US2429613A - Pulse multiplex communication system - Google Patents

Description

9 Sheets-Sheet 1 Oct.v 28, 1947. E. M. DELORAINE ETAL PULSE MULTIPLEX COMMUNICATION SYSTEM Filed oct. 19, 194:5

hib whk N Ski gh N H gwn. n,

l S, Sn v Oct. 28, 1947. E. M. DELoRAlNE ETAL PULSE MULTIPLEX kCOMMUNICA'IION SYSTEM Filed Oct. 19, 1943 9 Sheets-Sheet 2 l WZ.

r/Mf caf/rm mais rok's foMo/vo M. ofzo/Mw: JusT//v L. ffA/P//va Oct. 28, 1947. E. M. DELORAINE ErAL 2,429,613

PULSE MULTIPLEX COMMUNIATION SYSTEM Filed Oct. 19, 1943 9 sheets-Sheet s e. 114 :I 117 i.116 I INVENTUM Iilel k/ il( l" foMo/vo M. pao/PANE BY .fz/smv ffl/ww A TTRNE Y Oct. 28, 1947. E. M. DELORAINE ETAL 2,429,613

PULSE MULTIPLEX COMMUNICATION SYSTEM l Filed oct. 19, 194s s sheets-sheet 4 wfg: 6).

M DEMoDuLAToR (D M) INPUT Oct. 28, .1947.I E. M.- DELORAINE TAL 1 .2,429,613

PULSE IULTIPLEI COMMUNICATION SYSTEI Filed oct. 1s, 1943 9 sneetsf-sneqt 5 @figg-Z l '1a 2a sa y 1L .25 a5 1c- -//MW Aff/Mam Gm.' wwf fm' 'cwfwva Mo. I

1c, fue

' F16. s afnaauL/vmn c'a/vrm WAVL' qlf Amm/LY 'Oct 28, 1947. vE. M. DELoRAlNE E rAL PULSE HULTIPLEX COIUUNIGA'IION SYSTE Filed oct. 19, 1943 9 Sheets-Sheet 7 fm alf-nw 041) E w www mw, T; m w IL ,HKM mm/M/ Uil! w a tlm .r L

Oct. 28, i947., E. M. DELORAINE ErAL 2,429,513

PULSE MULTIPLEX coummIcATroN SYSTEM Filed Oct. 19, 1943 9 Sheets-Sheet 9 lSeptember 29, 1943,

Patented Oct. 28, 1947 OFFICE.

PULSE MULTIPLEX COMIWUNICATON SYSTEM Edmond M. Delai-aine, New York, and Justin L. Fearing, Scarsdale, N. Y., assignors to Federal Telephone and Radio Corporation, New York, N. Y., a corporation of Delaware Application October 19, 1943,'Serial No. 506,802

35 Claims. (Cl. 179-15) This invention relates to multiplex communication systemsl and methods utilizing electrical pulses, and is particularly adapted for telephone, either wire or wireless, and other signalling and pulse controlling systems.

In the United States patent to E. M. Deloraine- A. H. Reeves No. 2,262,838 and the corresponding British Patent No. 509,820, a multiplex signalling system utilizing time modulation of electrical pulses as distinguished from amplitude modulated pulses is disclosed. In the Deloraine- Reeves system, the signals are transmitted independent of variation of pulse amplitude, the pulses beingvof small constant width compared with the time interval between successive pulses for the same channel. The time displacement of the pulses is also small and is maintained Within limits, the interval of which is small compared to the time intervals between successive pulses. The time intervals between the pulses of a given channel are filled with pulses of other channels for multiplexingpurposes. To avoid overlap between the pulses of different channels, the pulses of the several channels are diierently timed by a distributor at the sending terminal and the receiving channels at the receiving terminal are synchronized therewith by a second distributor which is locked in step with the rst distributor by separate means.

In a copending application of E. M. Delorainc andN. H. Young, Jr., Serial No. 504,204, illed entitled circuit for multiplex systems, a multiplex system is disclosed which overcomes the necessity of separate synchronizing means between the transmitting and receiving terminals. 'I'his synchronizing is accomplished in the Deloraine- Young system by using the average timing or cadence frequency of the multi-channel pulses to generate a receiver controlling Wave of a frequency corresponding to the frequency of the control wave used at the transmitter. For monitoring purposes, the Deloraine-Young system provides a monitoring interval between groups of channel pulses, and Where desired, may provide the monitoring interval with a special monitoring impulse modulated with a desired signal.

In a co-pending E. M. Deloraine application, Serial No. 504,881, led Oct. 4, 1943, and entitled Communication and signaling systems, a multiplex system is disclosed using time modulated pulses both for ringing and transmission of voice without requiring transformation of the ringing signal as'he'retofore practiced in telephone systems.v

synchronizing common transmission medium they provide periodic time intervals useful for monitoring purposes.

Anotherl object is to provide a method and` means for monitoring and synchronizing the receiving stations of terminals and/or branch lines of a multiplex communication system with the proper sending stations thereof without using transmission means other than the transmission of channel pulses.

Another object is to provide a method and means for time modulating the pulses of each channel for both voice and ringing signals without requiring transformation ofthe ringing signals.

Another object is to provide a multiplex pulse signalling system with a repeater` amplifier for reshaping the pulses so as to minimize the introduction of cross-talk or hum from adjacent lines and other interference sources.

Still another object is to providea multiplex pulse transmitting system with means for selectively connecting one or more trunk or branch line circuits with any one of the channels of the system.

In accordance with the multiplexing principles of the present invention, a plurality of communication channels are provided by producing trains of electrical pulses one such train for each channel. The stations of yeach terminal of the system are each provided with a, transmitting circuit and a receiving circuit, the transmitting circuit being used to generate the pulses for one channel and the receiving circuit being used to receive the pulses of another or return channel. The pulses of each channel preferably are paired off, with the pulses of each pair, When unmodulated, having a time interval therebetween smaller than the time interval between succeeding pairs of pulses. The different trains of channel pulses from a terminal are differently timed so that when they are fed to a transmission line they sandwich together with a given time spacing be-v tween succeeding pulses thereby permitting a given amount of time modulation without interference from pulses of other channels. The

paired-of! relation lof the pulses operates to form distinct groups of the pulses with a characteristic monitoring interval between succeeding groups.

The receiving circuit of each sub-station and trunk line is provided with a gate circuit whereby only those pulses of the desired channel or channels aresegregated from the other channel pulses present in the common transmission medium. In

order to synchronize the receiving sub-stations .and trunk lines with the proper channels, the

or in conjunction with the oscillograph indicator. t

Regardless of the character of the monitoring device selected, a single monitoring device may be used for each sub-station or trunk line or to control -a band of sub-stations, whichever may be desired.

The time modulated pulses representing a plurality of channels may be transmitted directly between sending and receiving stations as video pulses over a line or coaxial cable without being ilrst converted into carrier waves, or the pulses may be employed to modulate carrier waves which are then transmitted through a common medium such as a line circuit, coaxial cable, a dielectric wave guide or through space between antennas.

Many advantages are present in the use of time modulation as compared with amplitude modulation in multiplexing. For example, since'the signals in all the channels are transmitted with time modulation, the pulses are maintained at substantially constant amplitude, thereby permitting the use of line-ampliers and various other devices.

which are oi.' simple, low cost construction without the necessity of4 providing for linearity of amplitude amplication or the use of elaborate lters such as are required in multiplex systems of the frequency selective type.

Another advantage is that noise and interfering disturbances originating outside or within the system such as the introduction of cross-talk and hum from adjacent lines are readily eliminated or counteracted by simple clipping and reshaping operations which serve to cut oil.' and reduce the unwanted components accompanying the desired pulses. ASince the pulses are relatively sharp in duration, they may be economically increased in amplitude to such a value that they considerably exceed the amplitude of ordinary noise and other disturbances in the common medium and no loss of signal modulation occurs when the upper and lower portions of the pulses are clipped. Line repeaters are thus enabled to regenerate all signals in the line without repeating appreciably any objectionable noise or other disturbances and to send out onthe line a signal pulse of given amplitude and width regardless of fluctuations in the amplitude of incoming signal pulses reach- 7 ing the repeater. The necessity for automatic gain control and compensating adjustments of amplifier gain, together with the cost of elaborate equipment usually provided for these all@ 931? purposes in systems employing varying amplitudes, are thus avoided.

A further feature oi.' the present invention is that the type of signal modulation and arrangement of pulses employed is adapted for maintenance of privacy of communication without being readily intercepted or obliterated. either by ordinary listening-in devices ot an unauthorized character or by jamming operations such as may completely obliterate ordinary amplitude modulated signals.

'I'he above and other objects and features of the invention will be understood more clearly upon reference to the following detailed description to be read in connection with the accompanying drawings, in which:

Fig. 1 is a block diagram of a multiplex signalling system provided with west and east terminals in accordance with the principles o! our invention;

Fig. 2 is a schematic wiring diagram of one of the forms of push-pull modulators that may be used in the multiplex signalling system;

Fig. 3 is a graphical illustration of a set of curves illustrating the pulse generation and modulation timing performed by a group of three modulators of the character shown in Fig. 2;

Fig. 4 is a block diagram of second form of push-pull modulator thatmay be used in the multiplex signalling system;

Fig. 5 is a graphical illustration of a set of curves illustrating the operation of the modulator of Fig. 4;

Fig. 6 is a schematic wiring diagram of ademodulator and timing circuit therefor adapted for selective reception of the pulses of a given channel separately from the pulses of other channels;

Fig. 7 is a graphical illustration of a set of curves useful for explaining the operation of the demodulator and timing circuit of Fig. 6;

Fig. 8 is a schematic wiring diagram of a modified form of demodulator and timing circuit;

Figs. 9 and 10 are schematic wiring diagrams of two forms of phase shifters used in the multiplex system;

Fig. 11 is a schematic wiring diagram of the circuit details of the operators supervisory unit for the east terminal which operates to produce the necessary timing wave energy for the stations for that terminal;

Fig. 12 is a schematic wiring diagram of a line ampliiier for the transmissionk link of the system;

Figs. 13 and 14 are diagrammatical showings of the screen of a cathode ray oscilloscope illustrating how the channels of communication may be monitored; and

Fig. 15 is a schematic block diagram of a branch line terminal connected to the transmission link of the system shown in'Fig. 1.

Referring to Fig. l of the drawings, an embodiment of the multiplexing system is shown for purposes of illustrating the principles of the invention. The system shown is provided with two (west and east) terminals interconnected by a transmission link 25. Additional terminals such as drop channel and/or channel inserter terminals may be included as will be made clear hereinafter.

Each terminal includes a plurality of terminal stations I, 2 n, and an operators supervisory unit such as unit 26 at the west terminal, which includes a. base wave source 21 and a monitoring visual indicator 28. Each terminal station includes a sub-station Il. a modulator 33,

a demodulator 34, "a hybrid connection 32 coni necting the modulator'and demodulator to the sub1-station, and two control'circuits 35 and 36. The circuit 35 contains a phase shifter 31 by which energy from the'master wave source 21 is properly phased for timing of the pulses gener- The sub-station 3| is of known character, such as commonly used for selectively switching in vtelephone lines and the like for two-way conversation. l

'I'he hybrid connection 32 is of known form having a balancing impedance Z whereby signal energy from the sub-station 3 Ils properly applied to the modulator 33 and signal energy from demodulator 34 is applied to the sub-station 3|.

Modulator circuits The modulator 33 may be any one of severa-l forms whereby a train of electrical pulses, either generated by the modulator or by a separatev source, is time modulated accordingto a signal wave. This time modulation of the pulses is pref erably biased so that when the pulsesfof all the channels are mixed together in a. common me dium, they form distinct groups separated by a monitoring interval of predetermined duration.

Monitoring intervals may be obtained by using balanced modulators with one channel omitted such as disclosed in the laforesaid Deloraine- Young application Serial No; 504,204. Such arrangement, however, produces two such intervals time control wave thereby saving timing space for the channels of communication. This is accomplished by properly biasing and controllingthe modulating operation -of a push-pull type of modulator.V 'Ihe push-pull modulator, for example,may be of the cusper .type disclosed in the A copending joint application of E..Labin-D. D. Grieg, Serial No. 455,897, filed August 24, 1942, or of. the gate clipping type ydisclosed ina second joint copending application ofE. Labin-D. D.

Grieg, Serial` No. 455,899, also led yAugust 24,

1942, or anyother suitableform.

1 i A circuit ofthe'cusper ,type'of modulation is` lshown in Fig. r2. The modulating. signal wave input from sub-station 3| 7isapplied `to 'an amplitude limiter AL whichv may. be of any well known tial `43having in shunt therewith a potentiometer i'. 44 whose sliding contact yisconnected-to ground. e

A condenser 45 offnegligiblytrlow impedance at speech frequencies isconnected between the slider of Ithe potentiometergandeach -terminal thereof, in order to by-pass components of the speech 6 oi' the cathodes 5I and 52 of a full-wave rectier 50. Connected across the connections to the cathodes is a capacitor by which the circuits including the transformer secondaries are tuned to the frequency of the time control wave. The anodes 53 and 54 of the rectier 50 are connected together and through an output resistance 55 to ground.

for each period of the time control wave.V In the present invention we have discovered that one such intervalmay be obtained per period' of the The operation of the cusper of Fig. 2 will b'e clear from reference to Fig. 3. The time control wave lw (curve a) fed over line 35 from phase shifter 31 (Fig. 1) is applied to the primary of the transformer 46 (Fig. 2). 'I'he setting of the potentiometer 44 controls the bias on the cusper at a potential level such as indicated at 59 in curve a. The full-wave rectification of the wave thus occurs with reference to the level 59 as the axis of rectication thereby producing an output wave 60 having cusps 6|, 62, 63, etc., (curve b) in the output voltage across resistor 55. These cusps are paired on' in time with the interval St between the cusps of each pair (see pulse pair 6|, 62) smaller than the interval 4t between succeeding pairs of cusps. The unit t is the time interval allowed for each pulse. This relationship of the two intervals represented by 3t and 4t is determined by the selected bias on the cusper. the pulse width and the number of channels to be contained in one period of the time control wave Iw. The relationship shown in Fig. 3 is selected for three channels and the pulse widths produced from the cusps are so selected as to provide for a given degree of time modulation and yet leave a safety interval S between the succeeding channel pulses as indicated at curve e. This timing feature will be described in more detail hereinafter.

The output orf the cusper is connected through a coupling condenser 61 of low impedance to the cusper waves (curve b), a grid leak 66 and a variable resistor 69 to a control grid 65 of vacuum tube 66 of a pulse shaper circuit. In the cathode to groundv circuit of the tube 66 is a variable output resistorv10. 'I'he screen and suppressor electrodes of the tube are connected in the usual manner, the screen being positively polarized by the anode source and the suppressor grid being connected directly to the cathode. 'I'he output conductors 1I for the time modulated pulses are connected across the resistor 16.

'I'he control grid resistor 69 of the shaper may be varied to adjust the positiveI saturation potential of the tube thereby producing a substantially flat-topped wave in the output, the locking potential of the control gridV being adjustable by varying the cathode resistor 10. By adjusting these two resistors in relation to each other it will be seen that the duration of the resulting :output pulses may be adjusted between wide limits. The full line rectangular pulse la in curve e of Fig. 3 shows the output pulse corresponding wavesand of the channel timing'wave around the a biasingsource. In. series with each ofthe -two secondary, windings 4| andr 42 is connected one. Y -of-ftwo-se'condary windings' 46-and 41r kof anl input transformen 48 forfthe time control wave lw (Fig. 3),.1 Each, of the secondary windings' 46 and 41 .,of the transformer 461s connectedwith'one to the cusp 6|. .The duration ofthe pulse, however, may be much smaller than shown depending .upon the adjustment of resistors 69 and 10. The

pulse Iaa ofcurve f is illustrative of a smaller widthpulse output. It will be understood that the pulse width may be reduced to as small as 1 or 2 microseconds for a timing wave of six kilocycles where 100 channels more or less are desired.

The maximumlimits of time displacement of "the pulses are controlled by adjustment of the amplitude limiter AL (Fig. 2) and are represented by the levels 12 and 13 in curve a. Thus the signal variation of a signal wave causing interval t.

7 the axis of rectification 59 to vary between limits 12 and 13 causes the output wave 80 to vary in phase between limits 12a and 13a. This signal variation causes the output pulse la produced from the cusp 6I to vary in displacement between the limits 12b and 13b. These limits are'so selected as to allow for the safety interval S between the limits of modulation of the pulses of adjacent channels.

The proportions of the curves of Fig. 3 are exaggerated for clearness of illustration, it being understood, of course, that the undulations of the time control waves have been flattened considerably and that in actual practice they are narrow so that the limits of modulation 12 and 13 extend over a portion ofthe wave that is substantiallylinear. It will also be understood that in practice the cusps lil, 62 etc., are sharp and elongated` affording a translation thereof into substantially rectangular pulses.

Assume that the west terminal has three stations and that time control waves iw,` Zw and Bw are supplied to the modulators thereof in the timing relation shown in curve a of Fig. 3. Three trains of pulses will be produced one for each of the time control'waves as indicated by the cusper waves of curves b, c and d: It will be observed that the timing relation of the control waves must be selected according to the pulse width produced and the limits of modulation so as to completely distribute the pulses of the three channels throughout the time intervals corresponding to the smaller undulations of the wave iw with regard to the rectification axis 59. It rfollows that because of the small and large undulations of waves on opposite sides of axis 59, the pulses of the three channels form groups of six pulses leaving an interval 15 therebetween. For synchronizing purposes it is important that the interval 15 be maintained equal tothe interval t required for each pulse or a small multiple thereof so that the pulse train may be used at a receiving terminal to control an oscillator or other wave generating circuit tuned to the average repetition rate of the pulses of each group without permitting the wave generating circuit rto pull out of step. 'I'he monitoring intervals .or other intervals caused by the de-energization of terminal stations etc., will not permit loss of synchronism so long as a large percentage of channel pulses are continued or the remaining channel pulses retain the repetition frequency component to which the oscillator at the receiving terminal is tuned.

It might be thought that the monitoring interval between groups of pulses could be made any particular value desired. It has been found, however, that in a multiplex system having economically spaced channels with equal spacings therebetween within a group of pulses, and where the channel oscillator at one end of the line is synchronized with that at the other end of the line, it is necessary to provide a spacing between the groups having a predetermined relation with the spacing between channels within a group. For this purpose it has been found that. the minimum space between the last pulse of a group and the rst pulse of the next group as indicated in curve e of Fig. 3 should be equal to the interval t allowed for each pulse within the group. The timing interval between groups may, however, be greater provided that it is precisely an integral number of the channel One reason for this requirement. is that each pulse must be so timed as regards its 8 average position that it tends to maintain a constant phase relation between one channel oscillator, acting as a secondary oscillator, and

the master oscillator controlling it from the other end of the line. Y

` While an example of the required timing relationship for the pulses of a three channel multiplex system is shown in Fig. it will be understood that these proportions may be varied considerably for dierent width pulses and for different numbers of channels without departing from the principles of the present system. The pulse width, for example, is shown in curve e is W, the pulse being displaceablev to the left and right maximum displacements D in response to modulating potentials. 'Ihis displacement D in each direction is chosen equal to one-half the width W for illustration purposes only, it being understood that many other dimensional relationships may be selected. The safety zone S is arbitrarily selected equal to the width D to provide a margin of safety against encroachment of one pulse upon another, and to provide adequate spacing for segregation of the pulses of one channel from those of other channels at the receiving terminal. The relation of these values for a system containing from 3 to as high as 100 channels, for example, may be expressed as:

For specific examples of timing see the section following entitled Channel timing.

It will be observed in Fig. 3 that for three channels the period of wave Iw is divided into with a sui-table frequency divider to reduce the frequency to that desired substantially as .provided at the supervisory unit 290 at the east terminal which is described hereinafter.

Another form of push-pull modulator that may be usedin place of the. cusper of Fig. 2 is shown in Fig, 4. This 'form of modulator operates on the gate clipping principle and includes a preliminary clipping amplifier to which the time control wave such as wave iw of Fig. 3 is rst applied. The preliminary clipping amplifier 80 is used to greatly amplify the time control wave and at the same time limit clip the upper and lower portions of the wave so that only the central portion remains as indicated by the wave 8| ofv curve a, Fig. 5. The preliminary clipping operation performs two desirable functions:

One of amplication whereby steep linear leading and trailing edges are provided for the undulations of wave 8i, and second, of limiting the peak power of the input wave.;

The main or gate clipping circuit 84 may be any one of several forms disclosed in the aforementioned copending application Serial No. 455,899. The form shown in Fig. 4 is of the type employing dry rectiers although pentode tubes may be used for the same purpose as explained in the above-mentioned application. The two dry rectiers 85 and 88 are connected back-toback and a small positive voltage is applied from battery through resistor 89 to the intermediate point 90. The resistor 89 is preferably at least severa1 times larger than the low forward resistance of the rectiilers although it should .be low in comparison with the high back resistance of the rectiers. The .input wave from amplifier 80 is applied through coupling condenser 92 to resistor 94 which is preferably several times lower than resistor 89. The output of the circuit is delivered across another resistor 95 which may be of the same order of magnitude as resistor 84.

For modulating the clipping action, speech signal input is supplied through the plug 96 to jack 91. 'I'he output of the main clipping circuit 84 is fed through rst and second differentiating circuits 98 and 99 and then a clipper or rectii'ler circuit |00.

The operation of the circuit shown in Fig. 4 can best be understood by consideration ofthe curves of Fig. 5. The wave 8| of curve a as here-r inbefore described represents the output of amplier 80, the initial time control wave having been converted into an approximately trapezoidal form. In order to bias the clipping operation according to the principles of this invention, the ground connection to the jack 91 is provided with a battery which supplies through resistor 94 a negative bias to the left hand side of rectifier 85. This results in a bias rectication' axis I 02 for the input wave. Thus, the input wave is made unsymmetrical with respect to this axis |02.

The upper clipping limit |03 is controlled by the potential of battery 88 at point 90. ,TheV rectier 85 operates only when the input wave exceeds the axis |02. Thus, the potential at point 90 will follow the input potential onlyrwhen it exceeds the negative bias or zero axis |02 and until the input potential equals the potential of batteryy 88. Any potential rise beyond this limit |03 will not be transmitted to point 90 so Athat only those portions of wave 8| lying below the limit |03 and above the axis |02 will pass to the point 90. In transmission from point 90 to the output of the main clipping circuit the rectifier action of elements 86 similarly eliminates all potential variations below the axis |02 since the point 90 cannot receive current from the right hand side of the circuit but can only deliver current thereto.v As a result all portions of the curve below the axis or lower clipping limit |02 will be blocked. The output of the main clipping circuit 84 is represented by the trapezoidal wave |05, curve b. The output wave is differentiated at 98 thereby producing a train of alternately positive and negative pulses according to curve c. The pulse i0, for example, corresponds to the leading edge of the trapezoid |06 while the pulse ||2 corresponds to the trailing edge of the trapezoid |06. The second differentiation performed at 99 produces the train of pulses of curve d, pulses 4 and 5 corresponding to the pulse ||0 and the pulses ||6 and 1 corresponding to the pulse ||2. Curve e represents the output of the clipper |00 which-operates to remove the negative pulses of curve d, It will be observed that the pulses of curve e are paired off the interval between'the pulses ||4 and being smaller than the interval between pulses and ||8 which constitute the second pulse of one pair and the rst pulse of thenext succeeding pair, respectively.

When a signal wave such as a voice is applied to the jack 91, it varies the`negative bias imposed by the battery |0| thereby varying. in

eiect the position of the gate limits |02, |03 with respect to the input wave 8|. The outer limits of this modulation is represented by limits |02a and |03a, curve a, and also by the broken lines of curves b. c, d vand c. The nal output pulses on curve e are thus capable of being modulated according to the intelligence applied at the jack v 91 between limits |20 and |2|. It will thus be apparent that when the modulator of Fig. 4 is used in the multiplexing system in Fig. l, that the pulse outputs of a plurality of such modulators will sandwich together in accordance with the illustration of curve e in Fig. 3.

Channel timing While a simple example of a three-channel The monitoring interval will, in such example,l

be equal to two pulse intervals (2t) and the duration of one cycle of the time control wave will equa1 2X99+2 or 200t. In other words, each cycle of the time control Wave would include 200 pulse intervals with two of such intervals as` the monitoring interval (see interval l5, Fig. 3).

Since there are two vpulses for each channel per cycle of the time control wave, there are for a 6 k. c. wave 2 6000 or 12,000 pulses per second for each channel. Also, since there are l166% microseconds per cycle in a 6 k. c. wave,

each pulse inverval t will equal of a microsecond. Referring for example, to the pulse relation illustrated in Fig. 3, W, the duration of each pulse will be 1/3 microsecond; D, the maximum displacement-in eachdirection will be ya microsecond and S,'the safety interval will be 1/6 microsecond.

It will be understood that many other channel systems. maybe worked out according to the principles of this invention, the pulse relation depending upon the number of channels, the wave frequency, the time spacing allowed for the monitoring intervals and the pulse duration, degree of modulation, and safety interval between adjacent channel pulses. For further comment on the timing of channels see the section labelled synchronizingl Demodulator circuits The demodulator 34 shown in Fig. 6 translates the intelligence conveyed by time modulated pulses into amplitude modulated waves for detection in the usual manner. The demodulator is provided with input terminals |20 for an input transformer I9, and output terminals |2| for the amplitude modulated signal waves. For use in multiplexing, the demodulator must be controlled to respond only to the pulses of a given channel. vThis may be accomplished by means of a gate wave and a demodulating wave applied to the demodulator circuit over input connections such as |22 and |23. The gate wave and demodulating wave are derived from a timer 39 having input terminals |25 for a time control Wave such as the wave iw of Fig. 3.

The timer 39 includes two units, a cusper and a pulse shaper. The cusper unit is quite similar to that of the cusper modulator in Fig. 2, but diiers therefrom in several respects, such as by the omission of the amplitude limiter AL and the 11 modulating signal wave input, and by the addition of circuit |28. 'I'he circuit |28 includes a variable resistor |21, a variable capacitor |28 and an inductor |28 in series, the circuit being connected at one end with the primary of transi'ormer 40a and connected at the other end by line |23 to the demodulator 24. A slider |30 is provided on the cusper output resistor 55a to vary the output voltage. The `tuning capacitance connected between the rectifier cathodes |a and 52a includes two equal capacitors |32 in series, the

`iunction of the capacitors being grounded. A

single capacitor may be employed as in Fig. 2 by omitting-the ground connection.

'I'he shaper of Fig. 6 has the same circuit as that in Fig. 2, but may be, if desired, oi' the form shown in Fig. 8 described hereinafter. The Voutput resistor 10a of the pulse shaper is connected by line |22 to the demodulator 34.

The demodulator 84 includes a vacuum tube |40 having a control grid |4| to which signal pulses are fed through the transformer ||9. A grid leak |42 is shunted across the secondary of the transformer through a source of negative potential |35 to ground |88. As will be made clear hereinafter, the transformer ||8 operates as a diilerentiator for translation of the signal pulses into sharp impulses of a character more easily segregated from the signal pulses of adjacent channels. The tube |40 has a cathode |45 connected to ground, three additional grid elements |48, |41 and |48 and an anode |48. To the screen grid |48 is coupled the'line |23 for impressing thereon the demodulating wave obtained from the primary winding of transformer 40a through circuit |28. Suppressor grid |41 is connected to the cathode |45. The grid element |48 is coupled to line |22 over whichrthe gate wave `is received from the timer 38. The grid element |48, however, is connected through a high resistance leak resistor |5| to a source of negative voltage |50 which maintains the tube |40 normally biased to cut-oil so that the energy of the signal pulses on vgrid |4I and the demodulating wave energy on the screen grid |48 will be insuilicient to cause the tube to pass current until they coincida, together with pulse energy of the gate wave on 'grid element |48.

A by-pass condenser |58 shunting the output circuit of anode |48 provides a path of relatively low impedance for components of a frequency above those of the signal waves. A resistor |58 in shunt with the primary of the output transformer |82 damps the output circuit to prevent or reduce objectionable transient oscillations which might otherwise result from shock excitation of the circuit by components of the pulses. An inductive reactance |80 may be connected in series with the secondary of the output transformer |82 for still further reducing the output l of undesired high frequency components above the signal wave frequency if desired. It will be understood that any suitable low-pass circuit means may be provided in either the primary or secondary transformer for accomplishing a similar result.

While the time control wave for the timer v38 may be used as the demodulator wave for the demodulator 34, it is preferred to pass the time control wave of circuit |28 through a known tuned amplifier |52 whereby a suitable harmonic of the time control wave is obtainable. It is preferable to employ as high a harmonic as possible in order to obtain the advantage of a steeper slope upon which the signal pulses and the gate pulses are to be superimposed for translation of the time modulation of the signal pulses into effective amplitude modulated pulses as will be seen by reference to Fig. 7.

Curve a of Fig. 7 shows a series of line or channel signal pulses for a system having three channels. The curve a shows six pulses in a group separated by a monitoring interval 15 from the first pulse of the next group the same as in curve e of Fig. 3. Curve b shows the signal pulses after differentiation by the input transformer I8, each signal pulse such as la being thereafter represented by two impulses, one positive and the other negative, such as la' and la. Curve c shows a series of gate wave pulses Iz, |11 and la in such alignment with the channel pulses of curve a as to pass those pulses representing channel l. As for other channels, gate waves similar to that shown in curve c are each aligned or synchronized with the pulses of the particular channel to be received by the corresponding demodulator.

Curve d shows, for purposes of illustration, a demodulating wave |10 such as would be obtained from the primary of transformer 40a, and which corresponds directly tothe frequency of the time control wave used for the three-channel system. This wave is of such phase relation to the gate wave and the pulses in channel No. as to align the sloping portions thereof with 'the gate pulses and the signal pulses of channel For the most effective translation of the timed modulation of the signal pulses into amplitude modulated pulses, a much higher frequency wave is preferred. This is desirable because the side slopes of each undulation of the higher frequency wave are much steeper as is clear from a comparison of the waves |10a and |10. When the gate pulses and the signal pulses are superimposed thereon, it will be clear that the steeper the slope, the greater is the amplitude variation obtainable for a given amount of pulse displacement. For a further understanding of the principles of demodulation of time modulated pulses, reference is made to the aforesaid Grieg application, Serial No. 459,959.

The different wave potentials applied to the grid elements of the demodulator tube include the impulses of curve b, the gate wave of curve c and the demodulating wave as may be |10, |10a or such other frequency wave desired. Curve e shows these dierent wave potentials combined to illustrate the gate and the threshold clipping functions of the tube |40 of Fig. 6. The normal negative D. C. bias on the grid element |48 may be adjusted so that anode current begins to iiow when the positive signal impulses reach a predetermined amplitude such as the threshold level |12. With this adjustment, anode current is produced in the tube |40 only for brief intervals corresponding to the peaks of signal impulses la', Ib' etc., the amplitude of the energy passed by the tube being substantially directly in proportion to the amount of displacement of the lsignal impulses from their unmodulated posiis illustrative of this principle, the steepnessofi the side slopes thereof being indicated at "b"2 in graph e for comparison with the top slopeoii gate pulse Iz.

To summarize the operation of the demodulator of Fig. 6, the anode current of tube |40 is 'nor-j mally blocked by the negative bias on grid ele,-v ment |418. When a gate pulse such as pulse la:`

of curve c, Fig. 7, is applied, the negative bias on the control grid is reduced to such a value that V the peak of the signal impulses will cause appreci` able anode current to flow through the tube.v The grid voltage, of course. is varied at thesame ticular line may have y"been capable of satlfa vy wily-.1 trensmittinge .In,f0t11 =1r-;vs'crds.y if! .a g given line V-b'e adapted'sfo transmitting a puls i?,l ,cor-.f5

" teirrivinimuni, duratontthe receiver. may b emade mzaesllld fb dliereiltiationfftma :considerably shorterspuls hanthe vlinenwas capable of .effec-f` tively t transmitting; Anl advantageof .transmite tingfthe longer duration-pulsethan that employed;

' at. thereceivergafter.r diilerentiation,l is that-morev time by the demodulating wave or |l0a as the case may be. Thus, the gate pulses operate Y to open the demodulator for reception of signal impulses. The demodulating wave produces the desired slope ,effect by which the time displacements of the signal impulses are translated linearly into amplitude modulated pulses. Since the signal impulses are all of uniform amplitude, the

effect of a time modulation of the impulses is to vary the eiective amplitude oi.' the anode current of tube |40 in proportion to the time modulation or displacement of the signal impulses. Thus, for illustration purposes, let it be assumed that the time modulations applied to impulses la', Ib' and lc; graph e, Fig. 7, are progressively greater as indicated by the broken linepositions Ila, ||'b `and ||c. lThe resulting anode current for such modulation of the signal impulses would be substantially as indicated by the impulses ||a lib and llc of curve f. These impulses of 'curve f represent in amplitude variations the time dis" placement between impulse positions la' to Ha; Ib to ||b; and lc' to llc, and the line |13 represents the envelope dei'lned by the anode current impulses.

It will be understood that these curves greatly exaggerate the relative proportions since there are for a 6 kc. system, 12,000 signal impulses per second. To represent more clearly the'envelopes defined by these impulses, curve g is provided at a reduced scale with three of the impulses thereof selected as representing impulses Ila', Hb andv llc' of curve f. Audio detection of the signal impulses of curve g is represented by thev resulting wave |13a. A

The advantage of differentiating the input signais before they are applied to the control grid is that differentiation' makes it possible in some cases to reduce the width of the gate opening to approximately 50 per cent of that necessary for allowing for the passage of undifferentiated pulses.

This result will be understood from the fact that" -a very sharp diierentiated pulse may be produced signalenergyfmay betransmittedat-a xed am l plitude thanl if pulses "of thevery short lduration z were actuallytransmittedontheline.l Thustherey would generallybe g less :likelihood of the; maxi--` mum signal `amplitude `falling '.below thedesired value Aby the time it reachesthe receiver.

It will be understood that thefsensitivity ofV response oi.' the demodulator increaseswlthfinycrease' of slopeof the demodulating wave and that the slope may be increased .byeither of two methods.` One by increasing the amplitude of the@` demodulating wave as indicatedat llllb,A the D. C.1 negative bias and the'gate'wave: applied `to vthe grid; of course, being correspondingly increased.r The other method as hereinbefore described is to employ a. higher odd harmonic of thechannel timing wave asindicatedbyiwave |l0a, provided the minimum gate opening `required rfor passing ksignal pulsesv does not extend over such a large fraction ofthe demodulating wave cycle that the portion vof demodulating wave within the limits of the gate opening 'departs appreciably from a straightline or departs sufllciently therefrom to introduce objectionable' distortionin the demodulating operation. The demodulator and timer of Fig. 8 maybe substituted for that of Fig. 6. The timer includes a cusperunit and a pulse Shaper unit as in Fig. 6,'

' the cusper vbeing of the same designy but the to represent the response producing edge, such as .i i

the leading edge of the signal pulse, and that it is necessary for the gate to be open only wide enough to equal the duration of the sharp pulse plus an,

interval corresponding approximately with t hey maximum time displacements caused by moduf lating signals.

.By dierentiating the signal pulses and reduc--v ing the time Width of the gate pulses, an advantage results that the signal-to-noise ratio may be considerably reduced as compared with that obtained when the undifferentiated pulse is applied -v lv70 ofl view, an advantage results from the fact that;

directly to the control grid. From another point the system operates at the received end ofthe line as though the line had satisfactorily transmitted pulses of shorter duration and correspondingly higher signal-to-noise ratio than the par-vy shaper differing by having atwo-stage ampliilen The input'of the shaper includes low impedance blocking condenser as shown and a grid leak resistance |82. The cathode |84 of the lrst tube |83 is connected through a .variablelresistor |86 with the negative terminal of the anode current source which is also grounded at i |86. The cathode resistor y|85 is shuntedI by a low impedance by-Dass condenser |88 to prevent feed-back of current variations vfrom the output to the input circuits. The anode current source ls connected l through a variable resistor |9|with the anode of tube |83, theresistor serving as a coupling. re-

sistor betweenA the tubes. Resistors |82a and |8'5a` Q and vcondensers |80a and |88afin the grid' circuit of tube |94` correspond respectively with resistors |82 'and'l85 and condensersy |80 and |88 in thelv grid circuit of tube |83. The cathode of the sec-vf ond tube |94 is connected to the negativeterminal Iof current source |95 and also grounded at/j `|81.-'-The lpositiveterminal of thefsource |95"isj. A connectedthrough'an output resistor |96 withthef anode of the secondftube |941.` -The outputicon-Q nection'* |91 across thejresistor' |96 is 'coupled t o--Y a grid element 20|of the demodulator"vacuumfy tube 200.v The grid-element 20| is'p'ivovidedwithy I a grid-resistor to'which is applied a-'negati've bias' |98 loi'fsutllcient' value tofnormally maintain' the 6L? kpentagrid converter type: lThe-inp'utmr si nalfpulses' Ais shown 'connected betweenv 'grid 208'y low impedance blocking condenser 209 with the aiaaeia upper line conductor and having a grid leak 2|! and negative potential source 2| I. The cathode and suppressor grid of tube 209 are connected to ground |99. Grid elements 202 and-204 are connected together and through tuned'circuit 2 Il to the positive terminal of the anode current source 2|0.

It will be noted that the Fig. 8 demodulator is shown as though the transformer ||9 of Fig. 6 were omitted thus applying the signal pulses without dierentiation directly to the control grid of the tube. Such arrangement would require that the gate pulse be made correspondingly wider than in the case of Fig. 6, in order to accommodate the passage of the line pulses I a, Ib, |c, etc., Fig. 7. It will be understood, however, that the transformer |I9 of Fig. 6 may be placed inthe i input 230 of Fig. 8, in which case, differentiated pulses will be applied to the control grid and the gate wave may then be made of the same width as in the case of Fig. 6.

In adjusting the pulse shaper of Fig.`8 for conl trol of the gate pulses, the maximum duration of the gate pulses may be obtained by adjusting the cathode resistor |85 which controls the cut-off level at which the base of the input wave is clipped. By raising the slider |30a on the output of the cusper, the width of the gate may be increased and by lowering the slider gate may be narrowed. vIn order that the gate wave may have `as nearly a vrectangular shape as possible, the slider of the cusper output resistor is preferably adjusted near its uppermost position with the cathode resistor |85 properly adjusted to provide the desired gate width. The resistor |9| is then adjusted to vary the clipping level at the top of the output gate wave to provide a gate wave of proper amplitude withrelation to the blocking bias normally acting on grid of the demodulator, so that the desired signal impulses superposed upon the top of the gate wave pulses,- produce anode current as hereinbefore described.

The demodulator tuned circuit 2|! is made resonant to the frequency of the time control wave or preferably to an odd harmonic thereof, thereby providing a demodulating wave frequency such that a substantially linear slope of proper steepness is provided at the top of each gate pulse.

The incoming signal pulses are applied to grid 203, those pulses of the proper channel being in proper timing with the gate pulses. Resonant oscillations are set up in circuit 2| 5 by shock excitation as when the tube 200 conducts anode current in response to signal pulses, the initiation of which may be brought about by decreasing the negative bias at |98. These oscillations are used as the demodulatlngwave since the amplitude thereof is directly related with the degree of time modulation of the signal pulses. This, as explained in the aforesaid Grieg application, Serial No. 459,959, results in a substantially'true translation of the time displacement of each signal pulse into corresponding amplitude for an output pulse.

The demodulators and timers of Figs. 6 `and 8 may be employed for the reception of time modu-`v lated pulses in systems other than those utilizing a double pulse subjected to push-pull displacement` For example, the demodulators and timlers may be used in the case of a time modulated pulse system wherein all pulses in any one channel have equal spacings between the average positions of successive pulses and the gate openings are equally spaced apart.

Signalling Ink addition to the speech frequency input, the modulator and demodulators of the multiplex system may be provided with input and outputs respectively, for direct current ringing, code, or other signals of a period longer than' that of the lowest utilized speech frequency component. In Fig. 2, a D. C. signal input circuit 240 is connected across the upper potentiometer condenser. the circuit being traced from the upper terminal of the potentiometer to the manually operable le` ver of the key 242 and to the armature of the relay 243 and from the back contacts of the key and relay through resistor 24|- to ground. The time constant of resistor 24| in combination with the upper potentiometer condenser 45 and its connected circuits is preferably slightly greater than the period of the lowest utilized `speech frequency component. The relay coil may be energized by any suitable A. C. current circuit 245 such as a telephone ringing control circuit for use with the telephone channel, or may be energized by any desired circuit for sending low frequency D. C. signals, such as a telegraph channel. The manual key 242 may be closed to send a signal or code message to an operator at the distant end of the channel.

Referring to Figs. 6 and 8, the anodes of the tubes of the demodulator circuits each connects conductively through an inductor or choke coil 250 in series with the coil of a marginal slow acting relay 25|. The main output for speech frequency waves in each circuit includes a transformer |62 whose primary is connected at one end by a blocking condenser 254 with the anode, the other end being connected directly with the positive terminal of the anode current source, so that the main output transformer is capacity coupled across the relay and choke coil which acts as a relatively high impedance to the speech frequency waves. y

The relay when operated closes its contacts to light the supervisory lamp 255 or to operate some other signal for the terminal supervisory operator, and/ or to transmit a ringing current or ringing control current for use in conjunction with the telephone circuit ofthe channel. The relay is preferably designed to have a minimum response period' slightly longer than that of the lowest utilized speech frequency component, and is adjusted to respond to closure of the modulator signal key or'relay contacts when the demodulator is properly aligned with the incoming channel pulses, but is adjusted to be non-responsive to the anode current produced by unmodulated pulses.` f

v Phiase shifters Figs. 9 and 10 show phase shifters previously referred to in connection with the adjustment of the time control circuits for the different channels of communication. The phase shifter of Fig. 9 is of the character used in the multiplex i system of Fig. 1 identified by reference character 31 and labelled PH. The phase shifter includes a transformer 260 whose primary serves output conductor 280.

is of simple form and is employed in parts of the circuit ofFig. 1 where only small range adjustments are required.

If desired, the phase shifter unit 310 shown in Fig.l may be used as the individual phase shifters PH in Fig. 1. This unit includes a transformer as in the case of Fig. 9 but differing from that of the Fig. 9 by having two branches 2'65 and 266 connected across the secondary thereof. The branch 265 includes a capacitor 261 in series with a variable resistor 268. includes a variable resistor 210 in series with a capacitor 21|. 'I'he output of the phase shifter has two terminals, one being connected between the capacitor and resistor of branch 265 and the other being connected between the corresponding elements of branch 266.

By varying both resistors simultaneously or both capacitors simultaneously, the phase angle between the output and input waves of the phase shifter may be varied with only a small variation in amplitude with change -of angle as compared with that obtained with the Fig. 9 form of phase shifter. Fig. 10 is, therefore, better adapted for convenient adjustment for thepdiilerent phase angles required for the different channel time control waves fed to the different channel modulators and demodulators in Fig. 1.

It will be noted that Fig. 10 includes a plurality of phase shifter units 310 the inputs of which are connected in parallel, while one terminal of the output circuit of each is connected with the upper output conductor 215 and the other terminal of each phase shifter unit is connected to a different contact on a selector switch216 whose contact arm 211 connects directly with the other This plural arrangement of phase Shifters represents the step type of phase shifter used in the system of Fig. 1 and identified by the label PH. Each unit 310 is differently adjusted, preferably by an amount equal to the diiference in timing between two'adjacent channels, so that by step actuation of arm 211 the time control waves from source 2'1 (Fig. 1) may be shifted in phase from one channel timing to the next. Y Y

The variable reactance 282 shown as an inductive reactance in series with the upper input conductor 213, permits small adjustments 'of the entire series of phase Shifters 310 as a group without affecting the phase angles therebetween.

One advantage of employing the transformer type of coupling within the phase shifter PH is that a 180 shift of phase may be obtained readily by merely reversing the primary or secondary terminal connections, so that when adjustment of the branch circuits vacross the secondary provides a maximum phase adjustment of 180, a reversal of the terminal connections on one of the transformer windings ,extends the range of phase angle adjustmentsv to 360. It will be understood that any other suitable form of phase shifter may be employed, provided the required steps of phase angle adjustment may be obtained by adjustment of the selector switch and provided that means are included for controlling the phase shifter PH as a whole without aiecting the angle between the small units 310. The phase shifters PH are located in the supervisory units of the system of Fig. 1 and their utility The branch 266 therein will be further described in the following description of the supervisory units.

Referring to the operators supervisory unit 26 at the west terminal of Fig. 1, themaster wave 18 source 21 thereof may comprise any known stable oscillator adjustable or fixed for producing a 'master time control wave of the frequency desired. The output of the master oscillator 21 is applied to the input of a phase shifter 213of the step type disclosed in Fig. 10. The output of the phase shifter 218 is applied to an amplifier 230 and thence over line 35 to the individual phase shif ters PH (31 for west terminal station No. 1) of the modulator circuits M. Referring to Fig. 3, it will be understood that the master oscillator provides a time control wave which when applied to the line 35 is transmitted to the modulators of the different stations with the energy thereof shifted according to the adjustment of the individual phase shifters PH thereby providing different time control waves` |10, 21a, 31o, etc., from which trains of pulses are produced at the different stations. 'Ihe ampliier 280 serves as a buffer amplifier between the individual phase shifters PH and the step phase shifter 213. The line 36 for providing time control waves for the demodulator circuits is connected to the output of amplifier 280 througha second step phase shifter 219 and a. buffer amplifier 23|. It will thus be understood that the wave energy supplied to the demodulator circuits may be of a timing different from the wave energy supplied over line 35 to the modulator circuits. 'Ihe reason for this will be made clear later on in the description.

synchronizing Referring to the operators supervisory unit 290 for the east terminal of Fig. 1, it will be ol'.- served that the source of time control waves for the stations thereof includes a frequency or pulse divider 29|, an oscillator 232, a step phase shifter 293 and a buffer amplier 234. The output of amplifier 294 is used for isolating the modulator and'demodulator circuits of the east terminal stations. The oscillator 232 is synchronized with the time control of oscillator 21 at the west terminal rby means of the average repetition rate of the channel pulses passed by the divider 29|.

The output of the modulators of the stations at the west terminal, for example, is applied through plug-jack connections 236 to a transmission line 291 which feeds into a repeater amplier 30| in one of the lines of a two-line transmission link 25.I As shown generally in Fig. 1, the transmission lines of the link each includes two repeater amplifiers, such as the repeaters 39| and 302 details of which are shown in Fig. 12 whereby the channel pulses after being transmitted a given distance are reshaped to a desired amplitude and width so that when' applied to the input line 304 at the east terminal they will have a given shape in spite of the attenuating characteristics of the line. The line 304 feeds the channel pulses to the demodulator circuits of each station and also to the input of the divider 29|.

In Fig. 11 is shown the wave generating and phase control units of the supervisory imit 230. The frequency or pulse divider 23| is preferably placed` at the input connection 305 to the oscillator 292, although it may follow the oscillator if desired. 'I'he divider 29| may be of any known form such as the mixing type or the hybrid coil type and preferably arranged to allow for division by even, odd or a combination of even and odd multiples whichever may be desired.

The oscillator 292 may also be of any known form but as chosen for purposes of illustration is `of D. D. Grieg, Serial No. 458,855, filed September 18, 1942. The oscillator 232 includes an ampliner tube 3|0 having an input connection including three capacitors C1, Ca and C: in series and three shunt resistors R1, Rz and R: and a feedback path 3|2 provided between the anode and control grid electrodes 3I| and 3|3 respectively. This feedback is adjusted so that the amplifier produces continuous oscillations with a high degree of stability. The input from the Adivider 29| is connected through a resistor 3M connected in circuit between the capacitors Cz and Cz, so that the incoming pulses reach the control grid through a capacity coupling,the pulse input circuit being thus partially isolated from the output circuit of the oscillator. The resulting circuit is responsive to incoming pulses which, when timed to the same period as that of the oscillator or a period very close thereto, pull the oscillator into synchronism therewith. The` oscillator should, however, preferably have suilicient ilywheel effect, or inertia. to respond with only a negligible change of phasemwhen the incoming pulses are push-pull modulated.

Since the oscillator 232 must be of such a Q as to permit its being pulled into step with the pulses from divider 23|, the resulting wave produced will be somewhat distorted. The tuned circuit 320 in the output of the oscillator overcomes the distortion. Circuit 320 is coupled to the anode circuit 32| through a high resistance 322, the circuit 320 being of a high Q and sharply tuned to the frequency of the time control wave desired for the demodulators M at the easty l terminal. Following thecircuit 320 is the phase shifter 233 shown in Figs. 1 and 10.

rIt will thus be vunderstood that the channel pulses received by the divider 23| from the west terminal controls the operation of the oscillator 232, and that the time control wave output to the phase shifter 233 will have a period corre- V-sponding substantially identically to the period of the time control wave applied tothe modulator and demodulator of the west terminal from the supervisory unit 26, thereby enabling an operator to eect proper synchronism of the sending and receiving terminals.

Repeater amplifiers Where the terminals of the system are some distance apart, line amplifiers or repeaters will be required to reshape and amplify the pulses during transmission. Fig. 12 shows such an amplifier for use in a one-way line. An ampliiler of this type may be employed at suitable points along the line between terminals and in the input and output connections thereof. The transmission line 291-304 of Fig. 1, for example, is shown with two such amplifiers at 30| and 302.

To the left in Fig. 12 is shown the grounded sheath 34| of a coaxial cable 340 having an inner conductor 342 coupled to the control grid 343 of amplifier tube 344 of stage A. The grid leak 346 of stage A is chosen of such resistance as to equal the characteristic impedance of the cable 340. The stage A, besides being used to match the characteristic impedance of the cable, is also used to threshold clip the channel pulses as indicated at 343 thereby eliminating at least part of the accompanying noise components that may be present.

Stage D is a known type of limit clipping amplifier which is useful for eliminating noise components superimposed on the channel pulses as indicated by the clipping level 350. The resulting pulse shape 3521s thus substantially freed of al1 noise components with `the possible exception of those noise components that may be superimposed on the substantially vertical leading and trailing edges of the channel pulses. This latter distortion while very small nevertheless is a source of interference.

This latter distortion, however, may be largely eliminated by reshaping the channel pulses. This is accomplished by diierentiating the pulses of the shape indicated at 352 by the differentiator stage C thereby obtaining for each input pulse two narrow width impulses 354 and 355 which correspond to the leading and trailing edges thereof. The impulse 354 is used to trigger the multivibrator stage D which preferably is of the type commonly known as flip-flop multivibrator adapted to be triggered from one state of operation to another and to return to said one state of operation after a predetermined interval. This return function is controlled by the bias imposed upon the multivibrator and by proper adjustment of the bias, the multivibrator may be made to produce output pulses 356 of a desired width corresponding exactly to the width of the channel pulses as originally produced at the sending terminal.

Stage E is similar to stage A in that the impedance of the output circuit thereof is arranged to match the characteristic impedance of the cable 360 or any other line to which the stage may be coupled. The stage E may also be used to further shape the channel pulses by limiting the amplitude thereof.

From the foregoing description it will be clear that the repeater amplifiers not only overcome attenuation of amplitude of the input pulses but also reshape the channel pulses to eliminate substantially all noise components .that may be present either due to outside influence or produced in the system itself. It will be readily apparent that the repeaters in reshaping the Operational multiplexing system Assuming that the system is properly monitored so that the modulators and demodulators of the corresponding sub-station units of the ltwo terminals are timed to avoid channel interference, multiple channel transmission from the west terminal to the east terminal of Fig. 1 will occur substantially as illustrated in Fig. 3.

Each modulator of the three stations shown in Fig. 1, for example, is supplied with a time control wave such as waves Iw, 2w and Sw over line 35 from the master oscillator 21. The phase shifters PH connected to line 35 are adjusted to time diiferently the wave energy as shown in curve a of Fig. 3. Since push-pull modulation is employed, the pulses produced by the modulators M are paired ofi with the interval 4t between successive pairs greaterthan the interval 3i between the pulses of each pair. When the trains of pulses from the modulators are mixed together in transmission line 231 for application to the transmission link 25, they form distinct` pulse groups with a monitoring interval between the groups as shown by curve e of Fig. 3 (see also curve a of Fig. 7). This characteristic grouping of the vchannel pulses may be controlled by proper selection of pulse widths and the bias maintained on the modulator circuit so that the interval 15 is equal to the same interval required by one or two signal pulses, as the case may be. This timing relationship is used, not only for monitoring purposes but also for synchronizing the timing waves for the demodulators at the receiving terminal with the time control Waves at the sending terminal.

At the east or other receiving terminal, the puise train of curve e, Fig. 3, or curve a, Fig. 7, is applied to each station unit alike. The Wave energy used for timing the demodulators of the receiving stations is derived vfrom -the frequency divider 29| and thence to oscillator 292. The frequency or pulse divider reduces by one or more stages the repetition frequency of the channel pulses to the frequency desired for oscillator 292. The resulting pulses pull the oscillator into step. The resulting wave produced by oscillator 292 is smoothed as hereinbefore stated by the sharply tuned resonant circuit 320, and is then adjusted in phase by phase shifter 293. The wave is then applied over line 325 to the individual phase shifters PH and thence to the 'timing circuits TI of the demodulators DM of the east terminal. The phase shiftersPH time diierently the wavl. energy similarly as in the case of the time control waves of curve a, Fig. 3, for the modulators M at the west terminal. The timers Tl each generate a train of gate pulses such as Irc, ly, Iz, etc., of curve c, Fig. 7, for gate control of the associ ated demodulator. The pulse energy produced by .the timer circuit operates to unblock the demodulator for intervals coinciding with the channel pulses to be received, the receiving station No. 1 at the east terminal being thus timed to receive pulses la, Ib, le, etc., to the exclusion of the pulses of the other channels. The unblocking pulses of the timers TI are of duration and timing so as to include reception of the pulses of the corresponding channels, regardless of the time displacement (modulation) of the channel pulses which, of course, may rst be differentiated to reduce their width as hereinbefore explained in connection with curve b, Fig. 7.

The return direction of communication from east terminal to West terminalis performed in the same manner as described above.A The timing of the modulators at the east terminal, however, is controlled by the timing waves applied to the timers TI This requires that the timing waves used for controlling the demodulators at the west terminal be separate from the time control waves for the modulators at such terminal. This vis taken care of by the separate phase shifter 219, feeding line 36 and the individual phase shifters PH connected thereto.

M onitorng that the monitoring indicating units are desirable to indicate on the one hand the phase relation between the pulses `transmitted at the sending terminal and on the other hand the timing of the received pulses with the gate pulses of the demod- 22 ulators at the receiving terminal. By this visual indication the operator can easily manipulate the phase shifters PH and PH, as the case may be, to obtain proper timing of the channel pulses and the desired alignment of a given channel with the demodulator ofk a given station.

Referring particularly to the visual indicator 28, it will be observed that the unit includes a cathode ray oscilloscope 460 having horizontal dellecting electrodes 46| and vertical deilecting electrodes 462, a sweep generator 464, a phase shifter 466, preferably of the character disclosed in Fig. 10 and a buffer amplifier 468. Thebull'er amplifier is fed with a time control wave over line 410 which may be connected by switch 412 to either of the output connections 413 or 414 0f phase shifter 31 and 38 respectively, of one of the stations such as No. l station. The control wave may be shifted in step fashion by the phase shifter 466 'so as to provide for the sweep potential the timing of not only station No. 1 but any of the other stations as may be desired. The output of phase shifter 466 is applied to a known saw-tooth generator 464 whereby a desired sweep potential is generated for the horizontal deiiecting electrodes 46|.

'I'he vertical deecting electrodes 462 of the oscilloscope 466 are connected across a line 415 which may be connected by switch 416 to line 411 which connects with the output transmission line 291 for the west terminal or to line 41? winch connects with the input line 480 from the transmission link 25.

'I'he screen of the oscilloscope 460 is preferably calibrated for the desired channel timing of the system. As shown in Fig. 13, for example, the calibration may include equally spaced index marks 482 which may be designated by numbers 1, 2, 3, etc., to correspond with the channels of communication. For illustration purposes, only three channels are indicated, The tracing made by the beam of the oscilloscope is located adjacent the calibration as indicated at 484 whereby the timing of the channels onewith respect to the other. and with respect to stations is readily observable. It will be noted that the channel pulses in Fig. 13 are each properly aligned with respect to the adjacent calibration marks 4'82 and .that the pulses la, Ib, Ic and I d are in alignment with the channel index marks for station No. 1. It will also be noted that the monitoring interval 15 occurs opposite thezero index marker. This alignment and monitoring arrangement is in accordance with the description of the channel pulses with respect to Figs. 3 and 7.

In Fig. 14, the screen` of the oscilloscope shown in Fig. 13 is shown to indicate the relationship of the pulses when the pulse train is shifted one step in phase accordingto the adjustment of either of the phase shifters 218, 219 and 466. To return the pulses into proper alignment with the corresponding stations, an adjustment of one of the aforementioned phase Shifters is required. Once the phase shifter 466 is properly adjusted to align the channel pulses with their respective station index markers, the phaseshif-ter 466 will normally be maintained at that adjustment. Should any re-alignment be required at the west terminal, for example, due to variation in the operation of the master oscillator or the oscillator 292, for example, the adjustment may be had by manipulation of one or the other of the phase Shifters 218 and 219, as the case may require.

In monitoring the system it will be desirable first to determine the propery timing of the output

Images