Professional Documents
Culture Documents
brightness of the scene. Insofar as this form due to a half-white half-black field
average value will change, the signal con- with a horizontal line of demarcation as frequency components, the problem of
tains low-frequency components which indicated in Figure 2(A). This is a video investigating transmission deviations in
extend in frequency upwards from zero. signal complete with horizontal syn- the low-frequency region is considerably
Figure 1(B) shows the same signal chronizing pulses, though for simplicity simplified. For this purposeit issufficient
after the blanking process. Here it may the vertical synchronizing pulses are not to consider the wave of Figure 2(C) only,
be seen that the effect of blanking is to shown. If the average value of this signal on the assumption that the wave of
multiply the original signal by a blanking is removed, due to lack of d-c transmis- Figure 2(B) will be transmitted faith-
wave form which is unity at all times ex- sion, the tips of the synchronizing pulses fully.
cept in the blanking interval, when it is will extend below the zero voltage line, Consider the effect of a single low-fre-
zero. This multiplication will be recog- in a negative direction. If the average quency cut-off consisting of a series capac-
nized as a modulation, where the blank- value of the signal changes, the extension itor and shunt resistance, of time-con-
ing wave form, shown in Figure 1(C), of the synchronizing pulses in a negative stant T. For a value of T moderately
plays the part of the carrier. By straight- direction also will change. This effect has
forward Fourier analysis it may be shown been called the "zero wander."
to consist of a d-c or average term plus It may be seen that any process which Papr 50-72,nrecomtmended rby tthe AiEE Televisio
the fundamental and harmonics of the lines up the tips of the synchronizing by the AIEE Technical Program Committee for
presentation at the AIEE Winter General Meeting,
line scanning frequency. -pulses at the zero, or any other fixed, New York, N. Y., January 30-February 3, 1950.
The result then of modulating the levrel reintroduces these low frequency Manuscript submitted November 16, 1948; made
original signal with this nonsinusoidlal components lOSt in transmission. aalbefrpitn eebr5 99
carrier is to obtain new frequency com- Moreover, it is evident that this re- Bell Tele'phoneLarLboratWoriesInc, aNew York,tN.he
5 1 1 \1 I9]
0 S Figure 2 (right). Waveform for J II H
> (B)
< IIA. The complete video '"0.85
0 ~~~~~~~~~~~~~waveform c
'J' ~~ ~ ~ ~~(B)B.
The high-frequency com- 0.2i
ponent °c
C. The low-frequency com-
ponent 0.32
D. Low-frequency cut-off dis- 0
o "
~~~~~~~~~~~~~~tortion of the low-freauency -0.32
(C) TIME component (D) TIME
large compared to the field (60-cycle) requires large coupling capacitors with a cuit or camera to another and is seen as a
frequency, the effect of applying the resultant high parasitic capacitance. low-frequency damped oscillation. For
wave of Figure 2(C) will appear as shown Furthermore, under these conditions, these reasons the benefits to be derived
in Figure 2(D), that is, the main charac- the use of transformers as coupling ele- from low-frequency compensations are
teristic of such distortion is to produce a ments has been impossible due to the limited.
slope, or tilt, in the wave form. difficulty of incorporating the extended
As the number of low-frequency cut-off low frequency cut-off and wide band Regeneration of Low Frequencies
circuits in the transmission system is in- width into one design.
creased, the resultant distortion remains One method of reducing the low-fre- The analysis of distortion has so far
essentially the same provided the time quency range is to utilize a circuit which neglected the distortion produced by the
constant of each is increased in proportion reduces phase shift at the field frequency. deletion of the d-c or average value of the
to the number of circuits. When viewed This is the well-known plate decoupling video signal. For a still picture this
on a picture monitor, this type of distor- or low-frequency compensation circuit, distortion could be eliminated by a man-
tion results in a shading which is most which is shown in Figure 3. The effect of ual adjustment of the background bright-
noticeable in the darker portions of the this circuit is to place a "bump" in the ness of the picture monitor.
picture. It is evidenced by a decrease in transmission characteristic, usually well When there is motion in the picture,
brightness in the region immediately below the field frequency. The phase however, an automatic means is re-
following the white portion of the picture, shift at the field frequency is materially quired. This need is met by the so-called
which decrease in brightness shades to an reduced. The net effect is that the "d-c restorer," a schematic of which is
increase in brightness in the region im- "slope" or tilt of the low-frequency signal shown in Figure 4. The operation of the
mediately preceding the white portion of is eliminated. If the gain bump produced d-c restorer is as follows. The diode is ar-
the figure. extends up to field frequency, a slight ranged so as to conduct on the synchro-
A change in brightness corresponding "bow" distortion may be produced. nizing pulses. With the diode resistance
to a slope in the low-frequency wave, With this artifice it is possible, at least low enough in the conducting direction,
Figure 2(D), of 1 per cent, for example, theoretically, to reduce the size of the the coupling capacitor will be charged to
would be produced by a single low-fre- coupling capacitor in the interstages and such a value as to just reduce the diode
quency cut-off element having a time con- compensate, by means of the plate de- current to zero. The voltage on the
stant of 0.4 second. In terms of the coupling circuits, for the distortion pro- following grid at this instant will be zero,
steady-state transmission characteristic, duced. and hence the video signal will be positive
the 3-decibel loss point occurs at 0.4 There are two difficulties attendant to for the picture time. If for any reason,
cycle and the phase shift at 60 cycles this procedure. The first lies essentially such as zero wander, the synchronizing
amounts to 0.38 degree. For large in a balance of distortions. This may be pulses tend to go more negative, the
numbers of cut-off circuits in tandem, the seen most readily by considering the coupling capacitor is quickly charged by
60-cycle phase shift adequately describes plate circuit as a predistorter and the the diode so that the minimum signal at
the resulting wave form slope. grid circuit as a restorer. Obviously if the grid is again zero. If on the other
The extremely low frequencies which the predistortion and restoration are con- hand the synchronizing pulses tend to go
must be transmitted, as indicated before siderable, a precise balance between the more pQsitive, the diode ceases to con-
causes considerable diffculties particu- two is required to limit the net distortion duct. However, the shunt resistance
larly when coupled with the 4-mega- to a threshold value. A further difficulty across the diode discharges the coupling
cycle band width required for the trans- arises from the magnitude of the gain capacitor to such a value as to cause the
mission of fine detail. This is a serious bump produced. This is serious only diode to conduct again on the synchroniz-
practical matter, since to obtain wide when a considerable number, ten or more, ing pulses. Hence the line-up of syn-
band widths and reasonable gain from of coupling circuits are employed. chronizing pulses is maintained. In
vacuum tube interstage circuits the para- The effect of a high gain bump is most effect, the d-c and those low frequencies
sitic capacitances must be made small. noticeable on switching transients such lost in transmission have been regener-
To pass the low frequencies, however, as are caused by switching from one cir- ated by the d-c restorer.
0
t ~~~reution
to~~~~~ by_5,.11 3
TIME -4FREQUENCY {N CYCLES PER SECONO
< -~~~~~~~~~~~~~~~~~~~~~~bpn 20 - 4 0002040 -0 40080
Figure (cright). el n-
Peekeint e -4c 5
reduction byc45of t - w-
tem i
clamping 20 40 60 100 200
400 1000
TIME ~**FREQUENCY IN CYCLES PER SFCOND
4000 8000
low-frequency wave form, Figure 2(D), advantages, as video coupling elements. quency nearly equal to scanning line fre-
which accumulates during the time of one The subtraction of the sampled en- quency and the resulting low recurrence
scanning line. The amplitude in the velope of the input wave from the input frequency of the derived envelope.
nonclamper case is the slope which ac- wave itself is as effective in reducing low- Reference to Figure 10 will serve to
cumulates over one-half frame scanning frequency interference as it is in reducing show how this generation of low beat fre-
time. For equal distortion amplitudes the distortion caused by the low-fre- quencies may degrade intolerbly a sys-
in the two cases, a possible reduction in quency cut-off of the transmission sys- tem which is otherwise acceptable. The
the required low-frequency time constant tem. The low-frequency interference may signal-to-noise ratio which is required of
due to the use of clampers is about 65 be comprised of any number of low-fre- an acceptable system falls off rapidly as
times. The lowered value of the cut-off quency sine waves which are added to the the frequency of the "noise" is increased.
time constant simplifies video interstage video wave fornn during transmission. If from a high-frequency interference
design and permits the use of When this wave is sampled at a rate having a low signal-to-noise requirement
transformers, with their attendant which is high compared to the interfering the clamper generates a beat frequency
frequency, the derived envelope is simnilar having a high signal-to-noise require-
in form to the interfering wave but pro.- ment, the effective requirement at the
gresses in the manner of a staircase. As high interference frequency is determined
the interfering frequency is lowered the by the requirement for the generated
sampled envelope becomes a better and beat frequency and becomes severe at all
better approximation to the input wave frequencies.
so that after substraction a smaller and When the interference is in the form of
smaller interference remains. The re- wide band random noise, the derived
sidual interference has a maximum value envelope has a random value from line to
given by the maximum change of the line which is unrelated to the noise during
Figure 7(A). A white rectangle with a grey input wave in the time of one scanning the picture time so that the envelope when
background. No low-frequency cut-off dis. line. Where the time of one scanning line added back into the video contributes an
tortion is To and the interference is of the form; additional random interference. This
e=sin ot, the remaining interference, R, interference has the appearance of streak-
is given by: R=sin w(t+To)-sin cot iness in the picture. It might be well to
and has a maximum value, (when wt= - point out that the clamper is not af-
coTo/2), equal to 2 sin wTo/2. This func- fected by the characteristic noise of the
tion is plotted in decibels in Figure 8(A) camera tube because this noise does not
for values of interference frequency, (f), appear in the synchronizing region.
from 20 cycles to 15750/2 cycles. This Random noise added into the system be-
curve represents the ratio of peak output yond the blanking stage will presumably
distortion to peak input distortion exist in the synchronizing region and will
achieved by ideal low-frequency clamping generate the streakiness in clamper output.
Figure 7(B). The effect of a large amount of of this form. The suppression of 60- A possibility for reducing the amplitude
low frequency cut-off distortion cycle power frequency interference by 33 of beat frequencies generated by the
decibels might alone justify the addition clamper exists in the finite duration of the
of a clamper in many situations. synchronizing pulse during which time
High-frequency interference is affected the envelope is derived by sampling. A
by clampers in a manner altogether dif- process of averaging over this duration in
ferent than low-frequency interference. the derivation of the envelope yields a
The derived envelope no longer resembles beat frequency amplitude which is equal
the interfering input wave but has a low to the interference wave amplitude mul-
recurrence frequency which is the beat tiplied by the function, sin ifT/lrfrTwhere
between the interfering frequency and r is the time duration of the synchroniz-
the nearest multiple of scanning line fre- ing pulse and f is the frequency of inter-
Figure 7(C). The effect of clamping applied quency. The process is illustrated in ference, (see Appendix II). This function
to the signal of Figure 7(B) Figure 9 which shows an interference fre- at high frequencies decreases as l/f.
Z200
Figure I11 (right). The (C2C.)tZ
ENVELOPE INTERFERENCE-s' reduction in amplitude 1 -25t heZ
n
of generated beat fre- in ba y im
quency as a function of ha -
Averaging of this kind will reduce the Theamp time constant IRE FQN INoCLE SteCrn
PER
nstat
me CYCLES
SEc oND
PER
from high-
that they will no longer affect the inter- beat frequency versus the interference region from 2.5ke to 100 kc the maximum
ference requirement at high frequencies. frequency is plotted in Figure 11. The amplitude of the generated low-frequency
At frequencies between 2.5 kc and 100 derivation of the curves is developed in beats is not much reduced by the time
kc the function sin 7rffq/nfp has not Appendix I. constant of the clamper so that in this
yet become effective in reducing the The addition of a time constant to re- region the transmission interference re-
generated heat frequency amplitude. In duce the generation of beats from high- quirements are about 6 decibels worse
this region the signal-to-noise require- frequency interference reduces somewhat when dampers are employed.
ment is perhaps 6 decibels worse with the efficiency with, which thecdamper
dampers than without. This seems to be acts to suppress low-frequency distortion Clamper Circuits
a minimum penalty to be paid for the use and interference. The increase in low-
of the dlamper. frequency dlamper output distortion as The usual form of damper, which we
The averaging of the values of the in- a function of the damper time constant is shall designate as the series capacitor
terference over the time duration of the shown in Figure 12, for the standard damper, is shown in Figure 5. In this
synchronizing pulses is accomplished by wave form having a synchronizing pulse case the envelope is obtained across
means of a time constant in the envelope width equal to 5 microseconds (see Ap- capacitor C by the periodic closing of the
detector circuit. A resistance-capaci- pendix II). The curves of low-frequency switch, S. The closed period corresponds
tance circuit has been employed thus far interference reduction, Figure 8, show the in time to the synchronizing pulse time
and modifies the generation of the en- same effects ofcamper time constant to of the video wave form so that the capac-
velope as follows. The envelope follows reduce the effectiveness of the lamper itor is charged to the value of the video
the low-frequency wander of the video as a low-frequency interference sup- wave at this time. The charge acquired
wave form abruptly at the beginning of pressor. during the closed period remains un-
the synchronizing pulse when there is no Reference to these curves and the altered during the open period, and there-
time constant. When a time constant is curves of beat frequency amplitudes, fore represents an envelope of the peak
employed, however, the envelope voltage Figure 11, show that great reductions in voltage of the video wave form during
cannot change suddenly, but during each beat frequency amplitude at high inter- the synchronizing pulse times. The out-
synchronizing pulse approaches expo- ference frequency are possible with only a put voltage is then obtained as the dif-
nentially the changed level of the video slight decrease in the low-frequency per- ference between the input voltage and the
Waet form teionstantainose nyelp ter- formance of the dlamper. Evaluation of sampled envelope.
deectrin follow awigh-frequteny tiontr
feringwaveformi attenuat
a
without ions-
these data make it possible to choose a
value for this time constant of about 2.5
The rapidly closing switch is achieved
electrically by a pair of diodes as shown
during the synchronizing pulse so that microseconds as being close to an opti- in Figure 13(A). These diodes are caused
at the end of the synchronizing pulse the mum value. With this choice, low-fre- to conduct during the synchronizing
instantaneous value of the interfering quency performance is degraded about 1.5 pulse intervals by driving pulses of the
wave remains as the envelope voltage, decibels while the beat frequency output same form as the synchronizing pulses.
When a time constant is employed the in- at 1 megacycle has been reduced 22 dec- The diodes and pulses taken together
terference is attenuated so that the en- ibels. This value of time constant has constitute a switch between the signal
velope voltage at the end of the syn- been incorporated into the dlamper cir- point and ground because periodically
chronizing pulse is less and less deter- cuits to be described. In the frequency they present a low impedance to ground.
mined by the interference as the fre-
quency of the interference is increaed. Figure 10 (left). The 22 - - - - -
The mplitude of this generated envelo relative visibility of 2 8 __
0~~~~~~~~~~~~~~~~~~~~
single-frequency inter- 1 _ f-
35
ferenc101.5 20 25 30 35 40 4550 icture ewingt disa c;
- t H t - - - staneis
(f tnheo°n 0c 111sl
ve tie
zr FRQECwNMGCCE E EODt iemcoscns LME IECNTNNMCOEOD
190 VOUM
- 9 Doba RikeClmPr in --e
Trnsisin 8
( I I microsecond capacitor charging time con- plifiers having amplification factors of
z
L o | stant required and the minimum value of - Al and +2. The outputs of the am-
| charging resistance which can be secured. plifiers are coupled to the diodes which
E Eo This charging resistance is comprised of are polanrzed to conduct during syn-
HFlFt t+ the resistance of the diodes and the re- chronizing pulses and remain nonconduct-
P Ap sistance of the signal source. The signal ing between synchronizing pulses.
I X 1I source, has, therefore, been chosen to be a When Al and A2 are of equal magnitude
(A) cathode follower and may have an im- this arrangement provides the equivalent
pedance of 100 to 200 ohms, The effec- of the switch, S, together with diode re-
Zr EO=E- tive resistance of the diodes when they sistance, r, and distortion, 'A, shown in
Eor+Z
are conducting is of the same order so that Figure 13(B). If A,i is greater than L2
E +A rAz
z
a total charging resistance of 200 to 300 a negative feedback circuit exists. The
ohms is obtained in this form of circuit. magnitude of the net feedback (1+,u) is
r________________________ Such a value yields 0.01 microfarad as equal to 1 +(Al -2)/2, and is useful in
(B) about the maximum value for the charg- reducing the effects of the diode distor-
ing capacitance. With this value of tion and resistance as in Figure 13(C).
z r Eo= r capacitance a leakage resistance of 10 When the diode driving pulses are
(Ts+,w) + Z megohms would yield the amount of derived from the video wave form there
EB3 (FfA) z horizontal shading equivalent to a 0.1 is a requirement that the maximum en-
|() \, ,)+ z second time constant and is considered velope distortion from one line to the
to be a tolerable value for the shading due next must not exceed the amplitude of the
(c) to a single circuit element. synchronizing pulse amplitude se-
since
°'~ ~ ~ ~ ~ ~ ~ 05
UF.
- S z UF. < 43 t | 3 | °° M 1
200 V.
GASTRIODES
DIODESAREARE TYPETYPE2C51OB'2 K VL0I_ OUTPUT AMP
_O__ MFM F M0V
V 3 V17 57 0 _V8A
300 0300J300
8.2K 2K
Z fO/
l_ V e (2)
decibel which is constant over the video The increase of the low-frequency cut-off fo fo
bandband o freqencie. If the
of frequencies. te impedance
impdance frequency means that reasonably sized units
and hence, with this substitution in equa-
may bcosrcted to cover the video
irregularity produced by this shunt is un- transmission band. tion 6
to
put leads. These resistors of about 7 Appendix i. D tor offo (1+jwT) (l- /r+i/fV)) (7)
ohms will provide 75-ohms impedance in EnvelopeIntiisacelo,henvophs
each direction and a loss of 1.6 decibels. Let the sampling frequency be Jo, and the same period as the input wave only if
Some experiments have been carried the duration of the sampling time infinites- the frequency is less than half the sampling
out using 75-ohm to 110-ohm transform- imally short. Then for an input wave frequency. In only this case is itproper to
ers at the dlamper terminals. With E =E(n/fo+t) where 0 <t <l1/Jo, and n is an speak of a residual distortion as being equal
this arrangement signals on balanced 110- integer, the resultant derived wave may be to the differehnce beitween the input and the
ohm circuits may be clamped. Using dsgae yevlp.Ti sgvnb
197A repeating coils the clamped output V-En (1) E-7t-n=e@>-7r
appeared to be free of low frequency dis- E(f-+I
-O / fo ~ 7~
R= _ fo -erT+1(10)
J analyzed by considering the clamper input
as made up of an undistorted signal plus Ec pT
The second factor of equation 10 an additional saw-tooth voltage. Then since E n{-f+t- o
Tfo the clamper leaves the undistorted signal \ fo / 1+pT
[1 7 1fT unmodified except for the addition of a d-c = (17)
1-- e ,- 1 -fJ component which may be disregarded, the 1+
~~~~~remaining
considered. saw-tooth term need only be l+pT
is a factor which .shows how the clamper
time constant reduces the effectiveness of It will be assumed for simplicity that Where Ec is the value of the derived
the clamper in reducing low-frequency dis- this saw-tooth wave is of constant slope and envelope across the capacitor C. As in
tortion. It has been plotted in decibels of sufficiently long duration compared *to Appendix I, Ec(n - 1)/fo is the value of the
as a function of T in Figure 12. It is the time of operation of the dcamper as to envelope for the preceding interval. Then
identical with an expression equation 16, for be essentially infinite in duration. since the difference between the input and
the effects of time constant upon the reduc- We take as our point of departure equa- output wave is the envelope, we have from
tion of cut-off distortion. h tion 5 of Appendix I, and let E =(n/f- equation 17, after the sampling interval
Values of R against f = l27r have been T+t). Substituting this value of E in equa-n- Tn
plotted in Figure 8 for the following values +tio 5 weu obtain
btiing athi valeg of E i e Ec~-n-i
p~ -'E(- o/)
fo l+,u 1+A fo-r+tj
tion after integration
of parameters n
fo=15,750 cycles V = V E-r/T+_(l1_E-/T)+ pT +T
r = 5 microseconds fo fo fo 1 +,u 1+,
T=0.8, 2.5. 10 microseconds aTe _/T-cT(l-CT/T) (14) (18)
When T- -0, equation 9 reduces to In the steady-state condition, the in- Comparing equation 4 we see that it is
crease in envelope voltage in proceeding entirely similar provided we recognize that
R-2 sin | -( --X ) (11) from one sampling interval to the next must the effective time-constant has been reduced
\2\ /be equal to the corresponding increase in by the factor 1 +,, and that the amplitude
When f>fo/2, the envelope generated is input voltage. That is of the input is modified by the term ,u/1l+.
extraneous and it is proper to speak of it n n-1 n n-1 ca Hence we conclude
as a new source of interference. As before, V--V = -E =-
let w=27r(mfo+F). The beat is then given fo fo fo fo fo Ec= V (19)
from equation 7, by Equation 14 may then be reduced to l+,u
j2Fn ~ -(1+jo T))v /n \ /1 \ rT
v = e (1-e ) T (12) VhaQ T)-T -T)l v/T (15) References
(1+jT
fs T(1(1-_ : (T
(1+JwT)( i)
('+iwF/s))JThe maximum value of the residual dis- 1. THE MARCONI-E.M.I. TELEVISION SYSTEM,
The absolute amplitude of equation 12 tortion is then given by Journal, Insttution of Electrical Engineers (Lon-
may be reduced to n+1 n don, England), December 1938, pages 758-66.
R= -r - V-= at --9 X 2. THE CLAMP CIRCUIT (PART I AND PART II),
l(sinh2 r/2T+sin2 7rfr) fo / fo L\ fo C. L. Townsend. The Broadcast Engineers'
A= l T F\ (13) / 1 foT Journal (New York, N. Y.), January 1947, pages
7-9, April 1947, pages 6-9.
+c2T2) ( sinh2+sin2
I( ) (16)
\ 2T ~~fo / r3. TELEVISION D-C COMPONENT, K. R. Wendt.
Values of A against frequency have been The second factor in equation 16 indi- 1948, pages 85-(N
u.
plotted in Figure 11 for the same range in cates the dependence of the residual distor- 4. A TV SYNC STRETCHER, R. C. Palmer. Com-
values of T. tion on the time-constant T, and has been munications (New York, N. Y.), March 1948,
Only the first few cyclic variations in A plotted in Figure 12. pages 10-11.
No DLiscussion