
(7) The feedback winding on the oscillator coil should be as small as possible,
consistent with correct oscillator operation, if good tracking is desired. For minimum
error the oscillator feedback winding should have its natural resonant frequency
well above the highest oscillator tuning frequency ; this means that stray capacitances
across the winding should be kept as low as possible. The natural resonant frequency
is important because the coefficient of coupling between the two oscillator coil wind-
ings is usually large, and the amount of reactance reflected into the tuned circuit not
only varies with frequency, but also sets a limit to the maximum possible tuning
range. The limitation in tuning range is a particularly serious factor in receivers
covering the short wave range in one band.
An additional and important reason for making the feedback winding as small as
possible, is that for good oscillator stability the highest possible coefficient of coupling
should be used, consistent with the smallest possible value of mutual inductance
(M = kVL4,,).
(8) When tracking a superheterodyne receiver at three points in the tuning range
(we will assume that the oscillator frequency is higher than the signal frequency)
it often happens that the centre tracking (crossover) frequency does not fall at the
frequency required. The question then arises as to how the oscillator series padder
value should be altered.
First the signal circuit trimmer capacitance (or inductance) is altered so as to give
maximum output. If the capacitance was increased then the oscillator series padder
should be increased. For a decrease in signal trimmer capacitance (or inductance)
the oscillator series padder should be decreased.
Complete re-alignment and re- checking is necessary after the padder value has
been altered and it is important to remember that the oscillator inductance and trim-
mer capacitance values will also require alteration. This is a simple process when the
oscillator coil is " slug " tuned, and the trimmer capacitance is adjustable.
Fig. 25.34 is also helpful in this regard, and the tracking error can be considered
in terms of Lo if so desired. It should be clear that if the value of Lo is increased
the padder value should be decreased (and vice versa) to retain the desired oscillator
frequency coverage.
A good alignment procedure is to set first the signal and oscillator circuits at the
band limits. Tracking is then obtained at the required points by setting the signal
generator to the required tracking frequency and rocking the receiver dial while
altering the oscillator trimmer capacitance (at the high frequency tracking point)
and the oscillator inductance (at the low frequency tracking point) until maximum
signal output is obtained. Tracking at the centre frequency can be checked either
by alteration of the signal circuit trimmers, which have not been altered after the
initial adjustment at the band limits, or, alternatively, by leaving the signal circuit
trimmers untouched and rocking the receiver dial while adjusting the oscillator
inductance (or capacitance) for maximum output. It is preferable to track the os-
cillator to the signal circuits since these give the required tuning law. A complete
re-check is always necessary after the initial alignment procedure is completed Of
course correct tracking can be obtained by a number of methods, but the procedure
suggested above has proved very satisfactory for receiver development.
25.3 (ii) (A) FORMULAE AND CHARTS 1005
1.4
IL
0.6
1.2
0.5
0.9
0.8 4
0.3
02 6
IX.18
a. 3 o
0.1
10 12 1.4 16 1.8 20 2.2 24 26 28 30
FIG. 25.35
Fig. 25.35. Charts giving T °ma for geometrical -mean tracking.
1006 (ii) (A) FORMULAE AND CHARTS 25.3
1111111111111111111 11
IIMMIN1111111 NM
HEIMMEMINOMMI
oramommorm
NOMMIBMIN
AMTEMINEELMEIR
MINIMEMINIENIE
imerammoomoss
mmummumma.
ullir Ñ
X uep
25.3 (ii) (A) FORMULAE AND CHARTS 1007
opm IL WNW
MEMO, 4
WESIM
10110 NINA la
mum ó
mu =
aÉ u
EMI EWE
MM MIME
ME ONE
iiimm
am maim
mom
imas
=mom maw
mm mom=
ME MMIERAI
EMMEIMUM
MME =EWE
NEMER11
Mileginj "Wb
uivad
Mr"
MA Alf
MINA 10ii, I
AMEN ipin
mum
..,
mmemEmmenri
HIMENEWAVAIMI
EMEIMPERNMEIN
MINENNISHIENI
1111111111111111111111
IMINEMEMINI N 8
ISECOMINHO III
REEMMEIMMUE
su1
T D1ü ( )
1
25.3 (ii) (A) FORMULAE AND CHARTS 1009
e
I I
CO_
O
1010 (ii) (A) FORMULAE AND CHARTS 25.3
á= W4'
w1
_ f4
fí
G' mas = total incremental capacitance of each section of the ganged capacitors
between signal frequencies 11' and 12'
T' _= T less the incremental capacitance of each section of the ganged tuning
capacitors between signal frequencies f2 and f,'
T,' - T, less the incremental capacitance of each section of the ganged tuning
capacitors between signal frequencies f2 and f2'.
Signal circuits :
Oscillator circuit :
2(a - ß)
-ß2 as (ß - 1)(a + Y)
2(a -ß)
and r
(ß
1+ - 1)(1 + y)
For the case of arithmetical-mean tracking
fl + f4
2
042 3+a 1 +3ß
r =P- 1 +3a +ß
For geometrical -mean tracking (i.e. fa = Vf1f2)
25.3 (ii) (A) FORMULAE AND CHARTS 1011
2ß + (1 -{- ß)al
aa
r =R -
2a +(1 +ß)al
-P
ßB
(a) General formulae for padding and trimming capacitances.
(1) Ta given.
P = Pmax m in
TL - Pmin LTQmax - T `]
(2) TL given.
= P2,n(1 Pmin
/
-
\/ll Pmin
4TL)l
T = Pmax
Formulae when TL
-P<=P Pmin + Te mam - P.
P = Pmin + TL
T, = Tamax
T
- T
a
error in P is given by -(PLL) x 100 per cent.
a
error in T is given by P x T, x 100 per cent.
(b) Formulae for inductance of oscillator coil
+
Pmax\a 1 (P,.\
Pm {APmn, 1
(a)a - 1
1 G'max
T'
T wa /
TL = T - T'
- Tc'examples
(B) Worked
To illustrate the application of the equations and charts, worked examples are
appended.
A tuning capacitor is available having a capacitance range of 12 -432 µµF. It is
required to tune over the range of frequencies 530 to 1620 Kc /s, and the i-f of the
receiver is 455 Kc /s. Determine the component values foi the signal and oscillator
circuits.
1012 (ii) (B) WORKED EXAMPLES 25.3
For ease of working the outer tracking frequencies will be made coincident with
the band limits, and the third tracking frequency (f,) will be taken as the geometrical -
mean of the outer frequencies ( /flf, = 926.6 Kc /s).
Signal circuits
Gmax= 432 -12= 420µµF
(162012
a2
l 530J = 9.342
420
T 9.342 - - 1
50.24 µµF (includes 12 µµF gang min. cap. so that
actual trimmer would be 38.24 µµF)
25 330
L
50.24 X 1.622 - 192.2 µH
Oscillator circuit
/162012
a2 9.342 ; a} = 1.748 ;
` 530 J
1620 + 455
2.107 ; [32 --- 4.440 ;
ß 530 + 455
(1 + ß)ai = 5.431
9.342 r(2 x 2.107) + 5.4311
r = R -
4.44 L(2 X 3.056) + 5.431J 1.758
432 12 -
Pmax 1.758 1 - -
554 µµF
(432 12) -
T,
(1.758 x 4.44) 1
61.7 µµF
- -
(a) A value for TL (about 8 µµF) will be assumed, after preliminary calculation,
as this is the usual case.
As obviously TL <
P (Compare 8 µµF with Pmax)
Pm:a = Pma¢ Ta
ma¢
- = 492.3 µµF
554 x 61.7
T L max 492.3
69.42 µµF. -
From this it is permissible to estimate TL as 8 µµF since this is about the order
of stray capacitances likely to be across L0. Any small error here is unimportant, as
long as the value chosen for TL is less than TL since this is taken up during circuit
alignment. ax
Hence P = 492.3 + 8 = 500.3 µµF
T, = 61.7 - 8 = 53.7 µµF (includes gang min. cap. of 12 µµF.
Actual trimmer would be 41.7 µµF)
25 330 X 492.3 X 554
(b) Lo =
61.7 X 500.32 x (1.62 + 0.455)2 - 103.9 µH
(c) and (d)
Suppose now that it is required to make a change in the value of one of the oscillator
components. What should be the new values of the other components ?
A good example is the padder ; a more suitable value may be 510 µµF (which lies
between the maximum and minimum values of 554 µµF and 492.3 µµF).
Then
ST, _ -9.7 µµF
STL =(1 -f- 500.38)l 9.7 = 10 µµF
2 x 103.9 x 9.7
SLo = - 500.3 -4.03 H. -
So that the new values for the oscillator components are P = 510 µµF
T, = 53.7 -9.7
=44µµF
TL =8-10 =18µµF
25.3 (ii) (B) WORKED EXAMPLES 1013
Lo = 103.9 -
4.03 = 99.87 µH
and these values all lie within the permissible range of values, set by the maximum
and minimum values calculated for P, TL and T, ; denoted by Pmax, P,aEa etc.
A table showing a series of values between the calculated maxima and minima is
often very useful. In this way the most suitable component values can be selected.
Also, three point tracking is still maintained at the selected frequencies with the range
of component values determined, as the change in values is independent of the track-
ing and intermediate frequencies provided these remain unaltered.
The same design problem, previously solved algebraically, will now be carried out
using the tracking charts.
Signal circuits
Procedure exactly as before.
Oscillator circuit
1620
= 530
= 3.056
1620 + 455
P 530 + 455 - 2.107
Gmax = 432 - 12 = 420 µµF
(a) From the charts:
T, max
Gmax
- 0.15 therefore; = 63 µµF max
Patin
Gmax
- 0.12, therefore = 504 µµF P,,..
TL
max
Gtaxa
- 0.17, therefore TLmax = 71 µµF
P,,,. = 504 + 63 = 567 µµF
Using TL = 8 µµF (as previously)
P= 504-1-8= 512µµF
T, = 63 -
8 = 55 µµF
25 330 x 504 x 567
(b) Lo = 102 µH
63 X 5122 x (1.62 + 0.455)a
(c) and (d) Any circuit component changes are made exactly as before.
Suitable practical values are then selected after allowing for strays.
The design problem is to select values for the padder P and the two trimming capaci-
tances TL and T, such that simultaneously the following conditions are fulfilled : --
(a) The desired frequency ratio is attained.
(b) The scale calibration is linear, or, more generally, corresponds to a desired
form.
(c) The L/C ratio is maintained at a high value in order to achieve adequate gain
in the signal circuits.
To simplify notation the virtual pilot circuit takes the same form as the original
signal circuit shown in Fig. 25.33 and the same notation is retained, for both the padded
signal and oscillator circuits, as was previously used for the oscillator circuit alone.
The design equations are limited to the cases of arithmetical -mean and geometrical-
mean tracking. Charts are available (in Ref. 103) covering the arithmetical -mean
case.
Summary of formulae
Pilot circuit :
Gmax
a2 1 -
C(y2/
l\T =
2
1 + (a2 - 1)
G
Padded circuit :
P + T, - Gmax
r - 1
Gmax
T' n+a z
rß2 - 1
TLmax - r(ß2...
P + T, = Pmax = Pmin
where r is limited to the values given by p and R.
T, specified :
P = Pmaz
P
T, -
TL T,
Pmin T`max
where P is first evaluated.
TL specified :
Pmin
2 Cl
L
+ 1 + 4T L
Pmin
Pmin
T, = T, ma - TL . P
where P is first evaluated.
When TL P <
T, = T, max - TL
P =Pmin+ TL
Gain is proportional to the dynamic impedance Z of the circuit and
Z =Ci
+ T, maz
P2 G
C = (effective capacitance across inductance).
Pmin G +Pmax
When G = Gmax then C = Cmax. The value of P is obtained from the appropriate
equations, according to whether Tc or TL is specified.
25.3 (iii) (A) PADDED SIGNAL CIRCUITS 1015
Inductance
Pmax
+ 1
Gmax
L°
1 P.a.
(wl + wr)' P' 1..
T°max
Gmax
25 330 Pmfa(Gmax
- + Pmax)
or for numerical computation L°
(fl fi)' P Gmaa + Tamax)
using µH, µµF, Mc /s.
in which (w1 + w') is the known low frequency tracking point of the padded signal
circuit. For a padded oscillator circuit replace (w1 + w!) by (w1 + w! ± wt). The
value for P is derived from the appropriate equation according as to whether T. or
TL is specified.
For an unpadded signal circuit
L 0w1'
as - 1
Gmax
The notation is identical with that given previously, with the addition of
w = 21r x any frequency in the tuning range
col = 21r x low frequency tracking point of pilot circuit
w2 = 21r x high frequency tracking point of pilot circuit
we = arithmetical- or geometrical -mean tracking point of pilot circuit
wf = 21r x true intermediate frequency of receiver
w5 = 27r x virtual intermediate frequency for the combination of a padded
signal circuit with a virtual pilot circuit
a = ws
w1
= f'f, = frequency ratio of pilot circuit
ß w'
w1
+
+
w°
w1
- frequency ratio of padded signal circuit
G = incremental capacitance of each section of the ganged tuning capacitors,
measured from its value at w2
Gmax = value of G at col
C = total effective capacitance across L°
Cmax = value of C when G = Gmax
Q = wL °/R where L° and R apply to the signal circuit, and this R is the
r -f resistance of the circuit.
(B) Worked example
The method will be applied to the design of a receiver tuning from 6 to 9 Mc /s,
using an i -f of 455 Kc /s, with a capacitor having a straight line frequency characteristic
and a capacitance range of 12 -432 µµF. Arithmetical -mean tracking will be used.
Pilot circuit
For the pilot circuit assume that the required scale calibration is given by
a = 2 (A sufficient range of values for most receiver calibration curves
is from about 3 to 1.5)
a2 =4
Signal circuits
ß = 9/6 = 15
P2 = 2.25
=
4 3 + 2 1+ (3 x 1.5)
- 1.552
P
= 2.25 3 + 1.5 1 + (3 x 2)
r
-P +T.- 432 - 12
Pmax 1.552 -1 760.8
µµF
432 12 -
T. max - (1.552 X 2.25) - 1
301.7 µµF
432 12 -
Temar - 1.552 (2.25 - 1)- 216.5 µµF
1016 (iii) (B) WORKED EXAMPLE 25.3
Ta TL P Lo
50 710.8
100 660.8
150 610.8
200 560.8
250 57.53 510.8 2.024
300 460.8
From this it seems that a convenient padder value is 510.8 µµF. Then find TL
and L° to complete the table. If the values so found, using the first selection for P,
are considered as being suitable, only those need be computed
TL
510.8
459.1
[301.7 -
250] = 57.3 µµF
25 330 X 459.1 x (420 -I- 760.8)
L0 6' X 510.8' x (420 . i 301.7) - 2.024 µH
So that the signal circuit components are
L° = 2.024 µH
TL= 57.3µµF
P= 510.8µµF
T,, = 250 µµF (which includes
12 µµF gang min. cap.)
(Suitable practical values are selected after due allowance for strays).
An estimate can now be made of the circuit dynamic impedance (Z), as this will
serve as a guide to the possible circuit gain. Improvement can often be effected by
a redistribution of the component values, and the most favourable L/C ratios can be
determined by completing the table set out earlier.
However the L/C ratio and the component values selected are usually a compro-
mise forced on the designer by considerations of practical convenience.
Oscillator circuit
It will be taken that the oscillator frequency is higher than the signal frequency.
9 + 0.455
ß
P2 =
6+0.455
2.145
- 1.464
4 3 + 2 1+ (3 x 1.464)
r = p
2.145 3 + 1.464 + (3 x 2) - 1.609
432 - 12
1
TL T, P L°
50 564.1
100 85.6 604 1.619
150 639.7
200 1072
25.3 (iii) (B) WORKED EXAMPLE 1017
SECTION 4 : REFERENCES
1. Armstrong, E. H. " A New System of Short-Wave Amplification " Proc. I.R.E. 9.1 (Feb. 1921) 3.
2. German Patent No. 324, 515.1918.
3. J. Scott- Taggart, German Patent No. 383,449.1919.
4. de Mare, J., R. Barthelemy H. de Bellescize and L. Levy " Use of double-grid valves in frequency-
changing circuits " L'Onde Electrique 5 (April 1926) 150.
5. " A four -electrode valve supersonic circuit " Experimental Wireless, 3.37 (Oct. 1926) 650.
6. Barthelemy, R. " Valve frequency- changers" Rev. Gen. d'Elec. 19 (April 1926) 663.
7. See also : M. Gausner, French Patent No. 639,028 : G. Thebault, French Patent 655,738 :
H.J.J.M. de Regauld de Bellescize, United States Patent No. 1,872,634.
8. Whitman, V. E., United States Patent No. 1,893,813: H. A. Wheeler, United States Patent No.
1,931,338.
9. Llewellyn, F. B., United States Patent No. 1,896,780.
10. Wheeler, H. A. " The hexode vacuum tube " Radio Eng. 13.4 (April 1933) 12.
11. Hassenberg, W. " The hexode " Funk. Tech. Monatshefte, (May 1933) 165.
12. Application Note No. 3 R.C.A. Radiotron Co., Inc.
13. Robinson, E. Y. British Patent No. 408,256.
14. Smith, J. C., Discussion on H. A. Wheeler paper " Image suppression and oscillator- modulators
in superheterodyne receivers " Proc. I.R.E. 23.6 (June 1935) 576.
15. Harris, W. A. ` The application of superheterodyne frequency conversion systems to multirange
receivers " Proc. I.R.E. 23.4 (April 1935) 279.
16. Shelton, E. E. " A new frequency changer " W.W. 35.21 (Oct. 5, 1934) 283.
17. Nesslage, C. F., E. W. Herold and W. A. Harris ' A new tube for use in superheterodyne frequency
conversion systems " Proc. I.R.E. 24.2 (Feb. 1936) 207.
18. Herold, E. W. W. A. Harris and T. J. Henry " A new converter tube for all -wave receivers "
R.C.A. Rev. 3.1 (July 1938) 67.
19. Harris, W. A. " A single -ended pentagrid converter " Presented, Rochester Fall Meeting, Rochester,
N.Y., Nov. 15, 1938. See Application Note No. 100, R.C.A. Manufacturing Co., Inc., Radiotron
Division, Harrison, N.J.
20. Klipsch, P. W. " Suppression of interlocking in first detector circuits " Proc. I.R.E. 22.6 (June
1934) 699.
21. " Suppressor grid circuit novelties " Radio World 22.15 (Dec. 24, 1932) 13.
22. Lukacs, E., F. Preisach and Z. Szepcsi " Noise in frequency changer valves " (Letter to Editor)
W.E. 15.182 (Nov. 1938) 611.
23. Herold, E. W. " Superheterodyne converter system considerations in television receivers " R.C.A.
Rev. 4.3 (Jan. 1940) 324.
24. Thompson, B. J., D. O. North and W. A. Harris " Fluctuations in space- charge-limited currents
at moderately high frequencies " R.C.A. Rev. 4.3 (Jan. 1940) 269 ; 4.4 (April 1940) 441 ; 5.1
(July 1940) 106 ; 5.2 (Oct. 1940) 244 ; 5.3 (Jan. 1941) 371 ; 5.4 (April 1941) 505 ; 6.1 (July 1941)
114.
26. Strutt, M. J. O. " Electron transit-time effects in multigrid valves " W.E. 15.177 (June 1938) 315.
27. de Bellescize, H., United States Patent No. 1,872,634 ; Gausner, M., French Patent No. 639,028.
28. Whitman, V. E. United States Patent No. 1,893,813.
29. Strutt, M. J. O. and A. van der Ziel " Simple circuit means for improving short-wave performance
of amplifier tubes " Elek. Nach. Tech. 13 (Aug. 1936) 260.
1018 REFERENCES 25.4
30. Freeman R. L. " Use of feedback to compensate for vacuum-tube input- capacitance variations
with grid bias " Proc. I.R.E. 26.11 (Nov. 1938) 1360.
32. Bakker, C. J. and G. de Vries " Amplification of small alternating tensions by an inductive action
of the electrons in a radio valve " Physica 1 (Oct. 1934) 1045.
33. Nicolson, A. M. United States Patent No. 1,255,211 (applied for in 1915).
34. Strutt, M. J. O. " Frequency changers in all-wave receivers " W.E. 14.163 (April 1937) 184.
35. Herold, E. W. " Frequency changers in all -wave receivers " (Letter to Editor) W.B. 14.168 (Sept.
1937) 488.
36. Herold, E. W. United Staten Patent No. 2,141,750.
37. Llewellyn, F. B. " Phase angle of vacuum tube transconductance at very high frequencies " Proc
I.R.E. 22.8 (Aug. 1934) 947.
38. Strutt, M. J. O. " Frequency changers in all -wave receivers " (Letter to Editor) W.E. 14.170
(Nov. 1937) 606.
40. Herold E. W. United States Patent No. 2,066,038.
41. SteimeÌ K." The influence of inertia and transit time of electrons in broadcast receiving tubes "
-Robre
Telefunken 5 (Nov. 1935) 213.
42. Knol, K. S. M. J. O. Strutt and A. van der Ziel " On the motion of electrons in an alternating
electric field " Physica 5 (May 1938) 325.
43. Strutt, M. J. O. and A. van der Ziel " Dynamic measurements of electron motion in multigrid
tubes " Elek. Nach. Tech. 15 (Sept. 1938) 277.
45. Jonker, J. L. H. and A. J. W. M. van Overbeek " A new converter valve " W.E. 15.179 (August
1938) 423.
48. Strutt, M. J. O. and A. van der Ziel " The causes for the increase of admittances of modern high-
frequency amplifier tubes on short waves " Proc. I.R.E. 26.8 (Aug. 1938) 1011.
49. An excellent historical summary of this work is found in W. E. Benham's " A contribution to tube
and amplifier theory " Proc. I.R.E. 26.7 (July 1935) 683.
50. Bakker, C. H. and C. de Vries " On vacuum tube electronics " Physica 2 (July 1935) 683.
51. Rothe, H. " The operation of electron tubes at high frequencies " Telefunken-Rohre 9 (April
1937) 33 : Proc. I.R.E. 28.7 (July 1940) 325.
52. Runge, I. " Transit -time effects in electron rubes " Zeit fur Tech. Phys. 18 (1937) 438.
53. Runge, I. " Multigrid tubes at high frequencies " Telefunken-Rohre 10 (Au 1937) 128.
54. Peterson, L. C. " Impedance properties of electron streams " B.S.T.J. 18.7 (July 1939) 465.
55. Strutt, M. J. O. " On conversion detectors " Proc. I.R.E. 22.8 (Aug. 1934) 981.
56. Howe, G. W. O. " Second channel and harmonic reception in superheterodynes " W.E. 11.132
(Sept. 1934) 461.
57. Wey, R. J. " Heptode frequency changers " W.E. 11.135 (Dec. 1934) 642.
58. Strutt, M. J. O. " Mixing valves " W.E. 12.137 (Feb. 1935) 59.
59. Stewart, J. " The operation of superheterodyne first detector valves " J.I.E.E. 76 (Feb. 1935) 227.
60. Strutt, M. J. O. " Whistling notes in superheterodyne receivers " W.E. 12.139 (April 1935) 194.
61. Strutt, M. J. O. " Diode frequency changers " W.E. 13.149 (Feb. 1936) 73.
62. Bell, D. A. " The diode as rectifier and frequency changer " W.E. 18.217 (Oct. 1941) 395.
63. Mitchell, C. J. " Heterodyning and modulation : do additive and multiplicative mixing amount
to the same thing ? " W.W. 53.10 (Oct. 1947) 359.
64. Whitehead, J. W. " Additive and multiplicative mixing " W.W. 53.12 (Dec. 1947) 486.
65. Strutt, M. J. O. " Noise figure reduction in mixer stages " Proc. I.R.E. 34.12 (Dec. 1946) 942.
66. Jonker, J. L. H. and A. J. W. M. van Overbeek " A new frequency- changing valve " Philips Tec.
Rev. 3.9 (Sept. 1938) 266.
67. " Application Note on Operation of the 6SA7 " R.C.A. Manufacturing Company, Inc. Radiotron
Division, Application Note No. 100 (Dec. 2, 1938).
68. " Characteristics of pentodes and triodes in mixer service " Radio Corporation of America, Tube
Dept. Application Note 139 (March 15, 1949).
69. Goddard, N. E. " Contour analysis of mixer valves " W.E. 26.314 (Nov. 1949) 350.
70. Herold, E. W. and L. Malter " Some aspects of radio reception at ultra -high frequency " Proc.
I.R.E. 31.10 (Oct. 1943) 567.
71. Tapp, C. E. " The application of super -regeneration to F -M receiver design " Proc. I.R.E. (Aust.)
8.4 (April 1947) 4.
72. " Use of the 6BA6 and 6BE6 miniature tubes in F -M receivers " Radio Corporation of America,
Tube Dept., Application Note 121. Reprinted Radiotronics 129 (Jan. 1948) 15.
73. Gladwin, A. S. Oscillation amplitude in simple valve oscillators " Part 1, W.E. 26.308 (May 1949)
159 ; Part 2 (contains section on squegging) W.B. 26.309 (June 1949) 201.
74. Stockman, H. " Frequency conversion circuit development " Comm. (Part 1) 25.4 (April 1945)
46 ; (Part 2) 25.5 (May 1945) 58.
75. Bedford, L. H. " The impulsive theory of the hexode frequency changer " W.E. 15.182 (Nov.
1938) 596.
76. " Compensation of frequency drift " (Particular reference to use of 6BE6 at 110 Mc /s) Radio
Corporatioonn of America, Tube Dept., Application Note AN -122. Reprinted Radiotronics 130
(March 36.
77. " Receiver microphonics caused by heater -cathode capacitance variations " Particular reference
to use of 6BE6 in F -M band, Radio Corporation of America, Tube Dept., Application Note AN-
123. Reprinted Radiotronics 130 (March 1948) 38.
78. Sturley, K R. " Radio Receiver Design " Part 1 (Chapman and Hall, 1943).
79. Zepler, E. E. " The Technique of Radio Design " (Chapman and Hall, 2nd edit. 1949) 151.
80. Erskine -Maconochie, H. C. C. " Regeneration in the Superheterodyne -where and how it may
be applied " W.W. 45.4 (July 27, 1939) 77.
81. Lukacs, E., and J. A. Sargrove (1) " Frequency changing problems : further development in mixer
valves " W.W. 44.4 (Jan. 26, 1939) 81. (2) " Frequency changing problems : The latest octodes "
W.W. 44.5 (Feb. 2, 1939) 119.
82. Cocking, W. T. " Short-wave oscillator problems : avoiding parasitic oscillation" W.W. 44.6
(Feb. 9, 1939) 127.
83. Sargrove, J. A. " Parasitic oscillation in frequency- changers: kinetic grid current in the anode-
heptode and triode hexode" W.W. 45.6 (Aug. 10, 1939) 121.
84. James, E. G. " Frequency changer valves : Little known effects on short waves " W.W. 44.13
(March 30 1939) 295.
85. The modern receiver stage by stage : (iv) The frequency changer " W.W. 44.12 (March 23,
1939) 281.
86. Aske, V. H. " Gain-doubling frequency converters " Elect. 24.1 (Jan. 1951) 92.
25.4 REFERENCES 1019
87. Dammers, B. G., J. Otte " Comments on interelectrode feedback in frequency changers " Philips
Tec. Corn. No. 5 (1950) 13 ; No. 6 (1950) 11 ; reprinted from Philips Electronic Application
Bulletin 11, 3 (March 1950).
101. Green, A. L. " Superheterodyne tracking charts " A.W.A. Tec. Rev. 5.3 (Feb. 1941) 77, Reprinted
W.E. 19.225 (June 1942) 243. (Introductory, gives extensive bibliography).
102. Payne -Scott, Ruby, and Green, A. L. " Superheterodyne tracking charts -II" A.W.A. Tec. Rev.
5.6 (Dec. 1941) 251. Reprinted W.E. 19.226 (July 1942) 290 (Derives general design equations and
arithmetical -mean tracking charts).
103. Green, A. L. " Superheterodyne tracking charts III " A.W.A. Tec. Rev. 6.2 (Feb. 1943) 97:
Reprinted W.E. 20.243 (Dec. 1943) 581 (Padded signal circuits and design charts).
104. Green, A. L. " Superheterodyne tracking charts-IV " A.W.A. Tec. Rev. 6.7 (July 1945) 423
(additional data on padded signal circuit but mainly confined to tracking R-C oscillators).
105. Green, A. L. " Superheterodyne tracking charts-V " A.W.A. Tec. Rev. 7.3 (April 1947) 295
(Tracking errors and design charts for geometrical-mean tracking).
106. Ross, H. A. and Miller P. M. " Measurement of Superheterodyne Tracking Errors " A.W.A. Tec.
Rev. 7.3 (April 1947) 323 (Experimental method of checking magnitude and distribution of tracking
errors).
107. Simon, A. W. " Tracking permeability tuned circuits " Elect. 19.9 (Sept. 1946) 138 (Method of
tracking identical inductors in signal and oscillator circuits).
108. Sowerby, A. L. " Ganging and tuning controls of a superheterodyne receiver " W.E. 9.101 (Feb.
1932) 70.
109. Cocking, W. T. " Wireless Servicing Manual " Text book (6th edit. 1944) Iliffe and Sons, London
(Gives general information on practical alignment of receivers etc.).
110. Sande!, B. " Tracking " Radiotronics No. 121 Sept.-Oct. 1946 page 101 (Gives practical design
details for typical receiver using arithmetical mean tracking. Method based on that of Payne-
Scott and Green (1941).
111. Yu, Y. P. " Superheterodyne tracking charts " Tele. Tech. 6.9 (Sept. 1947) 46. (Tracking Char
for permeability and capacitively tuned circuits).
112. Simon, A. W. " Tracking the permeability-tuned circuits " Elect. 20.11 (Nov. 1947) 142. (Method
of neutralizing effect of loop antenna reactance on tracking).
113. Simon, A. W. " Three point tracking method " Radio 30.11 (Nov. 1946) 20 (Simplified eqns. for
inductance or capacitance tuning ; broadcast band only).
114. Menzies, E. B. " Method of plotting tracking error " Elect. 20.1 (Jan. 1947) 128.
115. Strutt, M. J. O. " Noise -figure reduction in mixer stages " Proc. I.R.E. 34.12 (Dec. 1946) 942.
116. Stolze, W. J. " Input circuit noise calculations for F-M and television receivers " Comm. 27.2
(Feb. 1947) 12.
117. Strutt, M. J. O. " Properties of gain and noise figures at v.h.f. and u.h.f." Philips Tec. Com. 3
(March 1947) 3. Precis by E. G. Beard, Philips Tec. Com. 3 (March 1947) 17.
118. Miller, P. H. " Noise spectrum of crystal rectifiers ' Proc. I.R.E. 35.3 (March 1947) 252.
119. Roberts S. " Some considerations governing noise measurements on crystal mixers " Proc. I.R.E.
35.3 (March 1947) 257.
120. Tapp, C. E. " The application of super -regeneration to frequency- modulation receiver design "
Proc. I.R.E. (Australia) 8.4 (April 1947) 4.
121. Fairman, H. E. " Oscillator tracking methods in permeability tuning " Tele-Tech. 6.3 (March
1947) 48.
122. Middleton, D. " Detector noise " (review of article in Journal of Applied Physics, Oct. 1946)
Radio 31.1 (Jan. 1947) 4.
123. Kirkpatrick, C. B. " Three point tracking formulae " Proc. I.R.E. (Aust.) 8.7 (July 1947) 18.
124. Mueller, W. P. (extract) " H.F. Conversion with the 7F8 " Service 16.5 (May 1947) 28.
125. Edwards, C. F. " Microwave converters " Proc. I.R.E. 35.11 (Nov. 1947) 1181.
126. Whitehead, J. W. " Additive and multiplicative mixing' W.W. 53.121 (Dec. 1947) 486.
127. Stockman, H. " Superre enerative circuit applications " Elect. 21.2 (Feb. 1948) 81.
129. Haantjes, J. and B. D. H. Tellegen " The diode as converter and as detector " Philips Research
Reports, 2.6 (Dec. 1947) 401.
130. Bradley, W. E. " Superregenerative detection theory " Elect. 21.9 (Sept. 1948) 96.
131. Hazeltine, A., D. Richman and B. D. Loughlin " Superregenerator design " Elect. 21.9 (Sept.
1948) 99.
133. " Characteristics of pentodes and triodes in mixer service " R.C.A. Application Note AN-139
(March 15, 1949).
134. Coppin, K. J. " The tracking of superheterodyne receivers " Jour. Brit. I.R.E. 7.6 (Nov.-Dec.
1948) 265.
135. de Koe, H. S. " Tracking of superheterodyne receivers " W.E. 28.337 (Oct. 1951) 305.
136. Dammers, B. G., and L. J. Cock " Comments on frequency changers for 30 Mc /s to 120 Mc /a
for T.V. and F.M. receivers " Philips Tec. Com. Nos. 5,6,7 (1951). Reprinted from Electronic
Application Bulletin June /July 1950, Vol. 11 Nos. 6,7.
CHAPTER 26
INTERMEDIATE FREQUENCY AMPLIFIERS
BY B. SANDEL, A.S.T.C.
Section Page
1. Choice of frequency ... 1020
2. Number of stages ... 1021
3. Commonly used circuits 1022
4. Design methods 1025
5. Variable selectivity ... 1048
6. Variable bandwidth crystal filters 1050
7. Detuning due to a.v.c. ... 1061
8. Stability ... 1065
9. Distortion 1067
10. References 1069
1020
26.1 (ii) COMMONLY ACCEPTED INTERMEDIATE FREQUENCIES 1021
properties. Of course, almost any type of coupling can be used in i-f circuits but
are
transformers other than those using two windings coupled by mutual inductance
the exception rather than the rule in ordinary broadcast and communications receivers.
CONVER F.AMP.
B. AMC B+ B; ARC.
BIAS BIAS
COUPLED I.F. CIRCUIT
FIG 2S -I TYP CAL TRANSFORMER COUPLED FIG. 262 SHUNT CAPACITANCE
I.F. STAGE
9 Sect. 6(iii),
The type of circuit being considered is discussed also in Chapter 9 Sect. 10.
(iv), (v). Graphical methods of determining selectivity are given in Chapter
several trans-
Section 4 below gives detailed design methods and examples for
formers of this type.
1/Ci C2
Cm when CM > (C1, C2).
and also permits both transformers to be " slug " tuned without using elaborate
mechanical arrangements. In some cases where the transformers are both mounted
in one can, combinations of " slug " and capacitance adjustment are provided ; an
example of this is illustrated in Ref 12.
Many arrangements have been suggested for combined i -f transformers, and some
of these are described in Ref. 22. However, the circuit shown, or modifications of
it, has received fairly wide acceptance as providing quite a good solution to the pro-
blem.
(i) General
The design procedure for i -f transformers can be greatly simplified by the use of
charts and tables. If certain assumptions are made, which approximate to practical
conditions, the design procedure can be reduced to a few routine operations. Here
we will consider only the two winding transformer using mutual inductance coupling ;
the added capacitance coupling, which is always present, does not seriously affect the
results particularly as its presence is taken into account when setting the coefficient
of coupling (k).
The methods given below can be applied to inductive or capacitive coupling pro-
vided that the coefficient of coupling (k) is suitably interpreted [see appendix (vii)
below for methods of calculating coupling coefficients]. The practical two winding
transformer, as previously mentioned, has mixed coupling but this does not cause any
difficulty when the two forms of coupling assist each other [suitable connections are
given in (vi) below]. However, if the two forms of coupling are in opposition a
rejection frequency is possible which can occur within the pass band of the trans-
former. This effect is well known to receiver designers who have accidentally re-
versed the connections of one of the i -f transformer windings, and found that the i -f
stage gives practically no gain. This fact can also be made the basis of a useful method
for measuring k, the capacitive coupling being increased (using a calibrated capacitor)
until it equals the inductive coupling and zero voltage transfer then occurs at the
working frequency. For further data on mixed coupling see Ref. 4.
The initial assumptions will be that the primary and secondary inductances L1
and L, are given by L = i/L,L,. Also, it will be assumed that the values of Q do not
alter appreciably over the range in which the selectivity curves are taken. We will
not always take the primary and secondary Q's as being equal, and the advantages
1026 DESIGN METHODS-(i) GENERAL 26.4
As long as \V Q1Q2 .
design equations is negligible for most practical purposes.
(Q, + Q,) /2 the error in any of the usual approximate design
equations will be small. If Q1 and Q, differ by large amounts then the exact design
equations are necessary and can be obtained from Refs. 2, 3, 6 and 8, or the design
can be modified by using the universal selectivity curves to obtain the required results.
It is of interest to note that the simplified equations given by Kelly Johnson (Ref. 1).
Ross (Ref. 2) and Maynard (Ref. 6 and Figs. 9.17 and 9.18 of this book) are identical
when it is assumed that Q = Q1 = Q, and the various notations are made the same.
It may be thought that writing Q = y Q1Q, and L = y L1L, will be inconvenient
since the i -f transformer, as constructed, will have its primary and secondary in-
ductances and Q's equal. However, in the majority of cases L = L1 = Ls is applied,
and the method is extended to fulfil the condition that the unloaded primary and
secondary Q's should be equal while allowing the required Q = 1/Q1Q2 to be obtained
in the receiver, without further adjustment.
Critical- coupling, or a close approach to it, is most often employed in i -f trans-
formers but there is little difficulty in designing transformers for almost any degree
of coupling. All cases will be treated.
Universal selectivity and phase shift curves are given in Chapter 9, Sect. 10 (Figs.
9.17 and 9.18). Additional charts and tables are given, to be used as described in the
appropriate sections.
The design procedure generally consists of finding values of Q, k and stage gain
for given bandwidths at some value of i -f ; or of finding the required bandwidth for
values of Q and k previously determined. For clarity the cases of critical -, over -,
and under -coupled transformers will be dealt with separately. Single tuned circuits
are also included as they are sometimes required in i -f amplifiers. Additional data
for the design of F -M transformers will be given in Sect. 4(v).
Stagger tuning (Refs. 8, 13, 17 and 38) of i -f transformers (e.g. tuning primary and
secondary to different frequencies) to give substantially the same bandwidths as over-
coupled transformers, does not have a very wide application in F -M and A-M receivers
(except in cases where variable selectivity is to be used) and will not be discussed in
detail.
Stagger tuning of single and double tuned circuits is widely used in television
receivers, but this is a different application from the case where it is applied to the
comparatively narrow bandwidths of ordinary sound receivers. In this case the trans-
former primary and secondary are tuned to the same frequency, but this is not neces-
sarily the intermediate frequency.
.Since i -f transformers for television receivers involve special problems they are not
treated here (see Refs. 27, 28, 94, 95, 96 and 97) Also, triple tuned transformers
are not discussed, but the article of Ref. 26 gives an excellent treatment.
Finally, the design methods do not make allowance for regenerative effects, nor
should they be applied for finding the shape of resonance curves at frequencies far
removed from resonance. The selectivity curve shapes are assumed to be perfectly
symmetrical although in practice it will be found that this is seldom true.
(1) A transformer with equal values of primary and secondary Q's when un-
loaded. These will be denoted by Qu = Qui = Q,.3.
(2) The values of Q1 and Qu, to be such that when the transformer is connected
into the i -f amplifier, and loaded by the valve output and input resistances,
the desired value of Q = ,/Q1Q2 will be obtained.
(3) The required coefficient of coupling (k) (critical for this particular example)
to be unchanged. It will be described later how k can be pre -set to any
desired value for any two circuits coupled together.
(4) Excessive values of Qu1 and Qua are to be avoided (see the previous method
of determining Qu,) as far as possible, because of the practical difficulties
involved.
(5) The response curve of frequency versus attenuation to be that specified (or
very close to it).
All of these conditions can be fulfilled very closely, provided the approximations
made in deriving the design equations hold. A simple analysis of the circuits in-
volved, and including the required conditions, gives
a + V'a3 + 2Q'R1Raß
Qu - p
(7)
If
Qb
- 1 + Qu1Q2k2 (9)
Qu1 = Qu2 = Qu
then Qb - 1
Q.
+ (Q k)2 (9A)
where Q1 = primary Q (sec. o/c or detuned by large amount)
Q ,
= secondary Q (pri. o/c or detuned by large amount)
Qb = Q to be obtained when primary and secondary are coupled and the
secondary tuned to make the primary Q a minimum.
(Usual precautions as to can and earthy side of secondary winding being grounded
to be observed -it
is preferable to use the same order of connections for measurement
as those to be used in the receiver.)
When the transformer has different primary and secondary Q's, it is often advan-
tageous to use the higher Q winding as the primary when setting the coefficient of
coupling ; this applies particularly when the coupling is very loose.
The actual capacitance values tuning the primary and secondary for Q meter meas-
urements should include the allowance made for stray capacitance otherwise incorrect
slug positions (i.e. incorrect inductance values) will give rise to an error which can be
avoided.
In our example we desire a value for k = 0.0099 (for critical- coupling when Q =
1/Q,Q, = 101) ; Q = 131.
Then from (9A)
131
Qb 1 + (1.297)2 - 48.9.
All that is required is to adjust the spacing between the two resonant circuits until
the Q meter reads 48.9. The desired co- efficient of coupling has then been obtained.
Alternatively, by transposing terms in the equation, k is given for any values of Q1,
Q,,1, and Q
so that k /Q I - Q°
The method applies directly to under -, over -, or critically -coupled transformers
and is useful within the limits set by the usable range of the Q meter. For over-
coupled transformers additional methods are sometimes required, and the procedure
will be indicated in Sect. 5.
26.4 (ii) CRITICALLY-COUPLED TRANSFORMERS 1031
k /Ql - Qe (9C)
ke ,w/ Qb
and
k 2Q1Q2(Q1 - Qa) (9D)
ke Q#212 + Q2')
It might be noted that when Q1 = Q2 the expressions (9C) and (9D) are identical,
as would be expected.
a
a0 "T
I-
2
O
0
O 05 I IS 2 25 3 35 4_45 S 55 6
Qk
F10.26.4 GA N REDUCTION FACTORS FOR COUPLING
OTHER THAN CRITICAL
1/1 (Qky 1
db Times Down Qk (Qk)2 A
o 1.00 1.00 0.000 1.00 1.00
0.25 1.03 1.27 0.616 1.61 0.971
0.50 1.06 1.41 0.707 2.00 0.943
1 1.12 1.73 0.817 3.00 0.893
1.9 1.25 2.00 0.866 4.00 0.800
2 1.26 2.02 0.869 4.08 0.794
3 1.41 2.41 0.910 5.81 0.709
3.1 1.43 2.45 0.913 6.00 0.670
4 1.59 2.81 0.935 7.90 0.629
4.4 1.67 3.00 0.943 9.00 0.599
5 1.78 3.25 0.952 10.56 0.562
6 2.00 3.72 0.963 13.84 0.500
6.6 2.13 4.00 0.968 16.00 0.469
7 2.24 4.24 0.972 17.98 0.446
26.4 (iii) OVER -COUPLED TRANSFORMERS 1033
(B) Example
A 455 Kc /s i -f transformer is required to pass a band of frequencies 16 Kc /s wide
(i.e. ± 8 Kc /s). The variation in gain across the pass band is not to exceed 0.5 db.
(a) From Fig. 26.5 it is reasonable to take the total bandwidth as 16 Kc /s and
so the peak separation is 16/x2 = 11.3 Kc /s.
(b) 24f/f° = 11.3/455 = 0.0248.
(c) From Table 2, V1 -
1 /(Qk)2 = 0.707.
(d) From eqn. (12), k = 0.0248/0.707 = 0.035.
(e) From Table 2, Qk = 1.41
Q = 1.41/0.035 = 40.2.
(f) Assuming a valué for C, (= C2) of 80 µµF (including strays)
L = 25.33/0.4552 x 80 = 1.53 mH
and take L = Li = L2 = 1.53 mH.
(g) To determine the average stage gain in the pass band. From Fig. 26.4
(or eqn. 14), with Qk = 1.41, reading from the dotted curve, gain reduction
factor equals 0.97. Assuming we use a type 6J8 -G converter valve having a
conversion conductance of 290 mhos (0.29-mA /volt), then from eqn. (5)
and the gain reduction factor,
Average stage gain = 0.97 x i
x 0.29 x 40.2 x 0.455 x 1.53
= 24.7 times (or 27.9 db).
(h) Assume that the transformer is connected between a type 6J8 -G converter
and a type 6SK7 voltage amplifier and that both valves are working under
a particular set of operating conditions.
For type 6J8-G the conversion plate resistance r9 = 4Ml? = R, and for
the type 6SK7 the short circuit input resistance = 6.8 MD = R2 (as de-
termined from Chapter 23, Sect. 5).
From Equation (7)
a = 40.2 x 0.17 x 10.8 = 73.8 ; 012 = 0.544 x 10*
= 2[27.2 - 0.172] = 54.4 approximately
73.8 + V0.544 x 10* + 4.76 x 106
Q. - 54.4
41.4,
so that Q. = Q,.1 = Q,,, = 41.4 which is the unloaded value for primary and second-
ary Q's before the transformer is connected between the two valves ; the additional
refinement in design is hardly necessary here, and it would be sufficient to make
Q = Q, = Q2 = 40 (approx.).
PEAK SEPARATION
AVERAGE 2AFp
LEVEL A.
MAX. GAIN
VARIATION
9ANDIW IDTH
(f X 28r p)
fo
FREQUENCY
FIG 28.5 ILLUSTRATION OF TERMS USED FOR OVER - COUPLED YRANSFORMERS.
It can easily be checked, using the procedure set out for critically -coupled trans-
formers, that the geometrical-mean of the loaded values Ql and Q2 is 40.2 as required.
Complete resonance and phase shift curves are plotted as previously explained. The
values for k can be set in exactly the same way as explained in the section on critically -
coupled transformers, if this is convenient. If k is high it is preferable to apply
one of the following methods.
(C) k measurement (when k is high)
When k is high, one of the following methods may be used
The first of these uses the relationship
AC = C,k2 /(1 -
k2)
1034 (iii) OVER- COUPLED TRANSFORMERS 28.4
dC
or = d C + C1
k
where C, = capacitance required to tune the primary to resonance with the second-
ary open circuited.
k = coefficient of coupling required (say k greater than 0.1 or so)
and AC = increment in capacitance required to tune the primary to resonance
when the secondary is short circuited.
As an illustration suppose k = 0.2 and C, = 200 µµF (the exact working frequency
may not always be convenient) then AC = 8.34 µµF. Using a " Q" meter, the spac-
ing between primary and secondary is adjusted until this increment in capacitance is
obtained ; this gives the required value of k.
With some transformers, neither of the two methods given previously will be con-
venient, and a third method is required. In this case a " Q " meter is again used,
and the two transformer windings are connected firstly " parallel aiding " and then
" parallel opposing." Any convenient resonance frequency may be selected, and
the two capacitance values (C, and CE) required to resonate the coils with the two
different connections are noted. The coefficient of coupling is then given by
k - (1 2a+ aE/ (\ C, -+ )
C, C,
CE
where a2 = L, /L,. (This result was determined by J. B. Rudd).
For a pre -determined value of k the relationship
4akCE
dC- +a(a 1-2k)
is applied ; where d C = change in capacitance = C, -
CE.
As an example, suppose L, = LE (i.e. aE = 1) the capacitance CE = 100 µµF and
k = 0.1 is required. Then
4 x l x 0.1 x 100
dC =1 { 1(1 -2 x0.1) -22.2µµF.
This method should not be used for small values of k (say below about 0.02) as it
does not take into account capacitive coupling. For k equal to 0.02 or less the method
of Sect. (ii)E should always be used. For k greater than about 0.2 the short-circuit/
open -circuit method, given previously, is usually the most convenient. All measure-
ments must be made with the transformer in its can, and the can should be earthed.
In cases where a " Q " meter is not available, the operation of the transformer can
be checked using a single stage amplifier and measuring the selectivity curve with a
signal generator and a valve voltmeter. Alternatively, measurements can be made
in the receiver, and a typical case is illustrated in Chapter 27 Sect. 2(iv) in connection
with measurements on ratio detectors (see also Chapter 14 of Ref. 17).
and O = tan-1 f
fQ = tan-1 X (19)
Attenuation (p) N = 1 N = 2 N = 3
(B) Example
Two single tuned circuits are required to give an attenuation of 17 db for a total
bandwidth of 10 Kc /s. The i -f is 455 Kc /s.
(a) From Table 3, X = 2.45 (since N = 2).
(b) Q = 2.45 x 455/10 = 111.5 from eqn. (18).
(c) Assuming C = 200 µµF (including all strays)
L = 25.33/0.455' x 200 = 0.611 mH.
(d) The unloaded Q required depends on the combined effects of valve input
and output resistance. Take the loading for the two transformers as being
the same, for simplicity.
Then suppose r,, = 0.8 MQ and grid input resistance = 6.8 MQ the
effective shunt resistance is 0.715 MQ.
From eqn. (6)
-
Q = 111.5 x 0.715/(0.715 0.194) = 153.
(e) The gain of each stage (twice that for a critically- coupled transformer) is
gmQw,L, so that taking g,,, = 2 mA /volt (2000 µmhos) in each case, stage
gain = 27r x 2 x 111.5 x 0.455 x 0.611 = 390 times (or 51.8 db).
(f) Suppose we have the loaded Q given as 111.5 (as in our previous problem
using N = 2), and we require the bandwidth for 6 db attenuation ; from
table 3 obtain X = 1.0 and from eqn. (18) the total bandwidth (24f) is 4.08
Kc /s. In a similar manner the attenuation can be found when the band-
width is stated. The resonance curves could also be used to find X (= D in
Chapter 9, in this case) and p.
(C) Under- coupled transformers
If the transformers use very loose coupling, the methods for single tuned circuits
could be applied (N = 2 for each transformer) in conjunction with eqn. (7). This
1036 (iv) UNDER- COUPLED TRANSFORMERS 28.4
approach does not lead to very accurate results since the values of coupling are seldom
less than 0.1 of critical and more often are of the order of 0.5 to 0.8 of critical.
General design equations applicable to transformers having any degree of coupling
are given below, but it will be seen that they are not quite as tractable as in previous
cases unless an additional factor such as Q, k, or Qk (i.e. a given proportion of critical -
coupling) is specified. However, this will offer little difficulty.
X = [{(1 - as)2 + (p2/N
X2)2 + 4X2
1)(1 + a2)2 ).1 - (1 - (20)
(-(1 + as N12
P L (1 + as)2 ] (21)
X = (24f /fo)Q (22)
a
rX {X2 + (X2 + -
4)(p2 /N 1) }i - (X2 + p2 /N - 1)l+
elm -
Qk J (23)
= =L
= tan-1 2X/[1 XI +
O - 1
Modulation Index - dF
f
carrier frequency deviation
audio modulating frequency
dF
Values for may be interpolated with sufficient accuracy
f
dF /f = 0.01 -0.4 0.5 1.0 2.0 3.0 4.0 5.0 6.0
Bandwidth = 2f 4f 6f 8f 12f 14f 16f 18f
dF/f = 7.0 8.0 9.0 10.0 12.0 15.0 18.0 21.0
Bandwidth = 22f 24f 26f 28f 32f 38f 46f 52f
m (25)
p +
where m = amplitude modulation factor
and p = attenuation at a bandwidth of twice the deviation frequency (i.e. 24f).
Fig. 26.6 shows directly values of m for various values of X. Values of m are of
importance since they allow an estimate to be made of the amplitude limiting re-
quirements demanded from whatever device is incorporated in the receiver to " iron
out " amplitude variations. ao
Many F -M receiver designs allow anything
from 20% to 50% of introduced amplitude as
modulation, but these figures should always
be considered in connection with the amount
of non-linear distortion introduced by the
tuned circuits [see Sect. 9(8)]. Good designs
often allow considerably less than 20% of
introduced amplitude modulation.
O Os D a 1.2 1.6 20 as
It is also worth noting, before leaving this
PERCENTAGE OF INTRODUCED A -M
section, that the carrier frequency should FIG.26.6
IN AN F-M CURRENT FOR N
be regarded as a reference point only, since, CRITICALLY COUPLED TRANSFORMERS
L = inductance in µH
and B = experimentally determined constant.
As an example : A winding is to be made to have an inductance of 1.44 mH. Pre-
vious experiments have shown that, in the frequency range of 300 -900 Kc /s and for
*See also Chapter 11 Sect. 7.
1042 (vi) I -F TRANSFORMER CONSTRUCTION 26.4
an inductance of about 0.5 -2 mH, a two pie winding on a 9/16 in. former (each pie
5/32 in. wide, with 3/32 in. spacing between the pies and using 5/44 A.W.G. Litz
wire) has the factor B = 4.33. Then from eqn. (26)
N = 4.33A/1440 = 165 turns per pie ;
the two pies each of 165 turns, being connected series aiding. The same method
can be applied in cases where it is convenient, to any type of winding.
Since the presence of the iron " slug " will affect inductance (and r-f resistance)
it is necessary to determine its effect and also to calculate the variation in inductance
which can be made. The turns required are found for the condition with the " slug "
in the winding and in the position giving the mean inductance value. This means
that the value of L used in eqn. (26) will be less than the calculated value by the in-
crease due to the iron.
The value of the inductance (L) using a powdered iron core (e.g. magnetite) can
be found from
L = Lo [l -F a (r$) -
j2)(,ß ef, 1)] (27)
where Lo = inductance of air cored coil
rl = radius of iron core
r, = mean radius of coil
Il = length of core
12 = length of coil
¡Leff = effective permeability of iron core (Refs. 88 and 89 list values for µ
and p. eff for various types of iron powders ; typical values for p eff are
from 1.5 to 3, depending on the type of iron)
a = 0.8 when Il < I,
and a = 1 when 1, < 4.
The iron cores in common use are about in. to in. in diameter and range in length
from about in. to 1 in. An inductance change of about ± 10% when the " slug "
is moved through the winding, is generally sufficient for most requirements. The
dimensions required for solving eqn. (27) are available if the experimental procedure
previously suggested has been carried out on air cored coils ; or the whole procedure
can be carried out experimentally.
Some manufacturers make up 455 Kc /s i -f transformers completely enclosed in
powdered iron pots. There is often little difficulty with this construction in obtaining
Q's in the order of 150. Stray capacitances are often large, however, and sometimes
lead to very unsymmetrical resonance curves.
At the lower frequencies (up to about 1 or 2 Mc/s) Litz wire is advantageous for
obtaining high Q values and 3, 5, 7 and 9 strands of about 44 A.W.G. (or near S.W.G.
or A.W.G. gauges) wire are common ; Q values greater than about 120 will require
some care in the transformer construction, and the size of can selected will materially
affect the value obtainable (see Ref. 91 for illustration).
The required values of L1, L k, etc. for developmental purposes are conveniently
found using a Q meter. Methods for setting k, using a Q meter, have already been
outlined in this section. For an experimental transformer it is of assistance to place
the windings on strips of gummed paper (sticky side outwards) wrapped around the
former. In this way the windings can be moved quite readily along the former.
It should be noted that some variation in k can be expected when the transformer
is connected in the receiver because of alteration of" slug " position, added top capacit-
ance coupling (this occurs for example because of capacitance between the a.v.c. and
detector diodes) etc. Regeneration is also troublesome as it alters the effective Q
values, and hence the conditions for critical k. To avoid overcoupling it is sometimes
desirable to make the value of k somewhat less than is actually required (often about
0.8 to 0.95 of the critical value depending on the receiver construction and i-f). If
all added coupling is accounted for and the Q values are those specified then no diffi-
culty arises. Some slight increase in coupling, in the receiver, is not necessarily
serious, because the loaded primary and secondary Q's are not always equal. In this
26.4 (vi) I -F TRANSFORMER CONSTRUCTION 1043
case a double hump in the secondary voltage does not appear until transitional coupling
has been exceeded, and the k required for this to occur is always higher than k critical ;
transitional and critical k are the same, of course, when primary and secondary Q's
are equal.
For the capacitance and mutual inductance coupling to be aiding, the primary and
secondary windings are arranged so that if the plate connects to the start of the prim-
ary, then the grid (or diode plate) of the next stage connects to the finish of the second-
ary winding ; both coils being wound in the same direction. This method of con-
nection also assists in keeping the undesired capacitance coupling to a minimum.
The order of base connections is also important in reducing capacitance coupling and
the grid and plate connections should be as far from one another as possible.
The cans to be used with i-f transformers should be as large as is practicable. They
are generally made from aluminium, although copper was extensively used at one time.
Cans should preferably be round and seamless. Perfect screening is not obtained,
in general, and care is necessary in the layout of the various stages to ensure that the
transformers are not in close proximity to one another. When mounting the trans-
former into a can, if there is a choice as to the position of the leads (although this is
largely determined by the valve type available) it is always preferable to bring the con-
nections for each winding out to opposite ends, as this reduces stray capacitance
coupling. The effects of the coil shield on inductance and r -f resistance can be cal-
culated (see Chapter 10 and Ref. 4, p. 134) and the results serve as a useful guide, but
direct measurement on the complete transformer is the usual procedure. Mechanical
considerations generally ensure that the can is thick enough to provide adequate
shielding (for considerations of minimum thickness see Ref. 4, p. 135).
Methods for determining gear ratios, winding pitch and so on, for use with coil
winding machines are discussed in the literature (Refs. 29, 30, 31, 32) ; see also Chapter
11, Sect. 3(iv) in particular.
Finally, the measure of the success of any i -f transformer design will be how closely
the predicted performance approaches the actual results obtained when the transformer
is connected into the receiver.
/Cm
k= 1
C2 C2
(C1 + Cm)(Ce + Cm)
"V
iim/ \C$ ` + Cm/
For Fig. 9.13 using rule (b)
k
r (L, + Lm)(Ls + Lm)
+
It is sometimes helpful to make use of the obvious relationship in the form of k
for the two types of circuit, by drawing dual networks and writing down the value of
k for the duals as though capacitances were inductances and vice versa, e.g. the in-
ductances of Fig. 9.13 are the duals of the capacitances of Fig. 9.14, and so the value of
k is written down for, say, Fig. 9.14 and if k is required for Fig. 9.13 we merely sub-
stitute the symbols L, L, Lm for C, C, C,,.
As a more difficult example the coefficient of coupling for Fig. 9 16(A) will be cal-
culated. Firstly redraw the two coupled circuits as two T sections and then combine
the two directly connected series arms. The complete coupling circuit now com-
prises a series arm L1 -
M, ; a 17 section made up from M1, Lí
-
M1 -
+ L,' -MM, ; a series arm L, M,. Transforming the II section to a T
section and adding the two series arms of the previous circuit to the new series arms,
we have a single T section in which the first series arm is
M,(Lí -M, +Le -M,) _L,(Lí-I-Ls)- M,(M, +M,)
L,' + L,' + L1 - M1 Lí + L=
the second series arm is
M,(L,' - M1 + Lí - M,) -
L,(L,' + Lá) M,(M, + M,)
L,' + L,' + L, - M' L,' + L,'
M1Ma
and the shunt arm is Lí
L,' +
Applying rule (a), and writing L. = L,' + L2',
k -_/ M,M'
V [L1Lm - M,(M1 + M,) + M,M,][L,Lm - M,(M, + M,) + M1M,]
- M1M,
M,')
1/ (L1Lm M1')(L2Lm
/elk,
1.7(1 - k12)(1 - k,')
M,
where k1 - - /
M1
V L1Lm
andk,- L,Lm
As a further example a bridged T coupling section will be considered, made up
from inductances bridged by a capacitance. This corresponds to the familiar case
of a mutual inductance coupled transformer with added top capacitance coupling
(e.g Fig. 9.16D) and also refers to any practical i -f transformer because of the presence
of stray capacitance affecting the coupling between the primary and secondary wind-
ings.
The procedure in this case is to use rule (a) and determine the coupling coefficient
due to mutual inductance (the inductances are considered as forming a T network) ;
call this km. Next use rule (b) to determine the coupling coefficient due to capacit-
ance coupling (the capacitances form a II network) ; call this k,. Then if the con-
nections are such that the two forms of coupling aid each other, the resultant coupling
coefficient is
rvr
km +k,
1 + kmk,
and if kmk, < 1 then
26.4 (vii) CALCULATION OF COUPLING COEFFICIENTS 1045
kT = km + k4, (approximately).
For the case of the two forms of coupling being in opposition
km - k44
kT
1 - kmk,,
and if kmk,, < 1 then
kT = km - k4, (approximately).
The above expressions for kT can be derived quite simply from data given in Ref.
101, or directly using a similar method to that set out below. The results are most
helpful in determining rapidly the effect of additional top capacitance coupling e.g.
an i-f transformer is connected between the last i -f amplifier valve and a diode de-
tector, and the a.v.c. diode is connected to the primary of the transformer in the usual
manner. Then clearly there is top capacitance coupling added across the i -f trans-
former due to the direct capacitance between the two diodes, and a knowledge of the
.value of this capacitance will allow the added coupling to be taken into account when
designing the transformer. If the capacitance value is not known its effect can be
estimated when a response curve is taken.
This method for finding kT can be applied in exactly the same way to a bridged T
coupling network in which the two main series arms are inductances with mutual
inductance coupling between them, the shunt arm is a capacitance and the T is bridged
by a capacitance.
An alternative procedure, and a most important one, for determining k will now
be discussed. This method can be applied quite generally to determining coupling
coefficients for H and T networks which have L1C1 = L,C2. The method is due to
Howe and the reader should consult Refs. 99 and 100 for a more detailed explanation.
This procedure uses the relationship
k - 0)12
- 0)E2
CÚ12
+, G/22
where w1 is the higher angular frequency of free oscillation in the circuit and w2 is
the lower angular frequency of free oscillation.
w1is calculated from Fig. 26.6A, which shows a symmetrical T network with mixed
coupling, by considering the capacitors C to be charged as shown. During the dis-
charge of the capacitors it is clear that LmCm have no effect on the resonant frequency
in this case, and
w12 = 1 /LC.
w2 is calculated from Fig. 26.6B in which the capacitors C are charged as shown.
In this case, when the capacitors are discharging, it is convenient to consider the
shunt arm as being made up of two equal parts each carrying the primary and second-
ary current alone ; the reactances of Lm and Cm being doubled to allow for the single
current. Then
1
wEE
(L 2Lm)¡ CCm/2
`C+Cm/2f
so that =
LmCm LC
LmCm +LC +LCm
-
A somewhat similar procedure can be applied to a H network, or an equivalent T
network can be found and the above procedure applied.
1046 (vii) CALCULATION OF COUPLING COEFFICIENTS 26.4
For cases where the sections are unsymmetrical some modification in the procedure
is necessary. Taking the case of a T section using inductance coupling (with L1C1 =
L2C,) the shunt arm is replaced by two inductances (x1 and x,) in parallel such that
their combined inductance equals that of the shunt arm (L,,,) and such that the values
selected will give equal resonant frequencies for the two circuits
L, + x1 C2
i.e. -
L2 + x2 C1
The case of capacitance coupling is treated in a similar way. The procedure is then
exactly as before for the symmetrical coupling network. There would be little diffi-
culty in extending this procedure to the case of the unsymmetrical T section using
mixed coupling, but it is worth noting that numerical solutions are much easier to
handle than a general algebraic solution even in the case of simple coupling.
For the two mesh network just considered, using mixed coupling and in which
L1C, = L,C2 it is usually simpler to use the relationship
k, + k,
kT 1 + kmk,
as was done for the bridged T network. Taking the example just considered,
Lm
km - L + Lm
C
k, - C + Cm
Lm C
- L +Lm C + Cm
and kr LmC
1 -(L+Lm)(C+Cm)
LmCm LC -
LmCm + LC + LC,
exactly as before.
It should be carefully noted that in all cases of mixed coupling the sign of L,,, is
most important. For the example just given Lm was taken as being positive ; this
is the case of the two forms of coupling being in opposition, and zero voltage transfer
occurs when the shunt arm is series resonant. With Lm negative the two forms of
coupling aid each other and there is no series resonant frequency. The aiding con-
dition, which is the one usually required, is possible only when the coupling between
the two inductances is due to mutual induction, as a physical coupling inductance
would act so as to oppose the capacitive coupling.
The methods given so far, for determining the coupling coefficient, are useful for
the types of circuit considered. However, it should be observed that they can be
applied directly only to the following cases :
(1) Circuits having coupling elements all of the one kind i.e. all capacitances or all
inductances (simple coupling).
(2) Circuits having mixed coupling in which the primary and secondary circuits
are tuned to the same frequency. However, it is only necessary that L1C1 = L2C2
and the primary and secondary circuit elements need not be identical.
Determinauon of k for the cases just stated usually presents little difficulty and is
simplified by choosing one or other of the procedures outlined ;,these cases cover
most practical requirements. A real difficulty does arise when the primary and
secondary are tuned to different frequencies and the coupling is mixed.
It may now be of interest to consider the procedure given below which makes use
of the result obtained by Howe (Ref. 98) and follows from the definition that the
coupling between two circuits is the relation between the possible rate of transfer of
energy and the stored energy of the circuits ; by the possible rate of energy transfer
is meant the rate of energy transfer in the absence of all resistance other than that
utilized for coupling.
26.4 (vii) CALCULATION OF COUPLING COEFFICIENTS 1047
and k - 1
wlwe (L1
w1Lmw2Lm
+ Lm)(L2 + Lm)
- exactly as before.
1/(L1 + Lm)(L2 + Lm)
It can be seen that the primary and secondary frequencies need not be considered
in cases of this type involving simple coupling only.
Now consider an unsymmetrical T section with mixed coupling, and different
primary and secondary resonant frequencies :
- w1Lm - w1Cm
wi(L
- w18CZm12
1/
,n/
Z°`zl
= w2Lm - wßC,,, = wz(1 m- w 88 C m /
W1 = Ll + Lm
WS =L2+Lm
1 / wtw2(Lm - w18C,,, L. w22Cm/
wlwß (L1 + Lm)(L2 + Lm)
- Lm - w12Cm (L"`
1
wa2Cm)
(L1 + Lm)(L2 + Lm)
and since
Cl + Cm
w12 (L1 ± Lm)C1Cm
C2 +Cm
- (L2 + L,)CIC,
w8
2
k
L,,Cm LC -
(L + Lm)(C + Cm)
Since this case corresponds to that used in connection with Figs. 26.6A and 26.6B
it would be expected that the results would correspond. However, it is seen that
there is an additional term LmC in the denominator of the second solution. This is
explained by the method used for determining w, and cot in the latter case, and so
the method as given here cannot be considered exact. The approximation involved
in this particular example is that the result is equivalent to
km ± k,
kT = km ± k, and not kT - 1 ± kmk;
This is seen quite readily because
Lm C
km -
L + Lm' k, C + Cm
-
and so kT =
Lm(C + Cm) C(L + L,I,) - LmCm LC
(L + Lm)(C + Cm) (L+ Lm)(C + Cm)
which is the result just obtained.
A rather similar difficulty occurs when determining the central reference frequency
for the transformer of Fig. 26.2. The angular frequency (w) is determined exactly
as above by considering the secondary on open circuit, and for this case
w12 + w21
2
where w1 and w2 are now determined in the same way as for Figs. 26.6A and 26.6B.
TERTIARY
COIL
OTHER PURE
TYPES OFD. M
COUPLINGI,
fe
FIG.26 7
With any other type of coupling which has no mutual inductance component, one
of the two humps remains at the intermediate frequency. The mistuning is then
half the frequency bandwidth between humps.
The method of switching tertiary coils approximates far more closely to the curve
for pure mutual inductance (M in Fig. 26.7) than it does to the other curve. The
26.5 VARIABLE SELECTIVITY-(i) GENERAL 1049
tertiary coil may have no more than 5% of the turns on the main tuning coil and less
than 0.5% ratio of inductance. The symmetry of the overall selectivity curves is
usually good.
Variable capacitance coupling may be used and the coupling capacitor may be either
a small capacitor linking the top end of the primary to the top end of the secondary
or it may be a common capacitor in series with both primary and secondary circuits.
This latter arrangement is commonly known as " bottom coupling."
For " top coupling " very small capacitances are required and the effect of stray
capacitances is inclined to be serious, particularly in obtaining low minimum coupling.
It may be used, however, with a differential capacitor arrangement whereby con-
tinuously variable selectivity is obtained (Refs. 41, 42). The differential capacitance
in this case adds to or subtracts from the capacitance in the primary and secondary
circuits to give the requisite tuning. The disadvantage of this arrangement is that
sufficiently low minimum coupling is very difficult to obtain and the capacitor is a
non -standard type.
" Bottom coupling " has results similar to " top coupling," but is easier to handle
for switching and also has advantages for low coefficients of coupling. A two or three
step tapping switch may be used to give corresponding degrees of bandwidth pro-
vided that simultaneously other switch contacts insert the necessary capacitances in
the primary and secondary circuits, for each switch position, to give correct tuning.
LI
Q 100
8+
MF
002 --{O MrF
002pF
CONVERTER
A.V.C.LINE
TWO POSITION SELECTIVITY FIG. 26.7A
One such bottom coupling method which requires the minimum of switching for
two degrees of coupling is shown in Fig. 26.7A (Ref. 42a). The two coils Li and L,
should each be in a separate shield can. With switch S, in the position shown, the
selectivity is broad ; in the lower position the selectivity is normal. Capacitances
as shown are only illustrative. The same principle can be used for three values of
coupling, using seven capacitors.
In most cases, the second i-f transformer is slightly under- coupled and its peak is
used to fill in the " trough " between the peaks of the first transformer in its broad
position.
A wide variety of circuit arrangements for obtaining variable selectivity has been
described in the literature. Design methods and practical constructional details
for many of these can be found in the references listed at the end of this chapter.
Further discussion is also given in Chapter 11, Sect. 3(iii).
(ii) Automatic variable selectivity
Automatic variable selectivity allows the bandwidth of the i -f amplifier to be varied
with signal strength, the pass band being a maximum for the strongest signal.
Many types of circuits have been developed using variable circuit damping, varying
coupling reactance and circuit detuning. A number of the circuits are not very
practical arrangements since they involve the addition of a number of valves whose
sole function is to provide variable bandwidth.
One interesting circuit arrangement (Ref. 46), which does not require any additional
valves, is shown in Fig. 26.8.
To understand the operation of the circuit it is first necessary to consider what
happens in an ordinary mutual inductance coupled transformer. It is well known,
1050 (ii) AUTOMATIC VARIABLE SELECTIVITY 26.5
when two circuits are coupled together, that there is a back e.m.f. induced in the
primary winding from the secondary which will have a marked effect on the magni-
tude and phase characteristics of the primary current as a function of frequency.
The primary current response curve exhibits a double peak at values of k even less
than critical, because the back e.m.f. represents a voltage drop that decreases rapidly
on either side of the resonant frequency. Double peaking of the primary current is
masked in its effect on the secondary current by the selectivity of the secondary circuit.
When the coupling exceeds critical (actually transitional) coupling the secondary
current exhibits two peaks, but to a lesser extent than the primary current.
CONVERTER F VOLTAGE
(6A 8) AMPLIFIER
VI S--0--S (687)
Vs
LI- 1.18.01
L2' 114mH
0.05 M - 28.7pH
PF 3-15 OI- 02115
n pp F.
LI ¿L2 WOUND IN
0.05 lpoon SAME DIRECTION
(BUT CONNECTED
AS SHOWN)
AMC.
F IG.26e AUTOMATIC SELECTIVITY ARRANGEMENT
If the magnitude of the counter e.m.f. is increased by increasing the mutual in-
ductance, and at the same time a voltage of almost equal magnitude and opposite
phase is added in series with the primary circuit, a single peak of output current can
be obtained in the transformer secondary. The additional series voltage, so intro-
duced, will increase the stage gain and selectivity from the condition which existed
prior to its introduction ; in our case the prior condition is that the stage gain is less
and the secondary current has two peaks, i.e. the transformer is initially over -coupled.
In the circuit of Fig. 26.8 the valve V, performs the dual functions of amplification
and production of the series voltage to be added to the primary circuit. A resistance
in the cathode of V, produces an a.c. voltage drop, which is proportional to the g,,,
of the valve and the secondary current of the transformer ; this voltage drop across
the cathode resistor is the one applied back to the transformer primary. When a
strong signal is being received a.v.c. will increase the bias on the grid of V, and so its
g,,, is reduced. When the g,,, (and consequently the voltage drop across the cathode
resistor) is sufficiently small, peak separation is obtained ; the width between the
peaks depending on the value of mutual inductance. Weak signals reverse the pro-
cess and narrow down the overall response curve.
A quantitive analysis (setting out design data and instability conditions) is given
in the Ref. (46). For the circuit of Fig. 26.8 the peak separation is given as 8 Kc/s
foi an i -f of 450 Kc /s. Other designs are shown for peak separations of 15 Kc /s
and also for obtaining two position automatic selectivity ; the latter arrangement
necessitates the use of a relay.
Other types of automatic selectivity are discussed in the references listed.
In this network suppose we consider that R is zero, then between terminals 1 and 2
there is a reactance
X =
cuL - 1 /wC
co Co [wL - 1
ca
C1
C
1
Col)]
(28)
When wL = 1/ (DC the reactance is zero and the circuit is series resonant.
When wL = CC
+ Co) the reactance is infinitely large, and the circuit is
parallel resonant (anti -resonant).
The difference in frequency between f, and f, is given approximately by
df =
2 C fr.
From the values of the 455 Kc /s crystal constants which are given in (iii)E below,
it will be seen that Cif is about 250 c/s in a typical case.
The presence of R will slightly modify the conditions for which parallel resonance
occurs. For our purposes the conditions given are near enough.
It should be clear from the circuit that the parallel resonance frequency (f,) will be
higher than the series resonance frequency (f,). For frequencies above f, the resultant
reactance due to L and C in series is inductive, and this is connected in parallel with a
capacitive reactance due to Co. A typical curve of output voltage versus frequency
change, for this type of circuit, is shown in Fig. 26.10.
If the value of Co could be altered as required, then the position of f, would be
variable. Suppose by some means we are able to connect a negative capacitance
CN (or a parallel inductance will give somewhat similar results) across Co ; then the
value of the capacitance shunted across the series circuit will be
C' = Co Cs.- (29)
From this, it follows that C' can be reduced from its initial value of Co (when CN is
zero) until it becomes zero, and CN has then exactly neutralized Co. The response
curve would now be that for the series circuit (made up from L, R and C) alone, which
behaves as a pure series resistance at fr. If the magnitude of CN is further increased,
then C' becomes negative and the series circuit is shunted by a negative capacitance ;
which is equivalent to shunting an inductance across the series circuit (which behaves
like a capacitive reactance for frequencies below f,). This means that the parallel
resonance frequency (f,) will now be lower than the series resonance frequency (f,).
Any frequency above or below the series resonance frequency can now be chosen as
the rejection, or parallel resonance, frequency and the current through the circuit,
at this point, will be reduced ; the values for f,, and the current depending on the
circuit constants.
To achieve the variation in C' the circuit of Fig 26.9 (i.e. the crystal) is incorporated
in the bridge circuit of Fig. 26.11, in which CN is made variable to achieve the results
discussed. In this circuit ZL is the load impedance ; Z, is the impedance of the
voltage source ; Z, and Z_ are any two impedances used to make up the resultant
1052 (i) BEHAVIOUR OF EQUIVALENT CIRCUIT 26.6
I -F TRANS. 3 CN I -F TRANS. 3 CN
(A) (B)
FIG.26.12 PRACTICAL CIRCUIT ARRANGEMENTS FOR QUARTZ CRYSTAL
FILTERS
operation is confined to an i -f of 455 Kc /s, and to tap down on the circuit when this
is possible.
Two other points are worth mentioning. The first is that the crystal stage gives
high selectivity around resonance, but the " skirt " selectivity may be quite poor ;
for this reason the other stages in the receiver must provide the additional " skirt "
selectivity required. In addition, good " skirt " selectivity is a requirement of the
i-f amplifier to minimize any possible undesirable effects which may arise because of
crystal subsidiary resonances. The other point is that having CN as a variable control
is not necessarily a great advantage, and simpler operation is obtained when CN can
be pre -set to neutralize Co. This cannot be done with all types of variable bandwidth
circuits, since the conditions for neutralization may be altered as the bandwidth is
cli2nged.
(iii) Design of variable bandwidth i-f crystal filter circuits
(A) Simplifying assumptions
From Fig. 26.11, when a capacitance balance is obtained as far as Co is concerned,
CN Co(Z1 /Z2). (30)
Since Co is generally about 17 µµF it is convenient to make Z1 = Z2. Also, CN
must have very low losses if the attenuation at the rejection frequencies is to be high.
The general design will call for all capacitors to be of the low loss type. When the
condition of eqn. (30) is fulfilled, the equivalent circuit reduces to that of Fig. 26.13(A).
R L L R C
C 2 ,
I I
is
EO
4 4 4
where g, conductance of V,
Q =vmutual
Q,Q2 = /k, 1
= 2ar x i-f
coo
Stage gain =
E, E', E,, -¡ ZL 1¡gmQw,Ll.
E', x E, E, ZL + Z', + Z, ` 4 J (33)
Eqn. (33) gives a great deal of information about the circuit. For constancy of gain,
it follows that ZL should be very much greater than (Z', + Z,). Since Z,, (or R
at resonance) cannot be altered by the receiver designer, it is necessary to make Z',
as small as possible. However, as Z', is made smaller the overall gain will be affected
since it is a centre tap on the output of the input i -f transformer. The maximum value
for ZL will be limited by the maximum bandwidth requirements, and the permissible
values of circuit constants. The other factors affecting gain are g,, Q and L. For
a given valve under a fixed set of operating conditions, g, is practically outside the
designer's control; the value of L(= V/L,L2) is made as large as possible consistent
with the requirements of minimum permissible tuning capacitance ; Q = 1/Q1Q,
is adjusted so that Q, is made as high as possible, and since we desire Z', to be low a
suitable value is selected for Q2. The method of determining Q has a large effect
on the stage gain which can be obtained, as will be seen later in the illustrative example.
(C) Gain variation with bandwidth change
For constancy of maximum gain at f, (i.e. the i-f) it has been suggested that Z',
should be small. There is a limit, however, to the minimum gain variation that can
be obtained for a given change from maximum to minimum bandwidth. From the
preceding equations it may be deduced that the maximum stage gain (A, is ob- ,)
tained when the bandwidth is greatest, and the minimum gain (A,; when the band- )
width is least. This limit in gain variation (since R for the crystal is fixed) is the con-
dition for which the voltage source impedance (Z',) becomes zero, which occurs when
A RT2(RT, R) -
a0
- Amax RT2(RT2 - R) (34)
where = total series resistance (at f,,) when the bandwidth is least
RT2
(= R + Z', + ZL ; in which ZL has its smallest value)
and RT2 = total series resistance (at f0) when the bandwidth is greatest
(= R + Z',, + Zl ; in which ZL now takes up its largest value).
To determine the voltage source impedance (Z',) for the usual condition where
Z', is not zero, we use
- R) - aRT1(RT2 - R)
Z'. - RT2(RT,
R T2 aR T2 - (35)
which, for convenience, is rewritten in terms of eqn. (34) as
Z', - a - a
ao a (35A)
RT2 -R RT2 -R
where R T, and RT2 have the values given immediately above,
R = equivalent series resistance of crystal
and a = A mitt
Amas
- gain variation desired, and must always be less than one.
If a value is selected for Z', then
=
RT2(RTl - R) - Z',RT2
a
RT2(RT2 - R) - RT1Z', (36)
26.6 I-F CRYSTAL FILTER CIRCUITS 1055
(D) Selectivity
Selectivity around resonance can be calculated by the methods to be outlined, but
for most purposes it is sufficient to determine the bandwidth for a given attenuation
at the half power points (3 db attenuation or 1 PO- of the maximum voltage output)
on each of the required selectivity curves.
Considerable simplification in the design procedure is possible, if the bandwidths
fr,r 1 db attenuation (or less) are known. In this case it would be sufficiently accurate
to take ZL and Z, equal to their dynamic resistances at resonance, at least for most
practical conditions, without introducing appreciable error. The advantage obtained
being that, in what follows, R R' and RTz = R'T2 for all conditions. The
value of Q. would be given by QS = f0/4df for 1 db attenuation, and this expression
would be used in place of eqn. (38). However, the procedure given is more general
and there should be little difficulty in applying the simplified procedure, if necessary.
Further the method given illustrates (in reverse) how the bandwidths near resonance
can be calculated for various amounts of attenuation.
To determine the bandwidth at the half power points proceed as follows : first
it may be taken that the equivalent inductive reactance of the crystal is very much
greater than the inductive reactance of the load and source impedances. Also, pro-
vided the variations in gain are not allowed to become excessve, at and near resonance
the value of Z', is very closely R',. For very narrow bandwidths, near resonance,
it is also sufficiently close to take ZL = RL ; for large bandwidths the resistive com-
ponent (RL) of ZL at the actual working frequency will have to be found. From these
conditions, we have
04, w°L 1
(37)
Q: = RL + R', + R R'1 WOCR'T
where Q. = the equivalent Q of the circuit of Fig. 26.13(B)
L = equivalent inductance of the crystal
= equivalent series capacitance of the crystal
C
= equivalent series resistance of the crystal
R
RL = resistive component of ZL, the load impedance, at the frequency being
considered (= ZL at resonance)
R', = resistive component of Z', the voltage source impedance
R'T = RL + R', + R = total series resistance at frequencies away from fo
(for very narrow bandwidths RL ZL and R'T N RTl)
and wo = 2w x fo (where fo is the i-f).
From the principles of series resonant circuits we know that the total bandwidth
(24) for the half power points (3 db atten.) is given by
2df = fo /Qx. (38)
It follows from eqns. (38) and (37) that for large bandwidths Q. should be small
and so R'T should be large. For narrow bandwidths the reverse is true, and the
limiting case for the narrowest bandwidth would be when R' T = R (i.e for the crystal
alone) ; a condition impossible to achieve in practice because of circuit requirements.
Since ZL > (Z', + R) for constancy of maximum gain at fo, under conditions of
varying bandwidth, it also follows that Q. will mainly be determined by the magnitude
of the resistive component of the dynamic impedance of ZL.
Some additional selectivity is given by the input i -f transformer, and for more
exact results the attenuation for a particular bandwidth would be added to that found
for the crystal circuit. This additional selectivity is usually negligible around the
" nose " of the resonance curve, but can be determined from the universal resonance
curves of Chap 9 Sect. 10, evaluating D and b for unequal primary and secondary
Q's (Q1 and Q2) and noting that Q in these expressions is Q2.
The resonance curve for the complete circuit is seldom necessary for a preliminary
design. The procedure is rather lengthy but not very difficult. First it is necessary
to fmd the resistive component of ZL (i.e. RL) at frequencies off resonance. This is
carried out from a knowledge of 1ZLj(= ZL) at resonance, and by determining the
reduction factors from the universal selectivity curves. Multiplying IZL1 by the
1056 (iii) DESIGN OF VARIABLE BANDWIDTH 26.6
indicated attenuation factors gives the required magnitudes of load impedance IZ'L I.
To find the resistive component, values of O corresponding to the various bandwidths
(and values of Z'L I) are read from the universal phase -shift curves of Chapt. 9 Sect. 10.
I
Then
RL = IZ'LI cos B. (39)
The resistive component of Z', can be found in a similar way from
R', = Z', cos ¢
but since the impedance of Z', is generally small, and the circuit relatively unselective,
R', can often be taken as equal to Z', for a limited range of frequencies near resonance.
When Z', is very low, it can be neglected in comparison with RL + R.
Then, since the resistance (R) and the inductance (L) of the crystal are known,
eqn. (37) can be applied to determine the values of Q, corresponding to the various
bandwidths. (Strictly f, would be replaced by the actual operating frequency off
resonance, but this is hardly necessary).
Knowing the different values of Qs, and since the bandwidths corresponding to
each Q, value are known, the various points for the complete selectivity curve can be
found from the universal resonance curve for a single tuned circuit (it is unimportant
that this is for a parallel tuned circuit rather than a series tuned circuit, since the
conditions for parallel and series resonance are practically the same provided the Q
is not less than about 10 ; there is, however, a reversal in sign of the phase angles
for the two cases). Additional selectivity due to the input transformer can be taken
into account if this degree of accuracy is thought to be necessary.
(E) Crystal constants
Before a complete design can be carried out, some data on the equivalent electrical
constants of the quartz crystal must be available. In most cases details can be ob-
tained from the crystal manufacturer.
Typical values for the electrical constants of 455 Kc /s quartz crystals widely used
in Australian communications receivers are :
o- I
ol O
VERY LOW
05C. RRESISTANCEr VTVM
O O
(A) (s)
FIG.2615 APPROXIMATE METHODS FOR DETERMINING EQUIVALENT
ELECTRICAL CONSTANTS OF QUARTZ CRYSTALS
If the required data cannot be obtained, then details of methods for measuring,
firstly, the value of Q for the crystal, can be found in Ref. 65. The experimental
set -up is shown in Fig. 26.15(A) and Q is found from
=
f, E,
(40)
Q
2(J D - f.) E9
where f, = series resonance frequency
= parallel resonance (or antiresonance) frequency
fD
E, = voltage across terminating resistance at series resonance
and Ea = voltage across terminating resistance at parallel resonance.
A knowledge of Q will allow the filter circuit to be designed, but if values for C
and C. are required these can be readily measured (Ref. 64) by using the arrangement
of Fig. 26.15(B) and the relations
C -
CA+C6 (f0, f, (41)
26.6 I -F CRYSTAL FILTER CIRCUITS 1057
and fA - 217VLACA
1 (42)
455KCTS XTAL V
CN 160
M
3-35 uu
NNF
Ry
SELECTIVITY
CONTROL
01
B. A.IP
VI,V2 TYPE 65117 (MICA)
R,,.0 -40811 (FOR BANDWIDTHS STATED)
LLO'S86mH OL 140
LI'4.9741 QI_140
L2-0.32 mH 02.140
k,0.234 R, -246011
CP-241)JF (INCLUDING 51R405)
r 1 26:6 CRYSTAL FILTER CIRCUIT DESIGNED FROM ILLUSTRATIVE EXAMPLE
which can be varied from 200 c/s to 3 Kc /s. The variation in maximum stage gain
should not exceed about 2 db (f
1 db about the average gain) but it is desirable to
1058 (iv) DESIGN EXAMPLE 26.6
keep the stage gain as high as possible, consistent with stable operation with varying
signal input voltages (i.e. large detuning of the i -f circuits should not occur when the
signal voltages vary over a wide range).
To make the problem complete, it will be assumed that the filter is connected be-
tween two type 6SK7 pentode voltage amplifier valves. The complete circuit is
shown in Fig. 26.16.
(1') Since Z, = Z2 and Co 17 µµF, let us select a suitable capacitance range
for CN. The smallest residual capacitance for Cy will be about 3 µµF. From this,
8550 x 2275
and R'T, =
151.6 -0.128 MQ.
So that (since R = 1500 Q)
R Z', = 8550 - 1500 = 7050 Q
RL2
+Z',= 0.128 - 15 x 10- '=
0.126MQ.
For narrow bands of frequencies, RLI ZLI, and we may write
R'Tl = RTl and ZLI + Z', 7050 Q.
(4') To find ZL2. It should be clear, from the values just given (Z', remains
unchanged) that for fairly large bandwidths
RLZ > Z',.
Assume that, for the load circuit, QL = 140. Then by calculation, or from the
universal resonance curves, we have for a single tuned circuit and a frequency of
455 Kc/s,
ZL ac resonance
IZ'L2I at 3 Kc /s bandwidth -
137.
Also, the phase shift O = 42° 43'.
Using these factors in conjunction with eqn. (39),
RL 0.126
IZ'L21 - cos O 0.735
1.37 x 0.126
and so ZL2
= 0.735 -
0.235 MQ.
(If the maximum bandwidth required is too large it will be found that ZL2 cannot
be obtained with ordinary circuit components).
From this, since ZL2 = QLWOLL,
0.235 x 10'
LL 140 x 2a x 0.455 -
0.586 mH
25 330
and CL =
0.4552 x 586 - 208 µµF (see below).
(5') If it is possible to make ZL higher than 0.235 megohm then a voltage step -up
is possible using the arrangement of Fig. 26.17, which also reduces detuning effects
26.6 (iv) DESIGN EXAMPLE 1059
when CN is varied. Good circuit stability (which requires a large value for CL) is
most important since the circuit bandwidth and gain are very critical to detuning.
The main causes of detuning are capacitance changes at the input of V, due to grid
bias variations, and resetting of CN ; this latter capacitance change is offset, to some
extent, in the alternative circuit since it appears across only part of the tuned circuit
capacitance. Any form of tapping down is helpful in reducing detuning variations
due to CN, but this is limited by the impedance required for the filter load circuit.
A.V.C.
Methods of reducing detuning effects are discussed in Sect. 7(ii), and could be
applied here to offset changes in valve input-capacitance. Connecting the grid to a
tap on the tuned load is an obvious means of reducing the effects of valve input -
capacitance variation. This method involves a loss in gain, but this is not necessarily
serious as in some receivers the total gain is more than can usefully be employed.
The actual fixed capacitance for CL is found approximately as follows :
Valve input capacitance = 6.5 µµF
Strays across coil + wiring etc. = 8.5 µµF
CN + Co = 2C0 = 2 x 17 = 34 µµF (when Co is neutralized by
CN)
49 µµF
Total strays =
Fixed capacitance req. = 208 49 -
159 µµF (say 160 µµF).
=
(6') From eqn. (34) ; and since RT = 8550 Q and
RTI
= 0.237 MQ (approximately ; i.e. ZL' + R),
0.237 (7050)
°C08550 (0 235)
0.83 -
(which shows that the gain variation is practicable).
Then, since a = 2 db down = 0.794, we have from eqn. (35A)
Z
0.83 0.794 - 0.036 x 7050
306 12, -
' 0.83 0.794 0.83
7050 0.235 x 106
so that Z, = 4Z', = 1224 S2.
(7') To determine the approximate range of the selectivity control R,,. For maxi-
mum bandwidth R,, = 0. For minimum bandwidth, since ZL, (= RLi) = 7050 -
306 = 6744 Q,
6744 [This is an approximation only ; see end of
then =
QL1 ,LL
wZL1 1680 step (8')]
w °LL 1680
and R', = = = 420 Q.
4
Resistance already in circuit when Q is 140, is
RL = 1680/140 = 12 Q.
Range of R , required is from 0 408 Q. -
This is the range if R, is connected in series with LL. If R, is to be connected in
series with the capacitive arm of ZL it should be realized that it would actually be in
1060 (iv) DESIGN EXAMPLE 20.6
series with the fixed capacitance of 160 µµF only, and not the total capacitance, so
that new values for the range of R, would have to be calculated in this case.
(8') The design of the input transformer is the next step. For maximum gain the
transformer will be critically- coupled (this term is hardly a correct one in the design
which follows and for this particular case it might be preferable to use other methods).
Consider first the secondary circuit. The total secondary capacitance, if we select
100 µµF capacitors (connected in series) for the ratio arms of the bridge circuit, will be
100/2 + 14.5 = 64.5 µµF. The 14.5 µµF represents the approximate total for
Co Co
= (since CN = Co) = 8.5 µµF plus an allowance of 6 µµF
4N
for coil and circuit strays.
Then the apparent secondary inductance required is
25.33
LZ
0.455' x 64.5 -
1.9 mH.
From step (6') above, Z, = 1224 Q. Since critical- coupling (see remarks above)
will halve the actual value of Q2, we must use an uncoupled value for Z, of 2448 Q.
This allows us to determine the uncoupled secondary magnification factor
2448 2448
5440 =
QZ 0.45.
Because of the low value for Q2, the condition of w02L2C, = 1 is no longer suffici-
ently accurate. For cases such as this, where Q is less than about 10, proceed exactly
as before but modify the value of L_ by a factor Q2 /(1 + Q2). This gives the true
condition for resonance (unity power factor) if Q is assumed constant and L2 is variable
as it will be in most i -f transformers of the type being considered. The actual value
required for L2 is now
0.452 1
2 (actual) = 1.9 1 0.32 mH.
+ 0.452J -
(9') Since the secondary Q is very low, and L2 is fixed by other considerations, it
should be clear that if we require reasonably high stage gain the primary Q Q,)
and the primary inductance L, should be as high as possible ; since this will allow
QL = 1/Q1Q2L1L2 to be increased.
As the minimum capacitance across L, will be about 25 µµF (valve output + strays,
which will be fairly high for a large winding), then it is possible to make
25.33
L1 0.4552 x 25 -
4.9 mH.
Of course, it may not be advisable to resonate the primary with stray capacitances
only, but there are practically no detuning effects present (except those due to tem-
perature and humidity variations). The possibility of instability must not be over-
looked, however, when the grid to plate gain is high, and for the case given it would
probably be necessary to neutralize the i -f stage or to reduce the gain by increasing
the capacitance value above 25 µµF (and so reducing L1). However, L1 will be made
large here to illustrate the design procedure to be adopted in cases such as this, where
L, is not equal to L2 and also to bring out an additional useful point which could be
overlooked in connection with the design of a.v.c. systems ; see Chapter 27 Sect. 3(xv).
(10') Assume that unloaded Q values of 140 for the primary and secondary can be
obtained. The plate resistance (r2) for type 6SK7 is 0.8 MQ for a set of typical
operatingconditions. Then the actual value of Q, is
Qur 140 x 0.8
Q1 Qu w0L1 -i- rD 1.96 -{- 0.8
40.6. -
(11') In order that the actual value of Z, shall be 1224 Q (and so Z', = 306 Q) the
secondary circuit must be loaded with a resistance R, given by
Q0Q2w,L2 140 x 2448
R' Qua -- Q2 140
0.45 -
2460 Q
where L2 has the apparent value of 1.9 mH, and not its actual value, so as to make
allowance for the use of approximate expressions which are normally only applicable
28.6 (iv) DESIGN EXAMPLE 1061
operating at 100 Mc /s (see Ref. 68 ; also Chapter 2 Sect. 8 and Chapter 23, Sect. 5).
If all the factors which cause valve input capacitance and conductance changes
are considered the problem is rather involved as can be seen from the discussions
given in Chapter 2, Sect. 8 and Chapter 23, Sect. 5. We will confine attention here,
mainly, to input capacitance changes which are most troublesome in receiver design.
One of the principal causes of change in valve input capacitance is the variation in
position and density of the space charge distribution between grid and cathode,
brought about by the change in grid bias.
Variation in grid -plate transconductance (g,) is another principal cause of input
capacitance change as will be seen later.
Short- circuit input capacitance changes are not very much affected by frequency,
although the change in capacitance with transconductance is slightly greater at low
frequencies than at high. The slight variation with frequency will not usually be
important when compared with the changes due to alteration of transconductance,
so that the data given in Chapter 2, Sect. 8 can be used directly. From this data,
it is seen that the short- circuit input capacitance alters by about 1 to 2 µµF in typical
cases. It is interesting to compare these changes with that due to grid -to-plate
coupling i.e. Miller effect (see Chapter 2, Sect. 8).
Suppose, for simplicity, that the plate load acts as a pure resistance. Then the input
capacitance change is given by
dC= C,- C =C,D ( r gmrDRf )
(43
Taking
-}- R t
type 6SK7, with a plate load of 0.25 megohm, as a typical example. Then
a
since C = 0.003 µµF, r= 0.8 MO (this value is not constant, but it has been
assumed so for simplicity) and gm changes from 0 -2000 µmhos, the input capacitance
change is
2 x 0.8 x 0.25 x 103
AC = 0.003 -
1.14 µµF.
If C augmented by stray capacitances, such
is
socket, wiring etc., this figure
1.05
as those inevitably present due to the
for AC will be very appreciably increased. The
corresponding change in short- circuit input capacitance is 1.18 µµF (see table Chapter
2 Sect. 8).
The effects are additive and cause an increase in input capacitance as the grid bias
voltage becomes more positive.
(ii) Reduction of detuning effects
(A) General
There are a number of methods available for reducing the effects of valve input
capacitance variations, and these methods (in most cases) also reduce input conduct-
ance variations. Usually the reduction of these effects will lead to a loss in stage gain
unless additional circuit changes are made.
In the case of the ordinary broadcast receiver special circuits are not often used.
The only precautions taken are :
(1) The total capacitance tuning the i -f transformer secondary, in particular, is
made as large as is practicable consistent with other circuit requirements. Values
of 200 µµF, or so, should be used if possible, although values of the order of 100 µµF,
are generally used, since it is unlikely that the change in input signal will be such as
to cause the maximum input capacitance change. Large values of tuning capacitance
for the transformer are also helpful in increasing the attenuation at frequencies far
removed from resonance. Also, with some types of converter valves (e.g. 6A8-G)
a large tuning capacitance for the primary of the first i -f transformer can be of assist-
ance in reducing degeneration in the signal input circuit connected to the converter ;
this applies particularly to signal frequencies which approach the intermediate fre-
quency.
28.7 (ii) REDUCTION OF DETUNING EFFECTS 1063
(2) The receiver is aligned on small input signal voltages, since detuning effects
will be less serious when the input signal is large.
(3) The circuit layout is such as to minimize stray grid -to -plate capacitance.
The degree of circuit mistuning in terms of frequency shift can be determined by
comparing the input capacitance change with the total capacitance across the valve
input circuit. A number of helpful practical examples of circuit mistuning are given
in Ref. 69.
There are several additional factors which are of importance.
(1) Detuning will cause sideband asymmetry and so lead to the possibility of in-
creased distortion. This is of particular importance in high fidelity receivers.
When the carrier is received on the side of the i -f selectivity curve, one set of side -
bands is almost completely eliminated (or at least substantially reduced in amplitude),
and the amplitudes of the carrier and low frequency components of the other set of
sideband frequencies are reduced. The tuned circuits will introduce phase and fre-
quency distortion and the diode detector will give rise to non-linear distortion because
of the absence of one set of sidebands. Experiment has shown that with the usual
type of receiver the amount of detuning that can be tolerated before distortion becomes
noticeable to a critical listener is about + 1 Kc /s.
(2) Detuning of the r-f and converter input circuits can lead to a reduction in
signal -to -noise ratio. This effect is not often very serious, when compared with the
deterioration in signal -to -noise ratio caused by the reduction in gain of the r -f stages
as the a.v.c. bias is increased ; further discussion of this point will be given in con-
nection with the design of a.v.c. systems in Chapter 27.
(3) Detuning can result in considerable loss in adjacent channel selectivity. This
is not particularly serious if the receiver has been aligned on small signal input, since
the deterioration occurs when the desired signal is large.
(4) The i -f stages lend themselves more readily to compensation methods than the
r-f circuit* ; in the case of the r -f circuits exact compensation can usually be obtained
at one frequency only. Since the r-f stages are often relatively unselective, detuning
is not often a very serious factor and it is usual to compensate only in the i -f circuits.
(5) Even although exact compensation for input capacitance changes can be ob-
tained at the resonant frequency, the compensation is not complete at frequencies
removed from the i-f centre frequency, and there will always be some departure from
symmetry on the " skirts " of the overall response curve. However, since compensa-
tion is required mainly at the centre frequency, this is not serious.
(6) The neutralization methods to be described in Section 8, in connection with
stability, will affect input capacitance variations which are due to grid-to -plate coup-
ling. If the amplifier is neutralized, then the input capacitance changes with grid bias
will be due, mainly, to space charge effects.
(B) Circuits
Attention will be confined to the circuits of Fig. 26.18 (some discussion of which
has already been given in Chapter 23, Sect. 5), since these allow satisfactory results
to be obtained with a small number of components. Other types of compensating
circuits can be found in Refs. 67, 71, 72.
In Figs. 26.18A and B, Cb is a by -pass capacitor and may be neglected in the dis-
cussion which follows. It is possible to show that, for complete compensation of
the input capacitance variation, the condition required is
Rkgk = QC/C,k (44)
where Rk = unbypassed cathode resistance
gk = grid -to- cathode transconductance ; which is given by gm(Ik /Ib) in
which gm is the mutual conductance ; Ik is the total d.c. cathode cur-
rent (for a pentode this is usually the sum of the plate and screen -grid
currents) ; lb is the d.c. plate current
AC = valve input capacitance change
and C,k = grid -to- cathode capacitance.
for a typical case suppose A = 3 µµF, Cek = 6 µµF and gk = 2000 (11.8/9.2)
= 2560 µmhos, so that
1064 (ii) REDUCTION OF DETUNING EFFECTS 28.7
2000
"'n'n"u v e) 1 +
x 2.56 x 10 -6
195
1330 µmhos. -
If C,k is increased to 20 µµF then Rk = 58.5 S2 and the effective g. = 1740 µmhos.
The actual cathode bias resistor required is approximately 260 Q so that about 200 S2
( =R'5) would be used and this is then by- passed in the usual manner by Ck, as shown
in Fig. 26.18B. The increased value of C,k would need some corresponding re-
duction in C, since the total capacitance tuning the i -f transformer secondary includes
the resultant capacitive reactance due to C,k and Rk in series if the reactances of Ca
and Ck are neglected.
A.V.0
A.V.C.
FIG. 216 METHOD FOR REDUCING VARIATIONS OF INPUT
ADMITTANCE
SECTION 8 : STABILITY
(i) Design data (ii) Neutralizing circuits.
CN
where CN
= cCk
= neutralizing capacitance required
C = a.v.c. by -pass capacitance (usually about 0.01 µF)
(50)
A.V.C. Of
(A) (B)
FIG. 2619 I -F NEUTRALIZATION CIRCUIT
E'r
CDk
B1
L.
B
0.01
A.VC Bf A.V. C.
BY-PASS L
6+
Cj =CI C,y L C9D
C9D Cok
(I) (2)
BOTH CIRCUITS ARE DEGENERATIVE, BUT ARE RE-
GENERATIVE WHEN AA', BB' ARE REVERSED
F10.26-20 METHODS FOR STABILISING R -F AND -F AMPLIFIERS I
This circuit requires no additional components and C, is set to the value required
for neutralization. It will be noted that the capacitance of C. is somewhat larger
than is generally used for a by -pass capacitor. The only disadvantage is that an
additional lug is required on the i -f transformer base and care is necessary to keep
critical leads as short as possible. The advantage of the circuit in a r -f stage should
be apparent.
Receivers which combine the detector diode in the i -f amplifier valve sometimes
require additional neutralization because of feedback (which may be either regenera-
tive or degenerative) caused by the grid -to -diode capacitance. A typical circuit is
shown in Fig. 26.21. This requires two capacitors, CN, and CNa, in addition to the
usual circuit components.
26.8 (ii) NEUTRALIZING CIRCUITS 1067
Analysis of this circuit (Ref. 77) shows that sufficient accuracy can be obtained if
two bridge circuits are used to find the values of CNI and CN2. Although there is
interaction between the two circuits, which affects the values selected for CNI and CN,,
the calculated values are sufficiently close to allow the practical circuit to be satis-
factorily neutralized. The values of CNI and CN, are adjusted in the circuit until
complete neutralization is obtained. Fixed values can then be selected, since the
circuit is not very critical to small changes in the values of the neutralizing capacitors.
P
The equations required are (50) above, which is used to determine CNI, just as pre-
viously, and (52)
=
CN2 C(C,d /Cvt)
where C = a.v.c. by -pass capacitance (as before)
C,d = grid-to -diode capacitance including strays
and C,k = grid to cathode capacitance (as before).
For a typical case :
C = 0.007 µµF (including strays), C,d = 0.009 µµF (including strays),
C = 0.01 µF, C,k = 15 µµF (including strays),
CN, = 0.01 x 10 -0 x (0.007/15) = 3.3 µµF,
CN2 = 0.01 x 10 -6 x (0.009/15) = 6 µµF.
In an actual receiver, using a type 6SF7, the values required were CNI = 4 µµF ;
CN2 = 7 With wiring re-arrangement the values required were CN, = 3 µµF
and CN2 5 µµF
For methods of alignment and a suitable neutralizing procedure see Ref. 77.
SECTION 9 : DISTORTION
(i) Amplitude modulation i-f stages (ii) Frequency modulation i-f stages.
Ha =
4Aa
(1 + A20 L
r3Qaka 4A2 - 1l - (43)
ß Qaka)a
For a critically-coupled transformer Qk = 1 and so
A3
H, = (1 + A2)1(1 - 2A') (54)
AF
where A
fo Q
dF = frequency deviation
fo = central carrier frequency
Q = magnification factor (Q1 = Q,)
ß = modulation index = dF/f
and
f= a-f modulating frequency
k = coefficient of coupling (the coupling may be inductive or capacitive
and either shunt or series).
For 1 db attenuation it is known that, for a critically-coupled transformer with
identical primary and secondary circuits, Q = fo /total bandwidth. From this we
26.9 (ii) FREQUENCY MODULATION I -F STAGES 1069
may write A = 4F /total bandwidth. If it is taken that the total frequency deviation
(24F) is equal to the total bandwidth for 1 db attenuation, then we obtain
A = j- and H, =
32/3
= -ß %. (55)
The Q required for this condition is given from the previously stated relationship
Q = fo /total bandwidth.
Suppose Q = 71 is obtained, for a critically- coupled transformer, with fa = 10.7
Mc /s, 2dF = 150 Kc /s and ß = 5. Then the third harmonic audio distortion intro-
duced by each transformer is 1.4% (approx.) ; using either eqns. (54) or (55).
If Q = 80 is obtained from the procedure of Sect. 4, then eqn. (54) gives H, =
1.57%; eqn. (55) is not applicable in this case.
It should be clear that a simple approximate design procedure can be developed
from the conditions leading to eqn. (55).
The distortion given in the examples is somewhat higher than that experienced
under normal operating conditions. The third harmonic of 15 Kc /s will hardly
trouble the listener, but harmonics of the lower audio frequencies would be serious
if their amplitudes are appreciable, and the possibility of intermodulation effects
should not be neglected.
Applying eqn. (54) for Q = 71, AF = 50 Kc /s (more nearly the usual operating
condition) and a modulation frequency of 5 Kc /s, then H, = 0.074 %. (See also
Chapter 27, Sect. 2 under heading " Non -linear distortion ").
It should be carefully noted that eqn. (55) applies only for the special conditions
under which it was derived. Distortion is normally calculated from eqns. (53) and
(54).
Only a few of the significant factors have been mentioned here, and for more de-
tailed information the reader is referred to Refs. 82, 85, 86 and 87.
SECTION 10 : REFERENCES
(A) I -F
AMPLIFIER DESIGN
(1) Kelly Johnson, J. " Selectivity of superheterodyne receivers using high intermediate frequencies "
A.R.T.S. and P. Bulletin No. 12 (Reprint from Hazeltine Service Corp. Lab. Bulletin) April 1935.
(2) Ross. H. A. " The theory and design of intermediate -frequency transformers for frequency-
modulated signals " A.W.A. Tec. Rev. 6.8 (March 1946) 447.
(3) Everitt, W. L. (book) " Communication Engineering " 2nd ed., 1937, Chap. 16, pages 496 -504.
McGraw -Hill Book Co. Inc., New York and London.
(4) Sturley, K. R. (book) " Radio Receiver Design " Part 1, Chap. 7, Chapman and Hall, London, 1943.
(5) Espy, D. " Double -tuned transformer design " Elect. 17.10 (Oct. 1944) 142.
(6) Maynard, J. E. " Universal performance curves for tuned transformers " Elect. 10.2 (Feb. 1937) 15.
(7) Maynard, J. E. and P. C. Gardiner, " Aids in the design of intermediate- frequency systems "
Proc. I.R.E. 32.11 (Nov. 1944) 674.
(8) Aiken, C. B. " Two-mesh tuned coupled circuit filters " Proc. I.R.E. 25.2 (Feb. 1937) 230.
(9) Adams, J. J. " Undercoupling in tuned coupled circuits to realize optimum gain and selectivity "
Proc. I.R.E. 29.5 (May 1941) 277.
(10) Spaulding, F. E. " Design of superheterodyne intermediate Frequency Amplifiers " R.C.A. Rev.
6.4 (April 1940) 485.
(11a) Stern, E. " Measurements on intermediate -frequency transformers " Proc. I.R.E. (Aust.) 9.1
(Jan. 1948) 4 ; Reprinted Jour. Brit. I.R.E. 9.4 (April 1949) 157.
(lib) Rudd, J. B. " A design for double -tuned ; transformers " Proc. I.R.E. (Aust.) 10.1 (Jan. 1949)
3 ; A.W.A. Tec. Rev. 82 (April 1949) 147 Reprinted Jour. Brit. I.R.E. 9.8 (Aug. 1949) 306.
(1Ic) See also bibliography Chapter 9.
General I -F Amplifier Data
(12) Benin, Z. " Modern home receiver design " Elect. 19.8 (Aug. 1946) 94. -Hill Book Co. Inc.,
(13) Terman, (book) aRa4 o Engineers' Handbook " 1st edit. McGraw
New and London,'
(14) Kees, H. " Receiver with 2 Mc i -f " Elect. 18.4 (April 1945) 129.
(15) Adams, J. J. " Intermediate-frequency amplifiers for frequency -modulation receivers " Proc.
I.R.E. 35.9 (Sept. 1947) 960.
(16) Rust, Ramsay and Sturley " Broadcast receivers : A review " Jour. I.E.E. 88.2 Part 3
94)
(17) Zepler, E. E. (book) " The Technique of Radio Design " Chapman and Hall, London, 1943.
Wiley and Sons, New York, 1943.
(17a)BRidgI.R.E. E. S. " Practical aspects the design of intermediate frequency transformers " Jour.
S (March 1950)
1070 REFERENCES 26.10
Additional data for F -M i -f Amplifiers
(18) See (2), (12) and (15) above.
(19) Hund, A. (book) " Frequency Modulation " (1st edit.) McGraw -Hill Book Co. Inc., New York
and London, 1942.
(20) Foster, D. E. and J. A. Rankin " Intermediate frequency values for frequency- modulated -wave
receivers " Proc. I.R.E. 29.10 (Oct. 1941) 546.
(21) Landon, V. D. " Impulse noise in F -M reception " Elect. 14.2 (Feb. 1941) 26.
(22) Beard, E. G. " Intermediate frequency transformers for A -M /F-M receivers " Philips Tec. Corn.
No. 6 (Nov. 1946) 12.
(23) Sturley, K. R. (book) " Radio Receiver Design " Part 2 Chapman and Hall, London, 1945.
(24) Tibbs, C. E. (book) " Frequency Modulation Engineering " Chapman and Hall, London, 1947.
(25) Crosby, M. G. " A method of measuring frequency deviation " R.C.A. Rev. 6.4 (April 1940) 473.
I -F Transformer Construction
(29) " The design of the universal winding " A.R.T.S. and P. Bulletin No. 131 (Reprint Hazeltine
Corp. Bulletin) 13th Sept. 1943.
(30) Simon, A. W. " Universal coil design " Radio 31.2 (Feb. -Mch. 1947) 16.
(31) Simon, A. W. " On the theory of the progressive universal winding " Proc. I.R.E. 33.12 (Dec.
1945) 868.
(32) Kantor, M. " Theory and design of progressive and ordinary universal windings " Proc. I.R.E.
35.12 (Dec. 1947) 156.
(33) Scheer, F. H. " Notes on intermediate frequency transformer design " Proc. I.R.E. 23.12 (Dec.
1935) 1483.
(34) Callender, M. V. " Q of solenoid coils " (letter) W.E. 24.285 (June 1947) 185.
(35) Amos, S. W. " Calculating coupling co- efficients -useful formulae for finding the optimum spacing
of I -F transformer windings' W.W. 49.9 (Sept. 1943) 272.
(36a) Vergara, W. C. " Determining form factors of I -F transformers " Elect. 22.7 (July 1949) 168.
(36b) See also Refs. 88 to 93 below ; Chapters 10 and 11 and their bibliographies.
(62) Booth, C. F. " The application and use of quartz crystals in telecommunications " Jour. I.B.B.
Part 3, 88.2 (June 1941) 97.
(63) Grisdale, G. L., and R. B. Armstrong " Tendencies in the design of the communication type of
receiver " Jour. I.E.E. Part 3, 93.25 (Sept. 1946) 365.
Measurements on Crystals
(64) Builder, G. " A note on the determination of the equivalent electrical constants of a quartz- crystal
resonator" A.W.A. Tec. Rev. 5.1. (April 1940) 41.
(65) Mason, W. P., and I. E. Fair " A new direct crystal -controlled oscillator for ultra- short-wave
frequencies " Proc. I.R.E. 30.10 (Oct. 1942) 464.
(D) DETUNING EFFECTS AND COMPENSATION
(66) See above for Ref. 17, p. 171 : Ref. 4, p. 37 ; Chapter 2 Sect. 8 ; Chapter 23, Sect. 5.
(67) Ref. 13 above; page 472.
(68) " Input Admittance of Receiving Tubes " R.C.A. Application Note No. 118 (April 25th, 1947) ;
Reprinted Radiotronics No. 126 (July -Aug., 1947) 62.
(69) " Compensation for amplifier tube input capacitance variation " A.R.T.S. and P. Bulletin No. 67
(Dec. 22nd 1938) 5 (Reprint from Hazeltine Service Corp. Lab. Bulletin).
(70) Harvey A. F. (book) High Frequency Thermionic Tubes " (Chapter 2 and Bibliography).
John Miley, and Sons, New York, 1943 ; Chapman and Hall, London, 1943.
(71) ' Television topics -valve input capacity and resistance" W.W. 43.15 (Oct. 13th, 1938) 340.
(72) Freeman :R. L. " Input conductance neutralization " Elect. 12.10; (Oct. 1939) 22.
(E) STABILITY
(73) Chapter 23, Sect. 7 and Ref. B41 for that chapter.
(74) Jaffe, D. L. " Intermediate frequency amplifier stability factors " Radio 30.4 (April 1946) 26.
(75) " Effect of transconductance and grid -plate capacitance on the stability of i -f amplifiers " A.R.T.S.
and P. Bulletin No. 93 (July 8th, 1940) (Reprint of Hazeltine Service Corp. Bulletin).
(76) " I -F neutralization " Radiotronics No. 115 (July 1941) 54.
(77) Neutralization in circuits valve as a ombined i -f amplifier and diode detector "
" òroli 1 Ms employing a
(78) Ref. 17 above (Chapter 9). Ref. 13 above (p. 469).
(79) Root, C. S. " Method for stabilising R -F and I -F amplifiers " G. E. Docket 63495.
(80) " A neutralisation circuit for I.F. amplifiers " No. 122 (18th May, 1942) A.R.T.S. and P. Bulletin.
(81) Hultberg, C. A. " Neutralization of screen grid tubes to improve the stability of intermediate -
frequency amplifiers " Proc. I.R.E. 31.12 (Dec. 1943) 663.
(F) DISTORTION
Ref. 4 above (pages 154 and 335) Refs. B34, B35, B36, bibliography Chapter 23.
(82) Guttinger, P. The distortion of frequency modulated signals during transmission by band -pass
filters' The Brown Boveri Review 23.8 (Aug. 1946) 185.
(83) " Screen supply for the final I.F. amplifier " A.R.T.S. and P. Bulletin No. 88 (21st March 1940).
(84) " Screen dropping resistors-Restrictions with super -control valves " Radiotronics No. 96 (15th
March 1939).
(85) Jaffe, D. L. " A theoretical and experimental investigation of tuned- circuit distortion in F.M.
systems " Proc. I.R.E. 33.5 (May 1945) 318.
(86) Roder, H. " Effects of tuned circuits upon a frequency modulated signal " Proc. I.R.E. 25.12
(Dec. 1937) 1617.
(87) Cherry, E. C., and R. S. Rivlin, " Non -linear distortion with particular reference to the theory
of F.M. waves " Phil. Mag. (i) 32.213 (Oct. 1941) 265 ; (ii) 32.219 (April 1942) 272. Pt (i) covers
non-linear distortion due to valves ; Pt. (ii) tuned circuits.
0100E DETECTOR
Distortion V Modulation
0 10 20 30 40 50 60 60
FIG. 27.1 Modulation Pe «nage
(1) The a.v.c. system.
(2) The following grid resistor.
(3) An electron ray tuning indicator.
The circuit of a typical diode detector is shown in Fig. 27.2 in which the diode load
resistance is R2 together with R, ; the latter, in conjunction with C, and C,, form a
r-f filter so that the r -f voltage passed on to the a -f system may be a minimum. R,
is generally made about 10% of R, and typical values are 50 000 ohms and 0.5 megohm.
The capacitances of C, and C, depend upon the frequency of the carrier ; for an inter-
mediate frequency of 455 Kc /s they may both be 100 µµF.
If the volume control (R,) is turned to maximum the shunting effect due to R3
will be appreciable, since R3 cannot exceed 1 or 2 megohms with most types of valves.
If, however, grid leak bias is used on a high-mu triode valve, R, may be approximately
10 megohms and the input resistance of the valve will then be of the order of
5 megohms. This is sufficiently high to be unimportant but for lower values of R3
the distortion with the control set near maximum may be severe. It is found in most
conventional receivers that the a -f gain is considerably higher than that required for
strong carrier voltages and under these conditions the control will be turned to a low
setting. The a.c. shunting effect due to 122 is practically negligible provided the control
is below one -fifth of the maximum position. A good method of overcoming the
a.c. shunting due to the audio amplifier is shown in Fig. 27.3. The diode may be
in the same envelope as the i -f amplifier valve. Negative feedback is applied across
part of the cathode bias resistor for the a -f
voltage amplifier valve. For the component
values shown, the measured input resistance
between points A and B is in excess of 10
megohms for frequencies up to about 10 Kc /s.
A possible disadvantage is the reduction in
overall gain because of the negative feedback,
although this is not generally serious. In-
creasing the feedback resistance R increases
TO A.F.
the overall gain, but reduces the input re-
GRID sistance. Intermediate values of gain and
R2 Rj input resistance may be selected as desired.
For further details of receivers using this
arrangement see Refs. 6 and 7.
FIG.27.2 DIODE DETECTOR CIRCUIT
1074 (i) DIODES 27.1
LOAD 50001.1.
Bt B+
260V
FIG.27.3 AUDIO AMPLIFIER GIVING LOW A.C. SHUNTING ACROSS DIODE LOAD.
Another useful method of reducing a.c. shunting effects is to use a cathode follower
stage between the detector and the a -f voltage amplifier. This is often conveniently
accomplished by employing a double triode valve, one section serving as the cathode
follower and the other as an a -f voltage amplifier.
Distortion due to the a.v.c. system will be discussed in detail in Sect. 3 below. One
type of distortion often encountered is caused by delayed a.v.c. systems at the point
where the a.v.c. diode just starts to conduct. This form of distortion is called differ-
ential distortion and may be kept to low values by making the delay voltage small.
The conventional arrangement for obtaining a.v.c. voltage from the primary of the
last i -f transformer is generally preferred, as it reduces the a.c. shunting effect across
the detector diode circuit.
The a.c. shunting due to the addition of an electron ray tuning indicator to the diode
detector circuit is serious and difficult to avoid. In order to reduce the distortion to a
minimum the resistor feeding the grid of the tuning indicator may be made 2 megohms
and the effect will only then be apparent at high percentages of modulation. If the
tuning indicator is connected to the a.v.c. system it will not operate at low carrier
levels unless the delay voltage is extremely small. One possible method, where the
utmost fidelity is required, is to use the same circuit as for delayed a.v.c. but with a
delay voltage of zero, and to connect the tuning indicator to this a.v.c. circuit. With
this arrangement a.c shunting due to the a.v.c. circuit and the tuning indicator is
MQ eliminated, while differential
AVC loading no longer occurs.
IMQ An interesting arrangement
R4 (Ref. 8) for counteracting the
effect of a.c. loading, and so in-
creasing the maximum percent-
age of modulation which can be
handled without excessive dis-
tortion is accomplished by the
ozpr use of the circuit of Fig. 27.4.
In this arrangement a positive
bias is applied to the diode plate
IMn in such a way as to be propor-
R3
tional to the carrier input. A
fixed positive bias would not be
ìoopuT iJF satisfactory since it would only
FIG.27.4
- - y give low distortion at one carrier
COMPENSATED DIODE DETECTOR. level.
27.1 (i) (B) DIODE CURVES 1075
100
0
-30 -25 -20 -15 -10 -5 C o +5
PLATE VOLTS.
rIC.27 5 TYPICAL DIODE CHARACTERISTICS.
r
1
/
EfRATEO VALUE
40
3Ó
EWA S
a
-20 Mii
SOWN NNW'
r°
' . 20
0.1 S
.I_
/-
10
o
/!
, 2
5
0 20 40 60 80 100 120 140 160
FIG. 27.6 RECTIFIED MICROAMPERES
27.1 (i) (C) QUANTITATIVE DESIGN DATA 1077
bias voltage to the diode plate. This arrangement is only satisfactory if the bias can
be changed for each i -f input voltage (Er,,,) see Ref. 1 (page 348).
For this case
m
Ent. Rd, (5)
\ - Rao + R3 /F
The circuit of Fig. 27.4 shows a practical arrangement for increasing the critical
modulation ratio by applying positive bias to the diode plate, as mentioned previously
in Sect. (A) above.
(c) Equivalent damping across i -f transformer
The secondary circuit damping is found approximately from
RE = R,,(1/1
-
n2 -
n cos -1n)
n( COS -ln- -
n2)
RE /R,, can be read directly from Fig. 27.7 for various values of n.
Suppose Rao = 0.36 MQ and n = 0.793 then, since RE /Rao = 0.7, the equivalent
damping resistance is RE = 0.7 x 0.36 = 0.25 M.Q.
It is often taken in practice (as a " rule of thumb ") that the equivalent damping
resistance is half the d.c. diode load resistance. This gives an equivalent damping
resistance of Ra, /2 = 0.5/2 = 0.25 MD in this case. However, it should be realized
that if the diode efficiency is taken as unity, the equivalent damping resistance which
appears across the tuned circuit is half the a.c. load resistance ; this is indicated by
Fig. 27.7.
f I a
a
w
u
z
6..
ú1
0 2 04 0 6 0.8
F1G.27.7 DETECTION EFFICIENCY (ryj)
has a plate resistance of 0.8 megohm the primary damping resistance would be 0.23
megohm.
This means that, with detector and a.v.c. diodes conducting, both the secondary
and primary circuits of the i -f transformer are heavily damped and this damping must
be taken into account when the i -f transformer, connected between the last i-f voltage
amplifier valve and the diode detector, is being designed. Delayed a.v.c. will, of
course, affect the primary circuit damping and three conditions arise the diode not
:
conducting, the diode just starting to conduct, and the diode conducting when the
applied voltage is reasonably large. When conduction just starts the damping de-
pends very largely on the diode detection efficiency and will vary quite appreciably
for a small range of input voltages. It is usually sufficient, however, to know the
tuned circuit damping for the two conditions, diode not conducting, and diode con-
ducting with a reasonably large input voltage. For the diode not conducting the added
damping is approximately R., ; Rd, should not be used here. With the diode con-
ducting the added damping can be taken as Rd, /3 with sufficient accuracy in some
cases since in this case the fact that the diode detection efficiency is not unity offsets
to some extent the increase in damping which is obtained when R., is used in place
of Rd,.
Damping of the transformer tuned circuits as determined above assumes linear
diode plate -voltage -plate- current characteristics. For parabolic diode detection
the damping differs by only a small amount from that given by the linear characteristics
and so, in practice, it is seldom necessary to treat the two cases separately. In any
case the order of accuracy to be expected in the final results from the procedures set
out, would hardly justify any additional refinements in the design calculations.
(d) The actual degree of modulation
The degree of modulation applied to the detector is always less than the modulation
of the incoming carrier. The actual value of m which the detector will be required to
handle is given approximately by
R3(R a, + 2ZL)
m = m' (7)
Rd,R, + 2ZL(Ra, + R3)
where m' = original modulation percentage
Rd, = R, + R, (see Fig. 27.2)
R3 = grid resistor of following stage (see Fig. 27.2)
and ZL = load impedance presented to the i -f voltage amplifier preceding the
detector stage ; including the effects of all damping on the primary
and secondary of the i -f transformer.
In addition, at higher audio frequencies the reduction of side band amplitude by
the selectivity of the earlier stages in the receiver results in a further decrease in depth
of modulation.
(e) Audio frequency output voltage and response (frequency distortion)
The approximate r.m.s. audio frequency output voltage can be determined for a
particular arrangement by using the equivalent circuit of Fig. 27.8. It is often as-
sumed that all capacitive reactances are negligible, for simplicity in carrying out gain
calculations. However, the circuit is readily modified to include C C, and C, as
A
E OUT
shown by the dotted lines, and any additional capacitances in the circuit can be in-
cluded if necessary. For this diagram
m = modulation factor
i -f input voltage
n = detection efficiency determined from curves of Fig. 27.6
Z = total load impedance connected between the terminals AB. (This is
equal to R,,, over the middle audio frequency range)
123123
R,,,= R1--R2 +R,
and R'a = internal equivalent resistance of diode at audio frequencies. This is
given by the inverse slope of the curves of Fig. 27.6.
R1, R2 and R, are as shown in Fig. 27.2.
It should be noted that R' 4 is not the diode conduction resistance (Rd) but is related
to it by
nR a
R'a = cos -1n (8)
Also R'a and Rd are not constant for all operating conditions. For our previous
example we see from the 15 volt (r.m.s.) curve of Fig. 27.6, and taking the diode load
as being
R, + 123 = 0.5 MQ, for simplicity,
(17 - 16.5) x 106
(42 -
26)
31 200 Q.
It will be seen that R'a is determined in a rather similar manner to that used for
finding the plate resistance of multi- element valves. The diode conduction resistance
in this case is (approximately) 6500 ohms.
It is sometimes required to find the audio frequency response and this can readily
be determined from the complete circuit of Fig. 27.8. A suitable procedure would
be as for r.c. amplifiers in which the capacitances are ignored over the middle fre-
quency range (say around 400 to 3000 c /s), and the audio voltage output is calculated
for the resistance network only. At low frequencies the effects of C, and C, can
be ignored since they are usually about 50 to 100 µµF each, and only the effect of
C, would be considered At high frequencies C, would be neglected and the effects
of C, and C2 considered. Often only the maximum output voltage is required plus
the total output variation from say 50 c/s to 10 Kc /s, and this simplifies the calcu-
lations. If Fig. 27.8 is inspected it will be seen that it is a simple matter to apply
the generalized low frequency response curves given in Chapter 12. The high fre-
quency curves can also be applied directly if R, can be neglected (this is usually per-
missible) and C, and C2 are considered as being in parallel.
When using the circuit of Fig. 27.8 it is more convenient to determine the voltage
EAR since this is given directly by
EAR = +gym Ernie.
It follows that, over the middle frequency range,
Eoue = R,,, Rl/
and
Eoue
E, =EAa
Eou,` EAR R,,, R, -
E' R3, + Rd
If R, can be neglected in comparison with R2 and R, in parallel, then E,,,, = EAR
over the middle frequency range.
It is more accurate to determine-EAR from the curves of Fig. 27.6, since rl is not
constant for a modulated carrier voltage and the effects of contact potential are in-
cluded in the measured data. The procedure is to draw the a.c. loadline on the graph
and note the d.c. voltages E4,1 and Eao2 given by the maximum and minimum values
of E,me during modulation. From this
E 1 (Eae1 - Eae2).
AB 2 -/ 2
1080 (i) (C) QUANTITATIVE DESIGN DATA 27.1
The simplified graphical procedure is only suitable when the amplitudes of harmonics
higher than the second are negligible [see also (h) below]. More accurate graphical
methods are required if this is not true.
To compare the results obtained by the two methods, assume Erm, = 15 volts
and the percentage modulation is 50% (i.e. m = 0.5). As determined previously
7/ = 0.793, and it will be taken that Rae = Rd, = 0.5 megohm. Then
EAB = rim Erm, = 0.793 x 0.5 x 15 = 5.95 volts (r.m.s.).
From Fig. 27.6, Edc1 = 26.5 volts and Ed,a = 7.5 volts so that
EAB =
1
(26.5 - 7.5) = 6.7 volts.
The discrepancy between these two results is easily explained, as an examination of
typical diode circuits giving the d.c. voltage developed by the diode in the absence
of signal input shows that there is about -0.5 to -1 volt developed across a diode
load of 0.5 megohm. This difference is also illustrated by the curve of Fig. 27.5,
which shows that -1 volt is developed for the conditions being considered. It is
also seen that the variation in detection efficiency (n) over the range of voltage used
does not have a large effect on the final result. The presence of d.c. voltage across
the load resistor, in the absence of signal input voltage, is due to contact potential
in the diode.
(f) Effect of shunt capacitance on detection efficiency
Too low a value for C1 and C, will affect the detection efficiency. Provided the
12.72
total shunt capacitance across the diode load is not less than the ef-
.fR as
fect can be neglected where
f
= frequency (expressed in Mc/s)
R2Ra
R,, = R1 + (expressed in MQ).
Ra +Ra
For our previous example of Rai = 0.36 Ml?, and i -f of 455 Kc /s, the capacitance
due to C1 and Ca should not be less than about 78 µµF ; this includes all stray capacit-
ances. In the circuit of Fig. 27.2 typical values of C1 and C, are 100 µµF each and
if R1 is neglected it is seen that the total capacitance is more than twice the value
required even when strays are neglected. It would be feasible to reduce C1 and C,
to 50 µµF respectively with a reduction in attenuation at the higher audio frequencies.
(g)Non -linear distortion
If the value of total capacitance across the diode load resistance is too large, the
discharge time constant will be too long and the voltage across the diode load
will not follow the modulation envelope. This will give rise to non -linear distortion
and suggests that the rate of discharge of the load circuit should not be less than the
maximum rate of change of the modulation envelope. It can be shown that non-
wC,Rdc m.i
linear a -f distortion due to this cause can be almost completely avoided provided that
1
(9)
Suppose we neglect the decoupling resistance (R,) in Fig. 27.2, as this unnecessarily
complicates calculations when its value is about 1/10 of R2 (the usual case).
What is required, is to determine a value for C,Rd, which will allow m = 1, i.e.
the incoming signal to be 100% modulated. However, it should be clear from pre-
vious discussion that the modulation percentage is generally a good deal lower than this
figure, and it seldom happens that very high modulation percentages occur at high
audio frequencies.
Take Rd, as 0.5 megohm, then 1 /wC, = mRd, = 1 x 0.5 x 106.
Assume that the highest audio frequency is 10 Kc /s, then w = 27r x 10* and so
1032
C,
-2n x 102 x 0.5 x 102 -31.8µµF.
This is the largest value C, should have if the detector is to be capable of handling
audio frequencies of 10 Kc /s and 100% amplitude modulation.
It will be noticed from (e) above that the detector efficiency will be reduced if C, is
27.1 (i) (C) QUANTITATIVE DESIGN DATA 1081
earthing, or even leaving disconnected, the second diode will often effect an almost
complete cure. Obviously the diode connection nearest the plate of the voltage am-
plifier is the one which should be earthed.
A second effect which is less frequently encountered is residual volume due to
capacitance between a diode and the control grid of a combined detector and a -f
amplifier. Normally this capacitance is unimportant because with the volume control
turned right down the control grid is grounded through the grid coupling capacitor
(see Fig. 27.2). However if the audio amplifier is grid leak biased it may have a 10
megohm grid leak, and in this case a 0.001 pF grid coupling capacitor would give
adequate bass response for a small receiver. Under these conditions a voltage of the
order of one thousandth of the i -f input to the diode could appear on the control grid
and with a high gain receiver this is ample to give annoying minimum volume. The
remedy, of course, is to increase the size of the grid coupling capacitor.
(b) In general, diodes do not start to conduct at precisely the point where the 1páte
voltage exceeds zero. Contact potentials and other effects will sometimes allow
the valve to conduct when the diode plate vo taé is slightly ne rive (this is usual
with valves having indirectly heated cathodes), but in other cases e.g. some battery
valves) conduction does not occur until the plate voltage is appreciably positive. This
point can be appreciated by an examination of typical.dio curves. The effects of
the contact potentials will be further considered i Sect. (3) below when automatic
volume control is being discussed.
For a discussion of the effects of positive and negative start of plate current (which
may be deliberately introduced as mentioned in (C)(b) above) in diode detectors the
reader is referred to Ref. 1, Chapter 8.
(c) There are distinct advantages in having the a -f volume control as the diode
load. For the usual operating conditions on local signals the setting of the control
is fairly well down, and so the effects of a.c. loading are very much reduced. This
advantage is lost when the control is in the grid circuit of the a -f voltage amplifier.
The disadvantage is that many controls become noisy, usually after a fairly short
period, when they have the diode current passing through them. A compromise
arrangement may be best with the control in the grid circuit (R3 in Fig. 27.2) and
R, about 0.1 megohm and R2 say 0.5 megohm.
(d) As a summary of the characteristics of the diode detector, it may be stated that
its performance as regards frequency and non -linear distortion is excellent provided
the input voltage is high and the factors discussed above regarding a.c. shunting etc.
are incorporated in the detector design. All forms of detectors suffer from distortion
at low input levels, but the diode has the particular advantage that the input may be
increased to a very high level with consequent reduction of distortion, without any
overloading effect such as occurs with other forms of detectors.
(e) Crystal diodes are described in Section 7.
shown (as is usual with the diode) or at the point X is immaterial from the viewpoint
of operation. The diode is directly coupled to the triode and therefore the audio fre-
quency voltages developed in the diode detector are passed on to the triode grid, but at
the same time this grid is given a d.c. bias through the d.c. voltage developed in a
similar way to that by which a.v.c. is obtained. Consequently the operating point of
the triode varies along the eo in curve from zero towards more negative grid bias
voltages as the carrier voltage is increased. This is the same effect as that obtained
when the diode is omitted (Fig. 27.10) since the grid and cathode of the triode act as a
diode and produce the same results. The illustration given was purely to demonstrate
the derivation of the one from the other and not to be a practical form of detector
since no advantage is gained by retaining the diode in the circuit.
R
IIE
-
FIG.27.10 GRID LEAK DETECTOR WITH
TRANSFORMER COUPLING.
It will be seen that the operating point varies along the e-i
characteristic curve
I etween zero biasand the cut -off point (Fig. 27.11). There will be a certain strength
of carrier at which the detection will be most satisfactory, and at lower or higher
levels detection will not be so satisfactory on account of improper operating conditions.
If with a certain carrier Input voltage the d.c. bias on the grid is OA, then the point
corresponding to peak modulation is B where OB equals twice OA. If the point
B is on the curved part of the characteristic, or in the extreme case actually beyond
the cut -off, the distortion will be severe. A valve having low IL and low g,,, is capable
of operating with a higher carrier voltage than a valve with improved characteristics,
but the gain in the detector stage will be less. There is a further difficulty in that
the plate current at no signal, or at very weak signal, may be excessively high. If
transformer coupling is used this may, in extreme cases, damage the valve, cr pass too
much direct -current through the transformer, unless the plate supply voltage is re-
duced. If resistance coupling or parallel -feed is used the efficiency of the detector
is decreased. As with diode detection there is distortion at low levels due to the
" diode characteristics " but as distinct
from the diode, the overload point
occurs at quite a low carrier voltage.
This method of detection is therefore
P
very much limited in application.
With battery type valves used as
cumulative grid detectors it is often
advantageous to connect R1 (Fig. 27.10)
to filament positive ; C1 remains as
before. The advantage obtained is
that larger modulation percentages can
be handled and detection will start
with smaller input voltages. The
arrangement is equivalent to supplying
the grid, which is acting as the plate
t -E9
FIG.27.11
B A
A typical circuit for a grid leak detector as used in radio receivers is shown in
Fig. 27.12.
Damping of the input circuit occurs in the same manner as for a diode but additional
damping occurs because of grid -plate coupling. The arrangement shown in Fig.
27.12 for the grid resistor further increases the loading on the input circuit. Detailed
discussions of this type of circuit can be found in Refs. 2 (p. 414), 1 (p. 377) and 10.
B +
FIC.27.13 PLATE DETECTOR.
used as plate detectors
Screen grid and pentode valves with self-bias have been with the usual arrange-
very satisfactorily for a number of years, although the distortion
radio receivers at the
ment is too high for them to be used in any but the cheapest some types of short-wave
present time. Such a detector is, however, permissible for input impedance
reception and for amateur communication work where its high grid
results in higher sensitivity and selectivity.
(C2, Fig. 27.13) has
With the usual plate detector the cathode bypass capacitor
customarily been made sufficiently large to by -pass both audio mica and radio frequencies.
capacitor. This
Often C, is a 25 µF electrolytic in parallel with a 500 µµF one and leads to dis-
arrangement, although widely used in the past, is not a correct
similar to that de-
tortion when the modulation factor is at all high. The effect isappreciably different
scribed in connection with diode detectors having the a.c. loadline
for the resistor R$ to be
in slope from the d.c. loadline. The correct procedure calls in a very considerable
by- passed for radio frequencies only, and although this results
are excellent. In a
loss in gain because of negative feedback, the results obtainable
000 to 100 000 ohms and C2 as 500 to 100 µµF
typical case R2 might be taken as 10
; additional r-f
respectively, the plate load resistor would be about 0.25 megohmFig. 27.13. This
in
decoupling is usual in the plate circuit, although this is not shown
is sometimes reserved for
circuit is often called a reflex detector, although this term and the audio out-
cases where 100% negative feedback is used with a plate detector,
cathode circuit the plate is earthed for
put is then taken from the resistor in the ;
Best results are generally obtained using pentode valves and resistance-capacitance
coupling to the following audio stage. This arrangement minimizes " threshold
howl."
Feedback control is possible with a variety of circuits but least difficulty is usually
experienced with the variable capacitor arrangement, or the use of a variable resistor
in the screen circuit. The latter arrangement is advantageous as regards the re-
duction of damning effects.
Further details of this type of detector can be found in Refs. 1, 5, 16, 17, 18, 19
and 20.
(F) Superregenerative detectors
The superregenerative detector is a regenerative circuit in which the detector is
automatically switched in and out of oscillation at a very low radio
frequency rate
about 15 to 100 Kc /s). This switching frequency is called the " quench-
(usually
as the carrier
ing " frequency. In general the quenching frequency is increased using the
frequency becomes greater, but sensitivity and selectivity are improved byis usual to
lowest permissible value of quenching frequency. For
many cases it
contained in the
make the quench frequency about twice the highest audio frequency frequency
modulation envelope, and it is not considered good practice for the quench
is also important
to be lower than this value. The amplitude of the quench voltage In general
and it will have a very appreciable effect on the selectivity characteristic. circuit, the
if the quench voltage amplitude is increased in a separately quenched
selectivity is reduced.
which is possible
The advantage of this circuit is the extremely high sensitivity level in the absence of a
using a single valve. The disadvantages are the high noise
signal, the poor selectivity, and the high distortion. detector
The circuit of Fig. 27.14 can be made to operate as a superregenerative
capacitance combination,
by increasing the time constant of the grid resistance and frequency can
and making the amount of regeneration very large. The quenching
be simply adjusted by altering the value of the grid resistance.
detector arrange-
A separate quenching oscillator is often used with the regenerative an additional
ment, but as this requires additional circuit components and sometimes mentioned. The
valve, it is not so popular as the simple arrangement previously
largely overcomes
development of suitable types of double triode and converte! valves quenching.
this objection, however, and many modern circuits use separate
limiting occurs,
Several other points of interest arise with this detector. Amplitude
noises than when other
so there is less interference from car ignition and similar
greater than for weak
detectors are used ; the output on strong signals is not much by using low quench
signals. Noise quieting and limiting are improved however,
using this type of de-
frequencies. A r-f stage should be incorporated in receivers materially alter the signal-
tector to reduce radiation. This additional stage will not incorporating the prin-
to-noise ratio obtainable with the detector alone. Circuits ; this point will
ciple of superregeneration have been used in cheap F -M receivers
in the section on F -M receivers in Chapter 36.
receive some further consideration superregenerative
The method to be adopted for measuring the selectivity of a
are usually inadequate
receiver requires some consideration. Conventional methods
in most cases the problem is similar to that of taking the selectivity curve of
since
a receiver having a.v.c., but in this case the a.v.c.
cannot be disconnected. A suitable
procedure (Ref. 26) is as follows. A signal is ap-
With no input signal applied the audio noise output is measured. until the noise is sup-
plied at the resonance frequency and its amplitude adjusted The selectivity is then
pressed by about 10 to 20 db (or any convenient amount).
around resonance, in the
found by tuning the signal generator to various frequenciessame degree of quieting is
usual manner, and the input voltage is adjusted until the
signal inputs off and at
obtained as at resonance. The difference between the two at the particular fre-
resonance for the same degree of quieting, gives the attenuation
quency being considered. 662), 20 (page
Further details of this detection system are available in Refs. 5 (page
148), 21, 22, 23, 24, 25 and 26.
1088 (i) TYPES OF DETECTORS IN GENERAL USE 27.2
SECTION 2 : F -M DETECTORS
(i) Types of detectors in general use (ii) General principles (iii) Phase discrimina-
tors (A) General (B) Design data (C) Design example (iv) Ratio detectors (A) General
(B) Operation (C) Types of circuit (D) Design considerations (E) Practical circuits
(F) Measurements on ratio detectors.
B
FIG.27.15 BASIC PHASE DISCRIMINATOR CIRCUIT.
assumed constant for the moment) alters, the phase angle between E, and E, changes
from that at resonance. This leads to a change in the relative magnitudes of EA
and EB.
The vector relationship between the primary and half secondary voltages, for the
phase discriminator, are shown in Fig. 27.16. This assumes constant primary voltage
E,.
If the primary voltage E, has its
amplitude varied in a suitable man-
ner (e.g. by setting the coupling
between the transformer windings,
for given primary and secondary
Q's, so that two primary voltage
humps of the required amplitude
appear as the frequency is varied)
the linearity and sensitivity of the
discriminator can be very appreci-
ably improved.
This statement, regarding line-
arity and sensitivity, refers, of
course, to the relationship between
voltage output and frequency- devia-
tion from the central reference fre-
quency (i.e. the nominal intermedi-
ate frequency).
Analysis of the phase discrimina-
tor for F-M applications has been
B
FIG.27.16 VOLTAGE RELATIONSHIPS IN PHASE
made by K. R. Sturley (Refs. 27
DISCRIMINATOR WITH CONSTANT and 28), and the results of this
PRIMARY VOLTAGE. analysis will be used below It
might be mentioned that a number
of discriminator circuits for a wide variety of applications have been designed using the
data derived by Sturley and very satisfactory results have been obtained in practice.
Before proceeding to set out the design data, the circuit of Fig. 27.17 will be briefly
discussed.
The voltages applied to the plates of the double diode valve V, will be E' A and
E'B respectively. The output voltages from the diodes are developed across R,
and R and the circuit is arranged so that the available output voltage is equal to the
difference between the two separate voltages. This means that when the frequency
of the carrier voltage is exactly equal to the intermediate frequency, no output will be
obtained from the detector. Reference to Fig. 27.16 should help to make this point
quite clear, since the rectified voltages across R, and R4 in Fig. 27.17 are given by the
peak voltages applied to the diode plates multiplied by the detection efficiency of the
diodes. As the signal deviates from the central reference frequency (i.e. the inter-
mediate frequency), a voltage (E0,) will appear between the points P and N (Fig.
27.17), and its polarity will depend on the relative magnitudes of the voltages across
R, and R4 ; e.g. point P will be negative with respect to point N when the voltage
across R, is less than that across R4. It follows that it should be possible to calculate
100 (1 (iii) PHASE DISCRIMINATORS 27.2
DISCRIMINATOR
VI
C3 v2 DE-EMPHASIS
L I AA IT ER I
g+ 8+
FIG 27.17 TYPICAL PHASE DISCRIMINATOR ARRANGEMENT FOR F -M DETECTION.
the relationship between output voltage and frequency change from
where
Eoue = rl(E'A E'a) -
= the diode detection efficiency, and is assumed to be the same
(11)
r] for both
diodes of V2.
(B) Design data
For most designs it is usual to take the total secondary voltage E2 as being twice
the primary voltage E1 i.e.
EI/E1 = 2. (12)
Primary and secondary Q's are made equal, so that
Q = Q1 = Q5. (13)
The value for Q is determined from
Q = f /(2df)
where fr = intermediate frequency
(14)
and 24f = total frequency range for which substantially linear operation is re-
quired.
The coefficient of coupling for the transformer is found from
Qk = 1.5 (15)
when a good compromise between sensitivity and linearity is required (the usual case).
For special cases where linearity is the main requirement it is suggested that
Qk = 2 (16)
be used. The loss in sensitivity in this case is about 1.54 times as can be seen by
comparing eqns. (20) and (21).
From the principles of coupled circuits it is possible to show that
E= Q2k, / L1 (17)
24
::::::::::: 11: :.....................................................
'
9si92 ::i
...
. .
...................................... ......................
:::::
O
0
...........r
2
;. .6
..................:::............. .....::7C:.
.8 I
.:r.........e:::e:xx
O 1 2 14 l
1.:..
6
s...
::::.............
:x:::E
1.8 2O
:2I
2 2
X( a Srf)
CURVES Or RELATIVE VOLTAGE OUTPUT (Erra) VERSUS X
FIG. 2718(A) GENERALAED PHASE DISCRIMINATOR
WITH E2/E12; E).81.5 AND 2.
The two curves shown in Fig. 27.18(A) are calculated from the following equations.
For Qk = 1.5, the relative voltage output is
3.25A (22)
-x/(3.25
Erei
- X')' + 4X'
and for Qk = 2,
5A
Erel - V(5 - X2).2 + 4X'
(23)
where A = 1 :: X2 (1 + 1
X
+X') +\1
l' (
+X'
1 '
X = Q 241/fr
Q = magnification factor determined from eqn. (14). Equation (14) is the
condition for X = 1
df = frequency deviation from fr
and fr = intermediate frequency.
To illustrate the design procedure, and to bring out additional points, a worked
example is appended.
(C) Design example
A phase discriminator is required for use with a 10.7 Mc /s F -M i -f channel. The
maximum frequency deviation of the carrier is 75 Kc /s. To make the problem f
1092 (iii) (C) DESIGN EXAMPLE 27.2
complete it will be taken that the highest audio frequency is 15 Kc /s and the de-
emphasis time constant is 75 micro- seconds.
(a) The majority of applications call for reasonable discriminator sensitivity and
so the condition Qk = 1.5 (see eqn. 15) is practically always used. Also, high primary
dynamic impedance will increase the sensitivity of the discriminator. However,
since Q is fixed by other considerations (see eqn. 14) L1 should be large, but L1 is
limited by L2 which in turn is limited by the permissible minimum secondary capacit-
ance C2. A value is selected for C2 and the design can then proceed.
(b) A suitable value for C2, including all strays, is 50 µµF. Then
25 330
L2 10.72 x 50 - 4.41 H.
From eqn. (18),
L2 4.41
= 2.5 pH.
L1 1.77 1.77
From this
25 330
C1
10.72 x 2.5 -
88.5 µµF (including all strays).
(c) If the discriminator frequency-voltage characteristic were exactly linear then
it would be sufficient to make the total frequency range 24f = 2 x 75 = 150 Kc /s.
To this would be added an allowance for frequency drift due to the oscillator and the
discriminator tuned circuits. Since the discriminator characteristic is not exactly
linear (it is linear for about 80% of the total curve using the data given as can be seen
from Fig. 27.18) and the frequency drift is not always small it has become common
practice in broadcast F-M circuits (of the type being considered) to make 24f from
200 to 400 Kc /s.
As a practical compromise we will take the total bandwidth (24f) as 250 Kc /s, but
the design procedure is the same irrespective of what bandwidth is selected.
With 24f = 250 Kc /s we have from eqn. (14)
Q = 10.7/0.25 = 42.8.
From eqn. (15), Qk = 1.5
and so k = 1.5/42.8 = 0.035.
(d) Summarizing, for the discriminator transformer (see also Fig. 27.17)
L1= 2.5µH; L2= 4.41µH; C1= 88.5µµF; C2= 50µµF; Q= Q1 =Q2=
42.8 ; k = 0.035 ; M = 0.116 pH. Secondary winding to be centre- tapped. The
capacitance values include all strays.
14
12
10
4g6
i4
2
10
12
14
The procedure for finding the plotted points on the curve of Fig. 27.18(B) is as
follows. Draw up a table, as shown below, with values of X corresponding to values
that
of Era read from the curve of Fig. 27.18(A). Since it has just been determined
df = 125 Kc /s when X = 1, the column for df can be filled in has (e.g. when X = 0.8
been determined
then df = 125 x 0.8 = 100 Kc /s, and so on). The sensitivity
as 0.0875 volts per kilocycle deviation, and so for 25 Kc /s
deviation E ,
= 0.0875 x
be used to find
25 = 2.19 volts (the lowest deviation frequency in the table should
,
E in this case). It is now seen that the value of E
2.19/0.28 = 7.82.
,
= 2.19 volts corresponds to
If 7.82 is now multiplied
Ere/ = 0.28, and so the scale factor is
curve for
by Ent in each case, the column for E,,,, can be filled in. The complete
,
Ee versus df can now be plotted.
X df Eret E,.,
0.2 25 0.28 2.19
50 0.56 4.38
0.4
0.6 75 0.84 6.56
0.8 100 1.12 8.76
125 1.36 10.61
1.0
1.2 150 1.52 11.9
1.4 175 1.6 125
187.5 1.62 12.7
1.5
200 1.6 12.5
1.6
1.8 225 1.52 11.9
250 1.31 10.5
2.0
in the
(f) It now remains to determine suitable values for the other components
circuit of Fig. 27.17. to the secondary
C8 is for the purpose of connecting the transformer primary
and providing isolation of the secondary from the h.t. on the plate of V,.
centre -tap, insulation resistance.
A suitable value is 100 µµF. The capacitor should have high
C, is a by -pass capacitor ; a value of about 0.01 µF is usual.ohms each.
R, and R, are made equal, and are generally about 100 000 the audio frequency
Cg acts as a by -pass for i -f but must not appreciably
affect
response. The usual value is about 100 µµF. de- emphasis
The de- emphasis network consists of R, and Ce. For 75 microsecond If an improved
the nominal values would be 75 000 ohms and 0.001 µF respectively. be say, 0.25 MQ
a.c. /d.c. ratio is thought to be necessary then suitable values
would
resistor in the following a -f stage should not be less than
and 30 µµF. The grid
actual values used in a
1 MQ if excessive loss in gain is to be avoided. However, the
figures if the overall audio frequency
receiver will generally deviate from these nominal de-emphasis curve to
response of the receiver is made to follow the 75 microsecond in some cases to help
15 Kc /s, because of the presence of stray capacitances etc.
and
compensate for the overall a -f response. diode circuit,
To determine R, :-the transformer primary damping due to aR,single = 100000 ohms.
connected as shown, is Rd, /3. For this circuit Rd, =
R, =
1094 (iii) (C) DESIGN EXAMPLE 27.2
Also, because of the circuit arrangement, R, and R, are in parallel. The total
due to diode conduction currents in addition to R, and R4, is then damping,
Rao 100 000
3/ = 6 - 1660052.
If R, is not equal to R4, the damping resistance is given by R, /3 and R4/3 in parallel
i.e.
R,R4 /3(R, + R4). (When R, = R4 = Rd, the value Rae /6 is obtained as
above).
Take the undamped primary Q (written as Qu) as being 100, and neglect the
ad-
ditional damping due to the plate resistance of V,. Then the total damping resistance
(R) required to obtain a primary Q of 42.8 is
Q Q u,L, 100 x 42.8 x 2a x 10.7 x 2.5
R Qu - Q
= 12,500 D.
100-42.8
From this
12 500 x 16 600
R,
(16 600 -
12 500) -
50 600 D.
To determine R, : The transformer secondary damping is given by
R, = R4 = Rd, = 100 000 52.
(This follows because the damping across each half of the secondary is R, /2 and
R4 /2 respectively, and in each case there is a step up, due to the
transformer being
centre tapped, of 4 times. From this, across the whole of the transformer
secondary
there are two resistances 4R,/2 and 4R4/2 in parallel, and since R, = R4 = R4,
above result is obtained immediately). the
The total damping resistance (R') required to make Q, = Q, = 42.8 is
R' = 12 500 x 1.77 = 22 100 D.
From this
22 100 x 100 000
R' (100 000 -22 100) - 28 400 D.
The design of the limiter stage will be discussed in Chapter 29.
(g) Some causes of discriminator unbalance will be mentioned, before leaving
section, and it is helpful to consider the circuit as a bridge in which unbalance this
has to be
eliminated. Even when the transformer secondary is centre- tapped accurately,
balance can occur because the capacitive coupling between the two halves of un-
ary winding and the primary winding is not necessarily equal. This calls the second-
the method of arranging the windings. Two methods are in common for care in
use. The
first uses a bifilar method with the two halves of the secondary wound
side by side.
The second method arranges the secondary into two halves, placed on either side of the
primary winding, and the coupling of both sections, including that due to stray
ances, is made equal. capacit-
Capacitive unbalance will also occur when the input capacitances of the
are not equal. Suitable arrangement of stray capacitances can often be diodes
used to help
in offsetting this effect.
The use of a small capacitance connected across one of the diodes is helpful in
ducing capacitive unbalance effects. re-
Balancing of the conduction resistances of the two diode units is largely outside
the control of the receiver designer, but variation in the value of R, and R4
made to assist in cases where a very high degree of balance is thought to can be
be necessary.
Additional precautions of this nature are seldom carried out in commercial receivers.
It should be noted in Fig. 27.17 that, if the resistors R, and R4 are shunted
two separate capacitors, an additional series resistor or r-f choke will be required by
tween the junction of R, and R, and the centre tap on L,. Connecting the be-
of R, and R4 through a by -pass capacitor to ground would also effectivelyjunction
circuit the i -f primary voltage to ground, if the additional component is not short-
used.
For most circuits the arrangement shown is applicable, since it requires a minimum
of components consistent with satisfactory performance. The alternative arrangement
27.2 (iii) (C) DESIGN EXAMPLE 1095
Cg
LIMITER DISCRIMINATOR
V2 DE- EMPHASIS
B+
FIG.27.19 ALTERNATIVE ARRANGEMENT FOR PHASE DISCRIMINATOR.
is shown in Fig. 27.19. The design procedure to be used is exactly as before, except
that the primary circuit damping due to the diode circuits (the damping is actually
across the r-f choke) will be approximately Rdc /4 instead of Rdc/6 ; the secondary
circuit damping is Rd,, as previously, where Rd, = R, = R4. If a resistor is sub-
stituted for the r-f choke then the circuit damping is again modified depending on
the value of resistance used in the circuit. Considerations leading to the choice of
a suitable value for the inductance of the r-f choke will be given in detail in Chapter 29
Sect. 2, in connection with the discussion on a.f.c. discriminators.
It is perhaps worth mentioning that hum due to heater-cathode leakage is some-
times troublesome in discriminator circuits, particularly with miniature diodes, and
a simple and effective cure for this trouble is to make the cathode positive with respect
to the heater. This can be readily effected, for example in Fig. 27.19, by connecting
a 10 000 S2 resistor by- passed by, say, a 0.01 µF mica capacitor between the earthy
end of R4 and ground ; the junction of R4 and the 10 000 Q resistor is then connected
to B + via a series resistor whose value is selected so that about 10 to 15 volts appears
across the 10 000 D resistor (which now forms one arm of a voltage divider). In a
typical case the series resistor would be about 0.16 Ml? for a B + of 250 volts.
(iv) Ratio detectors
(A) General
Many of the details given below, regarding ratio detectors, have been taken from
Refs. 34, 35 and 36. Practical experience with several of the arrangements shown
has confirmed much of the data given in the design sections. However, a number of
additional factors will warrant discussion.
The principle underlying many circuits for F -M detection is the peak rectification
of two i -f voltages, the relative amplitudes of which are a function of frequency, to-
gether with means for combining the rectified voltages in reversed polarity. The
output is then equal to the difference between the two rectified voltages. This state-
ment is directly applicable to the phase discriminator circuits of Figs. 27.17 and 27.19,
the two i -f voltages being those applied to the diode plates, and the rectified voltages
combined in reversed polarity being those which appear across the load resistors
R3 and R4. The way in which the two i -f voltages, applied to the diode plates,
are
dependent on the instantaneous intermediate frequency has been discussed in con-
nection with Figs. 27.15 and 27.16. It was mentioned previously that, with the phase
discriminator, changes in the magnitude of the input signal will give rise to amplitude
changes in the resultant output voltage, and the need for some form of amplitude
limiting was emphasized.
In an attempt to eliminate the necessity for a limiter, a ratio type of detector has been
developed from the basic phase discriminator circuit. In this modified circuit the
rectified voltages are split into two parts in such a way that their ratio is proportional
to the ratio of the instantaneous i -f voltages applied to the detector diodes, and the
sum of the two rectified voltages is kept constant. It has been found in this type of
circuit that the useful output voltage, which is proportional to the difference between
the two rectified voltages developed by the diode detectors, tends to be independent
1096 (iv) RATIO DETECTORS 27.2
of amplitude variations superimposed on a frequency modulated voltage applied to
the discriminator circuit input terminals.
A basic ratio detector circuit is shown in Fig. 27.20(A). It can be seen that this is
similar to the conventional phase discriminator but one of the diodes is reversed, and
so the total voltage between the points P and N is equal to the sum of E, and E2.
The sum of E1 and E2 is held constant by means of a battery or a large capacitance.
This point will be further discussed as we proceed. The audio output voltage is
taken from the junction of Rs, R, and C5, C7 and is equal to (E1 -
E2) /2. The
voltage output with frequency change is seen from Fig. 27.20(B) to be similar to that
for a conventional discriminator circuit.
The sum voltage (E1 -}- E2) can be stabilized by using either a battery or by shunt-
ing a large capacitance across the load resistors Rs and R4. A battery would limit
the operation in such a way that the input signal would need to be at least strong
enough to overcome the fixed bias due to the battery voltage. A better solution is to
use a capacitor, since the voltage across it will vary in proportion to the average signal
amplitude and thus automatically adjust itself to the optimum operating level. This
allows amplitude rejection to be secured for a wide range of input signal voltages, the
lowest useful signal being determined by the ability of the diode rectifiers to conduct
with small input voltages.
When a capacitor is used to stabilize the rectified output voltage its capacitance
must be sufficiently large so that the sum of E1 -+ E2 cannot vary at an audio frequency
rate. This calls for a time constant in the circuit made up from R3, R4 and the
additional capacitance C of about 0.2 seconds. The effects of C5 and C7 on the time
constant can be neglected, since the value of the capacitance C will be of the order
of microfarads (usually about 8µF).
V2
D2
E2 R3 OUTPUT
A-F CS
L + EI-E2 EI+E2
2 2
CONSTANT)
EI R4
+ T`' I._
(A)
A -F OUTPUT
Ei -E2
2
a =FREQ. DEVIATION
(a)
FIG.2720 (A) BASIC CIRCUIT OF RATIO DETECTOR.
(B) TYPICAL OUTPUT CURVE.
27.2 (iv) RATIO DETECTORS 1097
E1 +E2
EL+E2 HELD CONSTANT AT VOLTAGE
EQUAL TO RECTIFIED OUTPUT WHEN
e -eo
IG.27.21 RATIO DETECTOR CIRCUIT SHOWING CONDITIONS FOR DIFFERENT SIGNAL
INPUT VOLTAGES.
It is worth noting that the a.v.c. voltages available from ratio detector circuits have
values which are not always directly suitable for application to controlled stages.
Some form of voltage divider arrangement is often necessary to obtain suitable volt-
ages for securing the desired a.v.c. characteristic.
(B) Operation
A brief qualitative description of the operation of the ratio detector will be given
here. A complete quantitative analysis has not been made, although many of the
significant factors have been investigated (see Refs. 34, 35 and 36).
Fig. 27.21 shows the same basic circuit as Fig. 27.20(A), rearranged in a more
convenient form for discussion. The main details of operation are also indicated,
and show what happens in the circuit when the amplitude of the signal input voltage
changes. For the case (A) where the amplitude of the signal input voltage is con-
stant, the stabilizing current is zero and the circuit has essentially the sanie output
characteristic as a phase discriminator. When the signal amplitude increases (B),
the average diode current also increases, and resultant direct current flows into the
stabilizing voltage source (i.e. the battery is being charged). From this since E, -F E,
remains constant; it is seen that the effective diode load resistance must decrease
and so the primary and secondary circuits are more heavily damped than for the case
(A) where the input voltage has a fixed value. The increased damping on the trans-
former circuits will tend to offset the increase in the amplitude of the input voltage,
and so the output voltage will also tend to be constant.
Similarly, when the signal amplitude decreases -case (C) -the stabilizing voltage
source supplies additional current (i.e. the battery is being discharged) to offset the
reduction in diode current, so as to maintain E, E2 constant ; the effective diode
load resistance is now increased, and so the circuit damping is reduced.
From this it is seen that stabilizing the rectified voltage results in the equivalent
load resistance varying in such a way as to offset changes in the amplitude of the signal
input voltage. It also follows that there must be optimum circuit conditions which
will allow the undesired amplitude variations to be offset most effectively (although
these conditions are not necessarily the same for all input voltages). The conditions
for obtaining the most satisfactory operation will be summarized below when dis-
cussing the design of the ratio detector.
1098 (iv) RATIO DETECTORS 27.2
(C) Types of circuit
There are two types of ratio detector circuit in common use. The first of these is
the so called balanced circuit and is illustrated by Fig. 27.22. At first glance this
does not seem to be the same as the basic circuit previously discussed. However,
if the tertiary winding is considered as L, for the moment (its presence will be ex-
plained presently), and it is taken that C, has a high reactance at audio frequencies
and negligible reactance at i-f, then a simple re- arrangement of the circuit will show
that it is the same as that of Figs. 27.20 and 27.21. Whether the junction of C5
and C, is connected to ground, as shown in Fig. 27.22, or C, is connected from the
junction of C, and C, to the junction of R. and R4 as would be expected (since C,
is part of the audio load circuit) is unimportant, as can be seen by redrawing the cir-
cuit. Cis the stabilizing capacitor.
AY.C.
microseconds or so) they can be used to derive a bias voltage for the grid of the valve
when the signal exceeds some pre -determined level, and the valve can then be used
to provide a degree of limiting. In this way the use of an additional valve is still
avoided, since most ratio detectors operate satisfactorily with as little as 10 millivolts
(or less) applied to the grid of the driver valve, but the advantages of additional am-
plitude limiting can be obtained on strong signals. Of course, satisfactory results
are obtainable with a ratio detector alone and the application of a.v.c. bias to the
earlier stages is helpful but improvement is possible with the arrangement suggested,
and with very little additional trouble. With the partial limiter arrangement a.v.c.
is sometimes applied to the first i -f voltage amplifier, the a.v.c. bias being derived from
the ratio detector circuit as indicated in Figs. 27.22 and 27.23. Some receiver manu-
facturers do not use a.v.c. at all, but use suitable grid resistor-capacitor arrangements
with both i -f valves to provide additional bias when the signal is sufficiently large.
Some care is necessary with the limiter arrangements, since detuning of the i-f trans-
former can have adverse effects particularly with some types of noise interference.
This point received consideration in Chapter 26 when the design of the F -M i -f
transformers was being discussed. MEAN SIGNAL I
+73Kci, I AMPLITUDE
+S01
AVE. Z +2S 1
01
_25
-501
O
-75j
FIG. 27.23
Unaalancet) Type Of Ratio Dctr<tor FIG 27.24
-i
Typtcol
SIGNAL AMPLITUDE
Input- Output Voltage Curves
-ar.
for Ratro Detector
value. Smaller values for R will increase the downward modulation handling capa-
bilities of the detector, but will also reduce its sensitivity.
Primary L/C ratio and Q
The primary L/C ratio should be as large as possible to increase the sensitivity.
The limiting factor is the maximum stable gain between the grid and plate of the ratio
detector driver valve (i.e. the last i -f valve preceding the detector). When determin-
ing the maximum gain it should be remembered that the gain may increase during
downward amplitude modulation, particularly on either side of the centre frequency.
If the primary Q is high enough so that the operating Q is determined mainly by
the diode loading, the grid -plate gain will rise during the A-M cycle. This is advan-
tageous since it increases the ability of the detector to reject downward modulation.
The primary Q is made as high as possible consistent with peak separation and stable
gain requirements.
Coupling
The value of coupling used, together with the number of turns in the tertiary wind-
ing, is the principal factor in determining the ratio of the tertiary voltage to the half
secondary voltage. This ratio should always be close to unity. The coupling is
generally adjusted to half critical in the actual circuit, since the degree of coupling is
dependent on the other circuit component values and these are selected before the
coupling is finally set to the desired amount.
Tertiary inductance
The number of turns on the tertiary winding is adjusted so that the required ratio
of tertiary voltage to half secondary voltage is obtained (see under Coupling).
Reducing unbalanced A -M component
There will always be some residual amplitude modulation in the detector output
because of the variation in the effective diode input capacitance during the A -M cycle,
and because of the unbalance introduced by the transformer and other circuit com-
ponents. Several methods are useful in overcoming this effect. One is to vary the
effective centre -tap on the secondary winding (this method is shown in Fig. 27.26)
and another is to make the resistors R, and R, shown in Fig. 27.25 unequal. In
addition a resistor R5 is used in series with the tertiary winding (see Fig. 27.25) to
modify the peak diode currents, which has the effect of appreciably reducing the un-
balanced A -M component particularly at high input voltages. The methods using
resistors to reduce unbalance also reduce the detector sensitivity.
The reduction of unbalance effects is carried out by observing the condition for
minimum A -M output when a signal which is simultaneously amplitude and frequency
modulated is applied to the detector.
Time constant of stabilizing voltage
The discharge time constant of the stabilizing capacitance and the load resistance
should be about 0.2 second. Larger time constants will give better amplitude re-
jection when the undesired modulating frequency is low, but have undesirable effects
on the tuning. The tuning effect is similar to that given by an ordinary A -M receiver
when the a.v.c. time constant is too long.
Experiments with the time constant of this circuit give some interesting effects.
If the stabilizing capacitance is made very large, say 100 µF or so, and the receiver is
tuned rapidly across a signal, it will be found that the point of maximum output is
very easily determined, and there is no effect from the usual side responses. When
the receiver is detuned the noise level rises, as the capacitor discharges, and the side
responses again become evident until the receiver has been tuned once through the
point of maximum output.
Additional details
It is not particularly easy to achieve a good balance with the ratio detector circuit,
and usually more care is required in this respect than with the conventional phase
discriminator. However, the better the balance obtained the better will be the
rejection of undesired amplitude modulation.
Although the side responses are normally well down on the response at the main
tuning point, with this type of circuit, considerable improvement has been noticed
27.2 (iv) RATIO DETECTORS 1101
CeI330yyFTÓ02 pF
R6-CS DEEMPHASIS CIRCUIT
FIG . 27. 25
when the bandwidth of the discriminator is fairly large (say 300 to 400 Kc /s) and the
i-f response is such as to give steep sides to the overall selectivity curve. With re-
ceivers using two i -f transformers (each critically -coupled and both having primary
and secondary Q's of about 75) the side responses are still sufficiently large to be notice-
able, although they are 12 -15 db down on the main tuning position. Receivers using
three transformers (each critically -coupled and having Q's of about 70) give a very
marked improvement, and the two undesired side responses can only be found after
careful searching. If the transformers are made more selective the side responses
are reduced, but the additional non -linear distortion introduced by the tuned circuits
may no longer be negligible, as can be determined from the data given in Chapter 26.
Further, the amount of amplitude modulation on a carrier goes up as the selectivity
of the transformers is increased, and this undesired A -M has to be removed by the
ratio detector thereby reducing its effectiveness to other undesired external noise.
Even in the case of the receiver with only two i -f transformers, the undesired side
responses are very much less noticeable than with receivers using the usual limiter
and discriminator combination
(E) Practical circuits
Two circuits which have been constructed and tested under working conditions
are shown in Figs. 27.25 and 27.26. The intermediate frequency is 10.7 Mc /s
Constructional details and performance data can be found in Refs. 34 and 37.
M A.V.0
6H6 50,0000
t
L= 4-30
yy
.I yF
B
yF
40,0000
AmO12NF
20,000
L
ff
F5M0
A.F.
B 4 (26sV)
FIG. 27.26 - Unbalanced Ratio Detector
De- emphasis is obtained in the circuit of Fig. 27.26 by adjusting the value of the
capacitor marked CA. The time constant (0.4 second) of the diode load circuit is
about twice as long as that previously recommended for typical cases, but the per-
formance is quite good, and, as suggested in Ref. 37, the individual designer can set
the constants to suit his own requirements.
The secondary winding for the circuit of Fig, 27.25 is a bifilar arrangement, while
that of Fig. 27.26 has the secondary split into two sections placed on either side of
the primary. In the latter case slug tuning is used for one half of the secondary so
that the effective centre tap can be set to any desired position.
1102 (iv) RATIO DETECTORS 27.2
kc.0
E_,
E1
- 1.
For example, if the signal voltage at the plate rises 25 per cent. (i.e. E, = 1.25 E1)
when the secondary is detuned, the coupling is 50 per cent of critical.
The ratio between the secondary and tertiary voltages can be measured indirectly
in terms of the rectified output voltage which is obtained (a) with the secondary tuned
and (b) with the secondary detuned. In both cases the input signal is adjusted so
that the primary voltage remains constant. If the ratio between the voltages read in
(a) and (b) is r, that is if
Rectified voltage (tuned secondary)
r Rectified voltage (detuned secondary)
then
P
=Vra - 1
type the secondary tuning is adjusted so that the d.c. voltage at the audio take -off
point is equal to half the total rectified voltage. As an alternative an amplitude
modulated signal can be used, the primary trimmer is adjusted for maximum d.c.
output voltage and the secondary trimmer adjusted to give minimum audio output ;
the residual output is measured in the conventional manner. Circuit unbalance is
indicated if the two methods do not give the same secondary trimmer setting.
If sweep alignment is used, the primary can be accurately aligned by using a corn -
paratively low deviation, and adjusting the primary trimmer for the maximum am-
plitude of output voltage. The secondary may be adjusted by using a deviation such
that the total frequency swing (twice the deviation) is equal to the peak separation.
This procedure makes it possible to adjust the secondary tuning so that a symmetrical
detector characteristic is obtained.
Peak separation
The separation between the peaks on the F-M output characteristic of a ratio de-
tector may be measured by applying a frequency modulated signal and increasing the
deviation until the response is just observed to flatten at the peaks. When this is
done, the peak separation is equal to twice the deviation.
If an attempt is made to measure the peak separation by plotting the output charac-
teristic point by point, the peak separation obtained will usually be considerably less
than that obtained under dynamic conditions with the output voltage stabilized. The
F-M detector characteristic may be plotted point by point provided a battery of the
proper voltage is connected across the stabilizing capacitor. The voltage of this
battery must be equal to the rectified voltage which exists at the centre frequency.
In practice it is convenient to use a 7.5 volt " C " battery and to adjust the signal
input so that the capacitor- stabilized voltage at the centre frequency is equal to the
battery voltage.
It is worth noting that the peak separation is not constant for all values of input
voltage, and measurements should be made for several representative voltages of the
magnitude likely to be encountered under practical conditions of reception. The
peak separation is wider for larger input voltages, as a result of the increased diode
loading on the tuned circuits. Similarly the peak separation is less for smaller input
voltages because of reduced circuit damping.
Measurement of A -M rejection
The measurement of A -M rejection can be carried out using a signal generator
which can be simultaneously frequency and amplitude modulated.
Ref. 34 shows a number of the typical wedge- shaped patterns which are obtained
when visual methods are used for determining the amplitude modulation present
in the audio output. A pattern of the type to be expected is shown in Fig. 27.27.
The measurement of the A -M rejection
of a ratio detector can be described in terms
of the pattern obtained for a given frequency
30 M
deviation and a given percentage of amplitude
modulation.
The measurements should be made for
several different values of input voltage.
Ideally the pattern obtained should be a
diagonal line regardless of the presence of
amplitude modulation. The amount by
which the pattern departs from a straight
line indicates the extent to which the de-
tector fails to reject amplitude modulation.
To measure A-M rejection with a genera-
tor which can only be frequency modulated
the procedure is as follows. A frequency
FIG 27.27 modulated signal fully deviated (+ 75 Kc /s
1104 (iv) RATIO DETECTORS 27.2
in the usual case) is applied, and a battery of which the voltage is equal to the rectified
output is shunted across the stabilizing capacitor. The magnitude of the input
signal is now reduced until the output becomes distorted as a result of the diodes
being biased by the stabilizing voltage, and finally the F -M output drops to zero.
The ratio between the initial input signal and the minimum input signal for which
the F-M output becomes distorted is then a measure of the amount of downward
modulation that the detector can reject. For example if the voltage ratio is r, then the
percentage of downward amplitude modulation which can be handled is 100 (r
To determine the rejection of the detector for upward amplitude modulation the
- 1) /r.
same set-up is used and the change in output is noted as the input signal is increased.
Since the amplitude of the input signal is not varied dynamically, this method will
not indicate any unbalanced component which may be present in the output. The
latter is more conveniently measured using simultaneous F-M and A-M as discussed
previously.
Another method which can be applied to measure the amplitude rejection pro-
perties of any type of detector for different signal input voltages may be worth dis-
cussing. The input signal is simultaneously frequency and amplitude modulated.
Any audio modulating frequencies can be selected which are not harmonically related,
and suitable values are say 400 c/s for F -M and 30 c/s for A -M. The percentage
A-M can be taken as 30% (or any value desired) and the deviation as f 22.5 Kc /s
(or any other value). The ratio of the desired and undesired audio outputs can then
be measured on a frequency selective voltmeter such as a wave analyser. Measure-
ments can be made at the centre frequency to indicate residual A-M, and for various
frequencies off the centre frequency to determine the rejection capabilities of the
detection system under operating conditions ; e.g. those which occur when oscillator
and other frequency drift occurs. Low audio frequencies are preferable for this
test (and also for the previous methods), particularly with ratio detectors, since the
amplitude rejection is least for this condition. Tests on a number of ratio detectors
have indicated similar performance when high audio modulating frequencies are
used, but considerable variation occurs in the amount of amplitude rejection obtained
for the lower audio frequencies using balanced and unbalanced circuits. This is
readily understood when the function of the stabilizing capacitor is considered.
Standard test procedures have been devised for F-M receivers, and standard methods
for evaluating the degree of A-M rejection, downward modulation handling capability,
and the effects of mistuning are described in Chapter 37.
Non -linear distortion
Distortion measurements are usually quite difficult to carry out with any degree of
accuracy because of the inherent non-linear distortion present in the signal source
and test equipment. Tests on typical F-M receivers using limiter- discriminator
combinations and ratio detectors have been made, using the low power stages of a
F-M broadcast transmitter as the signal source. The non -linear distortion in the
transmitter signal was less than 0.5% for frequencies from 100 c/s to 15 Kc /s, and
less than 1% from 30 to 100 c/s ; the frequency deviation was set to ± 75 Kc /s in
all cases.
Both types of receiver indicated a measured overall distortion of the order of 2 to
2.5% in the frequency range 100 c/s to 7.5 Kc /s, 3 to 5% at 50 c/s and 5% at 15 Kc /s
although this latter case was a visual check only. Above 10 Kc /s the non -linear
distortion with ratio detector receivers is generally somewhat higher than with limiter -
discriminator receivers. The input in all cases was approximately 101) µV and the
output was adjusted to 0.5 watt. The test frequency was 97.2 Mc /s and the i -f of
all receivers was 10.7 Mc /s. The measured distortion for the a -f amplifier alone was
0.5% for the range 100 c/s to 7.5 Kc /s and 2% at 50 c/s ; above 7.5 Kc /s, observa-
tion was made using an oscillograph and as there was no visible distortion it was as-
sumed that distortion was less than 3 %.
The non -linear distortion is increased with ratio detector receivers when the magni-
tude of the input voltage is made very large. With the other type of receiver the input
signal must be sufficiently large to ensure satisfactory limiter operation. Receivers
27.2 (iv) RATIO DETECTORS 1105
using a combination of partial limiting and a ratio detector give excellent results for a
wide range of signal input voltages, and do not require extra valves.
The test conditions used are very severe, as it is unlikely that a 15 Kc /s audio signal
will cause 75 Kc; s deviation ; although it should not be overlooked that pre -emphasis
in the transmitter will cause a 15 Kc /s modulating signal to be increased in magnitude
by about 17 db. Listening tests indicate that excellent results are obtainable with
any of the receiver arrangements. It will be appreciated that the possibility of errors
in the distortion measurements are very large, and the results are given to serve as an
indication of the magnitude of the distortion to be expected with typical F-M receivers
which do not include elaborate design features, but conform to good engineering
practice.
(i) Introduction
Automatic volume control is a device which automatically varies the total ampli-
fication of the signal in a radio receiver with changing strength of the received signal
carrier wave. In practice the usual arrangement is to employ valves having " remote
cut-off " or " variable -mu " grids and to apply to them a bias which is a function of
the strength of the carrier.
From the definition given, it would be correctly inferred that a more exact term to
describe the system would be automatic gain control (a.g.c.). The older term a.v.c.
has been retained here as the name is not likely to lead to any confusion, and is widely
used throughout the radio industry.
constant during charge, and C,(R, + R2) as the time constant during discharge.
It is seen that the charge time constant is shorter than the discharge time constant.
Time constants in a.v.c. circuits will be discussed in detail in (xii) below.
For 100% rectification efficiency the a.v.c. voltage would equal the peak i -f voltage
applied to the diode. Because of losses the a.v.c. voltage is always somewhat less
than the peak carrier voltage (this is readily seen from Figs. 27.5 and 27.6), and since,
in general, a higher voltage is required for the a.v.c. than for detection, this will be a
limitation to the use of simple a.v.c.
The design of the i -f transformer to couple the i -f voltage into the diode circuit is
quite straightforward. The damping to be expected across the windings due to the
diode circuits has already been discussed in Sect. 1(i)Cc of this chapter. Methods
of i-f transformer design are detailed in Chapter 26.
In the circuit of Fig. 27.28 the d.c. diode load consists of R1 and R2 in series, R,
being used in conjunction with capacitors C1 and C2 to form a filter to prevent the
major part of the undesired i -f voltages from appearing across R,. The volume
control is R, in this circuit, and the advantages and disadvantages of using the control
in this position were discussed in Sect. 1(i)Dc above. At audio frequencies R, is
shunted by R, and by R3 when the setting for R2 is at the maximum. It follows that
the diode load for a.c. is considerably different from that for d.c., and, as previously
discussed in Sect. 1(i), considerable non-linear distortion of the a -f output will result
when the modulation percentage is high. R, and C4 will substantially remove all
audio frequency variations from the a.v.c. bias applied to the controlled stages, as well
as any residual i -f voltages which may be present across R,.
Fig. 27.28 shows R4 connected to the junction of R, and R but in some cases R,
is connected to the top of R1. This latter arrangement will give a slightly higher
a.v.c. voltage, the amount depending on the size of RI, but the amount of i -f filtering
for the controlled stages is reduced. With the usual component values it is not very
important which arrangement is used.
Owing to the effects of contact potential in the diode, together with rectification
due to unavoidable noise voltages, there is a voltage developed across R, even with
no carrier input. Even with a weak carrier input this voltage is increased. Conse-
quently it will be seen that with the weakest carrier likely to be received there is an
appreciable negative voltage applied to the controlled grids. If no means were taken
to compensate for this, the overall sensitivity of the receiver would be decreased.
Compensation for the initial standing bias, before a carrier is received, can be carried
out by applying a lower minimum negative bias to the controlled stages. However,
a very real disadvantage of simple a.v.c. is now apparent in that the sensitivity of the
receiver may decrease as soon as a carrier is received. To overcome the loss in sensi-
tivity on weak signals a delayed a.v.c. system is used and this gives very much improved
control.
A special case of the effects of contact potential with zero bias valves is worth noting
before leaving this section. Suppose a negative bias voltage exists between the grids
and cathodes of the controlled valves due to positive grid current. This current can
be sufficiently large to cause an appreciable negative bias voltage to appear across the
resistance common to the controlled stages and the detector, and so give rise to a
negative bias voltage between plate and cathode of the diode. Under this condition,
if simple a.v.c. is being used, there will be no output from the receiver until the signal
is sufficiently large to overcome this delay bias on the detector. If a separate diode is
used for a.v.c. the effect will be to increase the delay bias. Typical examples of the
effect have occurred in practice with type 1P5 -GT voltage amplifiers used in con-
junction with the diode section of a type 1H5 -GT. An obvious remedy is to apply a
small positive bias voltage to the diode plate, but the grid current can often be ap-
preciably reduced by increasing the screen voltage.
carrier strength reaches a pre -determined level. The result is that no a.v.c. voltage
is applied to the grids of the controlled stages until a certain carrier strength is reached,
and the receiver will have its maximum sensitivity for signals with an amplitude below
this pre -determined level.
The circuit used for delayed a.v.c. also makes possible improved rectification
efficiency in the a.v.c. circuit, thus producing slightly greater a.v.c. voltage for the
same peak diode voltage. This is due to the higher value of a.v.c. diode load resistance
(R, in Fig. 27.29) permissible in this circuit. With the a.v.c. arrangement shown the
a.c. loading on the detectar circuit is reduced when compared with the arrangement of
Fig. 27.28 and this reduces the distortion of the audio output voltage. Some non-
linear distortion of the modulation envelope does occur at the primary of the trans-
former but this is considerably less than that introduced when the a.v.c. voltage is
olitained from the secondary circuit. In addition, the overall selectivity up to the
primary of the i -f transformer is less than that at the secondary, and the a.v.c. will
start to operate further from the carrier frequency than if fed from the secondary.
The advantage here is that the tendency to give shrill reproduction is reduced as the
receiver is detuned from a carrier.
To
CONTROLLED
GRIDS
However, in some cases this leads to difficulties with cross modulation. When cross
modulation is a serious problem it is preferable to control the r -f amplifier valves
fully. Often no great care is taken with broadcast receivers in the method of applying
a.v.c. With high quality communications receivers precautions are necessary, and
it is helpful if the r-f and i-f valves to be controlled by a.v.c. can be selected by the
operator. In this way optimum performance can be obtained under all conditions.
The usual arrangement, however, is to design an a.v.c. system with compromise
characteristics and fit manual switching for the selection of suitable time constants.
The signal -to-noise ratio of the receiver should be measured when the a.v.c. charac-
teristic is being determined, to ensure that the performance of the receiver is not
seriously impaired when the signal input voltage is increased. Cross modulation
tests are also necessary when the receiver performance is being determined. Further
discussion of these problems can be found in Ref. 43 (pages 169 and 179). A dual
a.v.c. system which has interesting possibilities is described in Ref. 50. Further dis-
cussion on a.v.c. and noise is given in Chapter 35 Sect. 3(i).
When a duo-diode triode (or pentode) valve is used with cathode bias, the value of
the bias is usually between 2 and 3 volts. Such a voltage is suitable for use as a.v.c.
delay bias ; a very simple arrangement is possible by returning R, to earth as shown
in Fig. 27.34. Alteration of the delay bias because of the presence of contact potentials
must be considered, and this effect has already been discussed in (ii) above.
A circuit which eliminates differential distortion, even with large delay voltages
and high modulation levels, is shown in Fig. 27.38B. Signal detection and a.v.c.
detection are carried out as before, but a diode, D,, and the three resistors R,, R,
and R1, are also used. R, and R1, form a voltage divider from B + to ground, the
junction of these resistors and R, having a potential of, say, +50 volts. Typical
values for R4 and R8 could be 1 megohm and 5 megohms respectively, and it will be
seen that R, and R8 form a voltage divider between +50 volts and the source of
a.v.c. potential.
The operation of the circuit can be explained most simply if the diode D, is ignored
initially. Under these conditions, with no signal input to the receiver, point X will
be practically at ground potential since there will be no i -f input to the diode Ds and
the voltage at Y will be approximately
R,
+ 50 x
R4 R,
- + 8 -1/3 volts.
Now as an increasing signal is applied to the input of the receiver, the available a.v.c.
voltage at X will increase, and when this voltage reaches -10 volts, the potential across
the voltage divider R,R, will be 50 -
( -10) volts, and the potential at Y will be
by R4 and R8. However by using a large value of resistance for R8 and a high positive
voltage the ratio can be made as close to unity as required by any practical considera-
tions.
The cost of the resistors Rs and R10 can often be avoided by using a source such
as the screen of a valve which has no a.v.c. applied, or a voltage divider used for some
other purpose, for the positive potential. To save the cost of a valve having an ad-
ditional diode for D, it is quite possible to use the signal diode D1 as the source of
a.v.c. voltages, and D, as the " sinking " diode. Alternatively D1 and D, can be used
as in Fig. 27.38B and the suppressor grid of say the i -f amplifier can be used as Ds,
although in this case with no signal applied the a.v.c. line will take up the potential
of the cathode of the i -f valve.
This method of providing delay either to the first stages or to all controlled stages
of a receiver is of course equally applicable whether manual or automatic volume
control is used
FIG. 27.30
(a) Series feed
One arrangement of series feed is shown in Fig. 27.30. It will be seen that in
each r -f tuned circuit a blocking capacitor (C,, CO is used so that the rotor of the
ganged tuning capacitors may be earthed and the a.v.c. voltage fed to the lower ends
of the coils. In the r-f stages the use of this blocking capacitor will reduce the fre-
quency coverage, and may also affect the tracking of the tuned circuits. To reduce
these effects the same capacitance value should be used in each of the t -f stages, and the
value selected should be fairly large consistent with other considerations. Also, if the
capacitors are not to affect the Q of the tuned circuits adversely, they must be of a
low loss type. A disadvantage of a high capacitance is that it increases the time con-
stant of the a.v.c. circuit.
For typical cases Cs and Ce would be about 0.05 sF if the maximum capacitance
of the tuning capacitor is about 400 µµF. The effect on the tuning range is easily
1110 (iv) METHODS OF FEED 27.3
TO a4
FIG. 27.31
With the r.f. stages an alternative arrangement is to insulate the rotor of the ganged
capacitors, and to by -pass it to earth by a single capacitor (Ca, Fig. 27.31). This
enables the a.v.c. voltage to be applied without using any blocking capacitors in the
r -f tuned circuits, and has the further advantage that the time constant of the a.v.c.
circuit may be made small. The arrangement also has obvious disadvantages and is
little used.
(b) Parallel feed
The " parallel (or shunt) feed " circuit is shown in Fig. 27.32. In this arrangement
a blocking capacitor is necessary to prevent a low resistance d.c. path being formed
from grid to earth by the tuning inductor. The resistors R8, Rs and R10 provide part
of the d.c. path from the valve grids back to the cathodes. Since the individual resis-
tors are shunted across their corresponding tuned circuits, the values selected must not
be too low as otherwise appreciable damping occurs. A 0.5 megohm resistance in
parallel with a typical r -f tuned circuit generally is not serious, but could appreciably
reduce the dynamic resistance of an i -f circuit. For these reasons " parallel feed "
is sometimes used with r -f stages, but it is seldom applied to i -f stages, and receivers
are often designed to use both systems of a.v.c. feed.
" Parallel feed " for the r -f stages is sometimes more convenient than " series feed,"
and appears tc be satisfactory in most respects, although it has been found that grid
blocking is more likely to occur with " parallel feed " than with " series feed."
With any method of feed it is important that the total resistance in the grid circuit
should not exceed the maximum for which the valves are rated. Depending on the
characteristics of the particular valve type and the effects on electrode dissipations
and total cathode current, as determined by equation (6) on page 82, the
following maximum values of grid resistor may be used as a general guide.
For one controlled stage 3 megohms
For two controlled stages 2.5 megohms
For three controlled stages 2 megohms
27.3 (iv) METHODS OF FEED 1111
These resistances are to be measured between the grid of any valve and its cathode.
The values above assume that the receiver is normally tuned to a station and that
the controlled valve or valves are operating at reduced cathode currents and trans -
conductance, as determined by the a.v.c. bias.
The circuit values shown are typical for an intermediate frequency of 455 Kc /s.
An a.v.c. resistor of 1.25 megohms is used so that the total resistance to earth from
any grid does not exceed 2 megohms. If there were only two controlled stages this
could be increased to about 1.75 megohms with a consequent decrease in the a.c.
shunting.
A typical delayed a.v.c. circuit is shown in Figs. 27.34. Cathode biasing is used
for the pentode section of the duo -diode pentode valve, and since the bias will usually
be about two or three volts this also provides a suitable a.v.c. delay voltage without
any further complication. With circuits using delay voltages on the a.v.c. diode the
bias due to automatic volume control on the controlled valves is zero until the peak
voltage on the diode exceeds the delay voltage. The controlled stages are arranged
to have a self-bias voltage equal to the recommended minimum grid -bias voltage.
it is frequently advantageous to operate valve types 1A7 -GT, 1C7 -G, 6A8, 6D8 -G
on fixed bias. Valve types 6J8 -G, 6K8, 6BE6, 6SA7 and all triode hexodes may
be used with a.v.c. on A-M broadcast and short wave bands *. For receivers operating
in the 88 -108 Mc /s F -M broadcast band, a.v.c. bias is not applied to the signal grid of
the converter valve ; very often on this band fixed bias is used for the r-f stage, and
a.v.c. is applied to the i -f stages only. Whether a.v.c. is used at all with these receivers
often depends largely on the type of detector, but it is advisable to avoid overloading
of the early stages so as to reduce the possibility of the generation of undesired spurious
responses ; this often calls for a.v.c. bias to be applied to the r -f stage.
Some oscillator frequency shift occurs with all types of converters when the signal
grid bias is altered. With valve types such as the 6A8 operating on the short wave
band the reception of a fading signal is difficult since the variations caused by a.v.c.
bias may cause the signal to swing in and out of the pass band of the receiver. An
even greater difficulty occurs when the receiver is being tuned to a strong signal,
since the magnitude of the output will be different when tuning in from either the
high frequency or the low frequency side of the signal. Very careful adjustment is
required to obtain the best tuning position. A rather similar effect has been noticed
on both the broadcast and short-wave bands, but in this case the effect is generally
due to faulty gang wipers. When this is the cause it can occur with any type of
receiver (t.r.f. or superheterodyne), but with superheterodynes the oscillator section
of the ganged capacitor should be checked first, as this is generally the one causing
most of the difficulty.
With ordinary broadcast receivers having a r -f stage, converter, and a single i -f
stage, automatic volume control is normally applied to all three stages. If decreased
modulation rise is required it is preferable to operate the i -f stage with about one -half
of the full a.v.c. bias or, alternatively, to supply the screen voltage by means of a series
resistor ; the latter arrangement is the more usual one in practice. Negligible modu-
lation rise is possible, with effective a.v.c. action, if all control is omitted from the
i -f stage, but difficulties such as those due to decreased signal-to -noise ratio must
not to be overlooked when this system of a.v.c. is used on a number of tuning ranges
(see (iii) above).
In a receiver without a r-f stage it is difficult to avoid overloading with large input
signals, and a.v.c. is applied to both stages even though modulation rise may be
objectionable with very high inputs. In order to obtain maximum control the screen
voltage for the i -f valve should be obtained from a voltage divider. The degree of
control must, however, be weighed against the possibility of non -linear distortion ;
this factor has already been discussed in Chapter 26 in connection with distortion in
i -f amplifiers.
For a receiver having two i -f stages the second stage should preferably be operated
at fixed bias in order to prevent modulation rise. It is possible that better overall
performance at high input signal levels will be obtained when the second i-f stage is
operated at a negative bias somewhat greater than the minimum bias. Removal
of control from the last i -f valve will often reduce the possibility of overloading this
stage, although this is not always a good solution as will be discussed in (xv) below.
B+
(b) If a common i -f channel is used, it
is possible to add a further i -f stage with
fixed bias for a.v.c. only, followed by a
separate a.v.c. rectifier. By this means
it is possible to avoid the distortion due
to shunting of the diode load resistor, or
to " differential loading " at a point
where the a.v.c. delay is just being over-
come. Avc
(c) A d.c. amplifier may be used to BIAS
amplify the voltages developed at the
rectifier. A typical circuit arrangement
giving d.c. amplified a.v.c. is shown in FIG. 27.35
Fig. 27.35. Design methods are given B-
Circuit Artanainwlt tar Obtaining D.C.ATPIIf1tA AMC
in Refs. 44 and 45.
See F_ 27.43 and the description given in Sect. 4(ii) later in this chapter.
iii).Audio a.v.ç.)
Audio a.v.c. is sometimes used in radio receivers in conjunction with a.v.c. applied
to the r-f and i -f stages to flatten out the overall a.v.c. characteristic. Whole or part
of the a.v.c. bias voltage is applied to an audio frequency voltage amplifier valve
having a grid with a variable-mu characteristic.
It should be clear that conventional a.v.c. systems cannot give a perfectly constant
a -f output with a varying input voltage, because conventional a.v.c. is a " back-acting "
device in which the effect on the controlled stages must follow the change occurring
at a later stage in the receiver. Audip_a vs-is " forward- acting " and so a constant
or even a drooping characteristic is possible for increasing signal input voltages.
Simple audio a.v.c. systems tend to introduce a considerable amount of non -linear
distortioninto the audio output voltage when the signal input voltage ís Targe. This
is due to the curvature of the g. -
eu characteristic of the controlled valve.
The amount of distortion can be reduced to negligible proportions by the use of
elaborate circuits, but as these systems generally require an additional a -f amplifier
valve they are not used to any extent in commercial radio receivers.
If it is decided that audio a.v.c. is desirable, the usual procedure is to obtain the
best possible a.v.c. characteristic, apart from the a -f amplifier, and then to add just
+20
M,1IIIIIIIIIII M
111111111111=111I1 M111111I1u111111M111111=_111111 I
M ,.I
1111 INMI11111IMI11111I 111111
ÌÌ
1a111111 1
=6111111M11111IM1111/1
lo
11
Ma111 11111111111111
I111EIIIIII In
MINI 2 111 MII1111111IIII1l=111111111
m11111111mIIIIIImsIIIIII
ÌÌÌÌÌÌ
+
M
=1I111m1111
111111111
8.t11
yp 11111 MI1111111111111111111
, IIIIII MIS11111II1=1111111=111111
O
IIIIIII
111lII1111
'ÌÌÌI ÌÌ ÌÌIÌÌÌÌÌÌÌÌÌÌ
/%M11111 1111111IIIIII11V': . III IS
IH
111
I
11/11111
C':':,,':
E' IIIII mH111111mIIIIII
`
1/í1111%111111u- 3 I11io/!;í11
0
20 tII1Il I/i
mt'IIIIIImw!5i
=I.111111 !WM
ML1I11I Ir/n/III
D
a.04
-a11
rntl 111111ía111111
III 1111 111111111111
MII1111111 NII111111.1.0111111
1=M1111111 11111 IIIIII
::: :
IIIIII IIaoPv 111.11111 111111IIIII
:I
IÌI
11I
ii IMaI1PAM, M1111111 1I11 1111
mIí1I// u"..
ÏÌÌ
1:::I1111MMIIII MII1111
WC':! MN11111MS1111
C:1 OOHS
-40 :: ÌÌÌ11ÌÌ
0
10
I
0 0
ÌÌÌ ÌÌÌ pool5
INPUT
Ì :1
MICROVOLTS
.
Ú;I
IÓ IO
FIG. 27.36
1114 (viii) AUDIO A.V.C. 27.3
+40d0
11uu1
/MUM C:Ì ICÌ
n11111 INEEuu"
=
4I11x =111111111111111
=ulll
1
=111111 I11111
=u
=11111 =
11111 =1111
n11111 =1 1111
::Ì CM:
c,
oa
C11:1iiii
MUNI
11
1!
+30db yp 11 C11111iÌÌi1
n11111
1
W.:WM111111
=11111 =M t.11 11111
11111 MUM
C:Ì CrÌ1i Ìij1!i
11111.Ir!Siii
*!
=wow MII11
4111 MIN11111
MMI/[A
Ì1I110CCH
01111111111111
1Ì11111 111
111
III
r=111111111111111
=111111111111111
111=111111111111111
11
á
E
Ñ +10db
CIi Cinä ii C CHi:H
Ci:i Ci::iiii
=MOWN
I 1 :1
111
I
MINIMA
I
C:;11=ilÌi "
,,
od
n1ii un n
=MUM M11i 11111i CMINIB i
11
I ENNUI
;'CC
1 CE::iiil
MIEN I
ó 1
C::"
C::HiBi®::Iiaó,l::_e:!!_:_
=mum= n :::iiiil
-20db
1.0 !1 10S OS - - 04 IOS 10
FIG. 27.37
INPUT -M ICROVOLTS
enough a.v.c. bias voltage to the a -f amplifier to give the required overall results.
The improvement in the overall a.v.c. characteristics, when audio a.v.c. is used, can
be seen from Figs. 27.36 and 27.37.
An example of combined a.v.c. occurs in receivers using a reflexed voltage amplifier
stage (see Chapter 28). In this case the a.v.c. bias voltage applied to the valve control
grid is effective in controlling both the i -f and a -f gain, and if properly used the system
can be made to assist very materially in obtaining good a v.c. characteristics from a
receiver having only one other controlled stage, namely the converter stage in the usual
receiver. Another example of a circuit using audio a.v.c. is shown in Fig. 27.45.
a.v.c. and its return is taken to filament negative. The diode at the negative end of
the filament is used for detection and its return is taken to filament positive.
It sometimes happens that the filaments of battery valves are operated in series or
series-parallel arrangements. Under these conditions it becomes more difficult to
design an efficient a.v.c. system than when parallel filament operation is used. In
the usual type of a.v.c. circuit the grid returns from the several controlled stages are
brought to a common point, and the zero signal grid voltage on these stages is the same.
With series or series -parallel operation, however, the filament voltages differ with
the result that the zero signal bias on one or more stages may differ from zero by a
multiple of the filament voltage. Circuit arrangements which allow the bias on the
grids to be zero for no signal input generally allow only a reduced proportion of the
a.v.c voltage developed by the diode to be applied to some of the controlled stages ;
full a v.c. bias can only be applied to one or two stages in most cases. Fig. 27.38A
shows a typical circuit arrangement for the case where the filaments are connected
in series and operated from the H.T. supply. The a.v.c. voltage divider is made up
from the resistors R8, R9, Rio and R11. A value of about 2 megohms is usual for
R10 while R11 is made about 2 megohms per 1.3 volt drop, in the case of 1.4 volt valves
(which are operated with 1.3 volts as a recommended value for the series filament
connection). In the circuit shown a higher proportion of the a.v.c. control voltage is
applied to valve V9 than to valve V3. The purpose of the resistors R,, Ra, R8, Rd
and R6 is to allow the correct filament voltage to be applied to each of the valves, since
it should be obvious that the valve V1, at the negative end of the chain, carries the
total cathode current for all the valves, and each of the other valves carries the cathode
currents for all the other valves which are nearer to the positive end of the chain than
the valve being considered. This statement is modified when R6 is used, because
this provides an alternative path for the cathode currents. From this it follows that
no shunting resistor is required for Vs
7 re2
R2 2 2
.111
i19 SmA
9+90V
Re R NRe
Ri -
R,- tathode
Filament dropping resister
current shunting resisters
R9- Filament current shunting resist*.
R e- Bias resistor
Re- Diode load
R,- filter resister
o
A.V.C.
across the a.v.c. diode load from being applied back to the grids of the controlled stages.
If r -f or i-f voltages are applied back to the grids then instability troubles are certain
to arise. A-F applied to the grids can lead to a reduction in the percentage of modu-
lation present on the carrier. Inter-stage filtering is also necessary because undesired
coupling can lead to instability and spurious whistle responses. From this it can be
seen that the prime function of the filter circuits is to allow only the d.c. voltage de-
veloped across the diode load to be applied as additional bias on the controlled valves.
The components making up the filter also serve the necessary function of completing
the r -f and i -f circuits through low impedance paths and, furthermore, the resistors
serve to complete the d.c. return path between grid and cathode of each controlled
valve so that the correct negative bias voltage can be applied.
It is not permissible to choose values of resistance and capacitance at random if
satisfactory circuit operation is to be obtained. Several conflicting factors must be
considered when the component values are being selected. The time constant (de-
fined below) of the filter network, must be low enough so that the a.v.c. bias voltage
can follow the changes in signal input voltage with sufficient rapidity to offset the
effects of fading. Limiting values for the total circuit resistance have been stated in
(iv) above, and the choice of suitable values for R and C will be discussed as we pro-
ceed, but first the expression " time constant " will be defined.
The " time constant " of a resistance -capacitance network is the time in seconds
required for the capacitor to acquire sufficient charge for the voltage between its plates
to be equal to 63.2% of the total voltage applied to the circuit. Alternatively, it is
the time taken for a charged capacitor to lose sufficient charge for the voltage across
its plates to fall to 36.8% of the initial voltage existing between the plates in the fully
charged condition. It is a simple matter to show that, for the conditions stated, the
time constant (T) is given by
T = RC seconds
where R = circuit resistance in ohms
and C = circuit capacitance in farads.
(It is usually more convenient to write
T = resistance in megohms x capacitance in microfarads = seconds).
If the total resistance in the circuit is altered in any way from its value during the
charging operation to a different value during the discharge of the capacitor, then it is
helpful to use the terms " charge time constant " and " discharge time constant."
In a.v.c. circuits there is always a difference between the charge and discharge time
constants, because during charge the diode is conducting and its conduction resistance
effectively short- circuits the d.c. diode load resistor. This also means that the " charge
time constant " is always more rapid than the " discharge time constant " in these
circuits.
Suitable values for " charge time constants " are
Broadcast good fidelity receivers 0.25 to 0.5 second
Broadcast receivers 0.1 to 0.3 second
Dual wave or multi -band receivers 0.1 to 0.2 second
For the reception of telegraphy longer time constants are often required to ensure
silence between signals, and a value of about 1 second is often used. In any good
quality communications receiver it is usual to provide facilities for selecting any one
of a number of a.v.c. time constants. This enables the operator to select the most
suitable condition to offset the particular type of fading being encountered. Too
rapid a time constant is not selected for high quality broadcast reception as rapid
fading would cause bass -frequency anti -modulation and so reduce the audio frequency
bass response. See also Chapter 35 Sect. 3(i)B2, page 1233.
In broadcast receivers the charge and discharge time constants are often very nearly
equal. However, for some types of reception it may be preferable for the charge time
constant to be rapid to prevent the beginning of a signal from being unduly loud,
but the discharge time constant is made comparatively slow to prevent a rapid rise
in noise output during intervals between signals (see also Ref. 43, p. 181).
27.3 (xii) THE A.V.C. FILTER AND ITS TIME CONSTANTS 1117
The procedure for calculating charge and discharge time constants is as follows.
Consider Fig. 27.30 in conjunction with Fig. 27.28, then
For 2 stages
Charge time constant =
R1(C2 + C4 + Cs) + R4(C4 + Cs) + R5C5
Discharge time constant =
CS(RS + R4 + R2) + C4(R4 + R2) + C1(R1 + R2) + C2R2.
These follow readily when it is considered that the diode is conducting during
charge, and R1 <
R2 ; the diode is non -conducting during discharge.
For 3 stages
Charge time constant =
R1(C2 + C4 + Cs + Cs) + R4(C4 + Cs + Cs) + Rs Cs + R6C6.
Discharge time constant =
Cs(Rs+ R4 + R2) + C5(R5 + R4 + R2) + C4(R4 + R2) + C,(R1 + R2) + C2R2.
If the circuit of Fig. 27.28 were considered alone the charge time constant would be
R1(C2 + C4) + R4C4 (assuming R2 >
R1), and the discharge time constant
C1(R1 + R2) + C2R2 + C4(R4 + R2). The way in which the total time constant
is built up for the more complicated cases will be readily seen, and the addition of
extra controlled stages should offer little difficulty when the new time constants are
to be determined.
For a more detailed discussion of time constants in a.v.c. circuits the reader is
referred particularly to Ref. 47.
C,
To
Controlled
Grids y
FIG. 27.380
Fig. 27.36 shows several a.v.c. curves, each corresponding to a particular condition.
In taking these curves two separate diodes were used to maintain constant transformer
loading and other conditions. Contact potential in the diode results in a slight in-
crease in the standing bias voltage on the controlled stages ; this effect has been dis-
cussed in (ii) above.
Curve A is the " no control " characteristic and is the curve which would be fol-
lowed, with the a.v.c. removed from the receiver, up to the point at which overloading
commences. This characteristic is a straight line and the slope is such that an increase
of ten times in the input voltage gives a 20 db increase in output.
Curve B is the a.v.c. characteristic for a delay of -9 volts. For inputs of 3 to
18 µV the experimental curve follows the no control line exactly, and then deviates
sharply for inputs above 18 V. From 18 to 500 000 µV the slope of the curve is
fairly constant, the output increasing by about 3.25 db for each 10 times increase in
the input voltage. Above 500 000 µV (i.e. 0.5 V) input the curve tends sharply up-
ward, indicating severe modulation rise.
Curve C is the a.v.c. characteristic for a delay of -3 volts. The a.v.c. comes into
operation at a lower input voltage, as would be expected, and the average slope is
steeper than for the higher delay voltage. In both cases, however, the " knee " of
the curve as it leaves the no control line is very clearly defined.
Curve D is the a.v.c. characteristic for a delay of zero voltage, with due compensa-
tion for the effect of contact potential on the standing bias of the controlled valves.
Curve E is typical of the characteristics obtained when audio a.v.c. is added to a
receiver. Over the range of inputs from 100 to 500 000 p.V the total rise in output
is only 3 db.
Curve B has been drawn according to the conventional method whereby the output
is adjusted to half the maximum undistorted output of the receiver for an input signal
of 1 volt. Curves C and D were then taken directly, without any further adjustment
to the volume control. Owing to a slight effect on the gain of the receiver when the
delay voltage is varied, Curves C and D fall slightly below the datum line at an input
of 1 volt. Curve E has been drawn to correspond to Curve C, since both have the
same delay voltage. The volume control, however, was advanced considerably for
Curve E. It should be noted that no conclusions should be drawn from the relative
vertical positions of a.v.c. characteristics drawn according to the conventional method
since the volume control settings are unknown.
3. The power output corresponding to any selected input voltage and any position
of the volume control may be obtained. For example, with a delay of -3 volts
(curve B1) the delay voltage is overcome at an output level of slightly less than 2 watts
with the volume control set at maximum. As a further example (again assuming
400 c/s and 30% modulation) take the same curve at an input of 1000 µV, where the
output is shown as approximately 10 watts with the volume control at maximum.
The setting of the volume control to give 4 watts output and using 100% modulation
instead of 30% is
V(30/100)3 x (4/10) = 0.11
of the maximum resistance of the control. The actual amount of rotation will depend
on whether the volume control resistance is related to rotation by a linear or a logarith-
mic law.
4. The voltage at the detector may be calculated from a knowledge of the a -f gain
and the detection efficiency. The a.v.c. bias voltage can also be determined in-
directly by calculation from the data obtained.
5. The slope of the initial part of the curve can show up an under- biased valve in
the receiver. Cheap receivers frequently use a common source of bias for two or
more valves, and under these conditions it can happen that one of the valves has too
little bias and gives less than its maximum gain when there is no signal input to the
receiver. As the input is increased, a.v.c. is applied to each controlled valve and this
gives a comparatively rapid increase in gain from the under- biased valve. Such an
increase shows up as an early section of the a.v.c. curve with a slope more vertical
than it would otherwise be- usually with a slope in excess of 6 db per octave-and
with a noticeable bend towards the horizontal at the point at which the valve receives
the bias required for maximum sensitivity.
To operate a valve under these conditions is of course undesirable, and in use the
receiver will emphasize the fading of any signals which fade through the range of
signal inputs over which the effect operates.
6. The signal -to-noise ratio for any input can be read directly from the graph, as
mentioned previously. In addition, the increments of signal -to-noise ratio with
increasing input can be checked, An ideal receiver would give a 20 db improvement
of signal-to -noise ratio for each 20 db increase in input, and it is possible for a normal
receiver to approach this value, at least over the first decades of the a.v.c. graph.
Any significant departure from the ideal can usually be traced to the application of
a.v.c. to the first valve in the receiver [see Chapter 35 Sect. 3(i)] and care should be
taken to keep the signal-to -noise ratio improving as rapidly as possible until it has
reached a value of at least 45 db.
Scroggie's method makes possible the measurement of ratios as great as this, or
much greater, because although one reading, the power output, may be beyond the
range of the output meter used, the volume control of the receiver is in effect used as
a multiplier.
7 The noise curve of the receiver might be expected to decrease indefinitely with
increasing a.v.c. voltage, but there are two reasons why it may not do so. The first is
that although the object of the test is to measure the signal -to-noise ratio of the re-
ceiver, in practice the signal -to -noise ratio of both the signal generator and the re-
ceiver are being measured. A 50 db ratio between 30% modulation and residual
noise with no modulation is a representative figure for a signal generator of reasonable
quality, and it is generator noise which causes the noise curve in Fig. 27.37 to become
approximately horizontal shortly after 1000 µV input.
The second reason for the noise curve not decreasing indefinitely is that in most
receivers in which power supply filtering has been kept to a minimum, for reasons
of economy, there is some modulation hum with high inputs. This hum shows up
as an increase in the level of the noise curve, and by plotting the curve over the full
range of inputs expected to be applied to the receiver the designer can assure himself
that the level of modulation hum does not become objectionable. An increase in
modulation hum can be seen in Fig. 27.37 as the input approaches 1 volt.
1120 (xiv) IMPROVED FORM OF A.V.C. CHARACTERISTIC 27.3
The a.v.c. characteristics in Fig. 27.37 can be taken as typical of curves taken by
this method. B1 and B2 show the signal and noise curves respectively for a 5 valve
receiver without audio a.v.c. and C1 and C2 show the results for the same receiver
with audio a.v.c. It will be seen that the delay is overcome at about 15 µV input
and that at say 0.1 volt input there is 811- db of audio control by the a.v.c. The noise
ratio (B1 - B2) should be identical with the ratio (C1 - C2) since the only difference
between the B curve and the C curves is that due to modified audio gain. Thus
the curve C2 would become approximately horizontal at 2000 µV input and at -25 db.
This method enables greater accuracy to be obtained for very small input signals
since the power output reading will be well up on the scale of a typical output meter.
With the conventional method the output is too small to measure accurately with a
standard type of output meter.
At extremely low input voltage levels the noise fluctuations make accurate output
measurements very difficult to carry out.
bias required if the detector output voltage is not to change by more than 6 db for a
change in signal input to the aerial stage of 50 db.
Assume first that the a.v.c. and signal detectors are linear, that equal signal voltages
are applied to the detector and a.v.c. diodes, and that the detection efficiency (71) is
unity. Since stage gain varies with change in mutual conductance (g,,,) it can be taken,
with typical voltage amplifier valves, that the gain alters by about 1.5 db for each ad-
ditional volt of negative grid bias (this can be determined with greater accuracy from
similar methods to those given later). Since there are three controlled stages using
similar valves, the total gain decrease is 4.5 db per volt of bias. (Suppose similar
valves are not used, one valve giving a gain change of 2 db per volt of bias and two
others giving 1.5 db per volt, then obviously 1 volt of a.v.c. bias change alters the re-
ceiver gain by 5 db).
The total decrease in receiver gain which is required equals 50 -6 = 44 db. To
achieve this the additional a.v.c. bias needed is 44/4.5 = 9.8 volts. That this is also
equal to the delay bias required for the stated conditions, can be appreciated because,
at the threshold of a.v.c. operation, the voltage on the detector diode will be 9.8 volts
peak, and for a further 9.8 volts (peak) increase in signal input the total voltage applied
to the detector is 19.6 volts giving an increase in a -f output voltage of 6 db. The
effect of 19.6 volts (peak) signal applied to the a.v.c. diode is to produce the required
9.8 volts d.c. bias since half the peak input voltage is required to overcome the delay
voltage.
To illustrate a further point, suppose it had been taken that an 80 db input change
should only give 6 db output voltage change. Then the delay bias required would be
(80- 6)/4.5 = 16.4 volts. This also means that the stage preceding the detector
must deliver 32.8 volts peak when the carrier is constant. During modulation of the
carrier the peak voltage delivered by the last i -f stage can reach 65.6 volts, and if
distortion is to be avoided the valve preceding the detector must be capable of de-
livering this amount of output voltage when the total bias is 16.4 volts plus the stand-
ing bias. Whether this is possible can be determined from curves of the signal
handling capability of the valve in question, or by direct measurements using a mod-
ulated carrier. It will be found, usually, that the maximum output voltage which can
be handled without severe distortion is very much reduced when the grid bias voltage
is large. This then suggests that reducing the a.v.c. bias on the last i -f amplifier
valve will allow the permissible plate voltage swing to be increased. Reducing the
bias voltage also results in the stage gain being increased, and so the same a.v.c. bias
voltage is produced with a smaller signal input voltage. However, the amount of
total control is reduced, and so partial or zero a.v.c. bias on the last i -f valve is not
necessarily a satisfactory solution in every case.
It should be observed that the last i -f stage is not necessarily the one in which over-
loading will occur first. In Chapter 26, when discussing the crystal filter stage,
it was pointed out that the grid to plate gain of the valve preceding the filter was very
high with the arrangement shown. Suppose the grid to plate gain of the first valve
is 400 times, but the gain from the grid of the first to the grid of the second valve
(into which the filter feeds) is 5 times, then if the grid to plate gain of the second valve
is less than 80 times it becomes immediately apparent that the first i -f valve will be
overloaded before the second one. Extreme cases of this type are not likely to occur,
but they present possibilities which should not be overlooked.
(2) A detailed design for a single controlled stage is to be carried out from valve
data. Simple a.v.c. is to be used.
The valve type selected for this example is the 6SK7 remote cut -off pentode.
The average characteristics are shown in Fig. 27.39. The standing bias is -3 volts
and the mutual conductance (g,,,) for this condition is 2000 micromhos. For a
total grid bias of -35 volts the g,,, is 10 micromhos. If cathode bias were used some
additional error would be introduced, although this is usually neglected.
The first step is to convert the g,,,- Ep curve of Fig. 27.39 to a curve showing the
change in gain of the controlled stage with variation in grid bias. The actual value
of stage gain is not required since it is assumed to be directly proportional to g,,,.
1122 (xv) DESIGN METHODS 27.3
The change in g,, is expressed in decibels, taking as the reference level (0 db) the
value of g
for maximum gain, i.e. the g,n with only the standing bias ( -3 V) applied
to the grid. In this case the gm used for zero reference level is 2000 micromhos.
Then using
= 3 V) -20log,o2000
Change in gain (db) _ -20 logea gm(Erl
gm gm
the following values are obtained (the values for g, are taken from Fig. 27.39).
III
Et -6.3 VOLTS
IL á
2500
Total Grid
Bias* g,
Change in
Gain (db)
SUPPRESSOR VOLTS O
J
PLATE VOLTS -250
-3 2000 0
SCREEN VOLTS.- 100
-4 1750 - 1.15
-5 1438 - 2.86
2000
-6 1200 - 4.44
N
g -7 925 - 6.69
4 0a -8 700 - 9.13
W V -9 520 -11.7
<
É
-10 383 -14.4
á
--12.5 225 -19
is moo ú -15 138 -23.2
-17.5 90 -26.9
u
ó
-20 70 -29.1
ó -22.5 50 -32
u
i -25 40 --34
á -27.5 30 -36.5
10 1000
o-
-30 25 -38
-32.5 18 -41
-35 10 -46
* _ - (3 + a.v.c. bias).
These results are plotted in Fig. 27.40 as
s soo curve 1. It might be noted that the aver-
age slope is 1.4 to 1.5 db per volt.
It is next required to determine the
amount of a.v.c. bias voltage developed by
the diode circuit, for various input voltages
applied to the diode. In a typical case the
d.c. diode load resistance would be 1 meg-
-So -40 -30 -20
CONTROL-GRID
-10
VOLTS
o
° ohm, and the d.c. voltage developed for
various signal input voltages can be taken
FIG. 27.39 65K7 Mutu61 Characteristics
directly from Fig. 27.6. The developed
d.c. voltages are tabulated in column (2) below against r.m.s. signal (E,m,) applied
to the diode.
- 0-50
240 -30 -20 0
Total G Id Bio Voltage
Curvet of °Variation in Stage Gain and
O Diode Input with Changing
FIG. 77.40 Grid Bias
It is now necessary to relate the total bias (a.v.c. + standing bias) to the change
in diode input voltage ; the change is conveniently expressed in decibels. Any signal
voltage applied to the diode may be taken as zero reference level (0 db), and in our
case Erm, = 2 volts will be convenient. Then the change in diode input voltage is
given by,
db change in diode input = 20 logo
Erm, (for total bias = -5.2 V)
Erm.
= 20 logo
2
The results are tabulated in column (4) of the table given above, and the total bias
versus diode signal input change is plotted as curve (2) in Fig. 27.40.
Finally, it is required to find the way in which the output changes with changes in
signal input to the amplifier. If we assume various values for the change in signal
voltage applied to the diode, it is a simple matter to determine the total grid bias, and
from this the change in stage gain, by using curves (2) and (1) Fig. 27.40. Also from
the curves, it is seen that, with zero change in diode input (0 db), the amplifier stage
gain is 3 db down on the maximum possible gain. (This is because we selected
Erm, = 2 volts as being the zero reference level). If now we assume, say, 2 db change
in diode input, the actual change in stage gain is -4.3 -
( -3) = -1.3 db. Because
voltage input applied to the diode has increased 2 db, and the amplifier stage gain has
decreased 1.3 db, it follows that the signal voltage applied to the amplifier grid must
have increased 2 + 1.3 = 3.3 db.
The direct output voltage change across the detector load, for 3.3 db change in sig-
nal input, is from -2.2 volts to -2.9 (i.e. from curve (1), Fig. 27.40, the total bias vol-
tage change is from -5.2 to -5.9 volts, which correspond respectively to -2.2 volts
and -2.9 volts across the diode load, when the standing bias is -3 volts). Ex-
pressing the direct output voltage change in decibels, we have
a.v.c. bias
Output change (db) = 20 logo
a.v.c. bias for Erm, = 2 V
a.v.c. bias
= 20 logo
-2 2
1124 (xv) DESIGN METHODS 27.3
From this we have for 3.3 db change in signal input voltage (2 db change in diode
input) a corresponding output change of 2.4 db. The complete a.v.c. characteristic
is tabulated below and plotted in Fig. 27.41.
0
0 10 20 30 40 50 60 70
db Change In Signal Input(CakuWted for hilmua
FIG. 27.41 Dlode Input Enns .2 Veltss)
Fig. 27.41. Calculated a.v.c. characteristic for a single controlled stage using type 6SK7.
The a.v.c. characteristic for diode input voltages smaller than Erm, = 2 volts can
be readily plotted, but as the useful range of signal input voltages, where control is
mainly required, is usually greater than this value in a reasonably sensitive receiver
(and because of the limitations of the diode curves) this value has been taken as afford-
ing a satisfactory example.
27.3 (xv) DESIGN METHODS 1125
The variation in gain given by simple a.v.c. is limited, and for very strong signals
a local -distance switch or some other method to prevent overloading of the receiver
is required.
(3) The voltage amplifier and diode detector used for the previous example are to
be used in a delayed a v.c. circuit, in which the delay bias applied to the diode is
-10 volts.
If they are not available, the diode characteristics relating r.m.s. input voltage to
d.c. output voltage, for various delay bias voltages, can be measured directly. The
rectified voltage, used for a.v.c. bias, is not completely independent of modulation
when a delay bias is used, and there is an increase in the available d.c. when the per-
centage modulation is increased. Usually the a.v.c. comes into operation for lower
carrier input voltages than would normally be expected, particularly when the per-
centage of modulation approaches 100 %.
The complete design procedure now follows that given previously for simple a.v.c.
using the data derived by the method given in the last paragraph. It should be clear
that curve (1) in Fig. 27.40 remains unchanged, but curve (2) must be replotted. The
resultant data are then available for plotting the a.v.c. characteristic. For signal
input voltages of which the peak value does not exceed the diode delay bias, the re-
ceiver operation is the same as for one which does not incorporate a.v.c. (this is clearly
shown in Fig. 27.36).
The slope of the no control line is readily determined since it is taken that the output
is directly proportional to the input.
It should be apparent that there will be a range of signal input voltages around the
threshold point over which the audio distortion will be severe.
c,
A,F
B+
AMC.
B+
A.M.C.
B+ To
A.F. Amplifier
FIG. 27.43 Biased Detector Muti g System
wite
0 C Amplified e Delayed A.V.C.
bias voltage only exists when the diode D2 is conducting, which occurs when the
cathode of V, is positive. As soon as the cathode of V, becomes negative, the a.v.c.
comes into operation, the diode 132 no longer conducts, and the negative bias is re-
moved from the plate of the detection diode D, which then begins to function in the
usual manner. Part of the detected voltage output will appear across Rk, Ck but
this is not serious if the value of R k is very much less than the resistance of R, (say
about one tenth or less). It is recommended that CkRk be made equal to C1R,.
(b) Inoperative audio amplifier
A typical arrangement of this type (Ref. 57) is shown in Fig. 27.44. V, is any double
diode, such as type 6H6, and 172 is a pentagrid mixer, such as the type 6L7, in which
27.4 MUTING -(ii) TYPICAL CIRCUITS 1127
grid 1 has a remote cut -off and grid 3 a sharp cut -off characteristic. Some systems
use two valves in place of V2 to effect muting and audio frequency amplification.
Improved systems are also available (Ref. 54) which use a duo -diode pentode to elimin-
ate the separate double diode valve, the pentode section being used as the intermediate
frequency voltage amplifier. Audio a.v.c. is readily applied in this latter case.
From Fig. 27.44 it is readily seen that the diode detector and simple a.v.c. arrange-
ments are quite conventional. The audio output from the detector is applied to grid 1
of the valve V,. Only part of the cathode bias is applied to grid 1 by using a tap on
the cathode resistor. In the absence of conduction through the diode D2 there is a
large negative bias applied to grid 3 due to the total cathode bias voltage developed
across R, and R8 in series. The d.c. path to grid 3 is via Rs, R5 and Re. Re has a
high value of resistance to prevent the plate of D2 from being connected to ground
through the comparatively low impedance of the cathode circuit of V, (which includes
the muting control as shown). The circuit made up from Re, C, and C, is merely
a filter to allow the application of direct bias voltage only to the muting grid (3) of
V, ; however, R, also performs the useful function of stabilizing the voltage applied
to grid 3 in the event of the anti -muting bias being sufficiently large to cause grid
current. R5 is the diode load resistor for D2, and C5 allows the application of i -f
voltage to the plate of the diode D,. When a signal is received the voltage developed
across R5 is sufficiently large, and of the correct polarity, to reduce the total negative
bias on grid 3 and so allow the valve V, to perform the function of voltage amplifier
for a -f voltages applied to grid 1.
An interesting arrangement (Ref. 54) is shown in Fig. 27.45, in which a duplex -
diode pentode and its associated circuit combine the functions of muting, detection,
conventional and audio a.v.c., and audio frequency amplification. The valve V, is
a duo -diode pentode in which grid 1 has a remote cut -off characteristic (the circuit
was developed around the Australian-made type 6G8 -G). The operation of this
circuit is somewhat different from those previously discussed. The maximum gain
obtainable from a resistance -capacitance coupled audio frequency amplifier occurs
when the negative grid bias has a particular value. For a negative bias less than the
optimum value (and for a particular screen voltage, which is generally made fairly
low), g, will drop and the stage will effectively be muted. When the negative bias
is made larger than the optimum value the gain of the stage will decrease at a com-
paratively slow rate. This principle is used in the circuit of Fig. 27.45 and the signal,
muting, and audio a.v.c. voltages are all applied simultaneously to the control grid.
The cathode resistors R1, and R. are used to set the bias voltage to give the optimum
value for maximum gain. Rt, is then adjusted to a lower value to reduce the total
bias voltage, and to cause the gain to drop to such a low value that muting has effec-
tively occurred. It is necessary to select the values of bias and screen voltage so that
the maximum screen dissipation is not exceeded. When a signal voltage is applied
to the a.v.c. diode, the rectified voltage developed across R, increases the grid bias
1128 MUTING-(ii) TYPICAL CIRCUITS 27.4
in a negative direction which increases g,,, (and so the stage gain) and the valve un-
mutes. For further increases in signal voltage the a.v.c. comes into operation, and
the gain of the audio stage (and also the other controlled stages) begins to fall off
in the usual manner. It is seen that only part of the total a.v.c. bias is applied to the
grid of
A.V.C.
FIG. 27.45 Single Vaho Arrangement to Pronide
Muting,Auaw AV.C., Detection and AYAmplification
2200011 1_ 47,00011
Rs
10 00011
B+
FIG. 27.46(A) e+
Muting System for F -M Receiver
27.4 (iii) CIRCUITS USED WITH F-M RECEIVERS 1129
When a signal voltage is impressed on the limiter V, the plate current is reduced,
and so the positive voltage applied to the cathode of triode (1) (V4) is increased. When
the signal input is sufficiently large the positive cathode voltage will exceed the positive
voltage on grid (1) by an amount which is sufficient to cut off the plate current. This
puts the points A and B at the same potential and so removes the negative bias from
the grid (2). Triode (2) of V4 then acts as an ordinary a -f voltage amplifier, and its
normal bias is obtained from the voltage drop across the cathode resistor Ra.
Limiter Discriminator 1st Limiter 2nd Limiter Discriminator
C, 10,000 n
SOMMF
zs,000 2,200
n MMF MMF
R 0IMfL
+
Relay
SOOn_ Muting
2,20034JF 10000 150v
R4
8+
R3
10M11. 025Mn I s,000 n
Rs C, Muting
0.5Mn O0I Control
MF
soon
FIG. 27.46(8) FIG. 27.46(C)
Fig. 27.46(B) shows a simple but effective arrangement for muting. Noise voltages,
in the abscene of a signal, which appear in the screen circuit of the second limiter
valve, are applied to the control grid of the triode section of the muting valve (e.g.
type 6AV6) and the amplified noise voltage is applied to the diode detector circuit.
The d.c. voltage across the diode load is applied as additional negative bias to the grid
of the a -f voltage amplifier valve and cuts off this stage. When a signal is received
the noise voltages are reduced and so the additional bias on the a -f valve falls and the
stage begins to function in the normal manner.
Another useful arrangement is shown in Fig. 27.46(C). In this circuit noise
voltages from the screen circuit of the second limiter are applied to the grid of the
muting valve (which could be a type 12AT7) via the high pass filter C, Cz R,. Triode
section (1) operates as an anode bend detector and the direct voltage developed across
R, cuts off triode section (2) ; the relay is then in the unoperated condition. When
a signal is received the noise voltages are reduced due to the receiver quieting, and
at the same time additional bias is applied to triode (1) by the rectified voltage appearing
at the first limiter grid. The voltage across R, is now insufficient to cut off triode (2)
and the relay is operated by the plate current. The relay can operate a contact to
short the grid of the output valve to earth additional contacts can also be utilized
:
should the receiver be used for special purposes. The circuit can readily be re-
arranged so that the relay is not required, and an additional control bias is made
available to cut off the a -f voltage amplifier by increasing the voltage in the cathode
circuit in the absence of a signal. The coupling network R3 R4 R5 C, has a suitably
selected time constant, and the circuit arrangement is such as to prevent amplitude
modulation or over deviation of the frequency modulated signal from muting the
receiver.
Both of the circuits of Fig. 27.46(B) and (C), or variations of them, have been widely
used in f-m mobile and V.H.F. link receivers. In some cases additional negative
bias is applied to both the a -f voltage amplifier valve and the a -f output valve ; this
has the advantage in the case of mobile receivers that the battery drain is reduced very
appreciably during stand -by periods since a large proportion of the battery drain is
due to the H.T. current drawn by the output valve.
1130 (iii) CIRCUITS USED WITH F-M RECEIVERS 27.4
(i) General
The purpose of a noise (or crash) limiter in a radio receiver is to assist in the reduction
of noise pulses, such as those due to ignition interference and crashes of static, so that
their effects may be minimized at the receiver output. The difference between muting
systems and noise limiters should be carefully distinguished, as they perform quite
different functions in connection with the type of noise which they are meant to
eliminate.
Noise limiters in general follow two trends. One group " punches a hole " in the
signal so that the receiver output is momentarily cut off. The other main group
functions by limiting the maximum output to a value which is not appreciably greater
than some pre -determined level.
It can be shown fairly readily (Ref. 63) that the best results for noise reduction can
be obtained by placing the limiter in the receiver at a point of low selectivity. The
Lamb silencer (Ref. 65) for example, is usually placed after the first i -f stage and its
operation is such as to cut off the plate current in the second i -f amplifier valve,
momentarily.
For many purposes elaborate noise limiting circuits (Refs. 65, 71) are not essential,
and the tendency in the usual communications receiver is to use simple diode limiting
to remove noise peaks. For amateur work sufficiently good results are often possible
using audio output limiters (see Refs. 66, 67, 68). With the A-M receivers used in
mobile communications systems a very elaborate noise limiter may be incorporated.
27.5 NOISE LIMITING -(i) GENERAL 1131
See also Chapter 16 Sect. 6 for speech clippers and Sect. 7 for noise peak and output
limiters.
The threshold of operation for noise limiters can be arranged for either automatic
or manual setting for different signal input voltages. Good results are possible using
automatic setting, but the more elaborate circuits practically always make some pro-
vision for manual control.
(ii) Typical circuit arrangements
The Lamb silencer can be made to give excellent results, but because it is rather
elaborate it is not extensively used in commercial communications receivers. For
a description of its operation, together with complete circuit data, the reader should
consult Ref. 65.
Fig. 27.47 shows a typical example of the " shunt " type
A
of diode limiter. It is seen that the circuit to the left of
De the dotted line AB is the usual diode detector circuit and
the component values are identical with those normally
selected. The negative direct bias voltage is set to prevent
the diode D, from conducting until the peak input voltage
a- exceeds the bias. When this occurs the diode D, acts as
Threshold a virtual short circuit, and there is practically no output
Control
from the detector circuit.
A " series " type of diode limiter is shown in Fig. 27.48.
A.F. - B
' FIG. 27.41 The diode D, is used for detection in the usual manner
Shunt Diode Noise Limiter Diode D, is biased to be conducting with normal
signal voltages applied to the detector circuit. For bursts of noise there is an in-
stantaneous negative voltage applied to the plate of D,, and when this negative voltage
exceeds the positive bias the diode stops conducting and so opens the audio output
circuit. A large amount of audio gain is required when this type of circuit is used,
because of the low audio voltage available.
These simple circuits have several disadvantages. Firstly they are both susceptible
to hum because of heater -cathode leakage when the cathode of D, is above earth.
For example this limits the size of the resistor R which can be used in the circuit
of Fig. 27.48, and so results in a loss in audio output. (An increase on the value shown
would be permissible however). A further disadvantage is that stray capacitances will
allow some of the high frequency noise components to appear in the receiver output,
even though the diodes are operating in the prescribed manner ; this can be largely
overcome with care in layout and wiring to minimize stray capacitances.
Automatic threshold control may be arranged with simple " series " and " shunt "
limiters. Suitable methods are given in the references (particularly Refs. 62A, 64 and
B+
'Series Diode Noise Limiter
FIG. 27.48
1132 (ü) TYPICAL CIRCUIT ARRANGEMENTS 27.5
70). However, a much more satisfactory circuit (Ref. 44) is shown in Fig. 27.49. The
diode D, is incorporated in the usual detector arrangement and this circuit provides a
variable bias voltage for the noise limiting diodes D2 and D,. A short burst of noise
voltage having a positive polarity will cause the diode D, to conduct and so flatten
out the audio waveform. A noise voltage having negative polarity causes D, to con-
duct which then flattens out the negative peak of the audio output. In the absence
of noise the positive and negative half cycles are equally damped and so distortion
is reduced.
Methods of noise reduction for use with reproduction from records are described
in Chapter 17 Sect. 7.
(i) Miscellaneous
A tuning indicator is a device which indicates, usually by means of a maximum
or minimum deflection, when a receiver is correctly tuned. These indicators have
taken a number of forms, particularly before the advent of the Electron Ray tuning
indicator, originally called the " Magic Eye," and a few of these will be briefly dis-
cussed before considering the Electron Ray Tube in detail.
INDICATOR LAMP
TO PLATE 6-3,&C.
SUPPLY B+
FIG. 27.50 Saturated Reactor Tuning Indicator
A further arrangement is to use a neon tube in which the length of the illuminated
column is proportional to the d.c. voltage obtained from a resistance connected in
series with the plate of an amplifier valve controlled by a.v.c.
of the tuning indicator is connected through a 2 megohm resistor to the detector diode
circuit. With the switch in the to a.v.c. " position the indicator is connected directly
to the a.v.c. line. This shows the two alternative connections just discussed. The
cathode of the tuning indicator is returned to a suitable tapping point on the cathode
bias resistor of the power valve as discussed below.
DE T.,...0"
OSyFm 05µF
J_- TUNING INDICATOR
Methods for Application of the Magic Eye Tuning Indicator
FIG. 27.51
1134 (ii) ELECTRON RAY TUNING INDICATORS 27.6
The cathode of the Electron Ray tube should be, as closely as possible, at the same
potential as the cathode of the diode. If its cathode is more negative than that of
the diode, grid current may occur thereby increasing the initial bias on the controlled
stages and reducing the sensitivity of the receiver. Consequently if the diode cathode
is earthed, the indicator tube cathode should also be earthed, but if the diode cathode
is positive then the indicator tube cathode should also be positive by an approxi-
mately equal amount. One satisfactory method of obtaining this positive voltage,
which however may only be used with a Class A power valve, is to connect the cathode
of the Magic Eye to a tapping on the cathode bias resistor of the power valve. With
this arrangement it is advisable for the tapping to be adjusted to make the cathode
of the Electron Ray tube about 0.5 volt less positive than that of the diode in order to
allow for contact potential in the indicator tube, the " delay " due to this small voltage
being negligible.
Alternatively, the cathode return of the Magic Eye may be taken to a suitable tap-
ping point on a voltage divider across the " B " supply. Due to the fairly heavy and
variable cathode currents drawn by the older types of indicator valves it is essential
that the voltage obtained from any voltage divider, or other source of voltage, should
not be affected appreciably by a current drain of from 0 to 8 mA. It is for this reason
that it is not satisfactory to tie the cathode of the indicator tube to that of the diode.
With the newer " space charge grid " construction the cathode currents remain more
nearly constant throughout life, and this allowance for change of current need not be
made. However, it is not advisable to base calculations for cathode bias resistors
on the published values of cathode currents since in some cases with valves of the newer
construction, these are higher than the average currents.
IMn
R +R,. 16,700.11
FIG. 27.52 CIrcu t for Wide -Angle Tuning
Overlapping of the two images is possible on very strong signals, whatever type
of indicator tube is selected, but it is generally found with a remote cut-off type (such
as the 6U5/6G5) that this is not often experienced under field conditions. Certain
arrangements have been devised to reduce the tendency to overlapping, but none is
free from criticism. Desensitization of the tuning indicator is readily applied, but
affects indications on weak signals. The use of two separate tuning indicators, or
a single type 6AF6 -G with two separate amplifiers, one for weak and one for strong
signals, is excellent but expensive. If the grid of the indicator tube is excited from
the moving contact of the volume control the deflection will depend upon the setting
of the control, and " silent tuning " will not be possible.
Wide Angle Tuning with a maximum angle of 180° is practicable if an external
triode amplifier is added (see Ref. 73). With this circuit (Fig. 27.52) the edges of
the pattern are sharp from 0° to about 150° to 180 °.
applied between grid and a cathode. Any suitable pre -amplifying stage may be used
to increase the sensitivity of the device if desired. When an a.c. voltage is applied
the sharp image will change to a blurred half-tone and as the null point is reached the
image will again become sharp. A heavy overload may cause overlapping of the
pattern, but this is not detrimental to the tube.
yF
eI F
R Mit
6Ú5/6G5
B+250V
104F IHF
6U516G5
Tuning InäB cator
02Ií'
FIG. 27.54 Tuning Indicator Operating from Discriminator Output
This latter feature is not available with similar arrangements operating from the dis-
criminator output alone.
Tuning aids are sometimes used in receivers with ratio detectors. The simplest
arrangement in this case is to take the control bias voltage from the a.v.c. take -off
point, or to use whatever fraction of this total voltage that is thought to be necessary
for operating the tuning indicator.
Voltage
There is little change in rectifier characteristics from 15 °C to 50 °C, but above this
temperature up to 100 °C, both forward and back slope resistances decrease slowly.
The characteristics of germanium crystal diodes vary considerably, depending on
the application and the manufacturer, but the following data are representative of
the majority of types.
Allowable direct current 20 to 60 mA.
Allowable surge current 100 to 600 mA.
Allowable reverse voltage 25 to 250 volts
Turnover voltage (reverse voltage for zero dynamic resistance) 40 to 275 volts
Minimum current with +1 volt 2.5 to 15 mA
(some are less than
2.5 mA).
Reverse current with -50 volts applied 40 to 1660 µA
(this only applies to types having an allowable reverse
voltage of 50 volts)
Capacitance 0.8 to 3.0 µµF.
Maximum ambient temperature +70° to +85 °C.
Minimum ambient temperature -40° to -55 °C.
Maximum frequency up to several hundred
Mc /s.
References to germanium detectors : 78, 79, 80, 81, 83, 84, 85, 85A, 85B.
(v) Transistors
Transistors are crystal devices with three or more electrodes, which are capable
of amplifying. A good introductory article is Ref. 86. See also Refs. 87, 88, 89, 90,
91, 92, 93, 94, 95, 96.
SECTION 8 : REFERENCES
(A) A-M DETECTORS
(1) Sturley, K. R. (book) " Radio Receiver Design " Part 1 (Chapter 8), Chapman and Hall, London,
1943 ; Wiley and Sons, New York, 1943.
(2) Everitt, W. L. (book) " Communication Engineering " (Chapters 13 and 14) McGraw -Hill, New
York and London, 1937.
(3) Zepler, E. E. (book) " Technique of Radio Design " (Chapters 4 and 11) Chapman and Hall,
London, 1943 Wiley and Sons, New York, 1943.
(4) Preisman, A. (book) " Graphical Construction for Vacuum Tube Circuits " (Chapter 6) McGraw -
Hill, New York and London, 1943.
(5) Terman, F. E. " Radio Engineer's Handbook " (Section 7) McGraw-Hill, New York and London ,
1943.
(6) " Radiotron Receiver RC52 " Radiotronics No. 117 Jan. /Feb. 1946.
(7) "Radiotron Receivers RC41 and RC42" Radiotronics No. 118 Mar. /April 1946.
(8) " Distortion in diode detectors " Radiotronics No. 73 (24th Feb. 1937) 11 ; Radiotronics No. 74
(31st March 1937) 20.
(9) Varrell, J. E. " Distortionless detection " W.W. 45.5 (3rd Aug., 1939) 94.
(10) Amos, S. W. " The mechanism of leaky grid detection " Electronic Eng. (1) 17.198 (Aug. 1944)
104 ; (2) 17.199 (Sept. 1944) 158.
(11) " New linear detector with high input impedance " A.R.T.S. and P. Bulletin No. 30 (22nd July
1936).
(12) " Reflex detector application " A.R.T.S. and P. Bulletin No. 32 (21st Sept. 1936).
(13) Healey, C. P., and H. A. Ross, " The linear reflex detector" A.W.A. Tec. Rev. 3.1 (July 1937) 1.
(14) Corfield, D. N. and R. V. Cundy " Valve Technique " Incorporated Radio Society of Great Britain,
London, April 1948.
(15) Weeden, W. N. " New detector circuit " W.W. 40.1 (1st Jan. 1937) 6.
(16) Robinson, H. A. " Regenerative detectors " Q.S.T. 17.2 (Feb. 1933) 26.
(17) " Reaction circuits " W.W. 46.2 (Dec. 1939) 40.
(18) Aider, L. S. B. " Threshold howl in reaction receivers " Exp. Wireless and Wireless Engineer
7.79 (April 1930) 197.
(19) Zepler, E. E. " Oscillation hysteresis in grid detectors " W.E. 23.275 (Aug. 1946) 222.
(20) " Radio Amateur's Handbook " (Chapter 7, p. 146) 1947 Edition, A.R.R.L., Connecticut.
(21) " Cathode Ray," " Super- regenerative receivers " W.W. 52.6 (June 1946) 176.
(22) Stockman, H. " Superregenerative circuit applications " Elect. 21.2 (Feb. 1948) 81.
(23) Frink, F. W. " The basic principles of superregenerative reception " Proc. I.R.E. 26.1 (Jan. 1938)
76.
(24) Easton, A. " Superregenerative detector selectivity " Elect. 19.3 (March 1946) 154.
(25) Bradley, W. E. " Superregenerative detection theory " Elect. 21.9 (Sept. 1948) 96.
(26) Hazeltine, Richman and Loughlin " Superregenerator design " Elect. 21.9 (Sept. 1948) 99
(B) F -M DETECTORS
See also bibliography covering A.F.C. given in Chapter 29 for further discriminator circuit data.
(27) Sturley, K. R. " The Phase Discriminator -its use as a frequency-amplitude converter for F -M
reception " W.E. 21.245 (Feb. 1944) 72.
(28) Sturley, K. R. (book) " Radio Receiver Design " Part 2, Chapters 13 and 15, Chapman and Hall,
London, 1945 ; Wiley and Sons, New York, 1945.
(29) Hund, A. (book) " Frequency Modulation " Chapters 2 and 4, McGraw-Hill Book Co., New York
and London, 1942.
(30) Lampitt, R. A. " The frequency discriminator " Elec. Eng. 17.200 (Oct. 1944) 196.
(31) Tibbs, C. " Frequency Modulation
(May 1943) 140.
5-
demodulation : theory of the discriminator " W.W. 49.5
(32) Maynard, J. E. " Coupled circuit design " (covers phase discriminator circuit) Comm. 25.1 (Jan.
1945) 38.
(33) Arguimbau, L. B. " Discriminator linearity " Elect. 18.3 (March 1945) 142.
(34) Seeley, S. W., and J. Avina, " The Ratio Detector " R.C.A. Rev. 8.2 (June 1947) 201.
(35) " Ratio Detectors for F -M receivers " Radiotronics No. 120 (July /Aug. 1946) 79.
(36) " Balanced phase shift discriminators " Radiotronics No. 120 (July /Aug. 1946) 83.
(37) Sandel, B. " F -M receiver for the 88 -108 Mc /s Band " Radiotronics 125 (May /June 1947) 39 ;
Radiotronics 127 (Sept. /Oct. 1947) 79. (Correction, Radiotronics 132 (July /Aug. 1948).)
(38) Maurice, D. and R. J. H. Slaughter " F.M. reception -comparison tests between phase discriminator
and ratio detectors" W.W. 54.3 (March 1948) 103.
(39) Kiver, M. S. (book) " F-M Simplified " Chapters 7 and 13, D. Van Nostrand Co. Inc., Toronto,
New York and London, 1947.
(40) Bradley, W. E. " Single -stage F -M detector " Elect. 19.10 (Oct. 1946) 88.
(41) Loughlin, B. D. " Performance characteristics of F-M detector systems " Tele-Tech, 7.1 (Jan.
1948) 30.
(42)" F -M detector systems " A.R.T.S. and P. Bulletin No. 139 (Reprint Hazeltine Bulletin and covers
practically the same data as Ref. 41.)
(42A) Loughhn, B. D. "The theory of amplitude - modulation rejection in the ratio detector" Proc.
I.R.E. 40.3 (March 1952) 289.
27.8 REFERENCES 1139
(i) Description
A reflex amplifier is one which is used to amplify at two frequencies -usually
intermediate and audio frequencies.
Reflex radio and audio frequency stages have also been used in very small T.R.F.
receivers (see Sect. 5).
Reflex receivers were fairly common in U.S.A. and Australia in the period 1934
to 1937, but suffered from serious distortion and high play-through. These re-
ceivers usually had high a -f plate load resistors giving comparatively high a -f gain,
low i -f gain and low operating plate voltages and currents. A.V.C. was not applied
to the reflex stage which was frequently a sharp- cut -off valve such as the 6B7 or
6B8. A considerable advance was made in Australia by the use of the remote cut -off
type 6G8-G with a.v.c., and a further advance was made in the adoption of a com-
paratively low a -f plate load resistor. The latter enabled nearly full i -f gain to be
obtained with normal voltages on the electrodes, thus increasing the maximum plate
voltage swing for a limited distortion. The most common application of a reflex
amplifier in Australia* at the present time is in a 3/4 valve receiver comprising con-
verter, reflex stage, power amplifier and rectifier. The reflex stage in such a re-
ceiver amplifies at intermediate frequency, provides detection and a.v.c. from its
diode or diodes, and then amplifies at audio frequency. It may be compared with
a straight receiver in which the second valve is used as i -f amplifier and detector,
the output from the detector being used to excite the power stage. Using the same
valves and components in both cases, the reflex receiver may have a sensitivity up
to 10 times that of the straight set.
1140
28.1 (ii) ADVANTAGES & DISADVANTAGES OF REFLEX RECEIVERS 1141
*A 3/4 valve receiver is one having three amplifying valves and a rectifier.
tFor 50 milliwatts output.
1142 (i) PLAY-THROUGH 28.2
(ii) Overloading
It is practicable to designa reflex receiver to handle input voltages up to 1 volt
without serious distortion (of the order of 10% at 100% modulation).
FIG. 28.1
To Heater, L Dial Lamp
Fig. 28.1. Reflex superhet. receiver with plate reflexing and fractional a.v.c. on both
controlled stages (A.W.A. Radiola Model 517 -M).
LOAD - 10,00011
6SA7 -GT 6SF7 200
)i.nr I.
C ¡---
bp 111
bT MS
OI
rHF HF
OSHF 20.00011
IHF 700055
25M55 2W
1500055;
vVvVr
IM11
6%5-GT jiii6H 35011
200/200
BHF 8HF
SOmA
-25 M55
FIG. 28.2
25055 6355
Fig. 28.2. Reflex superhet. receiver with plate reflexing, full a.v.c. on converter and
fractional a.v.c. on reflex stage (Ref. 1).
28.3 DESIGN OF PLATE REFLEX SUPERHET. RECEIVERS 1145
resistor was selected to provide optimum operation together with minimum hum ;
a certain amount of hum neutralization is possible between the reflex (a -f) and power
amplifier stages. Feedback in the reflex stage is neutralized by the very small fixed
capacitances C.ß1 and Cß.2, while a small degree of negative feedback is provided on
the output stage.
Considerably improved performance is obtainable by the use of type 6AR7-GT
in the reflex stage, together with high Q i -f transformers. This avoids the necessity
for neutralization and gives higher gain. Still higher gain is obtainable with a high-
slope output valve, although the play- through is increased thereby. This circuit
arrangement has considerable merits and there are prospects that, by suitable choice
of valve types and a.v.c. design, the play-through may be reduced to a low level.
A possible alternative form of the circuit would be to use type 6BE6 converter,
6BA6 reflex amplifier and a duo-diode- output pentode.
4ppF
- 010F 001pF
IpF _ 701IP ppF
50pTiF' i 3pp00 50000n 'IpFZ
,'2Mn 006pF 5,00011
5M11 2,500n
ÍsoPono
2 M .oSpFj'
440ppF
I Mn
Ì
Soon
I W. WV/.
íTIpF 25 Mn
loon
), 2 FIG. 28.3
Fig. 28.3. Reflex superhet. receiver with screen reflexing (Ref. 4).
1146 (i) SCREEN REFLEX RECEIVERS 28.4
through. The aerial sensitivity is approximately 5µV for 50 mW output, and the
signal to noise ratio is 10 to 13.5 db at 5µV input. The distortion is 6.9% with an
input of 1 volt, 30% modulated, but 12% with an input of 0.5 volt, 100% modulated.
An output of 0.5 watt is obtained with 20 µV input, 30% modulated. With an input
of 1 millivolt to the aerial terminal, 30% modulated at 400 c /s, an output of 2 watts
is obtained with 10% overall distortion.
Feedback is applied by returning the lower end of the volume control to the voice
coil. Maximum bass boosting is provided when the volume control is at the tap,
which is adjusted for the lowest listening level. The high frequency peak and bass
boosting are reduced as the volume control setting is increased towards maximum.
The response above 5000 c/s is cut sharply by a combination of shunt capacitances
together with negative feedback. The overall result is an automatic high frequency
tone control which gives a " mellow " tone with weak signals and a normal radio
tone on strong stations. The gain reduction due to feedback is 14 db with 1 mV
signal input and with the volume control at its minimum position. The gain re-
duction decreases as the volume control is advanced.
The conventional i -f filter in a reflex circuit is a series resistor from the volume con-
trol slider, connected to a by -pass capacitor at the cold end of the first i -f transformer
secondary. These two components give phase shift at some high frequency with
the volume control at its minimum setting, additional to the phase shifts already in
the circuit and may thus cause oscillation when heavy feedback is applied with the
volume control at zero. However if the resistor is wired in series with the hot end
of the volume control, as in Fig. 28.3, with the volume control at its zero setting there
is no series resistor in the feedback path, phase shift is reduced and high frequency
peaks (or oscillation) are avoided.
Resistance -capacitance filtering is used here for economy. The plate supply
voltage is 185 volts. One ninth of the a.v.c. voltage is applied to the reflex stage.
The play -through is as low as in some commercial receivers that do not use reflexing.
(ii) Comparison between plate and screen reflexing
Plate reflexing is capable of higher a -f gain than screen reflexing, since the latter
is limited by the " triode mu " of the reflex valve. For this reason, types 6AR7 -GT
and 6BA6 with grid -to- screen mu factors of 18 and 20 respectively are quite suitable
for use in screen reflex receivers. However, plate reflexing gives more play-through
than screen reflexing for similar conditions in both cases.
Type 6AR7-GT as a screen reflex amplifier allows a higher gain in the output
stage than a lower gain valve (e.g. 6G8 -G) with plate reflexing, for the same play -
through performance in both cases, and similar converter conditions.
SECTION 1 : LIMITERS
(i) General (ii) Typical circuits for F-M receivers (a) Single stage limiter (b) Cas-
caded limiters.
(i) General
In this section it is proposed to discuss only the conventional type of amplitude
limiter using a pentode valve operating as a saturated amplifier. Other circuits,
which combine the dual functions of detection and the removal of amplitude variations,
such as the ratio detector and locked oscillator, will not be treated here. The ratio
detector has been discussed in detail in Chapter 27 Sect. 2. Details of several other
alternative systems can be found in Refs. 2, 5, 6, 7 and those at the end of Chapter 27.
The need for some form of amplitude limiting was stressed in Chapter 27 Sect. 2,
when the phase discriminator was being discussed. It was pointed out that the diodes
in a phase discriminator are amplitude modulation detectors, and if undesired ampli-
tude variations are not to appear in the receiver output the amplitude of the voltage ap-
plied to the discriminator should be constant. This was emphasised by reference to
eqns. (20) and (21), where it was seen that the product g,,,E, should be held constant
if the discriminator sensitivity is to remain fixed for a given frequency deviation.
Circuit arrangements for limiters in F -M broadcast receivers have become quite
stereotyped. Sharp cut -off pentodes having fairly high values of g, are the usual
choice, and typical of these are types 6SJ7, 6SH7 and 6AU6. The limiting action
is brought about by a combination of grid -leak bias and low values of plate and screen
voltage. Grid -bias limiting is obtained by adding a resistor and capacitor, of suit-
able value, to the grid circuit and using zero or very small values of cathode bias.
The grid circuit arrangement is the same as for a grid -leak detector, and the operation
is almost identical since the average negative bias on the grid is determined by the
e, - i, characteristics of the valve in conjunction with the associated circuit. The
value of the average grid bias, together with the low screen voltage, determine the
condition for which plate current cut-off occurs. Any signal input voltage whose
amplitude is sufficiently large will cause the average negative bias to increase and so
tend to hold the output voltage constant. Because of the low values of plate and
screen voltage the plate voltage swing is limited to a comparatively small value and,
for an input signal of sufficient magnitude, there will be practically no corresponding
increase in output voltage when the signal input voltage is increased.
The added damping on the i -f transformer secondary, which connects to the limiter
grid, is given approximately by R, /2 or R, /3, depending on the circuit arrangement,
in exactly the same way as discussed previously for a diode detector ; R, being the
value of resistance selected for the grid leak. This should be taken into account (as
1147
1148 LIMITERS-(i) GENERAL 29.1
well as circuit detuning caused by changes in the valve input capacitance) when the
i -f transformer is being designed.
The necessity for high gain in the receiver stages preceding the limiter stage will
be appreciated when it is realized that a minimum of about 2 volts peak is required
at the limiter grid to obtain satisfactory operation with a typical circuit arrangement ;
it is preferable to have voltages of the order of 10 to 20 volts peak for best results
under adverse condition of reception. Since this limiter input voltage must be
obtained with the smallest signal input voltage likely to be met in the field, the gain
of the preceding stages in the receiver should be sufficient to give satisfactory limiter
operation with signal voltages as low as 2 or 3µV ; in F-M mobile communications
applications the limiter should saturate with signals of less than 1µV. Also, because
the conversion from F -M to A-M in the discriminator usually results in a low equival-
ent value of percentage amplitude modulation, it is necessary for the limiter output
voltage to be large if the detected audio voltage is to be sufficient to drive the audio
amplifier to full output ; this means that something in excess of 10 volts peak is
desirable at the plate circuit of the limiter.
For completely satisfactory amplitude limitation two limiter stages are necessary.
However, because of the cost factor, commercial domestic type receivers seldom use
more than one limiter stage. When two stages of limiting are used, it is essential
that the coupling circuit between them should not introduce any appreciable ampli-
tude modulation due to its selectivity characteristic. The selectivity should be suffici-
ent to attenuate harmonics of the intermediate frequency generated by the limiter,
although this is also accomplished by the primary of the discriminator transformer ;
with single stage limiters the discriminator transformer is relied upon to give the
necessary attenuation of the i -f harmonics. Transformer coupling between the two
limiter valves is the most satisfactory arrangement, but single tuned circuits are often
used.
The choice of the time constant for the grid resistance -capacitance combination
is important. It must be sufficiently short for the grid bias to be proportional to
changes in amplitude, but not so short as to prevent the bias change from being suffici-
ently large to control the amplification of the limiter stage, so as to offset any change
in signal input voltage. For a single stage, limiter time constants of 2.5 microseconds
are usual, although 10 to 20 microseconds and even higher have been used in some
receivers. For two stages the first limiter grid circuit uses a time constant of 1.25
to 5 microseconds and 2.5 to 10 microseconds or longer in the second stage, in typical
cases. The longer time constants of 10 to 20 microseconds are suitable for most
types of noise impulses, but some forms of motor car ignition noise are more com-
pletely suppressed when the shorter time constants are used. A careful choice of
the time constants is necessary if the noise is not to be heard in the receiver output
because the bias on the limiter valve must be able to follow the changes in the ampli-
tude of the input voltage.
No matter how effective the amplitude limiters may be and how carefully their
time constants are chosen it will often be found that the F -M receiver will not effec-
tively suppress bursts of noise such as those emitted by ignition systems. To
obtain the best results it is most important that the pass b d of the receiver be sym-
metrical and that the centre frequency of the i -f amplifier coincides exactly with the
centre frequency of the discriminator. A useful test is t align the receiver on an
unmodulated carrier at the signal frequency and then to switch off the carrier ; if
the alignment is correct, and the circuits symmetrical, a .c. vacuum tube voltmeter
connected across the discriminator output will give a rea ing of approximately zero
(of course the usual noise will be heard from the receiver o tput). Small inaccuracies
often arise when the receiver is aligned so as to give maximum grid current at the
limiter stage, or stages, even though the signal (unmodulated carrier) frequency is
such that zero d.c. output voltage is obtained from the discriminator. The grid
current reading is usually rather broad, and it will be found that the i -f and limiter
circuits, in particular, can often be realigned to give zero d.c. output voltage at the
discriminator on noise without reducing the limiter grid current on signals. As a
29.1 LIMITERS-(i) GENERAL 1149
further check on alignment and symmetry the receiver is tuned very carefully to an
unmodulated carrier so that the a.c. voltage at the output transformer is a minimum
(i.e. for maximum quieting) and for this condition the reading of the d.c. voltmeter
at the discriminator is noted ; then it will usually be found that this latter reading is
the same as that obtained from noise alone. (A casual reading of the text may not
bring out the full significance of this test, but a practical trial will lead to a better
appreciation of its possibilities). If minimum noise output, zero d.c. discriminator
output voltage, and maximum limiter grid current do not all occur at the same carrier
frequency, then in general the optimum conditions for impulse noise rejection have
not been obtained.
For applications other than domestic receivers the above factors require very care-
ful attention.
I-F Trans
so,000n
FIG. 29.2
The operation of both circuits is identical, but the damping of the i -f transformer
due to grid current is less with the arrangement of Fig. 29.1, being approximately
50 000/2 = 25 000 ohms. The damping in the alternative arrangement is 50 000/3
= 16 600 ohms. Whichever circuit is used will depend, largely, on the additional
damping required on thc transformer to achieve the required bandwidth or, perhaps,
upon considerations of practical convenience. The general method of operation
has been discussed in (i) above.
If complete valve characteristics are available then a preliminary design can be
made, but the work involved is hardly worth the effort because of the ease with which
the practical circuit can be made to give satisfactory results using experimental pro-
cedures. For a calculation of gain, or overall discriminator sensitivity, it is
necessary to determine the mutual conductance under the actual operating conditions.
This can usually be determined, with sufficient accuracy for a preliminary design,
from the valve data sheets by ignoring the change in plate voltage (provided the
change in plate current with plate voltage does not fall too far down on the knee of
the plate characteristics). For example, a type 6AU6 is to be operated with 50 volts
on the plate and screen, and zero grid bias. From the average characteristics relating
grid No. 1 volts to transconductance, the g,,, is 4000 micromhos for zero bias, 50 volts
on the screen and 250 volts on the plate. For typical cases of the type being con-
sidered, the g, so found is usually about 10% high, and so 3600 micromhos would be
a closer approximation. Alternatively, the valve can be set up as a straight amplifier
with the appropriate d.c. voltages applied, and a 1000 ohm resistor connected as the
plate load. Then with 1 volt of a -f input (from a source of low d.c. resistance) the
output voltage will equal the g, in mA/volt ; the actual input voltage is selected so
that the stage just starts to saturate. However, overall measurements of actual cir-
cuit performance are preferable.
1150 (ü) TYPICAL CIRCUITS FOR F-M RECEIVERS 29.1
- .2MA
OINF
10,00011
5D,000A
+250V +250v
The most useful application of the system in broadcast receivers is with those re-
ceivers having automatic tuning e.g. push- button station selection, and cam or motor
driven variable capacitors. In cases of this type the tuning may not be accurate
over extended periods of time, and a.f.c. may be used effectively to carry out the final
adjustment when the respective capacitances and /or inductances have been selected
by the automatic tuning system.
There are two devices necessary for any a.f.c. system. These are :-
(1) A frequency discriminator, which must be capable of changing a frequency
variation into a suitable direct voltage change which can be used for control purposes.
(2) A variable reactance, whose value can be controlled by the direct voltage changes
_due to the frequency discriminator. The variable reactance is connected to the
oscillator circuit in such a way as to control its frequency.
Signal
Voltage
Miller
t-F
Amplifier
Frequency
discriminator of
Audio
.Oscillato
Voltage
The complete arrangement can be visualized with the aid of the block schematic
of Fig. 29.4. It is seen that the additional elements to those normally found in a
superheterodyne receiver are the frequency discriminator and the variable reactance.
The variable reactance in the discussion to follow will be of the electronic type con-
sisting of a valve (which can be a pentode, a hexode or a heptode) and its associated
circuits. It will also be inferred from Fig. 29.4 that the frequency discriminator
can be used for normal detection, since there are suitable audio voltages developed
in this circuit by the applied modulated i.f. voltage.
Suitable voltages for a.v.c. are also available from the discriminator output, but
in this regard it is necessary to point out that a very efficient a.v.c. system is helpful
in obtaining satisfactory operation from the a.f.c. system. If reasonably constant
input voltage to the discriminator is not maintained, there will be a variation in the
" pull -in " and " throw -out " frequencies. Because of the stringent a.v.c. require-
ments it is fairly common practice to employ a separate diode coupled to the trans-
former primary, in the usual manner, to provide the a.v.c. bias voltage. There is
also a disadvantage in taking the a -f voltage from the discriminator output as the
distortion tends to be fairly high. However, in most commercial receivers the cost
factor leads to some arrangement such as that of Fig 29.8.
Signal I -F
Detector
Mixer Amplifier AV.C. etc.
Voltage 455Kt/s
A.F.C. A.F.C.
Local Variable OKris
Osciliotor Reactance Oiscrim Convertor
-Irqtor a Limiter
FIG. 29.S
The polarity of the output voltage with frequency change is all important, and, as
the electronic reactance is practically always inductive in a.f.c. systems used in tunable
broadcast radio receivers, the polarities of the direct output voltage with frequency
change as indicated on Fig. 29.6 will be correct. One purpose in making the re-
actance inductive is so as not to limit the frequency coverage of the receiver. This
follows because it is often difficult to keep the minimum capacitance of the tuning
unit to a sufficiently low value, and it is hardly wise to deliberately increase this capacit-
ance unnecessarily. However, if carefully designed, the undesired added capacitance
due to the capacitive reactance unit can be made very small.
The effect of the added parallel inductance can usually be offset fairly readily by
increasing the inductance of the oscillator coil. With receivers using push- button
tuning where the circuits are of the preset type either the inductive or capacitive form
of electronic reactance is satisfactory. For receivers with inductance tuning a capaci-
tive electronic reactance is usually preferable. The degree of frequency correction
is not constant over a band of frequencies, and this again suggests an inductive elec-
tronic reactance when the circuits are tuned by a variable capacitor, since the least
error is obtained at the low frequency end of the tuning range where the receiver is
most selective.
It may be helpful to follow through the steps leading to the polarities shown for
the direct voltage with frequency change. Suppose the receiver is tuned to a signal
of 1500 Kc /s. Then the oscillator will be set at 1955 Kc /s to give an i -f of 455 Kc /s,
and there will be no direct voltage output from the discriminator. If now the oscillator
drifts to 1957 Kc/s the i -f produced will be 457 Kç /s, and a direct voltage of negative
polarity will be produced by the discriminator and applied to the reactance valve.
The equivalent inductance of the electronic reactance is inversely proportional to
mutual conductance (gm) and so, as a more negative bias voltage reduces g,,, the
29.2 (ii) DISCRIMINATORS FOR A.F.C. 1153
shunt inductance across the oscillator coil is increased, the total inductance in the
oscillator circuit is now increased, thus lowering the oscillator frequency as required.
The two most commonly used types of discriminator circuits are shown in Figs.
29.7 and 29.8. The first is generally known as the Round- Travis circuit (see Ref.
10), and is a typical example of an amplitude discriminator. The secondary circuits
A and B are so tuned that A has its resonant frequency slightly above the intermediate
frequency and B is set slightly below the i-f (say +5 Kc /s in one case and -5 Kc /s
in the other). Each secondary circuit has its own diode detector and the diode loads
are connected in d.c. opposition, so that when the i -f is greater than the required value
of 455 Kc /s (and since the voltages developed across RI and R, have the polarities
shown) the voltage across R1 is greater than that across R2 and the a.f.c. bias voltage
is negative as is required for correction of the oscillator frequency. For example,
the voltage across R1 may be 4 volts when that across R2 is 3 volts, then the available
a.f.c. bias is -1 volt.
Discriminator
IFAmplifier
FIG. 29.7
A.VC. B+
The design procedure for an amplitude discriminator for a.f.c. is set out in detail
in Ref. 8. Since this type of circuit is seldom used in modern a.f.c. systems the details
of its operation will not be discussed further. There are a number of alternative
arrangements for the amplitude discriminator, and some of these are discussed in
Refs. 8, 9, 11, 15 and 13. It will be appreciated that a circuit of this type could be
used for F-M detection as an alternative to the phase discriminator, and several manu-
facturers have produced F-M receivers using modified amplitude discriminators
(see Ref. 39 given at end of Chapter 27, for typical examples).
A.F.0
C5 Discriminator
Bias
I-F Cs
Amplifier
4S5 Kc/s
Aud o
A.V.0 Output
FIG. 29.8 Bias
B+
A.V.C.
Fig. 29.8 shows the most commonly used circuit, the phase (or Foster - Seeley)
discriminator. In this circuit both primary and secondary circuits are tuned to the
intermediate frequency. It will be seen that the circuit arrangement is identical
with that of the F-M phase discriminator shown in Fig. 27.19, and the general dis-
cussion of its operation and the loading effects on the transformer apply equally well
here. The de-emphasis circuit is not required here, of course, but the same general
arrangements can be retained to provide r -f and a-f filtering, since only a direct control
voltage is required. Suitable values for the filter circuit are R, equal to 0.5 MQ, -2
and Co is, say, 0.05 -
0.1 F. The considerations governing the choice of the com-
ponent values can be compared with those for selecting filter components for auto-
1154 (ii) DISCRIMINATORS FOR A.F.C. 29.2
matic volume control circuits, and a time constant for R5C3 of about 0.1 second is
usual. The circuit arrangement of Fig. 27.17 could also be used here, and, if the
primary circuit is too heavily damped, an additional series resistor can be inserted
between the transformer centre -tap and the junction of the diode load resistors R,
and R4 ; however, this will lead to some loss in available output voltage, depending
on the value of series resistance selected. The difference in the by -pass capacitor
arrangements (i.e. C, and C,) for Figs. 27.17 and 27.19 (or Fig. 29.8) should be ob-
served, and the discussion in Chapter 27 Sect. 2, in connection with Fig. 27.19 will
be helpful, if the reasons for the arrangements are not immediately obvious. The
considerations leading to the choice of actual circuit values will differ because the
a.f.c. discriminator is usually designed for high sensitivity rather than for a very high
degree of linearity.
The theory of the circuit for a.f.c. use has been discussed by Roder (Ref. 14) and
design procedures have been treated in detail in Ref. 8. A brief discussion, which is
quite helpful, is given in Ref. 13.
The method of obtaining a -f output and a.v.c. bias from the a.f.c. discriminator
circuit is also shown in Fig. 29.8. To retain the same degree of selectivity as that
obtained with a similar receiver not incorporating a.f.c. one extra tuned circuit is
required. This is necessary because the discriminator provides very little selectivity
as a result of the heavy loading effects produced by the diodes. There is some in-
crease in gain over the usual arrangement where the a -f detector is connected across
the i -f transformer secondary only.
The frequency at which the a.f.c. will come into operation when tuning a signal is
called the " pull-in frequency." When tuning away from a carrier the frequency
at which the a.f.c. loses control is called the " throw -out frequency," and it is always
greater than the " pull -in frequency." It is desirable to make these two frequencies
as close as possible, because stations which are received when tuning-in may be passed
over when tuning -out. Many receivers incorporate arrangements for disconnecting
the a.f.c. until a carrier has been approximately tuned -in (see Refs. 8, 15).
Some of the general details of the discriminator design will be set out here, and the
reader is referred to Refs. 8, 9 and 14, in particular, for further information. Con-
sider Fig. 29.8. Typical values for R, and R4 are 0.5 MD. Capacitors C, and C,
must give adequate by- passing at the intermediate frequency of 455 Kc /s, and suit-
able values would be 100 to 200 µµF. C3 is usually 100 µµF. Values for R, and Co
have been discussed previously. The valve V, is any of the usual voltage amplifier
pentodes such as type 6SK7 etc., and V2 is a double diode such as the types 6H6 or
6AL5. The next step is to select a suitable value for the inductance of the r -f choke.
9mcg
HG. 29.9
As the inductance and capacitance of the choke will affect the resonant frequency
of the transformer primary (L1C1) and its Q, care is necessary. The choke, in associa-
tion with the transformer primary circuit, can be represented by the equivalent circuit
of Fig. 29.9 in which gme, represents the equivalent constant current generator for
the pentode voltage amplifier ; ra is the plate resistance ; C1 is the total primary
capacitance including strays ; L1 is the primary inductance and R1 its r-f resistance ;
C3 is the capacitor connecting the plate circuit to the choke L, (whose r -f resistance is
R0) ; Co represents any distributed or stray capacitances across Lo ; R ,. = R3 = R4
the diode load resistance. It should be apparent that three cases can arise with the
choke (Ref. 14) :-
(1) The capacitance Co negligibly small.
(2) The capacitance Co small enough to still allow the choke to be self resonant
well above the intermediate frequency, but the value of Co to be such as to
appreciably affect the resonant frequency of the complete primary circuit.
29.2 (ii) DISCRIMINATORS FOR A.F.C. 1155
= 1 - Xc,
XL.
(2)
Qv
1
Qo
+
Rae /4 ry
+ slX L(l
Q,XLi
o
+
1
-f
1 1
(3)
XL1XLo
X9 =
XLi + XLo (8)
The reciprocal of the equivalent loaded primary circuit Q is
1 r 1 1 1 1
= XPL + + -- + (9)
Qv Q0XL0 QX Li r,, Rad
All notation exactly as for case (1).
It should be noted in all cases that the loaded Q's refer to the case where the primary
and secondary are uncoupled from one another. This is the usual definition.
We are in a position to proceed with the determination of suitable values for the
primary and secondary circuits of the discriminator transformer, since all the external
effects can now be taken into account. What is required next are methods for de-
termining optimum values for Qy, Q
LD and L,(= L,) where these factors have
the meanings given by eqns. (1), (3), (4) etc. and L,(= L,) is the secondary inductance
(since its value is not changed by the presence of the choke). If the choke is not
used XL, = Xcl = X9 and so L, = L;
the procedure is then as for any other phase
discriminator. The design factors, as well as the bandwidth and sensitivity calcu-
lations given below, apply equally well for all phase discriminators provided the values
of Q, L and k (coefficient of coupling) so determined are those actually obtained in
the receiver.
Optimum values for k*/Q,Q, for various values of L,/L [see Refs. (8) and (9)]
can be found from
k(QQe)l =
r(Qn2Qea + 2Q ,,Q,'L, /L,)i Q ,,Qel+. -
L Q,'L, /LD -II
Oscillator Reactance
Volvo
Blas from
Discriminator
FIG. 29.10 B+
(2) Miller effect circuits. These circuits (Fig. 29.11) rely on the change of
input capacitance which occurs when the gain is varied. If the plate load can be
tuned so that it behaves as a pure resistance, then the valve input resistance can be
made very large. For the circuit shown there will be a resistive input component
due to Miller effect because of stray capacitance across the load resistor. The applica-
tion of the circuit is largely confined to use with fixed tuned oscillator circuits.
Reactance Bias from
I Volvo Discriminator
Oscillator IA
Vn
Bias from
Discriminator B+
FIG. 29.11 FIG.29.12
(3) Quadrature circuits (Figs. 29.12, 29.13). The grid is fed from a resistance
reactance network connected between plate and cathode. This provides a voltage
between the grid and cathode which is almost 90° out of phase with the plate to
cathode voltage. The source of the alternating plate to cathode voltage is the voltage
developed across the tank circuit of the oscillator (see Fig. 29.12). Since the plate
current is in phase with the grid voltage (for valves having high plate resistance) the
plate voltage and plate current will be approximately 90° out of phase. To the
external circuit (oscillator tank circuit in this case) connected between plate and cathode
of the reactance valve the behaviour is as though an additional reactance and re-
sistance had been connected in parallel. Whether the valve circuit behaves like
an inductive, or a capacitive reactance depends on the resistance -reactance network ar-
rangement, as can be seen from Fig. 29.13.
The value of the apparent reactance and resistance, due to the electronic reactance
(more correctly electronic impedance) depends on the mutual conductance (gm)
of the valve, and as the gm can be controlled by alteration of the grid voltage the
equivalent reactance and resistance can also be varied (the resistance variation is
usually undesirable). The required grid voltage variation is obtained by utilizing
the direct voltage changes at the discriminator output, when the reactance valve is
used in a.f.c. circuits. Frequency modulation, using this method, is obtained by
applying the audio frequency modulating voltage to the grid of the reactance valve,
in the same way as the direct voltage changes are applied for obtaining a.f.c. The
magnitude of the a -f voltage determines the change in the equivalent reactance shunted
across the oscillator tank circuit, and so determines the frequency deviation from the
nominal oscillator centre frequency. The number of times the frequency deviates
around the central reference frequency will be determined by the frequency of the a -f
modulating voltage. Variation in the value of the shunt resistance, due to the elec-
tronic reactance, across the oscillator circuit causes undesired amplitude modulation.
The quadrature circuits are the most widely used for a.f.c. and other purposes,
and attention will be confined to discussing some of the possible arrangements.
Fig. 29.12 shows the general circuit arrangement using a pentode valve. A hexode
or heptode valve can also be used, with the phase shifting network connected to the
signal grid and the control voltage to the oscillator grid. The impedances Zl and Z2
1158 (iii) ELECTRONIC REACTANCES 29.2
9mCiRa A
C¿.
I+(wC,R2Y C¿. TT
R,L!
9mR,Li
+(wC,RaY R?+uLj
R¿ 1
9m(wC,R2Y 9,0_ L3
For R, «XC¡ For XLtCR,
C.L. ^s 9mC,R2 Ct 9mL2
R,
R¿2, R¿x
9m(wC,Ra) 9mw L2
A
4+,1 L
9míR
R+w'L¡
For Ra «XL,
L¿4r L
9mRa
R^ w
9mR
FIG. 29.13
in Fig. 29.12 comprise the phase shifting network. Depending on the form these
impedances take it will be clear that an additional blocking capacitor may be required
between plate and grid of the reactance valve, and also, since there must be a d.c.
path between grid and cathode, an additional grid resistor may be necessary. The
values for these additional components should be such as to have negligible effect
on the performance of the circuit. Fig. 29.13 shows four possible arrangements for
quadrature circuits, together with the equations for the additional resistance (R,)
and inductance (Le), or capacitance (CO, connected in parallel with the oscillator tuned
circuit. Exact and approximate equations are given, but in most practical circuits
the approximate conditions will hold. The equations apply equally well to quad -
rature circuits using pentode, hexode or heptode valves. The angular frequency
w is that at which the oscillator circuit is meant to operate e.g. 27r x 1455 Kc /s etc.
(in the case of F-M this would be the nominal reference frequency). The circuit
of Fig. 29.13(c) is the one most commonly used in practical a.f.c. circuits in which
the oscillator is tuned by a variable capacitor. Care is necessary when using circuits
(b) and (d) as self resonance effects, due to stray capacitances across the inductances
L2 and L often lead to difficulties. Further, the r-f resistances of L2 and L, must
be low if the circuits are to behave as relatively pure reactances. Stray capacitances
across R1 and R2 can also affect performance, and should be kept small.
To carry out the design for an electronic reactance circuit it is necessary to know
the manner in which gM varies with grid bias. This information is generally available
on valve data sheets, for a given set of operating conditions. If other operating
conditions are required, then direct measurement of the g,,,
usual procedure.
- e, characteristic is the
(6) Select particular values for L,, R, or Co to conform to the circuit require-
ments ; remembering the previous restrictions of XL$ G< R1 and X c
This should lead to R1 being at least 5 times XL, or Xc2, but larger
:.R,.
ratios
are preferable (see below).
The most convenient procedure is to tabulate various values of R1, L, or C, and to
select the most suitable combination giving the product found in step (5) e.g. R, =
50 000 Q = 0.05 MD, then C, = 10.5 /0.05 = 210 µµF, and so X cs = 520 A.
This makes R1 > X ca as required.
Of course, the actual values of L, and R, can be determined directly from the ex-
pressions given in Fig. 29.13. Rt should always be so determined for the condition
of maximum gm, after the circuit values have been found, to ensure that the loading
on the oscillator circuit is not excessive. The value of g. to be used in these equations
corresponds to. the actual bias voltage for a particular operating frequency ; e.g. in
the above example, with no additional external bias applied, the operating frequency
is 1455 Kc /s, and the standing bias voltage can be taken as -6.5 volts, corresponding
to a mutual conductance of 925 µmhos. The total parallel inductance and resistance
changes can be found, using the gm values corresponding to -3 and -10 volts bias,
which are the values required when the operating frequencies are 1450 Kc /s and
1460 Kc /s respectively.
Before completing a design it is necessary to check the amplitude of the oscillator
voltage applied to the grid of the reactance valve by the phase shifting network.
This check is necessary as the possibility of grid current might be overlooked. In
1160 (iii) ELECTRONIC REACTANCES 29.2
our example the minimum bias is -3 volts and so the peak r -f grid voltage should
exceed about 2 volts if grid current is to be avoided. The proportion of the r-f not
voltage
developed across the oscillator tank circuit (and applied between plate and
cathode
of the reactance valve) which appears between grid and cathode of the reactance
valve is for Fig. 29.13(c)
Xca /V'X2c2 + R12. For our example, the voltage step
down is 0.0104, and so no possibility of grid current exists, as 2 volts peak r-f
at the
grid corresponds to 196 volts peak across the oscillator tank circuit. This is con-
siderably in excess of the voltage likely to be encountered in a receiver oscillator
circuit, where 60 volts peak is about the maximum to be expected with any of the
usual arrangements (it is usually considerably less than this value, depending on
the
type of circuit used).
Further details of the design of reactance valve circuits for a.f.c. can be found
in Refs. 8, 9 and 10. A graphical method for determining the " throw -out
" and
" pull -in " frequencies of an a.f.c. system is given in Ref. 8 (p. 260). The difference
in these two frequencies is reduced by using an electronic reactance which gives
correction for a limited range of discriminator voltages only ; outside the correction
range the added reactance should remain practically constant.
SECTION 3: REFERENCES
(A) LIMITERS
See also Refs. 40, 41, 42 Chapter 27.
1. Tibbs, C. E. " Frequency Modulation 3-Interference suppression, the limiter, and the capture
effect " W.W. 49.3 (March 1943) 82 (also gives bibliography).
2. Carnahan, C. W., and H. P. Kalmus " Synchronized oscillators as F.M.
17.8 (Aug. 1944) 108 (extensive bibliography). receiver limiters " Elect.
3. Tibbs, C. E. (book) " Frequency Modulation Engineering "
Chapman and Hall, London, 1947.
4. Hund, A. (book) " Frequency Modulation" Chapter 2, p. 209, McGraw
York and London, 1942. -Hill Book Co. Inc., New
5. Sturley, K. R. (book) " Radio Receiver Design " Part 2 Chapter
London, 1945. 15, p. 328 Chapman and Hall,
6. 'liver, M. S. (book) " F-M Simplified " Chapter 6 D. Van Nostrand Co.
and London, 1947. Inc., Toronto, New York
7. " Cascade dynamic limiter in F -M detector systems " A.R.T.S.
Hazeltine Electronics Corp. Bulletin). and P. Bulletin No. 143 (Reprint
1161
1162 (i) PRINCIPLES OF RECTIFICATION 30.1
-EPEAK
\ - --c
_
S7
/E__. EAVGE (FULL WAV
JEAVGE (HALF WAV
E."- -
-%//'
A VOLTAGE
g AT
r_
Ci Aj
- EPEAK
-//N'
-
.F"- T-
input and choke input rectifier systems.
Charge IDischorgr Dis.
The symbols on the left are those used in
I
I
---1
Section 2. E,c
_ _ __+_ _y_EPEAK-
I i
i I
WnuununWr,li nnnmh,
II
dumnmi
SPEAK
IE.I - \ ER btS.
EAVGE
Etc
IAVGE
naeNnaMIIAINAxmmD!innnnnnnaunnUlNlillll
FIG. 30.1
The mean level of ABA'B' is the effective direct voltage. The shaded area above
the curve AQB represents the voltage by which the transformer voltage exceeds that
of C1. The current through the plate circuit of the rectifier only flows for the interval
between A and B and between A' and B' because at other parts of the cycle the trans-
former voltage is below the voltage of Ct. The current through the rectifier, shown
in Fig. 30.1C, is similar in form to the difference in voltage between the curves APB
and AQB in Fig. 30.1B.
The ripple voltage may be determined from the ABA'B' curve, and the values of
the fundamental and harmonics may be determined by a Fourier analysis.
As the load resistance (RL in Fig. 30.1F) is increased, BA' becomes more nearly
horizontal and the area APB becomes smaller until in the extreme (theoretical) case
when the load resistance is infinite the direct voltage is equal to the peak voltage.
This graphical method may be applied to any rectifier with a condenser input filter
followed by a high inductance choke. The assumption is made that the current
through the inductance remains constant, that is to say that the lines BA', B'A" etc.
are straight.
With a (full wave) choke input filter* the conditions are as shown in Fig. 30.1D
assuming a very high inductance choke (Lt in Fig. 30.1G), although with practical
chokes there will necessarily be a certain amount of ripple in the load current.
The term "choke input" is used for convenience in this Handbook to indicate a series inductance
followed by a capacitance shunted across the load resistance.
30.1 (i) PRINCIPLES OF RECTIFICATION 1163
Maximum ratings
Rectifier valves are usually rated for maximum direct-current per plate, maximum
peak current per plate, maximum peak inverse plate voltage and maximum r.m.s.
supply voltage per plate. In some cases a maximum rated hot -switching transient
(or surge) plate current per plate (for a specified maximum time, e.g. 0.2 second) is
also given. It is important to ensure that no one of these ratings is exceeded under
the conditions of operation. If the ratings are design centre values -see pages 77
and 78 -they apply to nominal mains voltages.
direct plate current is reduced below its maximum rating and that the total effective
plate supply impedance per plate is increased for the higher voltage conditions (e.g.
Fig. 30.2A). With a choke input to the filter, the maximum r.m.s. supply voltage
per plate is sometimes only permissible provided that the direct plate current is
reduced below its maximum rating (e.g. Fig. 30.2C). With a choke input to the filter,
it is essential to have a choke inductance not less than the critical value for the
particular operating condition in question. The critical inductance is a function of
the load resistance and the frequency of the supply, as given by eqns. (1), (2) and (3)
in Sect. 3.
For any value of inductance, with constant r.m.s. supply voltage, there is a value
of current below which operation is not permitted. This is shown in Fig. 30.2C
where boundary lines for choke sizes are included.
,-
the input voltage minus the valve voltage drop
e.g. Output voltage = (0.9 E
-
The direct voltage output of such a system is equal to 0.9 times the r.m.s. value of
15) volts.
The choice of a rectifier valve for a particular service must take into account the
maximum permissible ratings for peak current, average current, and peak inverse
voltage. The design of the following filter will influence these last two factors par-
ticularly ; the type of filter, either choke or condenser input, will be determined partly
by the demands of power supply regulation. In supplies feeding Class A output
stages the choice will probably be a condenser input filter, but where Class ABl and
AB_ output stages are to be supplied, the regulation of the power supply becomes
a significant feature and choke input filters with low impedance rectifiers must be
used.
30.1 (ii) RECTIFIER VALVES AND TYPES OF SERVICE 1165
for the higher voltage curves (6 to 8). It is always permissible to adopt a higher value
of total effective plate supply voltage per plate than that shown on the curves, but the
direct voltage output will thereby be decreased somewhat. Operation is only per-
missible on, and below, the line formed by the highest curve and the " current and
voltage boundary line " ADK. These curves only apply to one specified value of
capacitance input to the filter, in this case 10 µF.
I I I
-8
III
Et SO VOLTS
I
1IIIÍ girlA
\
I I
CURRENT
BOUNDARY
IS INDICATED BY
I
I VOLTAGE
LINE
ADK
1
50
\t
` O. Íííi
Ry
\\
\ D
300
200
I 111111
1
-.....4
100
1
0
`K
20 40 60 80 IDO 120 140 160
O.C. LOAD (MILLIAMPERES)
FIG. 30.2A
Fig. 30.2A. Operation characteristics for a typical full -wave rectifier (5 Y3-G T) with
condenser input filter.
30.1 (iii) THE USE OF PUBLISHED CURVES 1167
The effect of change in the value of capacitance input to the filter is indicated
typically by Fig. 30.2B, where curves for 3 values of capacitance are drawn. A
higher capacitance gives better regulation and a higher output voltage, but increase
in capacitance beyond a certain value (here about 16 µF) has only a very slight effect.
If curves are only drawn for one specified value of input capacitance, operation with
a higher value is not permissible unless this has been demonstrated by measurement
or calculation to be within the peak current rating of the rectifier.
Curves for a typical full wave rectifier with choke input are given in Fig. 30.2C.
Operation is only permissible in the area to the right of the boundary line corres-
ponding to the proposed choke size, to the left of the current and voltage boundary
line CEK, and below the highest curve for a choke of infinite inductance. If the
direct current varies between two limits, it is important to select a value of inductance
at least equal to, or preferably higher than, the value required for the lower limit of
direct current ; the inductance should be measured at the lower limit of direct current.
The constant current curves (of which an example is given in Fig. 30.3) are very
helpful for deriving certain information. If it is required to find the input voltage
700 I
I
E ¡ 5.0 VOLTS A.C.
L 20 HENRYS
TOTAL EFFECTIVE PLATE SUPPLY
IMPEDANCE PER PLATE 150 ¡L
600 OL
TS
A4
STE
PE
C. 16 TO
\40p 40NF
8N
50 4 F
16 TO 40MF
4 F
300
16 TO 40pF
4MF
o 250
50 100 150
D.C. LOAD (MILLIAMPERES) FIG. 30.28
Fig. 30.2B. Operation characteristics for a typical full-wave rectifier (5 T4) with
condenser irrput filter, showing effect of input capacitance (C).
1168 (iii) THE USE OF PUBLISHED CURVES 30.1
to give 250 volts 100 mA direct current, a vertical is drawn upwards from an output
voltage of 250 volts to point P on the 100 mA curve. From here a horizontal line is
drawn which intersects the vertical axis at 277 volts, which is the desired value.
The line OD is for equal input and output voltages. Above and to the left of this
line the output voltage is less than the input voltage ; below and to the right of this
line the output voltage is greater than the input voltage.
Each of the points A, B, C etc. at which the several current lines cut the vertical
axis indicates the combined effective voltage drop in the valve and the transformer ;
700
E f 5.0 VOLTS A.C. SUPPLY FREQUENCY 60 C.P.S.
SOLID -LINE CURVES .. CHOKES OF INFINITE
INDUCTANCE
LONG -DASH LINES . BOUNDARY LINES FOR
% CHOKE SIZES AS SHOWN
600
el SMALL -DASH CURVES
FOR LAREPRESENTATIVE
CHOKE SIZES
Il j; 11 1111 CURRENT AND VOLTAGE BOUNDARY LINE
INDICATED BY
WEE
AS SHOWN 16PF
11111111111111
500 111111111111\s .
IIKEIIIIIIIINI
¢ 11111111,\,-,
°11
E 11111111
2400 LiI.irmau
AVM
11
inow:
g
1
fi_
111
500 s
i V
o
gri r
I,_TM
r
1
1
1
ll 00 '
talli'n
ú4 V.
I,l .II _
300
1 11
PER PLATE__
O
350
V
¢
Ó
300
M11
--
200 1ILIN/
CFAs 2D MIEN
100
mill
o
FiHI 20 40 60 BO 100
D.C. LOAD (MILLIAMPERES)
120 140
k
160
FIG. 30.2C
Fig. 30.2C. Operation characteristics for a typical full wave rectifier (5 Y3 -GT) with
choke input filter.
30.1 iii) THE USE OF PUBLISHED CURVES 1169
E
E D.C,` R.M.S. -iJ
300.-
277
' f---Y
F
/
ED.C.. 1414ERMS
o.
z
100
G ,
100 200 300 400
OUTPUT VOLTAGE (MC)
FIG. 30.3
Fig. 30.3. Constant current curves for a 5 Y3-GT rectifier; condenser -input filter with
capitance 8µF and effective plate supply impedance 80 ohms per plate.
in other words it is the input voltage required to maintain the specified load current
with the load terminals short- circuited.
The line OE is the theoretical limit of output voltage with no load-current, and is
drawn to correspond to an output voltage of 1.414 times the r.m.s. input voltage.
The regulation of the output voltage with varying load currents is indicated by Fig.
30.3. If the input voltage is 250 volts, a horizontal line may be drawn as in XY and
vertical lines may be drawn at each intersection with the constant current curves.
Line FG is for no load -current, while LM is for 100 mA load current. For this
example the output voltage will be seen to drop from 354 volts at no load -current
to 213 volts at 100 mA load-current.
It should be noted that the results given by any form of valve curves are only correct
for the set of conditions for which the curves were derived.
'..ÑÌÌ... : /, 4
...111111
1
-
MMMMM
:::u ..
m ..
rí.. .!s1'1IIIr.p
:.11.1
Inu/.777.í< 1 71
10 drL1111%'
.á ,,,,i%.ú.
../..:I.:
i11111'41
:i;ír.iïi::
:,11:.I'/.I.I.
I
11. .u.C.. 1 1
1
.IIII.11111111.1.1111
_......... .:.
-
f
.. 11 ...- .-_........
I
..r.
:.%Jii ii ¡'uwI
iii:%iiÌ.I/II/4.'.
I. %'/11r11111,.1
I/'//r//li.I.
.
L%2L/C/ilr'l. .
'111
151111
. 1./'.
I.'
HIM! /1111111111
II.
3 4 f 4 1 10 f f 4
100 2 1 400
DIODE ANODE VOLTAGE FIG. 30. 4
Fig. 30.4. Average anode characteristics of some rectifier valves (based on Ref. 6, with
additions). The value of r4 for any rectifier at any diode current may be determined by
dividing the diode anode voltage by the diode current.
Fig. 30.5. Ratio of rectified (direct) output voltage to peak a.c. rectifier input voltage
expressed as a percentage, as a function of wCR y for a half-wave rectifier with a condenser
input filter (Ref. 6).
30.2 (ii) RECTIFICATION WITH CONDENSER INPUT FILTER 1173
A
are curves from which can be found the relationship between Ed, and E for half-
wave, full -wave and full -wave voltage doubler circuits in terms of other circuit para-
A
meters. Curves of the ratios of effective i d and peak id diode currents to the direct
l l
current per anode Ta are given in Fig. 30.8 ; Fig. 30.9 gives details of the ripple factor
and Fig. 30.10, the peak inverse voltage (all based on Ref. 6).
The design considerations to be borne in mind when using these curves are
(1) The value of the capacitance C is usually chosen with regard to the maximum
:-
permissible ripple in the output (see below) but if IE_ is limited to a certain value
j
and the maximum Ed, is to be achieved, C may be increased above this value. In
doing this due regard must be given to the maximum permissible peak current of the
rectifier and, if necessary, limiting resistors placed in series with the anodes of the
rectifier.
(2) In order that the direct voltage should not be closely dependent upon the value
of C, the value of aiCRI must be on or to the right of the knee of the appropriate
curve in Fig. 30.5, 30.6 or 30.7 as required by the type of rectification.
O lop P°0
wCRL(C IN FARADS, PL IN OHMS) FIG. 30.6
Fig. 30.6. Ratio of rectified (direct) output voltage to peak a.c. rectifier input voltage,
expressed as a percentage, as a function of aCRL for a full-wave rectifier with a condenser
input filter (Ref. 6).
1174 (iii) TO DETERMINE PEAK AND AVERAGE DIODE CURRENTS 30.2
200
160
160
1=:
MIIMIINIUM=1=
..C:
IMIMIMINEM MI=
MI.IMI=ii
MINIMIMI .....iiiiiii
MIONINIMMIN =NM
IMMIIME MI= ____
IININ
_
Mi MIN
=NI MENEM=.
6 10 IOO 1,000
WCRL (C IN FARADS RI. IN OHMS) FIG. 30.7
Fig. 30.7. Ratio of rectified (direct) output voltage to peak a.c. rectifier input voltage,
expressed as a percentage, as a function of wCR L for a condenser-input voltage doubler
(Ref. 6).
30.2 (iii) TO DETERMINE PEAK AND AVERAGE DIODE CURRENTS 1175
Step 1. Determine
E /o =
350
- .
70.7%.
350
OO*
.1
YY7
.O
-
W
aa
aa
4°11.7
60
SO
40
30
7
6
s
4
3
C
R
R
..
C
RL IN
to
FOR
T. II2 FOR
ii FOR
HALF -WAVE SINGLE -PHASE RECTIFIER CIRCUITS
FULL -WAVE SINGLE-PHASE RECTIFIER CIRCUITS
VOLTAGE-DOUBLING
IN FARADS
OHMS
=21Tf
III1II!'%IIIIIII
111H11111111!11I1IMM %/!.G
>M111Eg111I%11
.n.i
R
.....-
Mf=.Iìiii.iii
11.. -¡¡::::
_.RR111MI111I11
111111111i
iiiii
RMR..RO.
CO::Ci7
mm..
===--------
...mw:::-Q1::'
__----'-
_.-
iiww=.. Nwim...
==MW..
CIRCUITS
"
PL.6101111111
I.1
.1RRRRR=RI1
iR11R.RR11.
C:C::CMC::7
.
wWw=wil
../111111=111010"=.11.11 -02
...M6Ii/.RRI1 a.ziii
tltll __.__1
-05
s
2
sc
I4°'
2"
10
30
IOO
111111111111
RRR.RRRIR111R111WWIIRI11R111RRRR.R1/1111
=0116n Imomm.u.
IMMIR1f111111111
RRRRRR111111fM111111
MIIIIIM11111111111111f111111
111M111111111111M1111111=111111111111111111111111111111111111111
10 20 50 100 200 500 'Poo
mmiu....
111111111111111111111
MININIIIM11111
IIMI1111
Fig. 30.8 (above). Ratio of r.m.s. diode current to average diode current for one anode,
expressed as a function of wCRL ; (below) Ratio of peak diode current to average diode
current for one anode, expressed as a function of wCRL (Ref. 6).
1176 (iii) TO DETERMINE PEAK AND AVERAGE DIODE CURRENTS 30.2
Step 6.
R8t
% = 2800
x 100 = 12.1%.
Step 7. Knowing that n = 2 for full wave rectification,
n
R, 12.1
= 2
% = 6.05 %.
Step 8. From Fig. 30.8 (lower) where nwCRL = 21.2 we may obtain
=d
loo
BO
60
40
20
10
e
8
6
4 0 B 2 4 6 e
10
20 40 60 e0
100
200 400 800
600 1,000
2,000
uCR(C IN FARADS ,RLIN OHMS)
FIG. 30.9
Fig. 30.9. Curves for the determination of the ripple factor of condenser input filter
rectifier circuits (Ref. 6).
30.2 (iv) TO DETERMINE RIPPLE PERCENTAGE 1177
20
II,1,1
é.. .:::
°°1lllllI 11 II I11IIl!!II il1111i1f1ti11I1U11_. 11111=i.
' 11:1 11111111111N
11
18 1 1
ii.::g11
Ió
: .
..:»in.mIllllw.11.Il I.l1-
11p11' I
'1_!1
IIIBi
II lli:I i®11ei i1
®PiiS:
E !oloni'limülll11111!!
E.,
14 =11111111Ii11! V! alfil! 11i:aiii®®11111
mum III 11IIIIIIIloo11mm
ONO 11111111112e1®111!!!!
I.2
::-:!!!nlll111111=!.ï
::.:=nim ':In:=..i.. ::menl
!!
C:::;llllllil1111111I111111I11+=111i11:!!!
.....
ro
OI 02 04 06 -08 I 4 4 6 8 10
YFIL FIG. 30. 10
Fig. 30.10. Ratio of operating peak inverse voltage to peak applied a.c. for rectifiers
used in condenser input filter circuits.
1178 (vi) PROCEDURE WHEN COMPLETE DATA NOT AVAILABLE 30.2
A A
Step 4. Calculate R, = rd + R,.
A
Step 5. Using Fig. 30.8 (lower curves), and knowing nWCRL, R. and nRL,
determine id/id. If this differs appreciably from the assumed value, repeat steps
3, 4 and 5.
A
Step 6. Calculate irai = ra /093.
Step 7. Calculate iR,Ì = R, + irai.
Step 8. Calculate the percentage 1R,1/RL and apply to Fig. 30.9 to determine
the ripple percentage.
Step 9. Calculate 1R,j /nRL and apply to Fig. 30.8 (upper curves) to determine
and thence lid! which is the transformer secondary r.m.s. current.
A
Step 10. Calculate r a = r x/0.88._
Step 11. Calculate R, = R. + ra.
24
20
16
5
u
« 12
é
n
0
20 40 60 !O loo
direct Output Voltage (Ed.c.)
%Voltage Ratio
n e Peak A.C. Voltage per Plate (r,)
FIG. 30.IOA
Fig. 30.10A. This curve is for full-wave, half-wave and voltage -doubler rectifiers with
condenser-input filters. It applies for any size of condenser so long as the condenser is
large enough to give maximum output voltage for the given output current and r.m.s.
voltage input. For half-wave and voltage -doubler rectifiers the direct current per anode
id = /L, ; for full wave rectifiers 171 = +I L.
n = 1 for full-wave and half-wave rectifiers.
n = 2 for voltage doubler rectifiers.
30.2 (vi) PROCEDURE WHEN COMPLETE DATA NOT AVAILABLE 1179
A
A
Imaa - A
R, rd,
where /,
A
=
s peak hot -switching transient plate current in amperes
E = peak alternating voltage per plate
R, = total effective plate supply resistance per plate
A
and rd, = diode resistance when hot- switching current is at its maximum.
A
The value of ra, may be derived from Fig. 30.4, by extending the curves upwards
if necessary. If the hot- switching current is greater than 1 ampere, but less than 10
amperes, the resistance may be read from the curves at one tenth of the current value,
and the resistance value so derived must then be multiplied by 0.47 (this has an
accuracy within about 2% for curves 1 to 17 inclusive). For example, type 5Y3 -GT
has a rated maximum hot -switching transient plate current of 2.2 amperes per plate.
The diode resistance at a plate current of 220 mA is given by Fig. 30.4 curve 14 as
A
85/0.22 = 386 ohms. The diode resistance ra, at a plate current of 2.2 amperes is
therefore 386 x 0.47 = 182 ohms. Continuing with the same example, if the peak
*This ratio is very commonly used in diode ratings.
30.2 (viii) PEAK HOT - SWITCHING TRANSIENT PLATE CURRENT 1181
alternating voltage per plate is 350/ and the peak hot -switching transient current
is not to exceed 2.2 amperes, then
A
R, + rd, = 350 x 1.41 /2.2 = 225 ohms.
A
But rd, = 182 ohms.
Therefore R, = 43 ohms minimum.
where w = 277f
f
= supply frequency
IER = ripple voltage r.m.s.
I
Based on English T.C.C. condensers, 450 volt (d.c.) working, ambient temperature
40 °C, (128/R L) = 10% and 50 c/s full wave rectification we have :
In the cases listed above, the load current is approximately 85% of the maximum
ripple current ; this only applies for (R ,/R L) = 10 %. Values for three conditions
are given below :
load current
73% 85% 102% approx.
ripple current
In practice, in radio receivers and a -f amplifiers, (R,/RL) is usually well within the
extreme limits 1% and 30 %.
If the ripple current for any desired condition is greater than the permissible limit,
the capacitance of a single unit condenser may be increased and /or the value of R,
may be increased either by selecting a different valve type or adding resistance in
series with each plate. Either method of increasing R, will require a higher trans-
former voltage. Alternatively two condensers may be connected in parallel, with the
total capacitance unchanged, each of which will carry part of the ripple current ;
however, equal sharing of current cannot be guaranteed and a large safety margin is
desirable. The parallel arrangement confers no appreciable benefits with the T.C.C.
ratings, as compared with an increase in capacitance of a single unit.
In general, load currents up to 120 mA may safely be employed with plain foil or
70 mA with etched foil without any investigation.
. -
,
ioo
_--.-----.:::.-:ÌM_ÌÌÌÌÌÌ
a---......;.._---_-- -
::-. .. .
.. =--ii71sIL1Ii:1"1`rVC=.i:l.::=---__.-:7:::-__-......
\0n/-........\--\.
. 1 ... ----
LI.O
MO...
MEMO. MM
L01() -\.1.///.=e1=--M
AES .MN --...IAII-.
was :=71==r4
M -.
MM..`.n11....4.111:M MG= BEMIS =mina
NON
-..... \\IÚ
60
40.0004
M 1M11..:
Ì1M11MMMMM1111S1IIí1MÌ
s11MMNMNÑ1g11E:/1111aM
'ONÑÌ Ì
MO
N Ulla\1\IIIGIIMMM.......
MMI
40
------3°,000:3 .--.
------ i..wv.ú-.:
___
-_C_
"11' S ._°'::.Z--
_...--
\...V.14Mp d..I.... i -..---
`-uo...
\\... ..--
,.
I.I
iiiiiiO'OOOn 1=1=1:¡'aaiOOíiiiiiii Ii%iii =n-
=I:--e7_iI:C=---=wO:C:
.-=
.... ......
M
' MIM1N11111.\I1I:IMO..a..../--M
--.._........
11N11111
.C11N11111..
N11111111111D1
1111MWMM!ALEO1FMlipii11ne
0 .;::----._:...... --
111h1....111P'. \\1E'
N_ . 111111 iii
. r./ouo.. vI]11 -..
---
ri
10,O0pi1
N11111111111111111 \i,\..A1III1.1\\11111
ílll:!í11111R:11111111/N
9
---
_-__--
MN---.
isemw...
-
-o
wommo.. n.--...
- . GCiGIif íil:iGiiìGiSii.W'mE.-
-o-iiñii-_.....ou.-..-.__'.G7..a__-___r
.. iiarWANó-
.o._._.-m. Noma,..7\W_1I---
4...ÌÌÌÌ .ii o .
Ì
6,000$,
MMN..111Y1 ..1M...
M...II1111MMMM.!",.u..6.4.n..M .A inr.
-- MMI 11111I11M1IIIIIMM11\S11II.\ 1
---
----
n=nl= .nn-....
nln. --.--..
-----
....... n.Mmnl.
141-01
:-._...."I.:......I
...\ i._:7n;
....1.....: mL-
_.i..`V./I./I ---- MMINI=IM. MUM
.
-GSiD .
41.NNAIMIAM.
a
MUM. MIME. =SOMME
MIN 1IIIIMMMM
/.so.MSe.
MM1111III111111111IIMM1111111I11NE11 d1111M1111
MM
ap00$, ....i-
.v.....1:Mill=.`MIN
L, a id
= L, 1 0.87 /L
= 2L, 0.5 0.75 /L
= 4L, 0.25 0.72 IL
= infinity 0 0.707 /L
(i) General
Where it is required to obtain a higher direct voltage from a given a.c. input than is
possible with normal rectifier circuits and where for reasons of weight, economy or
other factors it is not desired to use a transformer, voltage multiplying rectifier circuits
may be used.
These circuits involve the principle of charging condensers in parallel from the
input and adding them in series for the output, the switching being accomplished
by the rectifier valves.
v,
é R
FIG. 30.13
In both circuits, the larger the capacitance of the condensers the nearer the output
A
voltage will be to 2E_ and the better the voltage regulation, but care must be taken
that the peak current ratings of the rectifiers are not exceeded.
It can be shown (Ref. 10) that at values of &,CRL greater than 10, the values of
A - A
id/id, Ea , /E,.,, ripple etc. for a half-wave voltage doubler are for all practical pur-
poses the same as the values for the full-wave voltage doubler and calculations of both
types can therefore be made by means of the graphs in Sect. 2.
From these graphs it can be seen that if a voltage multiplication of 1.6 or greater
is required, the value of U CRL should not be less than 100, also that R,, /RL should not
be greater than 1.5 %. This means that if a voltage doubler is to give a high output
and to be of good regulation, the maximum output current is strictly limited.
FIG. 30.14
50µF
6JV FIG.30.17
Fig. 30.16.
valves.
f
Shunt diode bias supply suitable for the bias of r ; i-f ana a-f amplifier
Fig. 30.17. Voltage doubler bias supply suitable for the bias of output stages.
Il. M..
1189
1 .
600 :::
... I .. =M:C::::111: '0
I
MME
lila IIIÌ::::::111
Il.
I
1E
.u.u.uuu
......n
II
R
1
.1
1:..111'111M-.M:IU
1-...! AI.
nmv..r.u
.ME.II R125Q00ri 11'.IIIllurPVd11/'
.11
1
1i111...01
M
MM
111... ul.M..MMN1u I/..,lmlu'I
1
.M
1
M..EEN/lI1 1 I11'
...11..11..M.
..M II/ ul.
. 1
...
MM
1. .B1111111
1111
111....11I11......1.1..
....I.1IMM
11ul IIV
..1'mur l'110,1..1
11...'. 121.10,
Ì114u,/lllll
lu
4
EMME» .u
II r
No
M. 11
1
N
.111M.MM.00l
IM.EMEN00m
N01 u1u/1111
I
1 ,111.11r.11 lllII./'
1111M.MEIAINSI
MM ME
M. 1.IU.....\.uM. 1...M111111::
1111,111,1
I I..u111'111 /Al
111.1111/11111 Il11I.7M.ul
I.I.....1N
Eac. mina I
lllll MIEN
-u. .malin..MMMIu1I1.111.
1.111 Nome. I11 ./'MMEmO.
111 MME IMBU 11111 .1 11111' 1
Rr5p0On.
E.1 Il111/ 11/.11
-.. 111......101a
11....E..I.u..M.EE.l.l1'./.1m.111
M....mm lull./...1.1 1a.11
M.
MM
..uula
u.l....aflau
'11.1111.....1m111......ANS
MO SN' /AI uu1u', UNIS r.111/
IOWA ..11.0 MME/ A.»IO
'MM VA, lnl M.NiI
.M
.MM /
1
1 1..11....0011..
Iul1.M....u1M..E./ u.1...1
M
1
I..AM.M.../
..M.
MME
MMMMME 11
1
1111.../00001..../.'1NI II'
1l1l1MM.1M.
ml....uu1...I, l
.M.u1u....v.'.SU.
O u M...'IMN'. / JIl'.
,.M.ma
1.'Illll'111
M.
.
.1111
'1 Iml/A1....0.11
1/ .1111111111111111
IMI, AllIM.E1u KI11 MM.M1um
!
11
MiMi. Il
..
MME ll
I
I MENUISE .1I...1./.'/'01'Allllllll0/:rIIIIIMM..MNUI
l
MUNIE .IMM.I.I.' Il: u NEW u.n/
EMME EMBUS
MME
MINIM
....
i 11
1 I.MUNEUIUMruiAr
1
uuul11u1.
IEENE I..
1111
l.1I.0
1u1
111
111111
...111
_
lliuu.../r.1.
.1
.Nm1m..u II111.111..
11111NUllllul
111
lull1
05 FIG. 30. 19 I
IO
C IN yyF
5
Fig. 30.19. Output voltage of the circuit of Fig. 30.18 for various values of C, and R1.
250 Nom
Mi.
MME
.E.M
..
.
nul uI1M iMME.
11lll.iuu
IMBUE
u11MUUn
.1II.E.. .
11I
I
uu
lll
n.o
.....
ÌMN
11uuul..liiu mliii.'
..
M. uVUm11M.MMMI:I.
.II M
lull .5111
RIm5/000 :íi
.. ME..
.0
ul:ul .i1n o0
1
uI11.OM..MUI.IMEU.OUII
.M. a'- 1111MIAI.I;ill
.. u11..EE uu.u
n.uItm.u.lu1
I
.l.nun 7.M!I.LI
i
n 111111
l.
11111
.Mi. IM..MNIU...Mi
Ilul1.un'Nom
1111.MM
/\1.
RIIOPOOft II.IMiEEE 11./..
lIEIl1111.E..Mllm nom MMN lIt
1...
I
I...M ..M1:.1.1
.
M
Mi
1
lll11M.EM
...MEUIM.M
..u1.EEM.u./..
MM ul
m
AE'lE11
Ill'./íE.11
IH.. irr.lE.'uunu
1
..
150
..EM
.M
I
IEIIr
/01I1.MII1
'
Mile
le 1 Ull
I1lIMEEEE ... " 11.... ul,/AMU'. 1.M1r.1
u,íEMIU.
Illll.
uIU.....uu...MUUll1r.1111rA
OSSO
llll/i.Oluu
I
. I
0m umll... M. .
.M. I..u/'nluO:IU /m1'.u1r'.1 Po':
Rr50/0000.
N
uullllllE
IulmlllM.E
OMM
M 'r-.
,.I .0 I_Á..IUII.MUM111
Muu'd1.
.-ma
NI
Maill
MI
IIME
Iu1M..
lulmVU11MM/
1
.M..
M.M.E'JO.
MiMiU'.u/1.u:.1111
MMr..r.ínr:
1 IMM.M0lE10
00u
111M.M
50 . 1
...MM
.MM M.'i
Ií..í
.M7M!ar.