RESEARCH DEPARTMENT

Television sound transmission by modulated pulses in the line- blanking intervals

RESEARCH REPORT No. T - 176 UDC 621 . 397 . 238 1966/58

THE BRITISH BROADCASTING CORPORATION ENGINEERING DIVISION CORRIGENDA

RESEARCH DEPARTMENT - BRITISH BROADCASTING CORPORATION

Research Report No. T-176 (1966/58)

TELEVISION SOUND 7RANSMISSlON HY MODULATbD PULSES IN THE LlNb HLANKING INTERVALS

Page 4:

Equation (1) is shown in the wrong position (bottom of first column) and should follow the first paragraph of Section 3.2.1.

Page 28; second column, third paragraph, first line:

For "poses" read "posed".

SMW 10/11/66 RESEARCH DEPARTMENT

TELEVISION SOUND TRANSMISSION BY MODULATED PULSES IN THE LINE-BLANKING INTERVALS

Research Report No. T·176 ODe 621.397.23R 1966/58

A.V. Lord, B.Sc.(Tech.), A.\1.I.E.E. for Head of Re search Department This Report is the property of the British Broadcasting Corporation and may not be reproduced in any form without the written permission of the Corporation.

---_._------.--_._------[] Tbl8 Report use8 gI unite In &ctor­ dl'lnC'.e with R.S. dOOUOlflnt PO !')686. Research Report No. T-176

TELEVISION SOUND TRANSMISSION BY MODULATED PULSES IN THE LINE-BLANKING INTERVALS

Section Title Page

SUMMARY .•.• 1

1. INTRODUCTION . 1

2. GENERAL ••• 1

3. TYPES OF MODULATED PULSES .•• 2

3.1. Pulse-amplitude Modulation (P.A.M.) 3 3.2. Pulse-time Modulation (P.T.M.) 3 3.2.1. Pulse-position Modulation 4 3.2.2. Pulse-duration \1odulation . 6

4. GENERATION AND DEMODULATION 7

4.1. P.A.M •••.•.••••••• 7 4.2. P.T.M ••••••••....• 7 4.2.1. P.P.M. With Uniform ~ampling 7 4.2.2. P.P.M. With Natural Sampling 8

5. SIGNAL-TO-NOISE RATIO 13

5.1. P.A.M. • ..•.•. 13 5.2. P.T.M. : .••... 15 , 5.3. Comparisons With A.M. and F.Y1. 16 5.3.1. A.M. • 17 5.3.2. F.M. . 17 5.3.3. P.A.M. 17 5.3.4. P.T.M. 18

6. PRACTICAL CONSIDERATIONS . 18

6.1. 625-line Broadcasting 19 6.1.1. Form of Sound-Pulse Signal 22 6.1.2. Practical Demodulators . . • 24 6.2. ~ound Distribution using Vision Links 25 6.2.1. Form of Sound-Pulse Signal 26 6.2.2. Instrumentation •...... 28

7. CONCLUSIONS 29

8. REFERENCES 29

APPENDIX I 30

APPENDIX 11 32 October, 1966. Research Report No. T~176 UDC 621.397.238 1966/58

TELEVISION SOUND TRANSMISSION BY MODULATED PULSES IN THE LINE-BLANKING INTERVALS

SUMMARY

This report reviews the properties of modulated pulses with particular reference to their application to television sound transmission. It is shown that combinations of modulated pulses and video signals could prove practic­ able for both broadcasting and link purposes.

1. INTRODUCTION tion of modulated pulses as a means for television­ sound broadcasting.

Up to the present, broadcast television systems In recent years, however, the widespread have utilized sound channels based upon either development of televi sion in the United Kingdom ampli tude or frequency modulation of a continuous has led to a situation in which the space available carrier, although a number of proposals have been in the radio-frequency Bands I and III might not be made in the past for incorporating the sound wi thin adequate to provide a sufficient number of channels the video waveform by means of time-division multi­ for satisfactory coverage of the country with two plex, using modulated pulses or bursts occurring progra mmes if 625-line transmissions were used in during otherwise unoccupied intervals (e.g. line­ conjunction with a conventional a.m. or f.m. sound blanking intervals). Proposals of this nature channel. It has been suggested that this problem appeared in technical papers and patent specifica­ 1 could be eased, and perhaps solved, if a satis­ tions as far back as 1933. ,2 facto ry sound-pulse system were developed. It has also been suggested that, even if such a system In 1946 details of a proposed practical system were not us ed for broadcasting, it could enable. were published3 which was based upon the use of satisfactory sound signals to be conveyed, say, duration-modulated pulses occurring within the line­ from a programme-origination point to the main synchronizing intervals, the amplitude of the pulses transmitter sites; the sound signals would be deleted being some 3 dB greater than the video-signal from the vi sion waveform prior to emi ssion. This excursion from synchronizing level to white level. form of sound-s ignal distribution could appreciably The system was tested and demonstrated as applied reduce the cost of sound circuits and could lead to to the 405-line system, which resulted in a rather a reliability of service greater than that obtained at unsati sfactory upper audio-frequency limit (i.e. present, where breakdown of a separate sound link somewhat less than 5 kHz). A further and probably tends to occur m ore frequently than breakdown of a more important drawback a ssociated with this parti­ vision circuit; systems of this nature have been cular proposal was its performance in the presence tested in France 4 and the U.S.S.R. 5 of fairly high levels of noise. In an attempt to show a marked economic advantage over the con­ Thi s report outlines some of the general ventional a.m. sound system, the sound pulses at properties of systems whereby sound signals are the recei ver were not extracted from the video wave­ conveyed as modulated pulses located in line­ form by gating prior to demodulation. This necessi­ blanking intervals and discusses their advantages tated the use of a limiter operating at a level near and disadvantages in terms of the applications to that of the peak of the sound pulse and caused mentioned previously. the signal-to-noise ratio of the receiver soun'd out­ put to become very unsatisfactory at a receiver-in­ put signal level that was quite usable in terms of 2. GENERAL picture signal. The effect of sampling a single tone by regularly Since the appearance of the above-mentioned recurring pulses having durations small compared paper, little attention has been paid to the applica- with the recurrenc~ period is well known; the 2 process may be illustrated by Fig, 1 which shows Thus, in any combined sound and VI SlOn the lower-frequency part of the spectrum resulting system in which single samples of the sound wave­ from the sampling process, It will be seen that the form are transmitted once per line scanning period, spectrum includes a component at Wa representing the upper audio-frequency limit is equal to half the the single tone, together with symmetrical pairs of line-scan frequency; for the 625 line and 525 line components differing from the pulse recurrence standards, the practical upper limit is likely to be frequency Wr/27T and its harmonics by the frequency rather less than 7'8 kHz. If sampling of the sound of the tone, It is possible to recover the tone waveform occurs at twi ce the line-scan frequency, provided that two conditions are satisfied: and alternate samples are delayed so that pairs of samples are transmitted during each line-blanking (a) The frequency of the tone is restricted, before interval, the upper audio-frequency limit may be sampling, to a maximum of half the recurrence doubled. frequency. Although the characteristics of sampling have (b) The samples are passed either to a low-pass been outlined in terms of a single tone, they apply fil ter cutting off at half the recurrence fre­ equally well when many tones are present simul­ quency or to a band-pass filter and detector where the pa ss-band of the fi Iter is centred on taneously (i.e. a complex wave) provided that the the recurrence frequency, or one of its har­ above-mentioned condi tions (a) and (b) are sati sfied. monics, and extends either side of centre fre­ Any system of sound transmission involving quency by half the recurrence frequency. the use of pulses which, at least nominally, recur regularly, involves a sampling process similar to Failure to observe either of these two con­ that described. Some characteristic of each pulse ditions results in the appearance of an "inversion" is varied according to the amplitude of a sample of of the tone having a frequency equal to the dif­ the audio-frequency wave;. usually the sampling ference between the frequency of the tone and the action and the generation of the modulated pulses pul se-recurrence frequen cy. are coincident. Fig. 2 shows one form of pulse, occurring within the line-blanking interval, which could be modul ated by a.f. signals.

3. TYPES OF MODULA TED PULSES

Many forms of modul ated pul se are known. How­ ever, attention is confined in this report to the two general forms that are potentially suitable for both the applications mentioned in Section 1. The two forms of pulse modulation are:

(a) Pulse-amplitude modulation and

o Wo Wr-Wo Wr W r + Wo Fig. 1 - Lower frequencies of spectrum produced by (b) Pulse-time modulation, which includes pulse­ sampling an a.f. wave of frequency Wa/27T by narrow position modulation and pulse-duration modu­ pulses having a recurrence frequency of Wr/27T lation.

------~~~~~------

sound modulatad ~ pulsa

blanking lava I

Llina sync.~

L------.-- lina blanking Fig. 2 - Sound-modulated pulse within the line-blanking interval 3

(b)

Fig. 3 - P.A.M. (a) \1odulating Wave (b) Pulses ampli tude-modulated by (a)

3.1. Pulse-amplitude rvlodulation (P.A.\1.) 3.2. Pulse-time Modulation (P.T.rvl.)

In this arrangement, the amplitude of each In P. T.M. the ampli tude of each sample of the pulse of a regularly recurring train is linearly related modulating wave is used to vary the time of occur­ to the magnitude of the a.f. signal at the time of rence of some feature of a pulse. For example, in occurrence of the pulse. The pulses may, on the one form of P. T .M. the time of occurrence of one one hand, exactly describe the a.f. wave during edge of each pulse is varied; this results in assym- each pulse interval (i.e. the waveform of the pulse metric pulse-duration modulation (P.D.M.). In. between its leading and trailing edges follows the another form the modulating wave varies. the times variations of the a.f. wave) or, on the other hand, of occurrence of constant (usually short) duration only describe the magnitude of the modulating wave pulses with respec t to regularly recurring instants. at a certain regularly recurring instant within the If the modulating wave is used directly to vary the duration of each pulse; in the latter case the instant displacement of the pulse position, the system is of sampling usually coincides with the leading edge termed pulse-po sition modulation (P. P.M.) but if the of the pulse.· However, no significant difference modulating signal is first passed through a circuit between the two forms of P.A.rvl. arises when the whose gain varies inversely with frequency the duration of each pul se is very short compared wi th pulses are found to be modulated in frequency the period of the highest modulating frequency (i.e. (P.F.M.). half the pulse-recurrence frequency), as is the case for a train of modulated pulses suitable for inclusion The two forms of P. T.M. most applicable to in a video waveform. Fig. 3 illustrates a waveform combined vision-and-sound systems are P.P.M. and corresponding to pul se-ampli tude modulation. It P.D.M. Fig. 4(a) shows part of a typical modulating will be seen that a train of amplitude-modulated wave. In P.P.M., illustrated in Fig. 4(b), the pulses may be regarded as a train of samples of the deviation in "position" (i.e. change of time of a.f. wave; thus its spectrum is identical with that occurrence) of each con stant-duration pul se, with shown in Fig. 1. reference to a regular recurring time datum, is pro­ portional to the instantaneous a.f. signal amplitude; Wi thin the audio range s et by the recurrence in the example shown, the pulse is delayed with frequency and the design of practicable low-pass reference to the datum when the instantaneous filters, the system does not introduce any intrinsic value of the a.f. signal is "positive" and is ad­ di s tortion. vanced for "nega tive" values. Fig. 4(c) illustrates 4

P.D.M., in which the duration of each pulse is 3.2.1. Pulse-position Modulation determined by the instantaneous a.f. signal ampli­ tude; a positive value of the a.f. signal lengthens It has been shown 6 that, with uniform sam­ the pulse and vice versa. pling, the pul se-posi tion modulation of extremely short pulses of unit a rea by a single tone results in It will be appreciated that P.D.M. is closely a spectrum of the form: related to P.P.M.; each pulse of a P.D.M. train has two edges either, or both,* of which may be modu­ lated in position according to the instantaneous where wa/2l7 is the frequency of the audio modula­ magnitude of the a.f. wave. P.P.M. may be con­ tion (Hz) verted to asymmetric P .n.M. (i. e. where only one edge of the pul se is modulated) by arranging that each posi tion-m odulated pul se initiates one edge W r /217 -:: liT IS the mean pulse-recurrence (leading or trailing) of a further pulse whose other frequency (Hz) edge is initiated by a pulse that recurs regularly at the mean repetition frequency of the P.P.M. train and is timed so as to occur outside the interval d IS the peak pulse-position deviation (secs) (or "time slot") occupi ed by the posi tion-modulated pulse. Similarly, P.D.M. may be converted to P. P.M. by deriving a train of constant-duration m and n are integer parameters defining the pulses eachofwhich is coincident with the position­ orders of the pul se-recurrence-fre­ modulated edge of a duration-modulated pul se. quency components and the audio­ frequency components respectively. It is evident that both P.P.M. and P.D.M. involve a sampling process but, before considering the spectra corresponding to such form s of modu­ Wi th natural sampling: 6,7 lated-pulse train, it is important to know precisely the instants at which the a.f. modulating wave is sampled. If the a.f. wave is sampled at regularly recurring instants and the amplitude of each sample is then u sed to devi ate the po si tion ofa pul se form­ ing part of a P.P.M. train or used to control the T

00 00 + ;:2 """L.J £..J"'"' (1 + -.n -)Wo.. . In(m • --)27r d. • cos (mwr + IlCU a)t m= 1 n =-00 m Wr T timing of a pul se edge in a P. D. M. train, the sy stem Figs. 5(a) and 5(b) illustrate the lower-frequency is then said to employ uniform (or periodi c) sam­ portions of typical spectra corresponding to single­ pling. However, if the effective instants of sam­ tone, low-deviation P.P.M. with uniform and natural pling coincide wi th the actual timing s of the modu­ sampling re spectively. It will be seen that both lated pulses in P.P.M., or with the actual time­ types of sampling are characterised by sets of side­ modulated edges of the modulated pulses in P.D.M., bands, corresponding to the modulation frequency the system is then said to employ natural (or syn­ and its harmonics, associated with intj::ger multiples chronou s) sampling. of the pulse-recurrence frequency.

J n {n • -Wa • --}217d • co S IlWa t J. u.: r T

00 00 2 Wo.. 2rrd + I n {Vn + n. -} • --}. cos(mu.:r + IlWa) t. T Wr LL1 ::J:-I'l) T m = n

'" In such a case the two edges are modulated in oppo­ sing senses; if they are equally modulated this results in symmetrical P.D.M. 5

I ,I I I I I I I :I I I I I I Fig. 4· P.P.M. and P.D.M. i ----+ ___ t ima tdotum-----l : maon (a) Modulating wave __ -- I I :-'-____ : pulsa (b) Pulses position-modulated by (a) ..-/ I '~ pOSition (c) Pulses duration-modulated by (a)

, I I I I I : (b) I I maon pulsa durotlon--1., I

(c)

d.e. tarm fundomantol tarm

(a)

Fig. 5 - Lower frequency portions of

W r ,2C1J Q spectra corresponding to P.P.M. with r I ClJ ' :c.lQ---,-,t----- CIJ ~o 1'5Wr (a) Unifonn sampling (b) Natural sampling

(b)

o 6

The only significant difference between the results in lower sidebands of the carrier wave at two spectra lies in the baseband components. With pulse-recurrence frequency which correspond to uniform sampling these con si st of the modulation beats between the modulation tones. frequency and its harmonics; the amplitude of the baseband component representing the modulation is For a value of peak deviation typical of a proportional to the ratio of modulating frequency to P.P.M. signal suitable for combination with a 625- the pulse-recurrence frequency and the amplitudes line video signal, the amplitude of the second-order of the baseband components representing the har­ lower sideband of the carrier wave at pulse-recur­ monics of the modulation frequency are approxi­ renc e frequency (corre sponding to the second har­ mately proportional to the square, cube, etc. of the monic of the modulation tone) is fairly small com­ above-mentioned ratio. The nature of the spectrum pared to that of the baseband component describing also indi cates that, if more than one modulation the modulation and is approximately proportional to component is present, beats are produced between the square of the deviation; the third and higher­ the various m odulating tones. order sidebands of the pulse carrier wave have very small amplitudes approximately proportional to the cube, fourth power etc., of the deviation, and the However, if the relative pul se-po si tion devi a­ beats between modulation tones are negligible. tion (diD is low (a necessary condition if the modulated pulses are to be included within a video 3.2.2. Pulse-duration Modulation waveform) the amplitudes of the unwanted baseband components, and the beats produced if several modul ati on components are pres entsimul taneously, As in the case of pulse-position modulation are small compared to the amplitude of the wanted a train of pulse-duration modulated pulses may be sideband. derived using either natural or uniform sampling. However, as indicated earlier, modulation may also be carried out either symmetrically or asymmetric­ With natural sampling only one baseband com­ ally; in the former case the positions of both pulse ponent is produced and thi s occurs at the m odula­ edges are varied in opposite directions in accor­ tion frequency and its amplitude is proportional to dance with the modulation and, in the latter case, the ratio of its frequency to the pulse-recurrence the position of only one edge is varied. frequency. Two or more modulating signals result in an identical number of baseband components and With uniform sampling an asymmetrical P.D.M. beat-frequency signal s are not produced. pulse train may be expressed as: 6*

In(n .:: 2;8) . ] I ~[L(Xl Fu (t) I • Sin nWat T n = 1 n • '::3 27T8 wr'T

7T t (Xl (Xl (n

27T~ • T

The lower-sideband structures assQciated with the pulse-recurrence frequency are somewhat similar * A train of P.D.M. pulses may be regarded as the sum for both uniform and natural sampling. In both of a train of unmodulated pulses each having a duration cases lower sidebands are generated which corres­ greater than 8, together with a further train containing pond to the modulating signal and its harmonics, all information concerned with the modulation; the second and it is apparent that they can occur, for the pulse train alternates in polarity at the modulating fre­ higher modulation frequenci es, wi thin the band from quency. In Equations (3) and (4) the tenns describing the unmodulated pulses 0. e. a d.c. tenn and tenns repre­ zero tohalfthe pulse-recurrence frequency. Further, senting the pulse-recurrence frequency and its hannonics) simultaneous modulation by two or more tones have been discarded. 7

The corresponding expressions for symmetrical 4.2. P.T.M. P. D. M. may be easily derived by regarding a train of pulses with both edges modulated as a combina­ It has already been pointed out that P.P.M. tion of two asymmetrical P.D.M. pulse trains, the and P.D.M. are closely related and that one may be first having one edge modulated and the second easily derived from the other. As a consequence it consisting of a version of the first in which the is necessary to discuss only the generation and time variable is reversed in sign. demodulation of either P.P.M. or P.D.M. In the ensuing sections the generation and demodulation As with P.P.M. the spectrum corresponding to ofP.P.M., with uniform and natural sampling respec­ uniform sampling and single-tone modulation in­ tively, are discussed. cludes baseband components at multiples of the modulating frequency. However, for low values of olT, the amplitude of the baseband component 4.2.1. P.P.M. With Uniform Sampling at modulation frequency is substantially independent of the modulation frequency and the amplitudes of The use of uniform sampling can provide a those corresponding to harmonics of the modulating P.P.M. system free from unwanted distortion caused frequency are small; with several modulation fre­ by intrusion into the baseband of sidebands of the quencies present, low-amplitude beats can occur as pulse carrier corresponding to harmonics of the in P.P.M. with uniform sampling. With natural modulation. Techniques enabling such a result to sampling and single-tone modulation only one base­ be obtained involve the use of P.A.M. as an inter­ band component occurs at modulation frequency and medi ate step in both the generation and demodula­ its amplitude is independent of modulation fre­ tion of the signal. Fig. 6 illustrates the process of quency; two or more modulation tones produce generation. corresponding baseband components but none corresponding to beats between the modulating The a.f. wave (a) is first regularly sampled (at tones are produced. the pul se-recurrence frequency) at the in stants t1 , t~21 t 3, etc. and the samples so obtained may then With low values of 01 T the lower sidebands of be applied to a "sample-and-hold" circuit in which the pulse carrier are again fairly similar for both each sample initiates a flat-topped pulse whose uniform and natural sampling; as in the case of amplitude is equal to that of the sample and whose P.P.M., these lower sidebands correspond to the duration is equal to the interval between samples. modulating signal and its harmonics. However, The resulting waveform (b) is then compared, in unlike P.P.M. the amplitude of a sideband corres­ instantaneous magnitude, with regularly recurring ponding to the modulation or one of its harmonics is ramp waves, ini tiated at the time t10 t 2, t 3, etc. and substantially independent of modulating frequency having durations equal to twice the peak deviation for both forms of sampling. Nevertheless, the amp­ required in the output P.P.M. train, as shown in (c). litude of the second-order lower sideband of the At the instant at which the instantaneous magnitude pul se carri er is again fairly small compared wi th of a ramp wave equals the magnitude of the wave­ that of the basebattd component describing the form (b) one pulse of the P.P.M. train (d) is geneI,­ modulation tone and is approximately proportio nal ated. It will be seen that the time-datum of each to the square of olT; the third and higher-order position-modulated pulse is delayed, with respect lower sidebands have very small amplitudes approxi­ to the corresponding instant of sampling, by the mately proportional to the cube, fourth-power, etc. of peak devia tion d of the P.P.M. train. olT. Simultaneous modulation by two or more tones results in lower sidebands of the pulse carrier The process of demodulation is illustrated in corresponding to beats between the modulating tones Fig. 7; for convenience the input (a) consists of but, for low values of olT, their amplitudes are the P.P.M. train of Fig. 6(d). Regularly recurring very small. ramp waves (b), each having a duration not less than twice the peak positional deviation of the P.P.M. pulses (i.e. 2d), are sampled by the input 4. GENERATION AND DEMODULATION pulses. As a result each input pulse produces a sample whose amplitude is linearly related to its 4.1. P.A.M. deviation; this results in a further train of pulses modulated in both position and amplitude. The The generation of a P.A.M. pulse train and its position modulation may now be removed by apply­ demodulation have been outlined, in principle, in ing the pulses obtained by sampling the ramp Section 3.1. The pul se train may be produced using waves to a "hold" circuit which, as in the case of a conventional form of sampling gate, driven by Fig. 6(b), results in a series of flat-topped pulses constant-amplitude pulses, to which the a.f. wave (c) initiated by the samples and having amplitudes" is applied. The demodulation of the P.A.M. signal directly related to the ampli tudes of the samples. is carried out, as already mentioned, by means of a As before the duration of each flat-topped pul se is suitable low-pass filter. equal to the interval between the initiating sample 8 and its successor; however, in this case the sam­ 4.2.2. P.P.M. With Natural Sampling ples are not spaced regularly. The waveform (c) may now be sampled by regularly recurrent pulses The generation of P.P.M. with natural sam­ having the same recurrence frequency as the P.P.M. pling is illustrated in Fig. 8; in general, it is train (i.e. a period equal to t2 - t1) but so phased similar to the equivalent process used for unifonn that sampling cannot occur within the intervals sampling. However, the a.f. wave (a) is now com­ corresponding to the deviation range of the P.P.M. pared directly with regularly recurring ramp waves; pulses; this process is shown in (c) where the no intermediate sampling process is used. As sampling occurs at the instants t1 + e, t2 + e, etc. shown in (b), the ramp waves, whose mid-points The re sui t is to form a train of regularly recurring occur atTv T 2, T 8, etc., recur at a frequency equal pulses which are amplitude modulated by the a.f. to the mean recurrence frequency required for the wave (i.e. P.A.M. train). The a.f. wave may then output P.P.M. train. The instantaneous magnitude be recovered, free from di stortion, by applying the of each ramp wave is compared with that of the a.f. P.A.M. pulses to a suitable low-pass filter. wave and one pulse of the output P.P.M. train (c) is generate d when they are equal. It will be seen

that the time-datum of output pulses occurs at T1 , From Figs. 6 and 7 it will be seen that the a.f. T 2 , T 8 , etc., and that the deviation of each pulse wave has suffered a delay equal to or greater than describes the instantaneous magnitude of the a.f. twice the peak deviation of the P.P.M. pulses. wave at the instant at which the pulse is produced.

I I I I I I I I I ~~~_.~~ __~. __.. ~. __--'---~ ~~_. __ ~.~_.~ ___. .L~ ~~.~_ ·T--·--~·-~--- 1 12 13 14

fa)

fbJ ._._.,

! - 1

(e)

(d)

Fig. 6 - Generation of P.P.M. with uniform sampling (a) Modulating a.f. wave Cb) Result of "sampling-and-holding- Ca) (c) Comparison of Cb) with regular ramp waves Cd) Output pulse train 9

(b) I I I I t... a I t,. a 12 ·a t3·a I I -----' (c)

I I I I I I I .....-.--- ..---.-----c;~.----- ,.a ..______. __ --.--1.t2.a ______-----1.._ t).a -----""0.,--

(e)

Fig. 7 - nemodulation of P.P.M. with uniform sampling Ca) P.P.M. pulse train with uniform sampling Cb) Pulses of Ca) sampling regular ramp waves Cc) Samples of Cb) after "holding- (d) Samples produced by regularly sampling Cc) Ce) Output a.f. wave

Demodulation of P.P.M. with natural sampling earlier, unwanted "inversions" of modulation fre­ may be performed in several ways. One method quency signals, due to lower sidebands of the uses P.D.M. as an intermediate step. As mentioned carrier wave at pulse-recurrence frequency, may in Section 3.2, a P.P.M. pulse train may be used to intrude into the passband of the filter, causing form asymmetric P.D.M. by causing each input di stortion. pulse to initiate one edge (leading or trailing) of a further pulse whose other edge is initiated by one A train of P.P.M. pulses with natural sampling pulse of a regular train repeating at the mean pulse­ and parameters typical of a signal suitable for recurrence frequency of the input P. P.M. train. combination with a video signal (d -:: 1'5 fl s, The P.D.M. pulse train thus formed is characterised T -= 64 fl sand fT -:: 15,625 Hz) may be converted by natural sampling and a study of Equation 4 into a corresponding P.D.M. train (with 0= 1'5 flS) reveals that the corresponding spectrum contains whose spectrum contains a lower-sideband of the only baseband components describing the modula­ carrier at pulse-recurrence frequency having an ~ ting wave. Thus, the wanted modulation may be amplitude approximately -29 dB with respect to that derived by passing the P.D.M. signal through a of the baseband component representing the modu­ suitable low-pass filter. However, as mentioned lation tone and independent of modulation frequency. 10

J J t 1

----­ / ..... - / ,/

(b)

r, (c) Fig. 8 - Generation of P.P.M. with natural sampling (a) Modulating a.f. wave (b) Comparison of (a) with regular ramp waves (c) Output pulse train

d.e. t"rm fundamczntal tczrm (rczlativcz amplitudcz 10) (rczlativcz amplltudcz 2·0) 0·28

024

020

III "0 :J ~ 0·16 a. E "'a=0·1/ o uppczr modulation "'r;lf ~ frczQuczney limit / 012 .... )(''''a =0.2 ~III L .... "'r /./Na =0.3 ./ "'r .... ~"'a =OA 0·08 :~=0.5 1:: Wr -=05a "'a =0.4 "'r "'r ':2=0·3 c.>r 0·04

Fig. 9 - Amplitudes, as functions of modulating frequency, of lower frequency spectral components of typical P.P.M. train with natural sampling

---- baseband component --- first-order lower sideband of p.r.f. _____ first-order upper sideband of p.r.f. ______second-order lower sideband of p.r.f. - .. _ .. -second-order upper sideband of p.r.f. 11

As the modulation frequency is raised from frl4* to fr/2, whilst the peak deviation 8 remains constant, the frequency (wr - 2Wa)/27T, see Fig. 5, of the ------~ lower sideband of the carrier corresponding to the second harmonic of the modulation falls from fr/2 to zero, the amplitude remaining constant. a) A second method of demodulating P.P.M. with Q. natural sampling is apparent from consideration of E o the signal spectrum discussed in 'Section 3.2.1. ~ Fig. 9 illustrates certain properties of a typical :;; c spectrum and shows how the amplitudes of the base­ Cii ... modulation band components and the first and second order sidebands of the p.r.f. vary as a function of the ratio between the modulation frequency and the pulse-recurrence frequency. Fig. 9 also shows that the amplitude of the second-order lower sideband of the signal at pul se-recurrence frequency fall s, substantially linearly, as the modul ation frequency increases; thi s unwanted component does not intrude into the wanted band for modulation fre­ Fig. 10 - Effect of correcting network on P.P .M. quencies less than half the upper limit, frl4, and with natural sampling the ratio of the ampli tude of the unwanted inversion (a) Range of modulating frequency giving rise to to that of the wanted modulation is a maximum at inversion due to: this particular frequency. It is apparent that the (b) Second orde r lower sideband of p.r.f. modulation, described by the baseband component Input to correcting network may be derived by applying the modulated pulse ------Output from correcting network train to a corrected network whose response, as a function of frequency, fall s at 6 dB per octave from For the P.P.M. system-parameters described the lowest modulation frequency to half the pulse­ previously (i.e. d c= 1'5 J.L s, T c= 64 J.L sand fr c= recurrence frequency and is zero for higher fre­ 15,625 Hz) the ratio of the inversion amplitude to quencies. However, the effect of the co rrecting that of the wanted modulation is approximately network is progressively to reduce the amplitude of -29 dB at the output of the correcting network and the wanted component as its frequency is increased independent of modulation frequency; this is a whilst progressively increasing the amplitude of the result in agreement with that obtained using P.D.M. inversion (whose frequency falls as the modulation as an intermediate step. frequency rises from a quarter to half the pulse­ repetion frequency). A third method of demodulating P.P.M. with natural sampling is illustrated in Fig. 11. As in, Fig. 10 illustrates this effect of the correcting the case of P.P.M. with uniform sampling, Fig. 7, network. At the input to the network the wanted the input train (a) samples regul arly ,recurring ramp modulation has a frequency characteristic rising at waves (b) so phased that each ramp embraces the 6 dB per octave. Modulation-frequency signals in deviation range of the input pulses. However, in the frequency range between frl4 and fr/2 (Fig. 10 ~his case, the resulting train of samples, modulated (a» give rise to inversions (Fig. 10(b» between in both position and amplitude, is applied directly fr/2 and zero; the frequency characteri stic of the to a suitable low-pass filter. inversion also rises at 6 dB per octave. The characteri stics of the modulation and its inversion The operation of such a demodulator may be are shown as solid lines in the figure; It has been understood by considering the spectrum of the assumed that the correcting network has unity gain P.P.M. pulse train and that of the ramp wave; for at fr12 and its effects upon the modulation and the the sake of simplicity the ramp wave may be con­ inversion are shown as dotted lines; their ampli­ sidered as part of a sine wave at pulse recurrence tudes are now independent of modulation frequency. frequency. Fig. 12(a) shows qualitatively the However, modulation at a frequency frl4 is doubled lower-frequency components* of the spectrum of in amplitude while the amplitude of its inversion, at P.P.M. with natural sampling together with a spec­ fr/2 is unchanged. Thus, at the output of the tral line at the pulse-repetition frequency fr repre­ corre cting network, the ratio of the amplitude of the senting the ramp wave; in Fig. 12(a) it has been wanted modulation to that of its inversion is inde­ assumed that the modulation frequency fa lies pendent of modulation frequency. between frl4 and fr/2. • Sidebands of the carrier wave at pulse-recurrence fre­ • Ir = wr/2'TT, the pulse-recurrence frequency. quency of higher order than 2 have been neglected. 12

Fig. 12(b) illustrates qualitatively the output It will be seen that, for the three methods of of the ramp demodulator obtained by multiplying the demodulation for P.P.M. with natural sampling, sine wave by the P.P.M. pulses; only components distortion terms are produced. In all cases the lower in frequency than fr are shown. It will be level of the significant di stortion is approximately seen that several components of the P.P.M. spec­ proportional to the square of the pulse deviation, trum, after beating with the demodulating sine wave, which itself directly determines the output modula­ cause output components to appear. The principle tion-signal amplitude; for a modulation signal caus­ output component occurs, as desired, at fa but ing half the peak pulse deviation, the relative level unwanted components also appear at (fr - 2fa), of each of the two distortion contributions is halved. (fr - fa) and 2fa. Applying the output of the ramp The value of d (1'5 fJ- s) as sumed in the example demodulator to a low-pass filter removes (fr - fa) quoted in this section corresponds to the peak value and either (fr - 2fa) or 2fa. If the modulation fre­ of the deviation. As the average televi sion-sound quency fa were lower than fr/4 the (fr - 2fa) inver­ si gnal level is appreciably lower than the peak, the sion component would be rej ected but the s econd­ average level of distortion would be significantly harmonic di stortion component 2fa would be passed less than that indicated. to the output of the filter. As fa lies between fr/4 and fr/2 the second-harmonic component is rejected Further, as the distortion introduced by each but, in this case, the inversion component appears form of demodulator may be determined accurately, at the filter output. itis possible* to devise correction, at the generator of the modulated-pulse train, and thus to eliminate di stortion from the demodulator output. Where all Analysis based upon the use of a sine wave forms of demodulator are used with the same pulse representing the ramp wave shows that, for the train, a compromise correction may be devised P.P.M. system parameters described previously, the which significantly reduces output distortion from amplitudes of the wanted modulation, the second all demodulators. harmonic and the inversion components behave, as functions of frequency, in the manner shown in Fig. 13. * As pointed out by 0.0. Monteath.

T, (a)

(b)

(c)

Fig. 11 - Demodulation of P.P.M. with natural sampling, using ramp waves Ca) P.P.M. pulse train with natural sampling Cb) Regular ramp waves sampled by Ca) Cc) Samples produced as a result of Cb) Cd) Output a.f. wave 13

5. SIGNAL-TO-NOISE RATIO 5.1. P.A.M.

In a modulated pul se train, information describing In a P.A.M. system the maximum signal-to­ the modulation is tran smi tted during only a small noise ratio is obtained by gating the signal so as to interval during each pulse-recurrence period. Hence, remove all noise input to the demodulator during the in order to obtain maximum signal-to-noise ratio constant intervals between pulses; such gating also from such a system it is necessary to prevent noise removes any unwanted signals (e.g. video signals) reaching the demodulator except dur ing the above­ occurring during these intervals. mentioned small interval s. The following sub­ sections, describing the signal-to-noise charac­ If Vp is the peak vol tage excursion of the pul se teristics of P.A.M. and P.T.M., include outlines of amplitude at the input to the low-pass filt~r used the means used to achieve this. for demodulation,*

T is the pul se duration (secs), The form of noise considered consists of fluc­ T is the pulse-recurrence period (secs), tuation noise, as produced by thermal agitation in the input circuits of a receiver. No attempt is made • For 100% modulation, Vp is equal to the unmodulated­ to assess the effects of impulsive noise. pulse amplitude.

Co E o ~ :;:; o "ij L.

I I I I I 0 fr-2fa fa fr-fa fr I fr + fa fr + 2fa 2fr I 2fr+ fa f 2fr - 2fa 2fr -fa 3fr -2fa (a)

Cl '0 ~ Co E 0 ~ ;; 0 "ij L.

0 fa 'r 2fa fr f fr -2fa "2 fr-fa (b) Fig. 12 - Demodulation of P.P.M., with natural sampling, using a ramp wave (a) Spectral components of P.P.M. pulses with natural sampling, together with component representing ramp wave (fT/4

------component representing ramp wave ---- components of P.P.M. pulses 14

VN is the r.m. s. noi se vol tage at the filter thus: input during the pulse interval, ~ -= Vp • =. .!..... J2Tfo fo is the effective video bandwidth of the signal V T VN and noise applied to the filter (Hz) and the ratio, Rp.A.M., of the r.m.s. output-signal and fa -= 112T is the a.f. bandwidth (Hz), voltage to the r.m.s. output noise voltage:

the peak signal, Vs, obtained by filtering the pulses is ~Tfo ••••••. (5)

T Vs -= Vp T Certain deductions may be drawn from Equation (5). First, provided that the video bandwidth, fa, is The average noise power, "N'2, fed to the filter is: suffi cient to prevent significant di stortion of the modulated-pulse shape, the value of fa does not affect the output signal-to-noise ratio; an increase T "'N '2 = of fa is compensated by a correE'onding increase in T VN (which is proportional to ~fo). If, however, fa were reduced to a value such as to affect the pulse thus the average noise-power output, V'2, from the amplitude and duration, then: filter is:

T fa 'Vp er; fa, T er; V'2:: v~ T'fa fa and: T 1 :: V~ . -'-­ T 2Tfo

whence, from Equation 5, Rp .A.M. er; whence the r.m.s. noise-output voltage, V, is: Vc. Thus, it is desirable to ensure that the form of pulse used in P.A.M. has a spectrum confined well within the video bandwidth available.

0r------~

-10

III -20 u

~ (78kHz) (6'25kHz) a.:l r·--"'-·_.-£312kHz) +' ::l 0-30 (0) co i '--'--'-' > (7'8kHz) tii'"" '- -40

-50 I

-600~-----~------~----~~----~1~·------=-----~~----~----~82 3 4 5 6 7 modulation fr~qu~ncy, kHz

Fig. 13 - Uutput components of ramp demodulator of typical P.P.M., with natural sampling, as functions of modulating frequency ---- modulation component ------unwanted second harmonic component --_.- unwanted inversion component ( ) frequency of unwanted component 15

5.2. P.T.M. cribing the modulation. Thus, for the above-men­ tioned P.T.M. systems, the circuit arrangements In a P. T.M. system gating arrangements may prior to the demodulator should incorporate ampli­ be used, as in P.A.M., in order to remove unwanted tude limiting (or slicing) together wi th gating so as signals and noise occurring during the intervals to restri ct the noi se appearing at the demodulator between pulses; in this case, however, the gate input to that occurring during only a small portion must pass the required signal, together with nl'lise of the ri se (or fall) time of the pul se edge carrying for all positions of the modulated pulse edges. In modulation. P.P.M. and asymmetrical P.D.M. systems all essen­ tial information concerning the modulation is indi­ Fig. 14 illustrates this procedure. Fig. 14(a) cated by the position of one edge of each pulse;* shows a time-modulated pulse accompanied by ran­ further, pulse amplitude is not a parameter des- dom fluctuation noise. Noise accompanying all low­ slope regions of the pulse (e.g. the baseline and the • Only in the case of symmetrical P.D.M. is it necessary peak) is removed, by limiting, to form the pul s e to pass both edges to the demodulator. shown in Fig. 14(b).

J ~

/ (c) \

" (d)

(e)

(f) Fig. 14 - tHinimising the effects of noise prior to demodulation of P.T.M. (a) P.T.M. pulse with accompanying noise (b) Pulse produced by suitably limiting (a) (c) Asymmetric P. D.M. pulse with noise eliminated from unmodulated edge (d) Short, constant duration, P. P .M. pulse (e) Narrow pulses derived from edges of input pulse (f) Pulse produced by combining pulses of (e) 16

In the case of asymmetric P.D.M., a further where Ai s a constant, C< modulated pulse may be derived, one edge being t~ formed from the modulated edge of the pulse shown then the r.m. s. noise output V~ resulting from an it in Fig. 14(b) and the other, unmodulated edge, being edge-timing error Eis: tl formed using a regularly recurring pulse derived 5 from an oscillator that is locked, by flywheel means, to the unmodulated edge of the pulse shown in Fig. 14(a). The derived pulse, illustrated in and, for P.P.M., the r.m.s. signal to r.m.s. noise Fig. 14(c), is then passed to the demodulator. ratio, Rp.P.M., at the demodulator output is: v For P.P.M. the corresponding process consists S.d v Rp.P.M. == . . . . (6) r of using the pulse obtained after limiting to form a VN • ,J2 short modulated pul se, shown in Fig. 14(d), having t c a constant duration and a position that is perturbed Correspondingly: by the noise accompanying only one (say the lead­ r ing) edge of the pulse shown in Fig. 14(b);* this S . 8 pulse is then utilised by the demodulator. Rp.D.M. -:: . . . (6 a) VN .-G However, if the input P.P.M. pulses have durations such that the noise accompanying each If, however, the noise accompanying the two leading edge is substantially uncorrelated with that edges of each P.P.M. pulse is uncorrelated, the accompanying each trailing edge, a more efficient process outlined in Figs. 14(d), (e) and (f) results process may be used. In such a case two narrow in an improvement of 3 dB whence: pulses, as shown in Fig. 14(e), may be formed (one from the leading edge of each input pulse and one from the trailing edge) by, for exampl e, differen­ R'p.P.M.. -- VNSd •....• (7) tiation. The earlier of the two pulses may then be delayed so as to coinci de wi th the later pul se and combined with it to form a pulse of double amplitude Equations (6), (6a) and (7) show that, in each which is illustrated in Fig. 14(f); the associated case, the signal-to-noise ratio is directly propor­ increase in noise magnitude will, however, be only tional to S. If the pulse shape, at the input to the 3 dB. The pul se of Fig. 14(f) may now be limi ted limi ter, is determined by the bandwidth of the and used to generate a pulse of the form shown in circuit preceding the limiter, S is linearly related to Fig. 14(d) which is utilised, as before, by the the circuit bandwidth. In such circumstances, demodulator. increasing this bandwidth raises VN according to a square toot law; as a result the signal-to-noise The signal-to-noi se ratio obtained at the output ratio is proportional to the square root of the band­ of the demodulator for either asymmetric P.D.M. or width prior to limi ting. for P.P.M. (assuming that only one edge of each pulse is used) may be derived quite simply.** It will be appreciated that if the signal-to-noise ratio prior to the limiter is below a certain value, it If VN is the r.m.s. noise accompanying input may not be possible to prevent, by amplitude limit­ pulses (volts), ing, noise which occurs during low-slope regions of and S is the slope, prior to limi ting, of the the pulse waveform from reaching the demodulator. pulse edge at the slicing level (Vis), As the input signal-to-noise ratio is reduced and then the r.m. s. edge-timing error E, due to noi se, the critical or "threshold" value reached, the signal­ is 3,8 to-noise ratio at the demodulator output falls cata­ s trophi cally.

5.3. Comparisons With A.'VI. and F.M. If the peak signal output Vs from the demodulator is defined as: The signal-to-noise performances of pulse­ modulation systems may readily be compared with Vs == A. d (for P.P.M.) those of conventional A.M. and F.M. systems. In the ensuing discussion certain assumptions have or == A. 8 (for P.D.M.) been made wi th regard to th e parameters whi ch could be considered to be representative of typical television-sound systems; for example, it has been ... This may be carried out using a relaxation oscillator assumed that, in all cases, the a.f. bandwidth is triggered by the pulse of Fig. 14(b). 7'5 kHz. Further, it has been assumed that each .. A more general analysis, based upon that given in of the sound systems is associated with the same Reference 3, is given in Appendix I. television system whose vision-signal parameters 17

conform in all respects except one, with Standard I; the exception is the value of video bandwidth and (8) it has been assumed that the effective bandwidth at the output of the vi sion detector of the receiver is 5 MHz. where fd is the frequency deviation corresponding to 100% modulation. 5.3.1. A.M. If fd has the value 50 kHz and the ratio of the The signal-to-noise ratio of an A.M. tele­ unmodulated carrier power to the peak vision power vision-sound system is related to that of the tele­ is -7 dB (as in Standard I) then the ratio RF .M.! vision system by several factors. These are: the RA.M.has the value +20 dB. ratio of the video bandwidth to the a.f. bandwidth, the forms of transmission use d (i.e. whether double If, however, the receiver sound channel employs or asymmetric sideband), the ratio of the normalised the intercarrier method, in which the F.M. sound peak-to-peak amplitudes of the modulating signals carrier is heterodyned with the vision carrier so as and the ratio of the peak carrier powers. Table 1 to produce a further F.M. carrier whose frequency shows how these factors affect the ratio of RA.M. is independent of receiver tuning, the signal-to­ to RV where RA.M. is the ratio of r.m.s. sound noise performance of the sound system now depends signal at 100% modulation to r.m.s. sound noise and upon the signal-to-noise ratios, at the receiver in­ 9 Rv is the ratio of picture signal at white to r.m.s. put, of both the sound and vi sion carriers. The video noi se. vision carrier signal-to-noise ratio depends upon TABLE 1 the vision modulation characteristics of the tele­ vision sys tern and upon the actual picture-signal level; for Standard I, the vision-carrier signal-to­ Factor Effect on noise ratio, at the receiver input, for black level is RA.M.lRV dB some 12 dB greater than for white level. Bandwidth ratio (video: 5 \1Hz It can be shown*** that the signal-to-noise a.f.: 7'5 kfl z) +28 ratio, [~~ .M.' obtained from a sound channel em­ ploying the intercarrier principle is related to that Form of transmission (vision given by an ideal sound receiver, RF.M., by: asymmetric, sound double side­ band) + 3 R'F.M. ::: J1 + K'2 (9) Ratio of peak-to-peak ampli­ tudes of modulating signals + 5 where K is the r.m. s. value of the ratio of the Peak carrier power ratio* o sound-carrier voltage (assumed constant) to the vision-carrier voltage at the receiver inp ut. Correction of a.f. $.ignal level to give r.m.s. value - 9 For the television standards assumed, and a vision signal corresponding to white, K has a value TOTAL +27 of 2'05. Thus, in this case:

From Table 1 it will be seen that the ratio of ~ -7'5 dB the two signal-to-noi se ratios, RA.M.I Rv, for the parameters used has a value of +27 dB. I RF.M. wh ence -- := +12'5 dB approximately. 5.3.2. F.\1. RA .M• The signal-to-noise performance of an F.M. 5.3.3. P.A.M. television-sound system is very dependent upon the type of reception used. If it is as sumed that the Equation 5 of Section 5.1 permits the signal­ sound channel of the receiver utilises only the to-noise performance of a P.A.M. system to be sound carrier (i. e. a "separate" sound channel) then compared with that of A.M. In the following com­ the performance, RF .M., may be derived from that of parison it has been assured that the P.A.\1. pulse the corresponding A.M. system having the same train con si sts of cons tan t-duration pulses located unmodulated carrier power by the well-known in line-synchronizing intervals, in the manner shown . relation** in Fig. 2, and has the following parameters:

* Assuming the ratio of unmodulate d sound-carrier power Vp equal to half the excursion from sync level to to peak vision-carrier power is -6 dB (as in Standard IV. white level (i.e. 100<1c P.A.M. causes the

U This neglects pre-emphasis and de-emphasis. *** See Appendix n. 18

peak pulse amplitude to equal white level), Assuming "single-edge" demodulation of P.P.M. and Tequal to 3 fJ- s (a suitable value for Standard 1). (P.D.M. gives the same result) we have, from Equa­ tion 6: Thus: Esw 7r. d Rp.P.M. =Rv.- -r= Esw Ebw . 2x'-l2 c:: RV. Y2.-­ Ebw Rp.P.M. where Esw is the signal excursion from sync level ::: +22 dB to white level and Ebw is the signal excursion from l?v black level to white level. whence: Whence, from Equation (5): Rp.P.M. -5 dB -RA.M. Rp.A.M. The more efficient method for demodulating P.P.M. therefore: described in Section 5.2 would, from Equation (7), result in: , Rp.A.M. -= 11 Esw ~ Rp.P.M. 12'--.'-ITfo -: +9 dB -: +25 dB Rv Ebw RV , Thus, using the 27 dB figure from Table 1 and Rp . P •M• = -2 dB RA.M. Rp.A.M. = -18 dB RA.M. The threshold value of pulse signal-to-noise ratio, below which the ratio Rp.P.M.lRA.M. falls 5.3.4. P. T.M. catastrophically, occurs when the quasi peak-to­ peak noi se vol tage is approximately equal to the The signal-to-noise performances, above pulse amplitude. Por the system and pulse para­ threshold, of P.P.M. and P.D.M. for pulse para­ meters assumed the threshold occurs at a picture meters suitable for use with the television system signal-to-noise ratio of approximately 15 dB. assumed may be readily calculated using Equations (6), (6a) and (7) of Section 5.2. 6. PRACTICAL CONSIDERATIONS

Por a P.P.'v1. (or P.D.M.) pulse train in which In this Section of the report the use of sound­ each pulse is located within a line-synchronizing modulated pulses combined with the video signal is interval, as illus trated in Pig. 2, the following discussed in terms of the two main applications parameters are convenient: mentioned in the Introduction, and some practi cal problems associated with each application are Vp equal to the excursion from sync level to considered. white level It is as sumed, throughout the ensuing di scu s­ d (or 8) equal to 1'5 fJ- s sions, that the line-scan frequency of the television system is stable and free from phase and frequency S equal to the m aximum slope of the edge of a modulation a t audio frequency. Such a charac­ sine-squared pul se whose peak magnitude is teristic is desirable not only from the point of view Vp and whose half-magnitude duration, x, is of using sound-modulated pulses; it aids the design 0'2 fJ-S. and performance of receivers and also of equipment having long mechanical time-constants such as that Thus: used by a broadcasting organization. 7r " -- However, a broadcasting organization often 2x requires to synchronize televi sion signals obtained In thi s case: from widely separated sources and uses "genlock" and "slavelock" techniques for this purpose. These Esw - Rv processes fundamentally involve phase and fre­ .- quency modulation of the line-synchronizing signal 19 but, provided such modulation occurs at a suitably channels that are 8 MHz wide; Fig. 15(a) illustrates slow rate (which need not involve a significant the channel arrangements used. In order to provide increase in the time taken to perform the synchro­ satisfactory v.h.f. coverage of the U.K. with two nizing process), the effect upon the pulse-sound 625-line programmes it would be necessary to use system should be small. In the case of a properly appreciably narrower channels. designed slavelock system, the synchronization of a televi sion-signal source takes place before the If the coverage obtained wi th 7 MHz channels signal is used for programme purposes and the line­ were considered adequate, two alternative VlSiOn scan frequency is subject to only very small phase and sound signal arrangements could be used. and frequency modulation after synchronization has These are: occurred; wi th such a slavelock arrangement, the effect upon a pulse-sound system should be quite (i) A vision signal with a main-sideband band­ negligible. width of 5 MHz and a vestigial-sideband band­ width of 0'75 MHz, together with a sound signal 6.1. 625-line Broadcasting using a frequency-modulated carrier (i.e. trans­ missions according to Standards B or G). At the present time 625-line broadcasting in the U.K. is confined to u.h.f. using Standard I, In (ii) A vision signal with a main-sideband band-

lowQr adjacQnt upper adjacent channQI wanted channQI channQI

c s V c s V I . . I I 1'-'-\J1l i 'j I , \, ~ (--t------~ __ I . L -2'0 + --!O------:-4-'47::3=---+-t --:-6 , O'--~ 8, 0 -1,25 5,5 relativQ frQquQncy. MHz r. f. channals

relativQ 05 rQsponse

-125~ 0 1·25 rQlatlve frQquQncy. MHz vision racczlvar rasponsa fa)

wantQd channQI lowQr adjacant upper adj aCQnt channQI channQI

v eve I I i I -·-·-·-·-;· ...... r--+------+----r-r------I i I i i I -60 +-443 -50 ralativa frQquQncy, MHz r. f. channals

/-50

sound (bJ Fig. 15 - Current u.h.!. and possible v.h.!. channels (a) R.F. channels and receiver response used in u.h.f. (b) R. F. channels and receiver response possible, using pulse-sound, in v.h. f. V = Vision Carrier, S = Sound Carrier, C = Colour Carrier 20

width of 5'5 MHz and a vestigial-sideband contains strong 3 MHz components; 11 the sound of 1'25 MHz, together with a sound signal buzz is due to the single··sideband demodulation by using modulated pulses in the line-blanking the vision detector of the 3 MHz vision-modulation intervals (i.e. vision transmissions according components which, in turn, results in the production to Standard I with pulse sound). of 6 MHz harmonics that interfere severely with the sound intercarrier signal and may swamp it. Ex­ Although Standards Band G are widely used in peri ence with 625-line monochrome transmissions Europe, their narrower main and vestigial sideband shows that the design of present-day receivers is bandwidths render the video performance of these such that this effect can occur, although fairly standards inherently inferior to that of Standard I. infrequently, when the receiver is tuned correctly; 12 Within the context of 7 \1Hz channel spacing, there­ deliberate mistuning so as to increase the vision­ fore, a vision signal conforming to Standard I with carrier level at the vision detector with respect to pulse sound offers some advantage. the level of the 3 MHz sideband, thus inhibiting the production of 6 MHz harmonics, reduce s the tendency If, however, the degree of coverage obtainabl e to sound buzz by a large factor, at the cost of with 7 MHz channels were not considered adequate, pi cture sh arpne s s. it would be nec essary to consider the use of still narrower channels with, say, 6 MHz bandwidth. Further, the introduction of colour trans­ Within such channels a vision signal similar to missions may bring to light a further difficulty. It those of Standards B or G, together with pulse has been shown 13 that, unless the sound-carrier sound, is the only practicable arrangement that input to the vision detector is attenuated sufficient­ does not involve sacrificing the European sub­ ly (i.e. > about 30 dB) with respect to the vision carrier frequency of 4'43 MHz. This arrangement carrier, the beats between the sound carrier and the will be used as a basis for discussing the applica­ vision-signal side~ands corresponding to the chro­ tion of pulse sound to 625-line television broadcast minance signal (subcarriei frequency: 4'43 MHz) .transmissions. It should be emphasized, however, have an amplitude such as to impair the received ·that the use of pulse sound is in no way responsible picture significantly. These considerations assume for the fact that the video performance provided by that the receiver is correctly tuned and has the the vision signal is inferior to that provided by a conventional asymmetric-sideband characteristic signal conforming to Standard I. with a response at vision-carrier frequency -6 dB with respect to the peak response. Again, as in the Fig. 15(b) shows the relative disposition of case of sound buzz, the sound-to-chrominance beat adjacent channels. As will be seen, a certain may be considerably reduced in level (and thus degree of overlap between channels has been rendered unobtrusive) by deliberately mistuning the assumed; this appears practicable as the receiver receiver so as to decrease the level of the chro­ is only required to rej ect the lower-adj acen t vi sion minance signal with respect to that of the sound carrier beyond the edge of the full (lower) sideband signal; this may also result in a less sharp picture. and the upper-adjacent chrominance signal beyond the edge of the vestigial (upper) sideband. Such a The above-mentioned effects may be avoided performance is not unduly different from that by suita ble receiver design in which a separate required by the arrangements of Fig. 15(a) where it detector is us ed to provide the sound intercarrier is nec essary to attenuate severely the sound signal signal; the detector is fed from a point in the i.f. of the wanted channel beyond the edge of the full amplifier prior to the input circuit of the vision (upper) sideband and to rej ect the lower-adj a cent detector wh ich may now be [i tted wi th a sound­ sound signal beyond the edge of the vestigial carrier trap.12 Such an arrangement may well be (lower) sideband. used in colo ur receivers hut has not yet found favour in the U.K. wi th designers of monochrome In addition to a saving of channel bandwidth receivers, probably due to cost. the elimination of the conventional f.m. sound transmission could confer other advantages. It may be deduced that, although the chanm Present-day 625-line receivers employ intercarrier arrangements shown in Fig. 15(a) potentially provide sound reception in which the beat between the a picture sharpness corresponding to a video band­ sound and vision carriers is extracted from either width of 5'5 MHz. the effects of receiver mi stuning the vision detector or the video amplifier and, as it to avoid ei ther sound buzz or sound-to-chrominance is frequency modul ated by the wanted sound signal, beat (perhaps both!) may well result in a picture the beat is amplitude limited and demodulated in a that is no sharper than that provided by the channel conventional detector arrangement. Such a system arrangements and receiver response shown in Fig. demands that the sound carrier be passed, by the 15(b). Pul se-sound reception cannot suffer from the receiver Lf. amplifier, to the vision detector at a above-mentioned sound-buzz defect (sound and sui table level. If the level is attenuated by more picture signals are not present simultaneously) and than about 30 dB wi th respect to vi sion carri er, the beat that could occur between the wanted chro­ sound buzz occurs whenever the vision modulation minance signal and the lower-adjacent vision carrier 21 llnd is only likely under certain unusual reception A second method of achieving a single tele­ by conditions and, in addition, may be appreciably vision scanning standard appears feasible if modi­ fications to the existing u.h.f. transmissions were ion reduced in visibility by suitably offsetting the considered practicable. This follows from the ion vision-carrier frequencies of adjacent channels; concept that, at some future date, it may be possible the thus the tendency for the deliberate mistuning of receivers may be markedly reduced. to use transmissions in both v.h.f. and u.h.f. which ~x- are identical, with the possible exception that ms 8 MHz channels are used in u.h.f. and, say, 6 MHz is However, in considering the problems of 625- channels are used in v.h.f.; in such circumstances, rly line broadcasting in the v.h.f. Bands, the main 12 objective is to arrive at a plan whereby a single the only form of sound transmi ssion which could be televi sion-scanning standar d may be used for all used in both Bands is that employing pulses. In lO­ order to achieve this end, it would be necessary to to U.K. television broadcasting and it is important to introduce pulse-sound transmissions in u.h.f. in the he con si der how televi sion tran smi s sion s incorporating sound pulses may be introduced without unduly com­ near future, alongside the existing f.m. sound trans­ ICY mi ssion s, and "transi tional-s tage" receivers could of plicating an already complex situation. Existing dual-standard receivers accept in v.h.f. 405-line then be marketed which were capable of receiving: transmissions, with positive vision-modulation and Ca) 405-line transmi ssions, with po SI tJ ve vi sion­ s­ a.m. sound, and in u.h.f., 625-line transmissions modulation and a.m. sound, in v.h.f. It with negative vision-modulation and f.m. sound. er and t­ One possible approach to the problem of intro­ ducing pulse sound is to consider its possible VI lO (b) 625-line transmi ssions, with negative SlOn­ le application only to future v.h.f. television broad­ modulation and pulse sound, in both v.h.f. and casting. In this case it would be necessary, prior :)­ u.h.f. z) to the introduction of 625-line broadcasting in :d v.h.f., to ensure that all receivers were capable of It would be necessary to ensure that these recei ving: receivers ignore the f.m. sound transmissions e radiated in u.h.f. for the benefit of existing dual­ e Ca) 405-line transmissions, with positive vision­ standard receivers. After the replacement of the c modulation and a.m. sound, in v.h.f. 405-line transmissions in v.h.f. with 625-1ine trans­ B missions, it would be possible to make available, e Cb) 625-line transmi s sion s, with negative VI SlOn­ to the public, receivers operating on only 625 lines t modulation and pulse sound, conforming to, with negative vision-modulation and pulse-sound in s say, 6 MHz channels in v.h.f. e all Bands and, at this stage, the f.m. sound trans­ missions in the u.h.f. channels could be closed and :l down. Further, once the u.h.f. f.m. sound trans­ missions were closed down it might prove possible, Cc) 625-line transmi s sions, with negative vi sion­ subject to considerations of international inter­ modulation and f.T. sound, in u.h.f. ference, to replan the u.h.f. Bands on the basi s of, say, 6 MHz channels and thus provide scope for' e Thus it would be necessary to make available, more programmes. These "final stage" receivers to the public, recei vers containing facilities which would be suitable only for the transmissions des­ would not be used until some arbitrary future date cribed in Cb) above and although different channel when the 405-line transmissions in v.h.f. would be widths might be used in v.h.f. and u.h.f., the replaced by others using 625-lines; such receivers absence of sound carriers would simplify design. could well be appreciably more costly than existing If identical channel widths were used in all Bands dual-standard receivers and any failures in the after the close-down of the u.h.f. f.m. sound trans­ circui ts required only for v.h.f. 625-line transmi s­ mi ssions, this would enable truly single-s tandard sions would not be evident until the time of change­ receivers to be marketed; such receivers would over, which could precipitate a rather desperate probabl y prove appreciably cheaper than present­ service situation. After the replacement of the 405- day recei verso transmissions in v.h.f. by 625-line transmissions, receivers could then be marketed which were suit­ It will be appreciated that the practicability of able only for the two types of 625-line tran smi s sions utilizing this second method of achieving a single Cb) and Cc) mentioned above. However, such scanning standard is dependent upon it being recei vers would be of dual-standard nature in that possible to introduce pulse-sound transmissions they would have to accommodate two different into u.h.f. in the near future, wi thout interfering channel widths and two different types of sound with the operation of existing dual-standard re­ modulation. It may be concluded, therefore, that ceivers. This raises the issue of compatibility such a method for changing over from 405- to 625- which, in turn, must influence the pulse parameters line broadcasting in v.h.f. would not receive enthu­ chosen for the pulse-sound system. The choice of siastic support. a suitable form of pulse signal, bearing in mind 22 compatibility, and some of the factors in the design occurring during line-blanking intervals. Figs. 16(a) of pulse-sound demodulators for use in domestic and 16(b) assume that unidirectional pulses extend­ receivers are dealt with in the following two sub­ ing from blanking level to white level would be used section s. and thus no interference with receiver line synchro­ nization by sound pulses could occur; however, 6.1.1. Form of Sound-Pulse Signal although the modified field-synchronizing waveform would provide, in principle, adequate field-synchro­ Although Fig. 2 typifies one form of com­ nizing information, it would be necessary to ensure bined pul se-sound and vi sion signal, many factors that present-day receivers would be unaffe·cted. must be borne in mind when specifying a broadcast With regard to the visibility of the sound pulses on signal which is to be introduced in the manner the screens of receivers, part of the post-synchro­ described in Section 6.1. nizing line-blanking interval coinci des with the end of line-flyback in the receiver and it has been Principal requirements to be satisfied are: pointed out* that a sound pulse located at or near this point would occur when the scanning spot in (i) Existing 625-line receivers, when presented the display tube of a properly adjusted receiver is with the pulse-sound signal together with a behind the lefthand edge of the tube mask, and is conventional sound transmission, should pro­ thus hidden from the viewer. However, the early vide pictures and sound that are not signifi­ part of the post-synchronizing line-blanking interval cantly different from those obtained at present. has already been allocated to the colour burst and, in order to provide an adequate location for the (ii) The sound signal-to-noise ratio provided by sound pulse, it would probably be necessary to pulse sound, when demodulated by a pulse­ extend the interval somewhat, ** thus reducing the sound demodulator suitable for inclusion in a duration of the active line; thi s would demand that domestic receiver, must be satisfactory. the picture-width contr'ols of existing 625-line receivers be adjusted slightly in order to place the (iii) The sound quality, using a pulse-sound demodu­ sound pulse behind the tube mask. lator suitable for inclusion in a domestic receiver, must be satisfactory. With regard to the second requirement (i.e. a satisfactory signal-to-noise ratio) the results of The first of the above-mentioned requirements Section 5.3 lead to the conclusion that P. T.\1. influences, to a major extent, the location and shape should be used with the maximum possible pulse of the pulse within the waveform. Any excursion of amplitude and deviation; problems of compatibility the pulse waveform below blanking level must not influence the choice between P.D.\1. and P.P.M. result in interference with receiver synchronization In order to be able to use as high a pulse amplitude and, in order not to affect intercarrier sound recep­ as possible, the pulse waveform should, if possible, tion, it is probably desirable to limit the excursion traverse the region between blanking level and sync in the white direction at white level. Present-day level; however, disturbance of receiver synchro­ 625-line receivers incorporate line-synchronizing nization must be avoided. As may be deduced from circuits which may well be affected by additional Sections 3.2.1 and 3.2.2, the baseband-frequency pulses located within the line-synchronizing inter­ components of a P.P.M. spectrum are of lower vals, unless the fact that the pulses are sound amplitude than those of the corresponding P.D,M. modulated is undetectable at the output of the syn­ spectrum; as a result, a pulse train of the former chronizing-pulse separator. In such circumstances type would be less likely to cause interference than only P.A.M. could be used, with modulation that one of the latter type. Further, a pulse occupying causes the pulse amplitude to vary between the the preferred location whos e waveform excursion limits of blanking level and white level; as may be reached white level and traversed the region between deduced from Section 5.3.3, such an arrangement blanking and sync levels could be of such a shape would have a poor noise performance. Further, few as to give a spectrum wi th li ttle energy at the low­ present-day 625-line receivers incorporate line­ frequency end of the vi deo band. As it is cu s tomary, flyback suppression of the scanning beam in the in dom estic receivers, to restrict the bandwidth of picture display tube and, as a result, it is likely the circuit connecting the video stage to the syn­ that sound pulses, located within the line-synchro­ chronizing-pul se sep arator, the high-energy com­ nIZlng intervals with peak excursions reaching ponents of the pulse would be severely attenuated white level, would cause visible impairment of the prior to the separator, whose amplitude-lim iting picture. These difficulties could be avoided if the action would then prevent them affecting the receiver sound pulses were located within the post-synchro­ synchronization. Fig. 17 illustrates various forms nizing line-blanking intervals, as suggested in at of position-modulated pulse that could be used. In least one early proposal. 14 The resulting combined each case the pulse is bodily modulated in position; waveforms are illustrated in Figs. 16(a) and 16(b) where it will be seen that the field-synchronizing ... By J. R. Sanders. waveform has been modified in order to provide locations for the sound pulses similar to those ** As suggested by F.e. McLean. 23

6(a) ______~hit~~~ ______md­ Ised hra­ sound ver, mOdulatad -----+­ 'orm pu Isa HO­ lUre ~ed. on blanKing laval lra­ end

~en ear in ~-- lina sync. --~ is lina blanKing is faJ rJy __ yhit~~1 ral sound Id, I_modulatad he pulsa to he at !le l-623-i-624 ~625-+- 1 ~ 2 ~ 3 -l- 4 --i- 5 -J..--6 :he ~ ----- baginning of avan fiald white laval

a ~sound of modulatad 11. pulsa le ly 1- le 310--\- 311 -+-312 -1- 3\3 -..l-- 314 --i- 315 --i- 316 --I- 317-1-- 3113 -.l-319 I.. baginning of odd frald " (b) C I­ Fig. 16 - Sound-modulated pulses in post-sync line-blanking interval n Ca). waveform of line-blanking interval Cb) waveforms of field-blanking intervals y 'f

r __ ~!:!!.~ ~~a..!... _ ___ _

burst location burst burst location location

(a) (b) (c)

Fig. 17 - Some forms of position-modulated pulse suitable for broadcasting Ca) general form Cb) ·subtracted sine-squared" Cc) ·displaced sine-squared"

=-'- 24 a "burst" form ed from the modulation of a subcarrier 6.1.2. Practical Demodulators by a position-modulated unidirectional pulse would not provide as good a signal-to-noise ratio. Wave­ form (a) is of a general nature which satisfies the The basic principles of demodulation, for requirements that it should have a large magnitude P.P.M. with natural sampling, have been described and be characterized by little energy at the low in Section 4.2.2. However, the pulse-sound section video frequencies. It will be seen that the pulse of a domestic receiver must perform several func­ excurs ion extends well beyond sync level; thi s tions additional to actual demodulation. would probably be permi ssible using negative vision modulation. Waveform (b)* may be obtained by As already mentioned, the combined sound and subtracting the waveform of a broad sine -squared vision signal must be processed so as to derive a pulse from that of a narrower one. By choosing train of position-modulated pul ses as free as suitable values of magnitude and duration for the possible from extraneous noise; thus, it would be two pulses it is possible to confine the overall necessary to gate the combined signal using a waveform excursion within the limits of white level pulse, derived from the receiver line time-base, and synchronizing level whilst still forming a pulse timed so as to provide a time-slot embracing the with little energy at low video frequencies. Wave­ deviation range of the sound pulse. Further, in form (c)** is produced by combining two nominally order to utilize the fast edge extending over the identical sine-squared pulses; the second is delayed whole excursion of a pulse of the form shown in with respect to the first by approximately half the Fig. 17, it would be necessary to apply the gated half-magnitude duration and is of opposite polarity. pul se to a network (e.g. a differentiating network) whose output contains a major peak derived from The third requirement, that of satisfactory the long fast edge; this peak could then be applied quali ty, mus t be con si dered in reI ation to the to a relaxation oscillator, as described in Section problems of providing demo dui ator s suitable for 5.2, in order to obtain a pul se sui table for appli­ cation to the demodulator. inclusion in domestic receivers. Although, as described in Section 4.2.1, the use of uniform sampling could enable demodulators to be designed All three forms of demodulator described in which gave intrinsi cally di stortion-free sound wi thin Section 4.2.2 could, in principle, be used in domes­ the band 0 to 7'8 kHz, it is likely that they would tic receivers. The arrangement consisting of a low­ be appreciably more complex, and hence more pass filter and correcting network is simple, and expensi ve, than demodulators designed to operate apparently cheap, but its sensitivity is such that with a signal based upon natural sampling. In view considerable subsequent audio amplification would of the addi tional point that, wi th the relatively low be required in order to provide adequate output. value of pulse-position deviation which is possible For the other two forms of demodulator it would be in a combined vision and sound signal, the degree necessary to provide a train of pulses recurring of di stortion resulting from the use of natural regularly at the mean recurrence frequency of the sampling is probably quite tolerable, it would sound pulses and suitably timed with respect to appear desirable to use natural sampling in a broad­ them. In addition, it would be highly desirable that cast system. Further, as mentioned in Section the regular pulse train be as free as possible from 4.2.2, it appears practicable to devise pulse­ perturbation by noi se; otherwi se the signal-to-noi se modulation arrangements whereby the degree of ratio of the demodulator output would suffer. It distortion resulting from the use of relatively simple might be expected that this regular pulse train demodulators and natural sampling may be sub­ could be obtained from the receiver line time-bas e. stantially reduced. However some receivers utilize "hard-lock" time­ bases whose sensitivity to noise would render them From the foregoing con sideration s it would unsui table whil e "flywheel" circuits tend to be appear that, for a pul se-sound signal that is mo st perturbed, by field-synchronizing information, to an likely to prove both satisfactory and capable of extent which would also render them unsuitable. introduction alongside the existing system, P.P.M. Thus, it may be concluded that a pulse train derived based upon natural sampling should be used. The from a typical receiver line time-base would be pulse shape should be similar to one of those suitable only for providing time slots, as already illustrated in Fig. 17 with values of duration, discussed. magni tude and devi ation determined by many con­ flicting considerations, including compatibility, These difficulties could be overcome by deriv­ sound signal-to-noise ratio, sound-signal threshold ing the regularly recurring pulses from the sound conservation of the active-line period, and trans­ pulses themselves, using a suitable servo-controlled mitter peak-modulation characteristics. oscillator which is unresponsive to position modu­ lIition. Further, it would be possible to perform the • Proposed by E. R. Rout. process of ramp demodulation using the servo­ control circuits. Fig. 18(a) shows an example of .. Proposed by J.R. Sanders. such an arrangement. 15 25

sompl<2r low- pass controlJ<2d pulS<2S filt<2r, F1 OSC illotor

(a) oudio out put

------.J I ~Clviotion rang<2 I I I i position - modulat<2d I I pulS<2S

_ oscillator output

I I I -1------I I . (b) f.lOrT1flII,,5;t ~on 9 <2

Fig. 18 - Ramp demodulation using a servo-controlled oscillator Ca) Block schematic Cb) Sampling action and demodulation

Input sound pul s es are us ed to sample the out­ Fig. 18(b) illustrates the operation of the arrange­ put from, say, a sine-wave oscillator whose fre­ ment for the case where the frequency of the oscil­ quency and phase are varied according to an applied lator is equal to the pul se mean recurrence frequency_ control signal obtained by smoothing the output of the sampler in a low-pass filter F l ; such a servo­ One further function should be incorporated in controlled oscillator is, of course, similar to a the pulse-sound channel of a domestic receiver. It conventional flywheel time base. The oscillator is obvious that if the demodulator were presented operates at a frequency which is equal to (or is a with an erroneous input, such a s picture signal, the low multiple of) the 'sound-pulse mean recurrence output could cause loud and distressing noises from frequency and is sufficiently stable to ensure that, the loudspeaker; automatic muting circuits should in the absence of input pulses, its frequency does be used to prevent this effect which may we 11 occur not differ from the sound-pu I se mean recurrence whenever the receiver is grossly mistuned. ~any frequency (or a low multiple) by more than the cut­ methods of muting, of varying complexity and per­ off frequency of the filter F l _ Thus, on applying formance, may be devised, but simple and effective unmodulated input pulses, correction of the oscil­ operation may well be obtained if the audio output lator frequency and phase takes place until the of the demodulator were di sconnected from the loud­ output of the sampler consists of identical samples speaker whenever its peak-to-peak magnitude sub­ of small (or zero) magnitude; this occurs when each stantially exceeded that available when the demo du­ input pulse samples corresponding points on high lator input is correct. slope regions of the osci llator sine wave. However, if the input pulses are modulated, the phase of the oscillator varies in sympathy with the modula­ 6.2. Sound Distribution using Vision Unks tion unless the cut-off frequency of Fi is made less than the lowest modulation frequency. By using The ability to convey the associated sound such a low cut-off frequency the pha se relation ship signal along with the picture signal by means of between the oscillator output and the mean phase of time-di vision multiplex, using vi sion links which the input pul ses may be m ade constant. In such previously were used only for picture signals, would circumstances the sampler may now be regarded as confer advantages to broadcasters and "common­ part of a ramp demodulator and, by passing the carrier" authorities. The cost of separate sound .. samples to a low-pass filter F I' having a cut-off circui ts would be avoided and the frequency of fre quency equal to half the mean p ul se-recurrence occurrence of operational errors might be signifi­ frequency, the audio modulation may be recovered. cantly reduced. Links in which the picture and 26 sound signals are now transmitted by frequency­ The form shown in Fig. 19(a) employs two division multiplex, such as those involving communi­ unidirectional pulses located in each line-synchro­ cation satellites, could be reduced in cost as a nizing interval and in each corresponding interval result of the reduction in the bandwidth necessary within the field-blanking interval; such an arrange­ to convey both signals. Various systems of this ment may be used with a video signal whose wave­ nature are feasible, most of which require that the form is conventional except that the number of waveform of the video signal carried by the link be equalizing pulses is reduced to one,16 occurring made non-standard so as to accommoda te a pulse­ during alternate field-blanking intervals. The sound signal giving a good sound-channel perfor­ signal-to-noi se performance obtainable may be mance; in such circumstances, it would be neces­ estimated by assuming that two sine-squared pulses sary to modify the video waveform at the output of are used, each with a half-amplitude duration of the link so as to conform with the standards used 0'2 fJ- s and a peak deviation of 0'5 fJ- s. ** Reference for broadcasting. The following two sub-sections to Section 5.3.4 shows that, for such pul se para­ outline possible forms of pulse-sound signal suit­ meters and double-edge demodulation, the r.m. s. able for combination with link video signals and audio si gnal-to-r.m. s.-noi se ratio exceeds the pic­ discuss some of the instrumentation involved. ture signal-to-r.m. s.-noi se ratio by 15 dB; although the audio bandwidth of the arrangement has been 6.2.1. Form of Sound Pul se Signal increased by a factor of two as compared with the A pulse-sound system for use with vision example described in 5.3.4, this has been compen­ links should, if possible, satisfy the following sated by the increase in pulse recurrence frequency. As a typical good-quality vision link provides a requirement s: picture signal-to-noise ratio of at least 46 dB, the (i) The quality of the sound 0 btainable should be r.m. s. audio signal-to-r.m. s.-noise ratio at the out­ similar to that obtainable at present using put would be 61 dB. separate sound circuits. (ii) The signal-to-noise ratio at the output of the The second possible form of signal, illustrated sound channel should be similar to that obtain­ in Fig. 19(b), utilizes one unidirectional pulse able at present using separate sound circuits. located in each line-synchronizing in terval (and in (iii) The presence of the sound signal should not each corresponding interval in the field-synchro­ increase the total signal excursion. nizing inte rval) together wi th a further pul se in each post-synchronizing line-blanking interval It will be inferred, from the first requirement, which is characterized by one edge that extends that the pulse-repetition frequency should be greater from white level to synchronizing level and has a than the line-scan frequency of the 625-line 50-field similar form to the pulse illustrated in Fig. 17(c), system. Two pulses per line-blanking interval except for the excursion below blanking level. *** could be used, as mentioned in Section 2, thus Thus, for such a signal, it would be necessary to permitting a maximum audio bandwidth of 15 kHz. u'se a video waveform having only one equalizing As a very high degree of sound quality is desirable, pulse per field-blanking interval and field-synchro­ the use of uniform sampling would appear appro­ nizing pulses of the form shown in Fig. 16(b). The priate. signal-to-noise performance may be assessed by The second and third requirements infer that assuming that the first (synchronizing-interval) the chosen form of pulse should provide as high a pulse has a sine-squared shape, a half-amplitude signal-to-noise ratio as possible. This consider­ duration of 0'2 fJ- s and a peak de viation of 1'5 fJ- s ation, tog ether wi th the fact that the combined (as in the example of 5.3.4), and that the second vision-and-sound signal might be subjected to non­ pul se has a total duration of approximately 0'6 fJ- s, linear amplitude di stortion in the vi sion link, leads a long edge similar to the trailing edge of the first to the conclusion that P. T.M. should be used. pulse and a peak deviation of 0'5 fJ-s.**** In such Two of the many possible forms of signal are circumstances the r.m. s. audio sign al-to-r.m. S.­ illustrated in Fig. 19. It will be seen that, in both noise ratio is mainly determined by the peak devia­ cases, P.P.M. has been assumed. Other possible tion of the second pulse. Owing to the fact that the forms of signal could employ P.D.M. For example, peak deviation of the second pulse is only one-third the use of asymmetrical P.D.M. with leading-edge that of the first pulse, the r.m.s. audio signal-to­ modulation would ease the problem of deriving r.m. s.-noi se ratio isle s s than that which would correct line-synchronizing information 3 at the out­ have been obtained using the same peak deviation put of the link;* on the other hand, the use of by a factor of 7 dB. P.D.M. would substantially increase the amplitude of the audio-frequency component in the combined­ ** Probably the maximum practicable assuming that no signal spectrum which, in turn, could increase the increase of line-synchronizing pulse duration is permis­ possibility of interference with the picture signal in sible. the presence of certain forms of di stortion. *** Suggested by E.R. Rout. "'**'" Probably the maximum practicable assuming that a * In order to facilitate removal of the sound pulses and colour burst may be present and that no increase of line­ modification of the video waveform to the standard form. blanking duration is permitted. 27

white level ------,- -,,-­ ---r ------I , 1 11" , 1 ,1 1 I'" 1 d 1 1 11 1 1 1 1 ",I , 1 '1 I I 1 I I ,! I 1 ,1 ,1 , : burst line sync. I location , pulse ------, , " , I1 : 1, ....

deviation'

white level ------T---- , ----T------~- , 1 , 1 I, ,I 1 , 1 1 , 1 ,1 1 :1 ' , 1 I 1 1 , 1 1 1, 1 , : 1, I :, burst , 1 location 1 1 line sync. : 1 I pulse --- : 1 ,1 , 1 1 , : I --·-~7----- ~-T-.I Oaviation range daviatlan range of 1st pulscz of 2nd pulsa (b)

Fig. 19 - Two possible two-pulse arrangements (a) Two-pulses per line-synchronizing interval Cb) One pulse per line-synchronizing interval plus one pulse per post-sync line-blanking interval 28

Thus, for double-edge demodulation the dif­ the later of the two pulses shown in Fig. 19(a). ference between the r.m.s. audio and picture signal­ The remaining pulses are then delayed by an to-noi se ratios would be approximately 18 dB. For interval almost equal to half a line period, thus a picture signal-to-r.m.s.-noise ratio of 46 dB the providing the earlier pulse of each pair. r.m.s. audio signal-to-r.m.s.-noise ratio would approximate to 64 dB. Thus, both forms of two­ Immediately prior to demodulation it is neces­ pulse signal could provide fairly similar perfor­ sary to delay the later pulse of each pair by an mances in the presence of link noise. Both forms interval equal to the delay suffered, after gener­ could provide somewhat better performances if the ation, by the earlier pul ses, thus reforming a con­ pulses used were characterized by higher edge ventional pulse train. Demodulation may then be slopes; thi s would probably be pratti cable for a carried out by a method such as that described in link having a good performance with regard to band­ Section 4.2.1 and Fig. 7; the regularly recurring width and freedom from group-delay distortion and ramp waves, at twi ce the line-scan frequency, may might provide an improvement of 3 dB in each case. be derived using the line-synchronizing pul se s of the video waveform. However, in order to avoid Some further slight improvements could be difficulties due to possible slight phase modulation obtained, if it were permissible to restrict the of the line-synchronizing pulses, a demodulator may maximum audio frequency to, say, 10 kHz; in such be based upon the arrangement shown in Fig. 18 circumstances the peak deviation of each of the two which utilizes the sound pulses themselves in order pulses used in the first form of signal could be to provide the required regular ramp waves; as uni­ increased by 1 dB without the time interval separa­ form sampling has been assumed, the output of the ting the two pul ses ever falling below the previous sampler must be applied to a "hold" circuit and value (i.e. for a maximum audio frequency 15 kHz). resampled by a regular train of pulses, derived from However, it must be borne in mind that a further a suitably delayed version of the oscillator output, reduction in the maximum audio frequency to, say, before being applied to the iow-pass filter. 7'5 kHz would enable a signal in the form described in Section 5.3.4 to be used, with a corresponding The problem poses by such a system arises due improvement in signal-to-noise performance of 10 dB. to the need to delay the later pulse of each pair However, in assessing the significance of a change immediately prior to demoduLation. If this delay of signal-to-noise ratio due to a change of a.f. band­ does not equal that inserted in the earlier pulses width, it must be borne in mind that the subjective immedi ately after generation, the output of the effects of noise components having frequencies demodulator will include components at line-scan greater than 10 kHz are less than those with fre­ frequency together with unwanted inversions of the quencies between 1 kHz and 10 kHz. modulation. This could be avoided by applying the output from the demodulator, obtained from a point No consideration has been given to the prob­ prior to the low-pass filter, to a sampler whose lems, such as sound-system threshold, which could other input consists of regular pulses at line-scan arise in the presence of high link noise; threshold frequency derived, say, from line-synchronizing would not occur until the picture signal-to-noise pulses. The output of the sampler may then be ratio fell well below a value at which the picture smoothed in a circuit of long time-constant to give would be regarded as unsatisfactory. a control signal for adjusting the value of the delay. The smoothing ensures that 15 kHz components in 6.2.2. Instrumentation the modulation do not actuate the delay control; complete avoidance of this difficulty could, of Mos t of the problems of in strumentation course, be obtained by restricting the upper fre­ involved in a pulse-sound system for use with quency limit of the modulation to, say, 14 kHz. vision links are similar to those already described, in general terms, in Section 4. However, a system involving two pulses per line-blanking interval After the sound signal has been extracted at poses an additional problem not already mentioned; the receiving-end of the link, it will normally be this Section, which outlines the generation and necessary to derive, from the combined signal, a demodulation processes for sound signals of the video signal of the form sui table, say, for broad­ type described in the preceding Section, describes casting. This process may be carried out using the additional problem and poses a possible solution. conventional techniques in which a local generator of standard synchronizing pulses is locked to the Generation of a signal such as that illustrated synchronizing pulses of the combined signal so that in, say, Fig. 19(a), may be performed as described the line- and field-synchronizing pulses from both in Section 4.2.1 and Fig. 6, using a sampling fre­ sources are respectively coincident (e.g. by "gen­ quency equal to twice the line-scan frequency lock"). In the case of a composite signal, as illus­ (approximately 31 kHz) and free from any phase trated in Fig. 19(a), the original synchronizing modulation. The phase of the sampling frequency pulses may then be removed and replaced by the is adjusted so that alternate output pulses become standard form in a circuit which may be regarded as 29

a single-pole, double-throw, electronic switch; for sound-and-vision signal as one programme signal the form of composite signal shown in Fig. 19(b), it which may be treated, in broad terms, as a video would be necessary to replace both the synchro­ signal. nizing pulses and the blanking intervals. Experimental work, with particular reference to the application for broadcasting, is now well 7. CONCLUSIONS advanced and shows very promising results. Further work, directed to the link applications, appears to A review of the properties of modulated pulses be desirable. with particular reference to their application to televi sion-sound tran smi s sion shows that sui table systems may be devised which would have con­ 8. REFERENCES siderable advantages over current arrangements. 1. British Patent No. 434,890 (Zworykin). For U.K. broadcasting, the use of a pulse­ sound system employing one pulse per line-blanking 2. U.S. Patent No. 2,270,108 (Roosenstein). interval would enabl e the current 405-1ine trans­ missions in Band I and III to be replaced by 625- 3. LAWSON, D.I., LORD, A. V. and KHARBANDA, line transmissions without drastically reducing the S.R. 1945. A method of transmitting sound on number of channels available. Further, it would the vision carrier of a television system. J. appear feasible to introduce such a pulse-sound Instn.elect.Engrs., 1946, 93, Ill, 24, pp. 251 - system alongside the conventional sound system 274. used with 625-line transmissions thus enabling a smooth transition to be made from one sound system 4. LACHAISE, M. T. 1959. Multiplex son-image de to the other without rendering useless any of the television a 819 lignes. Centre National receivers in the hands of the public. For countries d'Rtudes des Telecommunications, 1959, Etude wishing to introduce bilingual transmissions, this No. 53IT. form of pulse-sound system could provide the "second-language" channel without affecting the 5. Y ASHCHENKO, K.A. and ZVEREV, Vu. B. con vention al sound channel. 1964. Television sound transmission by time multiplexing of the video signal using phase­ The di sadvantage s of the pul se-sound sy stem modulated pul ses. 1'elecomm. & Radio Fngng, considered for broadcas ting would be: Pt.l., 1964, 8, pp. 31 - 35.

(i) The audio-frequency bandwidth would be 6. MOSS, S.H. 1948. Frequency analysis of modu­ restricted to about 7 kHz. latedpulses. Phil.Mag., 1948, 39,296, pp. 663- 691. (ii) The audio signal-to-noise ratio under given conditions would be appreciably worse than 7. FITCH, E. 1947. The spectrum of modul ated that provided by' the conventional f.m. system pulses. instn. elect. Engrs. , 1947, 94, IlIA, 13, used at present wi th 625-lin e transmi s sion s. pp. 556 - 564.

Neither of these disadvantages is considered 8. JELONEK, Z. 1947. Noise problems in pulse serious. Very few present-day television receivers communications. J.instn.elect.Engrs., 1947, 94, are capable of an audio output with a frequency IIIA, 13, pp. 533 - 545. range extending beyond 7 kHz and the signal-to­ noi se ratio provided by pul se-sound would be similar 9. AVINS, J. 1963. Sound signal-to-noise ratio in to that provi ded by the conventional a.m. sy stem at intercarrier sound televi sion receivers. I. E.F. F.. present used with U.K. 405-line transmissions in Trans.Bcast & Telev.Receiv., 1963, BTR.J), 2, the v.h.f. band, except under conditions where the pp. 9 - 17. picture signal-to-noise ratio was so low that the picture would be considered worthless. 10. LORD, A.V. and ROUT, E.R. 1966. Pulse sound. Wireless Wld., 1966, 72, 2, pp. 56 - 57. The application of pulse-sound so as to enable the sound and vision signals to be transmitted, as a 11. Self-generated vision-on-sound interference in combined signal, through links appears quite 625-line television receivers using intercarrier feasible. The performance of the sound channel in sound. Research Department Report No. 0-095, such a system could be made comparable wi th that Serial No. 1964/73. already provided using separate circuits and very significant economies could be made both by broad­ 12. CALLENDER, M.V. 1965. Sound buzz in inter- ~ casters and by common-carrier authorities. Opera­ carrier televi sion receivers. I.E. R.E. Conference tional advantages might also be obtained due to the Proceedings No. 6, International Conference on ability to consider, at all times, the combined U.H.F. Television, 1965, paper No. 25. r

30

13. The effect of the sound modulation system upon television sound on the picture carrier. Proc. the monochrome reception of NTSC-type colour Inst. Radio Fngrs., 1946, 34, 2, pp. 49 - 61. televi si on signal s. Research Department Report No. T-087, Serial No. 1962/3. 15. British Patent Application 53931/65.

14. FREDENDALL, G.L., SCHLESINGER, K. and 16. Equalizing pulses. Research Department Tech­ SCHROEDER, A.C. 1945. Transmission of nical Memorandum No. T-I077.

APPENDIX I

The Signal-To-Noise Ratios of P.P.M. and Asym­ Further, the noise VN may be considered to be the metric P.D.M. resultant of a very large number of sinusoids having equal amplitudes 6N, different frequencies ranging In thi s analy si sit is as sumed that each pul se from zero to fo, and random phase relationships. of a P.P.~. train, after the removal of all noise VN is related to 6N by: located on all low-slope regions of the waveform, is used to initiate one edge of a pulse whose other edge is free from disturbance and repeats at the mean pul se-recurrence frequency, thu s forming the corresponding P.D.M. train. The P.D.M. pulse train is then demodulated using a low-pass filter cutting off at half the mean pulse-recurrence frequency. As suming that the noi se vfilltage in a narrow fre­ This process is illustrated in Fig. 20(a) wherein it quency band df may be represented by a sinusoid of is assumed that the input P.P.:vI. pulses have unit amplitude 6N, then, from Equation (10), magnitude and are accompanied by noise of r.m.s. value VN (volts). Slicing and conversion to P. D.:vI. (11) results in the waveform, again of unit magnitude, which is illustrated in Fig. 20(b). From Fig. 20(b) it may be seen that the r.m.s. pulse-edge deviation E, due to noise, is related to The r.m.s. noise V is related to the mean N VN by: power density N, which is assumed uniform, and the video bandwidth fo by the relationship; VN :;. - E :: (12) YN :: ,v . fo (10) S

waveform (bJ PPM

(unit magn~dll) DI--~---r71--T---1~~"""::~- a. f. 7' ) lL.J ) I::::::::J > output VN slicllr PP M to low -pass (0 ) PD.M. filt(2r convrzrslon

Slopll S

(: I I 1 I I

V\N : / I1 I1 11 (b) __ .1

Fig. 20 - P.P.M. demodulator assumed for signal-to-noise calculation (a) Slicing and demodulation: block schematic (b) Waveform of P.D.M. derived from P.P.M. 31

Hence the peak pulse-edge deviation due to one 2 N frl). sinusoid of amplitude ,0,N is: No -= -- J . df S2 T2 0

N fr (5) S2 T 2 ' 2

From Equation (11) we have: Owing to the "folding", the noise power contributed by each of the remaining terms (except the last) ~2N • df makes a double contribution, thus: ,0,€== (13) s

If the sinusoid of amplitude 6 N has a frequency fO ~ N fr w/21T, then the spectrum of the pulse train, with a 2 --1--.- (16) ( fr S2 T2 2 deviation of 6 € is, from Equation (4):

where ';m is an abbreviation for 21Tm,0,€/T

and wN may have any value between 0 and Wo.

Assuming that ,;m is very small compared with The noise power contributed by the last term is the unity for all relevant values of m, we may take same as that contributed by the first: Jo(';m) as unity, Jl(~m) -= -J-1(';m) -= ;m/2 and neglect entirely all Bessel functions of higher order. (17)

F~(6€,t) thus simplifies to:

{sin(mwr + WN)t- sin(mwr - WN)t}] (14)

The first term of Equation (14) has a uniform spec­ Whence the total noi se power wi thin the band zero trum of amplitude ,0,€/T at all frequencies between to fr/2- i s o and WO/21T. The typical term sin(mwr + ~)t likewise has a uniform spectrum of amplitude ,0,€/T 2 N fr co 2fo. at all frequencies between mWr/21T and (mwr + wO> VN' -.- fr S2. T2 2 /21T. The term sin(mwr - GcN)t is a similar and balancing sideband equally spaced in the negative Nfo -= frequency sense, but it must be noted that wo> mWr S2 T2 except for the highest permissible value of m, so that for all values of m

After demodulation by the low-pass filter Thus the r.m.s. noise output from the low-pass filter cutting off at the frequency fr/2(or Wr/41T) the total is: noise power at the output may be derived as follows: VN V,v' -= The noise power contributed by the first term S T of Equation (14) is: The peak output signal corresponding to a peak deviation of 0 (which corresponds to a peak devia­ 2 fr/2 tion of d for P.P.\1.) is, from Equation (4): No -:: L: l~ • o Vs = Substituting for ,0,€ from Equation (13) we have: T 32

Whence the ratio of the r.m.s. signal to the r.m.s. We may also write: noise is: d S Rp.D.M. Rp.P.M. -= c:: VN'G

APPENDIX 1I

The Signal- To-Noise Ratio Obtained Using Inter­ However, the factor 1/k varies as a function of the carrier F.M. Reception, as Compared to That picture-signal magnitude. The mean noise power Obtained Using Only the Sound Carrier accompanying the vi sion-carrier component at the demodulator is: If 1/k is the instantaneous ratio of the vision­ carri er voltage to the sound-carri er vol tage, *

VNV is the instantaneous noise voltage accom­ panying the vision-carrier component at the Hence the total mean noise power at the demodulator intercarrier demodulator, is: and VNS is the r.m.s. noise voltage* accompanying vRrs + vRrs co 'V&sO + k!) the sound-carrier component at the intercarrier demodulator. and the total r.m.s. noise input, V.~T, to the de­ modulator is: The relative effects of "Nv and VNS upon the noise output of the demodulator may be expressed VNT VNS ~1 + P as: vNV c:: k • VNS . VNS~l +K'2 R' whence K'2 ... Both the sound-carrier voltage and its accompanying ~ = ~1 + r.m.s. noise voltage are assumed constant. RF.M.

SMW

Printed by BBC Research Department, Kingswood Warren, Tadworth, Surrey