US2447233A

US2447233A - Pulse time modulation multiplex receiver

Info

Publication number: US2447233A
Application number: US526566A
Authority: US
Inventors: Chatterjea Prafulla Kumar; Houghton Leslie Wilfred
Original assignee: Standard Telephone and Cables PLC
Current assignee: STC PLC
Priority date: 1943-04-07
Filing date: 1944-03-15
Publication date: 1948-08-17
Anticipated expiration: 1965-08-17

Description

Aug. 17, 1948. P. K. CHATTERJEA ETAL' 2,447,233

PULSE TIME MODULATION MULTIPLEX RECEIVER [nu ntors A ttorn ug 17, 1948.- P. K. CHATERJEA lai-Al. 2,447,233

PULSE TIME MODULATION MULTIPLEX RECEIVER l Filed March 15, 1944 v sheets-sheet a Attory ETAL 2,447,233

'7 Sheets-Shef 5 P. K. CHATTERJ'EA PULSE TIME MODULATION MULTIPLEX RECEIVER Filed Marqh l5, 1944 Aug. 17,1948.

BACI( C672 PULSE- l H H HH sff/Frcrrs. F/Gfa Aug- 17, 1948 P. K. CHATTERJEA ETAL 2,447,233

PULSE TIME MODULATION MULTIPLEX RECEIVER Filed March l5, 1944 7 Sheets-Sheet H all lll rm). C C H 'cl/H cl//H A [torn Aug. 17, 1948. P. K. CHATTERJEA ET AL PULsE nmz MonuLA'rIon MULTIPLEX nEcEIvEa Filed March 15, 1944 '7 Sheets-Sheet 5 Ffa/.2. wm

u s 8 n. 9. 4 wm., UT/. 2. 3 .4. PMM .0. 0. a Mu. N N N A m m B 8 ,DE 7 M. ww. 9 l 6 e n we 7% EN 2K U C E o M 0 C UG M mm P/ MM f *IL .a mM 6 Wu T CT DP M ma oa a B Tx C rc P.

EFFEC 7'/ VE' SVG/VAL Assi/VCE of im. spa-'cl Aug. 17, 194s.

P. K. cHATTERJr-:A ETAL 2,447,233

PULSE. TINE Monunurou MULTIPLEX RECEIVER Filed March l5, 1944 7 Sheets-Sheet 6 cfM/v/VEL TERM/NAL l4 5 C D A @C D c'o/zPEcr SEQUENCE 2 34 2 34 wmv nvrffpfA/cs pms! mfr 23 -r 4 23 CHAN/VEL TERM/NAL A PULSE mA//v 7A TUMA VOL TAGE /09 CONr/eol. voL MGE F3n/w /04 Aug. 17, 1948. P. K. CHATTERJEA ETAL 2,447,233

l PULSE TIME MODULATION MLTIPLEX RECEIVER Filed March 15, 1944 v 'r ,sheets-sheet 7 2nd. BUILD AcK C] BEH- Asc Ell/e7- FILTER W A ttorny- Patented Aug. 17, 1948 Y A .2,447,233 PULSETIME M O DEKIIKJA'LLION 'MULTIPLEX 'RECEIVER Pra'fulia -Kuinar Chatterjea and Leslie Wilfred England. .assieme t0 Standard Telephones and Cables Limited, Lon- I -Y -don,England,-a Brit-ish company y Y Y Application March-515, 19454, Sria-*l No. 552.6;5'66 v,

Greatritain 'April '1, 194'3 "l The present invention relates to 'multiplex `signal high frequency `'transmission systems "ojf "the kind utilising electrical pulses.

It is now Well known that signal Waves bear*- inff Vintelligence can loe 'transmitted by means ci.. a train of 'electrical pulses 1occurring fat equal time intervals and whose dura-tions are varied in vaccordance with the instantaneous amplitude of the signal. Such pulses are 'herein referred to as duration modulated pulses. In 1a modification of this system of signal Wave transmission, shcrt electrical pul'sefs'cffcoiistant'duraticn are transmitted to mark the lleading and trailing edges of duration 'modulation Apulses,'and Wlre'n either the leading or trailing Yedge occurs at'equal intervals of time, onlythe pulse Amar-King 'the edges of the duration trnodulated pulses Which occur at varying -intervals of timefneedibe trans*- mitted. This type of transmission `is 'sometimes referred to as ia single pulse or time 'modul-ated pulse transmission 'cr 'pulsed phase'fmdulation, the pulses o1" constant -short duration `occurring at varying intervals o'f time in accordance with the instantaneous amplitude of thesignalrwave to be transmitted.

When both the 4ixed and -movirig'f'malrking pulses of a duration modulated 'system 4are'traitsinitted, the system is referred lto 'as ya fdoilble pulse system. In the single pulse system, :the Xed pulse may be rei'n'sert-e'd `at 'the receiver and the solid pulses dr'duatn mdulatedpulses obtained from `the 'iixed movable ,.'pa'irs *of Lcn'don,

it claims. (otite-i5) l timed orfphasedto-occur pulsesfor example lcymearis'cf 'knownfbuildback circuits which serve `to rproduce the original rectangular duration yn'ioclulated pulses. It 'in the single pulse sys-tem the pulse duration lis -only a small "fraction of the time interval between the occurrence oi 'successive pulses. fafnd 'if the .variation of this time interval 'taken up lfor 'fa manie inuin time cr phase modulation of the pulses is also small, then other trains of similar zp'ulses may be transmitted having 'the fsanie'pulse repel titicn frequency, but vvhselpulses occur'in the intervals between the pulses 'of f'ano'tlier train?. Thus each pulse train `may lhear time or Aphase 'modulation of its respective 'sign'f'a'l 'wave and 'a multiplex channel transnisscn v'syste'ni is prbvided. It 'is the object of thisin'ventio'n to p 'ro' vide 'practical arrangements 'whereby such `la system may be put into practice.

In systems of this type fo'r main 'operations are essential. Firstly, producing the Gramsci unn'iodulated or uniphase'd pulses, one train "for each channel, the pulses o'f ach train idr-each channel, the nuisesio'f each being corrctiy 2 with respectto the` pulses 'voi other tra-ins of the system, and modulating the `.pulses of each chan-nel Awith the vrespective signal Wave. Secondly combiningall the channels into one transmission. Thirdy, -at the re '-ceiving enf-lof the system directing the pulsesgoi a channel to respective receiving apparatus. Fourthly, obtaining the signal Waves from the ytrain o-fpulses -in fthe respective channel. v twill be understood that Amore than one of these x-opera-tions may be performed by the same equipment; that `the pulses-may be transmitted by line, Wave guide or radio,` and `may comprise di- 1rect currentnpulses 'or trains oiwaves `of higher ireq-uency than the pulse repetition frequency.

-Apparatus for' carrying out these loperations in practice Will now ice described, yreference being made to the accompanying drawings.

-In the drawings-z n Fig. -l is an. explanatory 'diagram relating to an Varran'gement for producing timeaphased lpulses;

Fig. 2 shows in blockischematic a circuit arrangement `of a `multi-channel pulse :signalling Fig. '3 is lan v errplainatory diag-ram used in con nection with Fig. 2; n r Y 'Figxfi shows in block schema-tic another `circuilt Karrangement :of la lr'nulti-channel pulse signalling system; v

`Fig. 5 :shows "in blockschematic a lfurther circuit farraiigern'ent of Ta multi-channel Vpulse signalling Isystem;

xFig. =6 is an Fexplanatory `diagram used in ccnlu Anecticn with Fig. 5; Y

vFig -'7 vshows diagrammatically a cathode ray tube arrangement for producing time-phased modulated -pulse's in "a'lplurality off channels;

Fig. 8 rshows lin block :schematic vform circuit arrangements fof 'a receiving equipment; Fig 9 -sliiovsexplariatcry curves used in `the descripti'onf AFig. 8; Y

Fig. "1a 's'hvvs 'in block schematic Afunn Athe Are Caving circuit 6r rtg, 1.1 incftran'ng corrective devices in accordance vvt'h features f tl'ieinm vention; A

Figs. I3 and 14 show various curves used in -t-he description; Y

Fig. 15 shows lin `block schematic form receiving arrangements incorporating further corrective devices;

Figs. 16 and 17 are explanatory data;

Fig. 18 shows curves used in the explanation of the operation of the corrective devices used in Fig. l;

Fig. 19 is a circuit arrangement of the unit |05 of Fig. 15;

Fig. 20 shows a circuit arrangement of the systems shown in Fig. 15.

From one aspect, a system of the type specified comprises a series or train of short constant duration pulses, every nth pulse of which is time or phase modulated in accordance with a desired signal wave, there being n different channels. From another aspect, a system of the type specified comprises a plurality of pulse generating devices of the same pulse repetition frequency, one for each channel, and which are properly timephased with respect to each other so that the generated pulses of one device or channel occur during the intervals between the generated pulsesof the other devices or channels, the duration of the pulses and the intervals between successive pulses of the combined channels leaving sufficient time between successive occurrences of pulses in the combined channel train to allow for time modulation of the pulses.

The pulses of a plurality of channels are shown in Fig. 1 of the drawings. By way of example, four channels are illustrated. Channel I consists of pulses a, a1, all, etc. Channel 2 consists of pulses b, b1, D11, etc., and so on for channels c and d which are representative of

channels

3 and 4 respectively. Each pulse is liable to occur at any position within the time interval indicated by Yits adjacent dotted lines, depending on the amplitude of the modulating signal.

vArrangements for producing a time-phased pulse train for one channel will now be described.

Firstly, a train of pulses are duration modulated according to known technique, and difierentiated for -example by passing through a high pass lter to produce pulses of short constant duration of opposite signs respectively at the instants of the leading and trailing edges of the duration modulated pulses.

There are many ways of phase modulating a pulse train, and similarly there is also a variety of methods which may be employed forcombining individually modulated pulse trains, and/or the producing of pulse trains having the correct unmodulated relationship prior to their being modulated by their respective modulating signals. A basic circuit arrangement is shown schematically in Fig. 2 where I represents a sinusoidal oscillator of the pulse repetition frequency of each channel. Its output is fed to

units

2, 3, 4 and 5, which phase the sinusoidal wave the correct amount in each channel corresponding to the unmodulated relative pulse phases of the different channels as shown at a, b, c and d Fig. 3.

Units

6, 1, 8 and 9 are further phasing circuits in which the degree of phasing is controlled by the applied signal voltage applied at terminals I5, I6, I1 and I8, respectively, according to known practice such as is employed in frequency modulation systems. If desired, these two stages 2-6, 3--'l, 4 8 or 5 9, may conveniently be combined. Units I0, Il, I2 and I3 are pulse forming circuits for example biassed amplifiers and amplitude limiters which convert the applied phased sinusoidal waveforms into short duration pulses, which pulses from all the channels are fed to a common amplier so as to form the required single pulse Itrain in amplifier or other unit I4. This whole operation is further illustrated by Fig. 3 where a. represents the output of I, Fig. 2. It is usual though not necessary to allow this to provide one of the channels so that here no phase displacement is given by 2, Fig. 2. Phase displacements given by

units

3, 4 and 5, Fig. 2, provide outputs as shown at b, c and d, Fig. 3, respectively.

Neglecting for the moment the variable phase shifting produced by the modulation, signal units IB, Il, I2 and I3 produce pulses as shown at e. f, y and h, Fig. 3, which on combining at unit I4, Fig. 2, produces the pulse train y, Fig. 3, which may then be employed to modulate a carrier wave transmitter according to known practice.

Another basic arrangement for producing the phased pulse trains is shown in Fig. 4 where I9 is any pulse generator and 20, 2|, 22 and 23 are time delay networks which may each be divided into two sections 20a, 20h, 2Ia, 2lb, 22a, 22h, 23a. and 23h, respectively, if desired, parts 20A, etc.,

Vgiving the requisite fixed delay so as to phase the pulses of each channel appropriately and parts 20B, etc., providing a small Variable delay, controlled by the modulating signal which would be? applied at

terminals

24, 25, 23 and 21 respectively. Such variable time delay device may, for example, comprise an artificial line network having an electron `discharge device arranged so that by means of its reactance Variation, known as the Miller effect, the inductance or capacity of the net work can be varied to modify the delay produced by the line network. The Miller effect is controlled in known manner by the signal voltages. This circuit may be of any well known form as is used in ordinary phase modulation circuits, it being only necessary to substitute a pulse t-rain for the usual sine wave input of such phasing units. A suitable circuit for this purpose is illustrated in U. S. Patent No. 2,259,392, grantedV October 14, 1941. Then these pulse trains would be combined at amplier 23 and the resulting pulse train obtained at terminal 29 used for modulating a transmitter.

A third basic arrangement for producing the phased pulse trains is illustrated by Fig. 5 where 3|] is a sawtooth generator or generator oi any waveform from which it is possible to produce a variable width pulse. Y This is then fed into pulse

former units

3l, 32, 33 and 34 respectively each containing an amplifier, and these ampliers are each biassed or otherwise adjusted so as to pass varying portions of the wave form and thus to produce progressively longer duration pulses from channel to channel (see Fig. 6). By applying also a modifying voltage determined by the channel modulation signal to

terminals

35, 36, 31 and 38 respectively, operating for example on the amplifier kcontrol grids, these various width pulses may be duration modulated andthcn after dierentiation by

units

39, 4B, 4I and 42 which comprise, for example, high pass lters and amplitude limiting by

units

43, 44, 45 and 46 respectively, to eliminate one of the differential pulses, the differential pulses due to the moving edges of the duration modulated pulses are obtained and are the required phased pulse trains which maybe combined together in amplier unit 41 prior to being used to modulate a transmitter. If the initial waveform is a sawtooth the resulting pulses will have one xed edge the differential pulses resulting from which are eliminated, and one variable edge, but in the case of, say, a sinewave'both edges will vary. However, this and-3b the rectified mea-n. voltage. ItI will-bessen that pulses-` |140Bjcorresponding to transmitted pulses,l I 40; may Ybe easily segregated 'by amplitude clipping to provide .the lsynchronizing waves. -gThe same'efectfcanof course be obtained by omitting Vone channeL: a. negative 'dip being obtainedasshown dotted at b, Fig. 9, which may be usedas thesynchronisingp'ulse. f- 1 Y iis-fAn entirely different method of reception 'makes use of-.thadoublepulse build backcircuit `principle and may be termed Progressive selection method. HA-build backfcircuit is a multiavibratoritype :circuitwhich-,has a free `.oscillation period at least-asgreat as vthe greatest pulse sepa- .rationand has, double stability, or in other words, two, conditions ofrest, and-,isgarranged to remain in either condition of rest until acted upon by a pulse, whereupon it shifts to theother condition otrest `in which condition it remains until acted uponibyapulseof the same polarity as the previous pulse, whereupon it shifts backV again to its first condition of rest. The method using such a circuit consists in passing the received pulse train obtained from the receiving unit after rectiilcation of the H. F. carrier signal, through a vb uildback'circuit with the result that alternate pairs oi pulses form alonger duration pulse, the leading edge being formed by one pulse and the trailing edge bythe next, the position .of these edges being dependent upon the positions of the corresponding pulses. On passing this newly formed pulse through a diierentiating circuit a positive and a negative pulse will be obtained from the respective edges and these may readily be separated vby passing through la limiter stage. This therefore, has the result of separating the received pulse train into two separate trains. A further division of these two trains may now be carried out until eventually pulse trains corresponding to `individual channel pulse trains are obtained. Theoperation of this arrangement may be rreadily seen by referring to Fig. l where e represents a received pulse train containing for example four channels, namely a, b, c and d, channel a consisting of pulse a, a1, al1, etc., channel b of pulses b, b1, b11,et etc. Y v

- On feeding this pulse train e to a build back circuit the pulse train shown at f is obtained which on passing through a differentiating circuit, for example a high pass iilter, produces pulse train g consisting of alternate positive and negative pulses from which bypassing through correctly adjusted limiting circuits may be obtained pulse trains hand i. O-n repeating these operations with build-back and diierentiating circuits utilising the pulse train shown at h, pulse train y is obtained and thence Vpulse trains 7c, l and m from which it will be seen that pulse train l consists of the pulses Yforming channel a only i. e. a, a1, a11,etc. and pulse train m consists of channel' c pulses only namely c, c1, etc. A similar operationparried out utilising pulse train i would result in the separation of pulses of channels b kand d.

The great advantage of this system of reception is that no synchronising mechanism need be employed at Vthe receiver but the number of channels required in this system must be 2 where 11, can have any integral value between 1 and inlnity. A schematic circuit arrangement for such a system is shown in Fig. 1l in which 12 represents the receiving unit from which the multi-channel pulse train is obtained. 13 represents the first build back circuit and differentiation circuit followed by

limiter circuits

14 and 15 respectively adjusted to pass vOli-'ly one o ffthe differentiated pulses from the output of 13. Each Vof these then feeds into thesecondbuild back and differentiation circuits 16 and 11 respectively, from whichthe two further pulse -trains forming the required channel 4pulse trains are obtained via amplitude limiter circuits 18, 19, and 8l at terminals B2, 83, 84 and 85 respectively. Whilst an aerial pickup has been shown in Fig. 11 it will be understood the pulses may bereceivedover a transmission line. p

More specific circuit detailsywill be disclose hereinafter in connection with Fig. 20.-vr` Y Y The demodulation oftheA individualchannel pulse trains may be performed by the employment ofa circuit as described in the specification of United States application Nlo. 374,660, issued Patent No. 2,406,790, September 3,1946, or by any other known pulse demodulation circuit such as a low pass lter. The circuit disclosed in said application `is a differentiating Ydetector circuit having a three element discharge tube provided with a grid leak and condenser input circuit. The Whole system may be made to include all the advantages inherent in pulse systems of transmission such as improved signal to noise ratio and economical power consumption. With the synchronised systems of reception each channel is made unresponsive to any unwanted signal such as noise interference except during the period at which the wanted pulse may occur. In the case of the progressive selection method, by operating all amplifiers in the pulse circuits from anode current cut yoi to overload or saturation value, unwanted signals are in this case also eliminated except for the period occupied by the pulse edge as with all pulse systems of transmission.

In its basic form the progressive selection system just described is liable to produce unwanted results if the received signal is not entirely free, especially where radio links are employed, from external influence such as interference or fading etc. or intentional jamming Yby insertingextra pulses to change the recurrence frequency.

Under such circumstances it is possible that any individual channel signal may appear at different selecting circuits in a random fashion.

The-following are the methods and arrangements according to features of this invention by which such occurrences may be prevented--it will be realised that these methods will also secure the correct starting operation or selection, in proper rotation, of the channels of the communication system.

In describing the following methods it will be assumed that such details as limiting, differentiating and other such items will be inserted where necessary, `as will be understood by those versed in the art. The description will first Vbe oi the principles of methods which may be used. These can be enumerated as follows:

(a) Correction of functioning may be obtained by the use lof build back circuits oi the multivibrator type wherein their freel or natural running frequency periods correspond to maximum permissible time modulation of the received pulse train in the manner shown in Fig. 13 in which a, b, c, etc., represent the pulses,V al, a2, b1, b2, c1, c2, etc., the respective limitsof time modulation. The second curve in this figure shows the pulses when the multivibrator circuit is running free at the pulse repetition frequency of each channel multiplied by the number of channels, the pulses having the maximum durations. The third curve shows the pulses in the multivibrator outq ..9 put circuit'when controlled by signal pulses, there being no external eiTect to upset the proper and correct reception. To ensure correct occurrence at equally spaced intervals of time, the leading edges of the pulses would be synchronised to a local master oscillator, which would in turn be synchronized with the average repetition frequency of the incoming pulses. Trigger pulses from the master oscillator may be suitably phased in a known manner and used t operate the multivibrator circuit in one sense to produce the leading edges of the pulses shown in the third curve of Fig. 13, while the incoming signal pulses are used to Ioperate the multi-vibrator in the other sense to produce the trailing edges of the pulses.v

(b) By converting the time interval of the interference into an amplitude transient which latter may then be used to provide the required correction. For example, referring to Fig. 14, a train of sharp signal pulses would produce in a circuit a mean D. C. level of height hm whichin the absence of external eil'ects will be constant but when any external interference arrives the value .of hm will vary and this variation may be used to provide the required correction, asby utilizing the value hm to control the blocking of the receiver so that the interference p-ulse will block the receiver and prevent the response of the build back circuit at that instant.

(c) A distinguishing feature may be given to one channel pulser train such as a pilot frequency modulation fc which may occupy the same channel as Ya Ysignal or may occupy a channel itself. Alternatively one channel pulse train may consist of pulses of a duration different from the pulses in the remaining channels. Other methods not so satisfactory consist in giving the pulses in one channel a different amplitude or a differently shaped edge from the pulses of other channels.

Use is then made of this distinguishing feature or channel characteristic at the sections of the system where pulse trains are separated by providing an additional unit capable of indicating whether the special pulse train is in its correct path of the equipment. The output from this unit may then be arranged to provide the necessary correction such as to insert or eliminate a pulse from the incoming signal, or modify Vthe operation of a build back circuit to produce the equivalent eiect.

For the case of a pilot frequency this unit will consist of a pulse demodulation circuit incorporating afilter capable of selecting the control frequency fc, and for the case of a diierent pulse width or duration, Va `demodulation circuit givingv a D. C. output voltage at a constant level for correct operation, and a variationfor incorrect operation. This D. C. output may be applied directly or by meansrof a bridge circuit arrangement balanced at Vthe Voltage of correct operation to provide the requisite correcting control. f

A control unit must be incorporated at the output of each pulse separating stage, i. e. 74, (Fig. 11),-i. e. at the output of the differentiating yand limiting stages, in order to ensure correct-channel selection' on starting up the receiver, or ifan interfering signal lasts for a large number of received pulses, or alternatively if the receiver signal fades badly.

It will be appreciated that although itis only necessaryV to employ one channel' for control purposes, more than one may beus'ed for providing greater scope for detail design considerations.

Referring again to Fig. 10 and to Fig. l2 in which parts are given like reference to corresponding parts in Fig. 11, let it be assumed that pulse train a has a distinguishing feature, such as a modulation frequency fc. Now the lirst stage of selection occurs after pulse train g has been obtained and results in two separate pulse trains h and i the former containing pulses a and c and the latter b and ld. If the circuits for accepting a and c, i. e. 14 in Fig. 12, do so in fact, the control unit 86, Fig. 12, then extracts from the a and c pulse train the component of frequency fc which is applied to prevent correcting unit 81, Fig. 12, from functioning. If the pulse train fed tol'fl, Fig. 12, consists of pulses b and d then no fc is obtained and unit 81, Fig. 12, applies, as explained fully hereinafter, the necessary correction to '13, Fig. 12, for example by inverting the pulse train f, Fig. 10, or by inserting a pulse in the received pulse train,

It will be seen that the pulse train accepted by 'M as correct may be a, c, a, c, etc., or c, a, c, a, etc., only one of which will provide the correct signals at

outputs

82 and 83 etc., Fig. 12, hence the same :control operation must 4be repeated by

units

88 and 89, Fig. 12, which provide a signal capable of causing the output from unit 'I3 to be advanced or retarded Vtwo signal pulse periods, by the insertion or elimination of two received pulses. By causing circuit 89 to control the rst build back unit 13 and not the later-'ones such as 'I6 it is possible to ensure that'the correct signal is obtained at the

output terminals

82, 83,

` 84 and 85 respectively, by the use of only; one

pilot frequency. n

(d) By providing a distinctive modulation for each channel so that at the receiver means may be provided for converting'this modulation into voltage to be usedv for controlling correction circuits if a channel other than the correct one `appears at a selected channel terminal equipment. The type of correction provided depends on the cause and duration of received signal interruption. For simplicity correction will be described for overcoming interference appertaining to the receivedsignal and not to circuits at a later stage of the receiving equipment, Which'if deemed necessary could be a duplication of parts of these arrangements. This method has many advantages over the more straight forward methods referred to under (c) and typies a type of synchronising method for multichannel systems employing a modulating component as opposed to a special pulse characteristic atchannel pulse repetition frequency.

A schematic diagramrof such an arrangementv is shown in Fig. 15 and in more detail in Fig. 20. The partsin Fig. 20 'which correspond to the blocks in Fig. 15 are given the same references as the blocks. As there are known build back circuits represented at BB available which willv operate with positive or negative pulses, it is assumed that the appropriate ones are used, otherwise additional phase reversal devices will be necessary. The units comprising capacity CII` and resistance R together form a -pulse differentiation circuit. DN. is a delay network and F a filter or other arrangement for ,producing/a D'. C. control voltage from the characteristic of the channel, for example, the pilot frequency. The block 96' represents the ilrst stages of a receivergiving an output similar to the train of phased pulses used for modulating the transmitter. These are then passed through correcting unit |06 to the rst build back and diil'erentiation circuit 97 after which the rst division takes place at

units

98 and 99 and the process is repeated giving two further divisions at units and |0| from 93 and |02 and |03 from 99. This process ymay be continued further, but by way of example, a four channel system will be described herein, as shown in Fig. 15. Units |04, |05 and |06 are the circuits provided for maintaining correct channel selection.

For the case where a pilot frequency is used outside the required signal frequency band, the pilot frequency wave may have a different amplitude or frequency for each channel. In the latter case the filter for separating out the pilot frequency may be such as to provide at the output thereof different amplitudes for different input frequencies received. In one case, such a filter would be connected to the output of one o channel and be provided with a rectifier circuit for producing a D. C. voltage which has a predetermined value when the correct signal is received and would vary when the wrong signal is received by that channel. Y The variation in voltage may then be used to provide the correction to the incoming pulse train so` that the pulses fed to the channel having the filter provide the said predetermined datum voltage which exerts no control or variation in the operation of the correcting unit. 'I'his pilot frequency selection filter is indicated in Fig. 15 by |04 the output from which is fed to the two correcting units |05 and |06. Unit |05 is capable of extracting pulses from the received pulse train and unit |05 provides means for inserting pulses in the received pulse train. Such units will be described more in detail in connection with Fig. 19.

The necessity for providing two such types of correction units |05, |06 is in order to prevent continuous over correction which is a possibility when employing control voltages whose varia,

tions are slow compared to the pulse repetition frequency. Fig. 16 indicates in the third row down the effect of the introduction of a short interfering signal (INT), or in the fourth row down the loss of va pulse (the second pulse) due to fading and it will be seen that these occurrences are equivalent to altering the channel selection sequence.

For example, let it be assumed that unit |04 Fig. 15 is connected to channel terminal B and that pulse train 2 provides in the outputJ of unit |04 a voltage equal to the predetermined or datum voltage, and then as shown in Fig. 17, the voltage produced at terminal B by the channel train of pulses will be less and Voltage produced by pulse trains of

channelsV

3 and 4 greater respectively than the voltage produced by the pulse trainV of channel 2. If an interference pulse is received pulse train I may be received at channel B and then output Voltage of unit |04, Fig. 15, will be reduced from that produced by the correct train 2. From a study of Fig. 17 it can be seen that a pulse missed will cause an increased voltage output while 2 pulses missed will cause a greater output voltage of |04, acondition which, as will be seen hereinafter will cause a quicker correction and is especially useful where more than four channels are employed. 'Ihe keffect of a long period of noise, or fading may leave the system in any condition, but if circumstances are such as to require it there will immediately be produced in the filter output the necessary correcting voltage.

Units |05 and |06 may consist of a variable impedance shown as an electron discharge tube H0, Fig. 19, whose vcontrol grid G is provided with bias voltage from unit |04 applied at terminal |09. The received pulse train is applied between terminals |01 and |99 to the anode and control ,grid via capacities Cl and C2 respectively.

The operation of unit |05, the pulse extractor unit, Fig. 15, can best be explained in conjunction with Fig. 18 and Fig. 19. (a) Fig. 18 represents the input pulse train to |05, Fig. 15, at terminals |01 and |03, Fig. 19, yobtained from receiver 96, Fig. 15. vUnder normal conditions the output voltage of unit lili-Fig. 15, which is fed to |09, Fig. 19, to control 'the impedance of the tube is such as to cause the output of 55, Fig. 15, obtained at terminals and H2, Fig. 22, to be as shown at (b) Fig. 18.

On the occasion of an additional interference pulse being received it will be seen that channel pulse train No. l will be received at B (see Fig. 16) so that the voltage output of |04, Fig. 15, will be reduced which will cause the impedance of H0, Fig. 19, to increase. The tube H5, Fig. 19, is initially biassed beyond positive cut-off, i. e. beyond saturation point so that the resulting increased negative bias voltage on valve Vl le, Fig. 19, will cause this valve to feed an out of phase pulse to terminals H2 and lll, and the succeeding output pulses from unit |05 will tend to decrease in amplitude. However, as the impedance of valve H0, Fig. 19, increases, so does the time of the trailing edges of output pulses also increase. These ,successive pulses will still operate unit 91, Fig; 15, until an extra large spacing between pulses occurs when unit Sl, Fig. 15, will not be operated, owing to the fact that the succeeding pulse occurring after the extra large spacing occurs after the preceding pulse has terminated and the amplitude of the said succeeding pulse is not sufficient by itself to operate the build back circuit 01.

Due to the variable spacing which occurs between the pulses and the consequent reduction in voltage at a relatively slow rate compared to the mean pulse repetition frequency, pulses will only be eliminated as just described after the longest intervals which are liable to occur between pulses. The chances, however, of pulse elimination will increase until the requisite number of pulses have been eliminated and the control voltage from |04 is increased to its normal value.

Unit |06 of Fig. 15 provides means for inserting an additional pulse or pulses into the received train of pulses and consists of a similar circuit'to that of Fig. 19, or an impedance may be used in place of valve ||0 so as to produce at the terminals III, H2, a varying amplitude pulse train which is fed to an amplifier |29 (Fig. 20) that is Anormally biassed beyond out off. Its grid bias is, however, controlled by the output voltage of |04, Fig. 15, so that in the event Yof this output becoming positiveY due to missing pulses in the received pulse train, the amplifers bias is reduced and a maximum amplitude pulse, obtained from the second of two pulses occurring close together, is amplified and passed through a delay circuit DN (Fig. 20) from which it is fed to the first build back circuit 91.v A nominaldelay Vshould be used in DN, and if the rst pulse coincides with another pulse, the operation will be repeated. The frequency of this operation will be increased as the control voltage increases. In curve c, Fig. 18, the dotted line (i) shows the .cut off bias to which the amplifier |29, Fig. drops as the output from |04 increases and the continuous line shows the varying amplitude pulses in the output of |'|l. It will be seen in this illustrative case that only one pulse passes to amplifier |29.

y'Ihe random action of this addition and subtraction of pulses by units |06 and |05 respectively provides time for the control voltage to be restored to normal and so eliminates the possibility of excessive over-correction. Although a low frequency control signal must inherently tend to be relatively slow in operation, by correct adjustment of the circuit time constants involved, it is possible to obtain very rapid action with the arrangement described as use is made of the iirst initialchange only. fi .i ,1

The advantages of employing a distinctive feature of the channel modulations as compared with the employment of additional pulses for synchronisation are a reduction of the overall band width of transmission with a saving in radiated energy, and also it would most probably permit of `an increase inthe maximum permissible modulation per channel.

A further point applicable to the system herein described -is that it is possible to adjust unit |05, Figs. 15 and 20, so that pulses of a shorter duration than thecorrect pulses will not be passed through it with su'icient amplitude to operate thefirst build back circuit.

All that is necessary to convert a phased pulse system as hereinbeiore described, into a double pulse system is the provision at the receiving end of the fixed marking pulses which may readily be accomplished by any of the following methods.

Firstly, to the master oscillator ci the channel pulse frequency, obtained by any of the previously described methods may be added further phasing circuits'for generating the fixed marking pulses.

Secondly,v in the system requiring a pulse-for controlling vpass circuits, this pulse should be increased in duration and phased so as to permit both the received phased pulses and the inserted Xed pulses to be passed. The fixed pulses are obtained by multiplication from the master oscillator i. e. if four channels are required, then the fourth harmonic is generated and the correspondingl pulses obtained therefrom applied simultaneously to each channel circuit, one pulse of every four being accepted by each channel in turn.

Whilst an antenna has been shown, the pulses may, it will be understood, be received over a transmission line.

Another method, employing the progressive selection principle has one channel unmodulated and then the pulses in this Vchannel are either phased by time delay networks or by extracting a Wave of channel pulse repetition frequency and obtaining from it pulses which are properly phased by phasing circuits for the respective channels.

This principle could, of course, be employed with any system of phased pulses.

In the case where a large number of channels are required, and a frequency multiplication stage is employed, it may be found an advantage to sub-divide the channels into groups.

It will be understood that the methods and arrangements described hereinbefore for directing the pulses to their respective channels at the receiver may also equally well be employed in a` multiplex system utilising duration modulated pulses in place of the time modulated or phased pulses, with the exception of the progressive selection system which involves differentiation circuits. The signals may be obtained in each channel from the duration modulated pulses in known manner.

` What isclaimed is:

1. In a multichannel electrical signallingl sys-V tem utilizing the-transmission of electrical pulses, each channel comprising a train of time-phase modulated pulses, time-phased with respect to the pulses of other channels, means for directing the pulse trains of the-channels to respective receiving apparatus including progressive selection means for deriving from the received total pulse train of all .the channels a particular channel pulse train comprising a iirst stage including build back circuit means for deriving from consecutive pairs of received pulses further pulses of durations equal to the time intervals between the received pulses of the respective pairs, means for differentiating said further pulses, means for eliminating pulses otone sign from the pulses resulting from the differentiation means, and further stages including bui-ld back circuit means for obtaining in like manner from consecutive pairs of the selected diierentiated pulse train vof the preceding stage further pulse trains which comprise pulses of a single channel only, and means in each receiving apparatus for deriving as an amplitude modulated wave from the time-phase modulation of the train of pulses` receive the intelligence carried thereby. v

2. In a multichannel electrical signalling system as claimed in claim 1, progressive selection means wherein said means for deriving from the received total pulse train or from a train of resulting differentiated pulses comprises a build I back circuit arrangement for producing pulses whose durations'are equal to the time intervals between the `pulses of the respective pairs of pulses. f f

3. Progressive selection means asr claimed in claim l furthercomprising correcting means for ensuring correct channel selection of pulses from the received pulse train. Y l f 4. Progressive selection means as claimed in claim 1` further rcomprising `correcting means ins corporated at the output of each pulse separating stagefor ensuring correct channel selection of pulses from the received pulse train.

5. Progressive selection means as claimed in claim 1 further comprising correcting means incorporated at the output of each pulse separating stage for ensuring correct channel selection of pulses from the received ypulse train, said means being designed to insert a pulse or to eliminate a pulse from the received pulse train at the Y input to a pulse build back circuit means.

6. In a multi-channel electrical signalling system utilizing the transmission of electrical pulses, each channel comprising a train of time-phase modulated pulses, time-phased with respect to the pulses of other channels, receiving apparatus comprising means for directing the pulse trains of the channels to respective channel receiving circuits including'progressive selection means f-or deriving from the received total pulse train of all the channels a particular channel pulse train comprising a iirst stage including build back circuit means for deriving from consecutive pairs of` received pulses further pulses of durations equal to the time intervals between the received pulses of the respective pairs, means for differentiating said further pulses, means for eliminating pulses of one sign from the pulses resulting from the differentiation means, further stages including build back circuit `means for obtaining in like manner from consecutive pairs of the selected differentiated pulse train of the preceding stage further pulse trains which comprise pulses of a single channel only, and' correcting means incorporated at the output of each pulse separating stage for ensuring correct channel selection of pulses from the received pulse train, said correcting means comprising a control unit and a correcting unit, said control unit being responsive to a characteristic of the received pulses and said correcting unit comprising means for inserting a pulse inthe received pulsetrain and means for eliminating a pulse in the received pulse train under the control of said control unit, and means in each channel receiving circuit for deriving as an'amplitude modulated Wave from the timephase modulation of the train of pulses received the intelligence carried thereby.

7. Progressive selection means as claimed in claim 6 wherein said control unit comprises means for giving a variable output voltage depending upon the variability of predetermined characteristic of the pulses in a channel.

8. Progressive selection means as claimed in claim 6 wherein said means for eliminating a pulse from the received pulse train comprises a variable impedance, means for varying said impedance under the control of said control unit, means for applying the received pulses to the input terminals of said impedance, and means for applying the output pulses to the build back circuit means.

9. ProgressiveV selection means as claimed in claim 6 wherein said means for inserting a pulse into the received pulse train comprises an impedance, means for applying the received pulses to the input terminals of said impedance, an amplifier, 'means for varying the gain of said aml plier under the control of said control unit, means for feeding the output from said imped ance to said amplifier, and means for feeding the output of said ampliiier through a delay network of appropriate delay constant to the build back circuit means.

v10'. Progressive selection means as claimed in claim 6 wherein the means for eliminating a pulse from the received pulse train comprises a var- 16 iable impedance electron discharge tube, and wherein is provided means for varying the impedance oi said tube under the control of said control unit, means for applying the received pulses of the input terminals of said tube, and means for applying the pulses from the output of said tube to the build back circuit means.

' PRAFULLA KUMAR CHATTERJEA.

LESLIE WILFRED HOUGHTON.

REFERENCES CITED Number Name Date 1,928,093 Coyle Sept. 26, 1933 2,021,743 Nicolson Nov. 19, 1935 2,036,350 Montani Apr. 7, 1936 2,048,081 Riggs July 21, 1936 2,057,773 Finch Oct. 20, 1936 2,061,734 Kell Nov. 24, 1936 2,085,418 Crosby June 29, 1937 2,100,156 Bushbeck Nov. 23, 1937 2,110,548 Finch Mar. 8, 1938 2,113,214 Luck Apr. 5, 1938 2,146,876 Zworykin Feb. 14, 1939 2,171,150 Shelby Aug. 29, 1939 2,172,354 Blumlein Sept. 12, 1939 2,185,693 Mertz Jan. 2, 1940 2,199,634 Koch May 7, 1940 2,213,941 Peterson Sept. 3, 1940 2,256,336 Beatty Sept. 16, 1941 2,259,392 Roberts Oct. 14, 1941 $2,262,838 Deloraine et al Nov. 18, 1941 2,265,216 Wolf Dec. 9, 1941 2,266,401 Reeves Dec. 16, 1941 2,275,974 Mathes Mar. 10, 1942 2,277,192 Wilson Mar. 24, 1942 2,282,046 Goldsmith May 5, 1942 2,284,401 Manley et al May 26, 1942 2,308,639 Beatty et al Jan. 19, 1943 2,328,944 Beatty Sept. 7, 1943 2,403,210 Butement et al July 2, 1946 FOREIGN PATENTS Number Country Date 362,943 Great Britain Dec. 11, 1931