the In which one of our redoubtable English cousins does a Professor Higgins on Heath's metered sine- square oscillator and turns an Eliza Doolittle into a lady of ravishing style and performance. WW WOULD YOU like to own a gen- uuinely high grade sine and Square wave audio generator? Just look at these specifications: RANGE: 1 Hz - 1O0GkHz sine or sa. RISE TIME: 12 nanoseconds T.HsD.: 04078; typically, 0-002- 90-0053, ACCURACY: better than 49-5% PLITIDE STABILITY: #1 dB, Yes, I can hear you say, ani where'do I rustle up the $400 or nore for what is obviously a labor- Story standan instrument? Well, you lucky Audio Amtour subscriber, read on, for you can be the proud omer of an instruzent in this class and for a bare fraction of that price tag. (he of sy professional tasks Is evaluation of new products for the ost popular of English audio mga- dines Hi-Fi ews, Last year, T had the good fortune to undertake a re- view of the nev Hesthkit model. TG 18 audio generator, This transis~ tor version of their pepalar cubed model Ig-72 has identical perfor- mance and ranges, but also includes 4 square wave section, Heath's specifications of the ne model are modest, with a clain for Tess than 0-18 distortion, 35% fre- quency accuracy and a square wave Copyright © 1971 by Edward T. Dell Jnr. All rights reserved. By REGINALD WILLIAMSON rise tine of better than SD nano- seconds. In fact, a tolerably good performance for Something aimed at the service shop or the not-so-eri- tical amateur. The Kit had a couple of errors when I came to check it out, but these were corrected with the’ willing cooperation of the British comany.(I'Il be mentioning these later on.) In ny subsequent appraisal of the product, therefore, I was pleased to give it a wam welcome, with a nild qualification about its total harmonic distortion which it barely pet. What about ny warranty? Later, hovever, while idly studying the schematic, 'I began to realize Heath had given us a laboratory standard instrument in embryo, Since the basic circult desi gn philosophy was sound and needed just a few refinements here and there, to bring about quite a startling perfomance inprovenent. To begin, you will need a Heath 1-18 audio generator kit costing $67.50. One word of warming, hox= P1G.1. Sampliried schematic or the operational amplifier Meath used in the 16-18 design. The triangle is the amplifier; its- input the in~ verting, the + the noninverting, ever. Quite rightly, our friends in Benton Harbor are perfectly en- titled to invalidate their warranty if the ower makes unauthorized changes in their kit. Their excell- ent construction mnuals usually make this quite clear, so first, you should build the kit as the manual directs and get it working exectly as they specify, In this Way you can be sure the kit is free Of basic flaws before making chang: es in it, If subsequent to modifying the kit, you have trouble and want to return it to lleath, then you must be prepared co remove the modifica tios 1 shall be describing, On the other hand, I'm prepared to give assistance to any Audio Amiceur reader who may write to me in care of the magazin How these osci ators work ‘The 16-18, in common, with most au- dic signal generators in use today, is a resistance/copacitance (RC) type, which means the frequency of oscillation is determined by sharp- Iy frequency selective resistance/ capacitance networls, Sosetines, the networks are in the bridge for as in the so-called Nein bridge os clllators: or, as in the case of the 16-18, the frequency determin ing elements are in a "bridged T" confi gurati This network type simlates a broadly tuned resonant circuit, and in the IG-18 is inserted in the de- generative, or negstive feedback Path of a high gair amplifier, Nog- tive feedback is at a naximin, ox fept at its notch" frequency--so if oscillation occurs, it can only be at this frequency, So by simply altering the values of Ror C, the frequency of oscillation is deter- mined very accurately, to a toler lance defined by the accuracy of the bridge components, What small er- rors that do occur, could be caused by phase shifts within the amplifi- er; s0 this mst be designed care- fully and have a bandwidth well be- yond the generator's frequency coverage. ‘The operational amp and loops. Heath's designers have come up with What is essentially, a very high gain operational emlifier, vith Tow output impedance and two in— puts, non-inverting and inverting. he ic bridge is in the inverting (or degenerative) path; the non-in- vexting path is the main regenera tive loop and is also the amplitude controlling network. (See Fig.l) T should explain at this point, that because of the low "Q" of RC tuned networks, it is essential to include them in an active filter configuration to insure low harmon- Issue 1/1971 7Copyright © 1971 by Edward T. Dell Jnr. All rights reserved. 8 ic generation in an oscillator of this type, It is highly desirable tat the forward amplityiag path should be as high gain as possible, but at all tines, irrespective of the insertion losses of the fre- quency determining networks, the Slosed loep gsin should be just above unity. The positive loop has in it « positive temperature coefficient thermistor so as the amplitude of oscillations rises, it heate, re Sistace increases ad the ve feedback is reduced, thereby close Ay controlling the Closed loop gaia at all frequency control element settings. Checkout before modification As it stands, the 1G-18 checked out well on this’ count, over the entire Tange 1 Hz up to 100KH2 4% met their specification of #148 varia~ Hon. Distortion wasn't quite as good as. specified, typically around +12 to O-14%, rising appreciably At the ends of the spectrum. Har- tonic products were 2 disconcerting mixture of mainly second, third, and higher orders. Hum also contri~ buted to the residue hen the pro- ducts were checked on an ofcillo- ‘scope. . Accuracy was poor st sone points and white even the 25% isn't too wonderful for a decade instrument, fat sone scale settings the 16-18 was 128 out! Finally, the meter Grove me nad at frequencies below Lolz, when ie started to road indi— FIG.2 Jotens ad muir sitches P16.2. Heath designe ©1968, The Heath Co, Used by perm vidual half cycles in wild needle oscillations, becoming virtually useless in these rangos. Modifications: first step I decided, first of all, to deal with the serious blenishes--the Frequency inaccurscies, This tumed out ‘to be an error in assembling the kit--very rare for Heath, in wy experience. Heath improved the 1G- AS over its ancestor, the 16-72 by providing a continuously variable third significant figure, a mst useful feature when one is checking resonant circuits. The change vir- tually removes the major disadvan- tage of a switched decade type in- strument, ‘Unfortunately, the design re quizes the extra resistive elengpt ‘to be svitched out altogether then an the "0" position, Te ordinary: gauged potentiometer used made it Impossible to achieve this condi- tion, So when the resistive. compo- nents in the bridge were relatively high in value, an error of up to 108 was inevitsble, What Heath should have used was a rheostat, After their attention had been dram to this, I was assured the errer has been corrected, But if you have nade the kit already or will be getting one, it is a wise Precaution to check’ that this com ponent--a 1 negohm ganged control, R, in the Heath schematic~-does 0 open cireuit when in its zero or fully anticlockwise position, (See Figs 20) Added 2 variable rheontat to the three muitchen of the oscillator. The dual INQ Rt goes open circuit in its minimm position, author Williamson says the unit works better if the #2xi resistors which heath supplies to go across the rheostat’s termnais are replaced wich 42000 units. The control adds a variable cne cycle to whatever Frequency the network's three switches produce. The: Auntie Anateur: ‘Step two. One other small layout: error Heath has also accepted, and corrected, was piving rise to’ ex- cessive crosstalk between the square and sine wave outputs. AIT that was necessary, vas to shift 2 common ground retum for the sine and square output attenuator sec tions, The one for the square wave section is now separste and an ox- tra wire is taken from the attenu- ator direct to the ground line on the wave generator etched cireut board, ALT Heath's mnuals subse quent to Dec, 5, 1968 are amended, You'll find the dave of issue on the front cover, bottom right. MODIFICATION PARTS LIST Step one: 1 dual Ima linear theostat (check your kit before ordering) 2 100s, 42% We rest ators Step two; One short length of hookup wire Step three: 30V bead tantalum cape: 1 100uF @ 6V electrolytic caps Step fiver A7ka 4 resistor 45% 1OKa $M resistor 25% twa Iv resistor 25% 1OuF @ 16V Electrolytic cap, W0OuF P 16V Electrolytic e 4owO8 transistor (RCA) step six: 2 1000uF @ 64Y Electrolyele cap. Step seven 1 S:6Kn JW resistor 25% 1 250uF € GV Electrolytic cap, 1 Germanium diode (1N34, 1191) Stop eight 1 2N5087 (Mot) transistor 1 220KR sw resistor 352 Step nine 1 40361 (RCA) transistor (04) 1 49362 (RCA) transietor (@5) Step ten AIK AV resistor 25%Copyright © 1971 by Edward T. Dell Jnr. All rights reserved. FIGS Inplifier attacked These changes move frequency accur~ acy to vithin specification. The actual check-out on a laboratory comter measures 0 to 15%; quite good, of course, but--vell, for a Suitched generator it could be a ot better, At this stage, having got ny review out of the way, 7 felt more ang pore convinced’ the G18 could be mich inproved. T needed a really accurate, low dis- tortion genezator for my’ own use anyway. Wy original, also one oF ay om designs, was getting a little an- cient afd its accuracy a little warward, A bright and sunny English Sunday, "late in the sumer, there- fore, found ne aitting in my den surrounded by frequency counter, distortion measuring set capable of reading distertion products down to D+0018,, osci lloscepe and sundry other bits and pieces of test gear. Step three, I made my first attack gon the amplifier section of the sine wave generator (See Fig. 3), since the harmonic content of the generated waveforn was closely de- Pendent upon the open loop gain, First decided to increase the of a bypass capacitor an the base bias ciresir for Q2 of the differ- ential input stage. Heath thenselues Seemed to have had a bit of a prob- Jem deciding on the value for this Lopacitos, which goes to the ground Line from the slider of 89. The etched circuit board diagram indi- cates it should be a 0-005uF, whereas a slip in sy manual aid O-6EuF, which is what was supplied, My caleulations indicated both were wrong. had a hunch that in Fact it should be a b-BuF, 30 1 Fitted a bead tantalum type (not less than 30V, rating) aad got the Hirst big improvement, Accuracy was now within #0+3% over the entire Tange and total harmonic distortion dropped to below O+1% over nearly all the range. Step four, With the bit now well between my teoth, I decided to a tempt to further increase the for- ward gain of the amplifier stege. 1 \ddea'a 100ut (0V.) bypass. capac: tor across the emitter resistor of Q3 (R11) on the underside of the etched cireuit boart, Modified retering By now, however, I noticed the dis- SCHEMATIC OF THE SINE-SQUARE AUDIO GENERATOR copyrlgnt © 196% Heath company MAN Fights reserved, Used by parmisston, tortica Was very dependent upon the setting of the "Amplitude Fine" control, with high order products predoniAant, The reason for this was obvious, the loading of the teter circult and the nonlinear detion of the diodes, The only ms- ver wes a buffer amplifier and this Ys the only major modification, Stop five, The added cireuit, Fig. 4, 4s a simple emitter follover, ghite easily made up on perforated Yoard which can be suspended below the etched circuit board of the sine generator, The transistor T Used isn't too critical, so long as At has a Veeo rating greater thn 45, Tused‘an ACA 40408, This Left the distortion products quite clean and mainly low order, Step six, At this point, the level of harmonics wexe getting sifficult to read because of hun and noise, $0 the power section was isproved by the simple addition of extra smoothing prior to the rogulat: sdded two capacitors, 1,000uF. ach, 60V, working, across the ex- isting Cland C2 (See Fig. 3), sol- Goring then dizectly scroce the power supply etched circuit board. “Continued overleaf red Issue 1/1971 9Copyright © 1971 by Edward T. Dell Jnr. All rights reserved. 10 Step 7--Damping sone waggles Before leaving the meter circuit, I decided to do something about the poorly damped novenent. and added = Tow voltage working, 250uF, capaci- tor across the meter terminals. The meter action linearity siso left a Little to be desired, due to the nonlinearity of the tectifiers, D3 fand 04, This vas corrected by add- ing 2 5-6X resistor and a gemaniun diode across the meter terminals ax well, Almost any snall diode will do, provided it 1s a germanium ype, ‘fake care in adding the diode and capacitor to connect the proper polarities according to Fig. 4. ‘This tightens up the meter accuracy incidentally, to within 228 as ag- alnst Heath’S specified 25%, The metor nov reads without that vio~ Jent needte waggle that irritated me so mich, dow to IH: where the fluctuation in reading is less than #103, Step efgit, 1 increased the current gain of QS by converting 1t into a Compound pair, i.e, adding another transistor, The adiitional transis- tor, Qda, shown in Fig, 5, should be a high hee, low noise device With a Vego 8f not less than 45V. and mist, of course, be a pnp. One has a variety to choose from and I used a sptorola 2Ns087, You'll find it quite easy to add the extra transistor, Q5a to Q3. Just unsolder Q8's base lead (with a heat sink attached) and then lift Le up out of the hole in the board, Bend this up ina "U" shape. Push the base lead of the Q3a transistor ‘through this ole, and solder. The collecter lead of QSa stould bbe put through the sane hole as that of the original Q3's collector, and also soldered, Now join the base lead of QS to the emitter lead ‘of QSa, along with one ond of a 220k resistor, The other end of this resistor goes fo a convenient ‘tye (sie) point on the etched circuit board. T found it @ good idea to drill a small hole (#58 drill) near the +ve end of R1l and solder the 2201 to the same section of the copper Iaminate, Step nine, T changed the output transisters, Qt and Q5 to larger chip type, substituting the mst inexpensive available: an RCA 40361 for QS, and RCA's 40362 for QS. (See Fig. 3) Step ten, Change RS in the differ~ ential input from its original val- tue of 4-7knto 472, The new specifications These are about all the modifica tons any hone constructor is able to do, ‘The oscillater should now be The Audio Anateur FIG. 5, Details of adding a trans Astor to 93 to make 1¢ a compound pair. See text. F404, Additional circuitry for isolating che yeter from che oxci amplifier and for danping low frequency cscillations in neter rea: set up in accordance with the in- structions on manual pages 41 the rough 43, Please keep one minor point in mind, The adjustuent of fhe feedback control, R7, Le new going to be rather shazp, and hence A little more difficult to set, Mso, when adjusting the bias con- trol, RQ, have the feedback control ser to the point where clipping is just noticeable on the -ve and sve half cycles of the waveform, You All achleve the best results by Setting the bias so these clipping points are as symmetrical as possi- Ble, Now, after doing all this, what are the pevulte? You may rest a sured, performance of your 16-18 Vill now be up to laboratory stand- ard, The distortion products of sy unit are typically 0-038, with noise contributing an equal anount so that the total is alvays less than 0-01%, Accuracy, as I said sarlier ie 40-S¥ as the instrument Stands, If you can get the use of a digital counter, you could do as T have and with a litcie brigge cir- cuit padding here and there, get it ts close as) O-2¥, Getting the best out Hecause of the circuit design, the unit Will perform best when the FiG, 6.Rise time of the square wave section of the 16-18, Distance be~ tween craticule divisione=0-luser. The author says the tiny ring at the top is test Jead inductance, foarse Amplitude" control is in its highest position, permitting appreciable attenuation with the "Fine" control. You will get mini- mun noise products wheaever you can use the "Tens!" frequency switch, also. The nature of the discortion products is now mich mre clearly defined and almost pure second har- menic--ideul for amplifier measure ‘So far, T have said nothing about the squate wave section, T do not need to, Heath have undersold then selves here, because it is way, vay better than their modest claim of 50 nanosecond rise tim, I checked this on a Tektronix oscilloscope which has an inherent rise time on its verticel amplifiers of 12 nand- seconds, My "scope picture (see 7) shows it is at least equal, if not very mich better than this-- so absolutely no changes are neces- sary. Now T wonder if Heath might be thinking of updating their Harmonie Distortion Analyzer? While ve have no inside inform tion from Beaton Karbor, we do have a modification of Heath's 158 in fie works for publication coe. Tape butts will be particularly interested,--£4,)