US2262838A - Electric signaling system - Google Patents

Description

Nov. 18, 1941- E. M. DELORAINE ETAL ELECTRIC SIGNALING SYSTEM Filed NOV. 8, 1938 5 Sheets-Sheet 1 Fig. 1.

A ttorne y Nov. 13, 1941.

E. M. DELORAINE ETAL ELECTRIC S IGNALING SYSTEM Filed Nov. 8, 1938 s Sheets- Sheet 2 Fig 7.

Fig.1]

Er E1 Er El-C) In ven for pIADAA/NE AM- 1 1 WES.

Attorney 1941- E. M. DELORAINE ETAL 2,262,833

ELECTRIC SIGNALING SYSTEM Filed Nov. 8, 1938 5 sheets-sheet s Fig. 4.

Fig. 5.

CABLE- a In ventor A itorney 5/7 0.42 ORB/NE L K64 PA'EVES.

1941+ E. M. DELORAINEI ETAL 2,262,838

ELECTRIC S IGNALING SYSTEM Filed Nov. 8, 1938 5 Sheets-Sheet 4 w z t N m? n W5 I V 5 k 5 5 1 a E 1 m N 5 .LSQ .8 E 5 Attorney Fig 12 Nova 18, 1941- E. M. DELORAINE ETAL 2,262,338

' ELECTRIC SIGNALING- SYSTEM Filed Nov. 8, 1938 Y 5 Sheets-Sheet 5 000 sec.

Inventor if? REEVES b Attorney Patented Nov.- 18, 1941 uNlraosrAr Edmond Maurice Deloraine and Aleo llarle'y Reeves, Paris, France, assignors to Interna-; .tional Standard Electric Corporation, New

,York, N. Y.

Application November a, 1938, Serial to. 239,454 f '1 In Great Britain Novemberv 19, 1937 l 8 Claims. (Cl. 178-71) I changes in the circuitequivalents of the cable The present invention aims at providing sigsections themselves due to variations in temperanaling systems, particularlysuitable for high and very high frequencies which will be at least as 1 eflicient as systems now much cheaper to install and to maintain.

Signaling systems used heretofore, especially telephone systems had to be constructed and installed with a large degree of accuracy, thus entailing either precision in manufacture and maintenance, or the necessity of making many adjustments and measurements.

In the prior art, it wasgenerally admitted that in use, but which will be decibels above reference level would be difficult to obtain from simple repeatersja drop of about 60 db. per repeater section hasbeenconsidered. Thislimits the length of such a section to some 10 miles for a maximum frequency of two mega cycles and to less iorthe higher frequencies required in television. Long" distance cables have therefore a considerable number of repeaters in tandem and the requirementsas regards stability, distortion and noise are very stringent. Besides, particular attention has to be paid to arrange the installation in such a way that the breakdown of a repeater tube does not interrupt the whole service. This means: sparefrepeaters or even a spare cable. The repeaters have'tojbe complicated with arrangements for keeping the levels constant notwithstanding temperature or tube variations. Even if a satisfactory service can be obtained, all this means a relativelyhigh price for the.repeaters andtheaccessory equipments, and an increased maintenance. I

In many signaling systems now in useand ture or other causes.

In the second place, not only variations in level must be guarded against, but also changes in the distorting properties oi the cable sections; other* wise it is impossible to correct for the totaldis-f tortligon at the receivingstation' by any fixed networ An object of the present invention is therefore to overcome the above difficulties'by means considerably cheaper and more reliable than existing methods and this by converting the signals or intelligence to be transmitted into signals of special nature, thus enabling the use as repeaters of devices of simple and robust design, the

details of which will be apparent from the following description: r

In outline, according to the-present invention, the repeaters arechosen of a type suchthat their peak amplitude outputs remain substantially constant inspite of variations, within prescribed limits, of the peak amplitudes at their inputs;

while the speech (or other) wave-form to be I transmitted is according to the invention, conveyedby a type of signal" such that the nonlinearity in amplification characteristics inherent in the above form of repeater producesv negligible distortion in the final wave-form given out by the receiving terminal equipment.

, According to one aspect of the invention, fre

projected for the future, a ratherlarge number of repeaters are involved. A5 the transmission equivalent .must, in general,fbe constant within,

close limits, it'is therefore necessary for'the gain of each repeater, or series of repeaters, to be kept within limits of variation that are rverysmall irideed, thus involving rather/expensive and com-,

quency or phase modulation of a carrier rre-= quency is used, in conjunction with a repeater as described above. It is well knownthat considerableamplitude distortion may be applied to a carrier having either of theseforms of modurlation without causing appreciabledistortion in" This'iact isused in the present the final output. instance, in a manner hitherto not knownin the art, by applying one or both of these types 'of modulation to a carrier to be conveyed byfthe f transmlssionsystem (e. g; of a cable) .under con-1 sideration; a transmission syster'n containing fre'- peaters which distort in amplitude'by limiting, the-peak outputs.

According to another, H the invention the wave-form to .be transmitted is conveyed by a series of pulsesof-one or several amplitudes at the transmitter'outpuh of constant or variable duration, and each. pulse being either aperiodic in one or both'currentidirections, .or com'rising {a train or waves' oi ra carrier-fre quency; in combination 'ag'ain,withrepeaters for I o these fpulsesl of "the ak' outputs are sub and preferred, form of tions, or of one or more television channels.

stantially constant whatever the pulse input voltages within prearranged limits.

ing to the burning out of a tube. The simple paralleling of the amplifying element by a reserve but in which close limits of phase variations and/or amplitude variations at different frequencies, have to be met. An example of the former is'the case of the simultaneous transmission of a large number of telephone conversa- In both these instances band widths of several megacycles have frequently to be dealt with. In the former art, to avoid serious inequalities in transmission characteristics at the extreme ends of the frequency band, it is usual to employ in-addition to the speech or television itself at least two pilot channels, one at each end of the frequency range, arranged to cause automatic compensation for gain changes throughout the entire band width. As this compensating apparatus had-to be applied not only at the receiving station but also at every three or four repeater stations (in general), the resulting expense and complexity is considerable; four or more repeater tubes, each with its associated apparatus, having often to be employed at these repeater points. Similar complexity of apparatus had formerly to be used in the case of a narrower band, but one requirin'g particularly close limitsof working. According to the present invention both these cases are solved by the use of repeater apparatus that is much simpler, cheaper, and more reliable: in general only one active amplifying element is required. The preferred solution consists in the use of pulses as explained above,- each pulse or group of pulses representing (according to a prearranged law) the momentary characteristic (e. g. amplitude) of the wave-form of one or of a group of communication channels. In the case of multiplex operation, the pulse or group of pulses representing each' channel or channel group recurs at regular intervals, these time intervals being at least as short as one period of the highest component frequency in the channel or channel group to be transmitted. For either single or multi-channel working, the repeater outputs consist of pulses having quite definite voltages and wave-forms independent of the input (within the working limits) they are, in fact, more of the nature of the simple relays as used in telegraphy than that of the usual -quantitative amplifier. Any wave-form distortion occurring, therefore. in the transmission medium will be corrected automatically at each repeater tube gives rise to difiiculties, as it is not easy to ensure having .the same gain with either two tubes 'or a single'tube in circuit. The alternative planof switching in automatically a spare tube, or spare, repeater, when failure occurs introduces an extra element of unreliability on account of the relay contacts involved. These dimculties have in the past been considered so important that in some cases it had been planned to instal a complete spare cable together with its repeaters,

a very expensive solution. a

. The present invention entirely overcomes the above troubles inherent in the former art; the

' method of the simple paralleling of the amplifier struction, their cost may now be very low, thus tube, or the complete repeater, is now free from objections, as the change in repeater output of 3 db. or more due to the failure of one tube makes no difference at all to the output of the subsequent repeater, as long as this reduced level is still sufilcient to operate the latter. At each repeater point, therefore, one or more spare repeaters or tubes may be connected in parallel, and in such a manner that all the tubes or repeater elements may be very easily removed for routine testing, and replacement. As the repeaters further are of very simple and robust conmaking their duplication by reserve elements a comparatively cheap matter.

Further advantages arising from the use of the invention will be apparent from the description (below) in detail of some of the apparatus coming within its scope; and in particular, the use of the application termed the fsynchronous type of repeater for carrier frequencies of the order of 10,000 megacycles or more (in free space or on dielectric guide) frequencies for which no direct means of amplification were known in the former art.

A more detailed description will now be given of some embodiments of the, present invention with reference to application to certain types of transmission problems.

Figure 1 shows one form of blocked repeater suitable when frequency or phase modulation is used. The terminal apparatus for these forms of modulation being well known in the art will therefore not be described. Such a repeater is also suitable for single pulse" or double pulse station-and without the use of any auxiliary 1 apparatus.

According to a further feature of the invention, by the use of one form of device conforming to the above conditions a repeater is obtained capable of two-way operation without singing and byan additional slight modification of it an echo suppressor is obtained at each repeater station without the use of any supplementary ap- Another advantage of the invention is the greatly improved facility it gives for the introduction of spare amplifier tubes, or complete spare units, at the repeater points. In the former art considerable difliculty has been experienced in solving this problem satisfactorily; a problem it is quite essential to overcome, as it is inadmissible to allow the possibility of the simultaneous failure'of a complete group of circuits ow- 75 of Figure 7 suitable for frequencies of 10,000-

modulation.

Figure 2 shows a second form of repeater-of the trigger type suitable for either single pulse or double pulse operation.

Figure 3 shows another trigger" type of repeater circuit. It is suitable for "double pulse" working only; the duration as well as the amplitude, of the output pulse has a fixed value irrespective of the input within prescribed 'limits.

Figure 3a shows a diagram used in the explanation of the action of the synchronous type of repeater.

Figure 4 shows a repeater circuit R branched on a two-way line L.

Figure 5 shows a two-way repeater installation somewhat analogous to a d-wire system.

Figure 6 shows an application of the arrangement of Figure 4 utilising a double triode valve.

Figure 7 shows a synchronous type of repeater suitable for operation with double pulse modulation with a can'ier wave.

Figure 8 shows a variation of the arrangement megacycles or even higher. The active element is a magnetron of the transit time form." a

Figure 9 shows one form of terminal equipment suitable for operation with the repeaters shown in Figures '7 and 8. It is capable of giving automatic echo suppression a: the Vodas" type at all the repeater points in the transmission system.

Figure 10 showsone form of terminal equipment for 12 channel operation on the distributor principle, using "double pulse" operation without carrier. (The equipment for two channels only is shown.)

Figure 11 shows a multi-repeater installation and is used to explain the 2-way action of the synchronoustype of repeaters.

Figure 12 shows diagram used in explaining the action of the circuit otFigure- 10.

In the simple repeater of Figure 1, phase or frequency modulated carrier waves arrive from the cable or antenna by the line B, are stepped up by transformer C to the appropriate impedance, and applied as shown to the grid of amplifier tube A. By means of transformer E the amplified output signals are applied to the output antenna circuit or cable F.

Resistance D, high compared. with the grid impedance when this grid is appreciably positive, causes the positive input peaks to be out of! sharply when they exceed the value of the bias battery G.

The value of G is arranged to be such that the tube acts as a class B or class C amplifier, the gain being thus small enough, in the absence of signals, to prevent the whole system from either singing or overloading the subsequent. repeaters by noisea condition that might otherwise occur.

As the peaks are cut off, for input voltages exceeding the value of G the peak output voltage is substantially constant and independent of the input, within wide limits of the latter, limits wide enough to take into account the maximum fiuctuations of the input voltage at the input of the transmission system considered.

The minimum value of this input voltage when fluctuations occur is arranged-to be slightly in excess of that of the battery G; the amplifier, therefore, always operates on the flat portion of its output v. input characteristic. In cases where the input voltage in practice is less than that required to meet these conditions,one or more extra stages of amplification, using conventional amplifier circuits, are placed between the input from B and the transformer C.

A number of possible modifications of the circuit of Figure 1, giving a limiting action on the peaks at the same time as reduced gain in the absence of signals, will be immediately apparent to those skilled in the art.

The invention is not limited to frequency or phase modulation but on the contrary the invention is applicable to all processes of modulation implying the use of a carrier wave.

In the transmission of an electroemagnetic wave, four variables are involved, viz, frequency, phase, amplitude and time. Mostordinarymodulation systems transmit intelligence by variation of amplitude against time, phaseand frequency being kept constant. So-called frequency-modulation systems transmitintelligence by variation of frequency against time, amplitude and, perhaps, phase being kept'lconstant. So-

called phase-modulation systems transmit intelli constant. 'I'he-fpulse-modulation! system transmits intelligenceby variation'of a time-interval gence by variation of phaseagainst time, amplitude and, in a certain sense,- irequency, being kept against time, any one or more of the -variables.-. frequency, phase and amplitude, being changed suddenly to indicate the'beginningand end of the time-intervaL- 1 In other words, pulse-modulation" consists in the transmission and reception of a number of pulses, the latter occurring at a frequency at' least as high as, and normally higher than, the

highest frequency component in the wave-form to be conveyed. In the simplest examples, every pulse has the same amplitude. In "single pulse" working substantially rectangular impulses are transmitted, the duration of each pulse being linearly proportional to the momentary amplitude of the wave-form to be sent. With "double pulse operation, the duration of every pulse is constant and small compared with the time interval betweenadiacent pulses; the impulses occur in pairs, the time interval between the members of each pair being-in this case proportional to the transmitted amplitude;

As a repeater for such a system, therefore, it is clear that the circuit of Figure 1 maybe used if desired; the necessary constancy of output peak voltage, irrespective of input value within limits will be obtained by it. The circuit ofFigure 1,

however, has the disadvantage of giving an out-' put pulse steepness of wave-front proportionnal to the input voltage; after a number of repeater stages therefore, under conditions when thegains of all the repeater sections are'low the received pulse wave-fronts may be very considerably less steepthan the original pulses from the transmitter. 1 I

In such systems, the signal-to-noise ratio for single channel working, as well as the interchannel interference when multiplexed on the "distributor principle, is dependent on this steepness of pulse wave-front, the steeper the wave-front the higher the signal/noise ratio, and the less the cross-talk between channels. If,

therefore, at the receiver the pulse wave-front is thus made appreciably less steep than at the.

transmitter, the signal/noise ratio and/or the cross-talk may lie outside the'requiredlimits; to

avoid this, the output pulse from each repeater should have a constant wave-front steepness (as well-as constant peak amplitude), of a value at least as high as that of the original pulses from the transmitter. i I v To accomplish this result according to the present invention, instead of a simple amplifier as shown in Figure l, a trigger device is used; the requirements for such a trigger circuit are: (a) an output peak amplitude constant and independent of the input voltage giving the triggering action (within limits of this input value) (b) toavoid singing and noise operation, the trigger must be inoperative for E. M. F. appreciably lower than that of the minimum pulse value; and (c) the steepness of wave-front of the output pulses both on triggering and on restoring, must be constant and independent of the input waveform or voltage (within limits of the latter) and of a value at least as high as that of the original pulses from the transmitter.

Figure 2 shows one form of trigger circuit complying with the above conditions.

The double triode' A has cross-connections from each plate to the opposite-grid as shown in the diagram; the effect of theseconnectlons is to produce a circuit of the same type, in'principle, as that described in elementary form in Figure 3a,-

equilibrium, one with the left-hand plate current being practically at cut-off, with'the right-hand plate having a current corresponding to nearly zero grid bias, and the second stable position having the above conditions reversed quite symmetrically.

The reason for the use, in practice of the circuit of Figure 2' rather than that of Figure 3a is that it is generally not convenient commercially, to employ two separate sources of high tension (as in Figure 3a) especially as the latter are floatingwith respect to ground, and therefore must have their capacities to ground of very low values.

In Figure 2, a source of high tension F is used having its ends grounded through condensers. A potentiometer G (of low value compared with the plate resistance of A) is connected across F with the slider (approximately at the centre point) connected to the filament and ground. In this way half the voltage value of F (a source of doublethe usual plate voltage required) is used as the-plate supply to A, and the negative half used by means of resistances D and E to compensate on the grids of A for the high tension that would otherwise arrive at these grids.

Resistances B and C are added to allow the plates to remain positive while the grids have the correct small negative values.'

For the present purpose an extra negative voltage H is added to the left-hand grid, of such a value that only equilibrium position No. 1 (with the left-hand plate current of a low value-the right-hand current being high), can be maintained, in the absence of input voltage from the line. The arriving pulses from the cable or antenna line U are stepped up in transformer K to match the grid impedance, and coupled to the left-hand-grid by condenser L; a condenser of suitable value (for single pulse working) such that, in conjunction with the eifective gridresistance the original rectangular wave-form of the pulses is maintained at the grid, the active grid pulses, moreover, being of positive sign.

The arrival of apositive pulse through L of .suflicient amplitude reverses the equilibrium position of A; causing the circuit to trigger over to the equilibrium position No. 2, the left-hand plate now having the high current and the right-hand plate the low current value.

Thesteepness of wave-front of the changingwave-front that is independent of that of the input pulses, and may be-arranged to be as high as desired withinreasonable limits.

The output from A is taken through the coupling tube M, by condenser'N and decoupling resistance P, as shown-in-the diagram.

Although the circuit of Figure 2 maybe used as a repeater for both single or double pulse working, there will now be described a modification of it that is rathermore advantageous for the latter case. 1

Although, according to Figure 2 the peak voltage and steepness of wave-front in the output pulses are constant, the duration of these output pulses is not constant, but is substantially the same as that of the input wave-form. This condition is necessary for single pulse operation, but is not required, and is, in fact, a disadvantage when using the double pulse system, a system in which all the pulses should be identical in waveform.

The circuit of Figure 3 fulfills this new condition. Resistance C of Figure 2 is replaced by the condenser S; and the right-hand grid instead of being connected to the negative end of F (Figure 2) through resistance D, is now connected direct.

to ground through grid leak T and the variable negative bias R. In the stationary state, therefore, the right-hand grid voltage can have only one value, that of R, which is normally small or zero. Even in, the absence of the extra battery H, therefore, the circuit now can have only one position of stable equilibrium, position No. 1 in which the left-hand plate current is low, and the right hand current high. on the arrival of a positive input pulse at the left-hand grid, condenser S enables the sudden rise in current in the left-hand plate to cause a momentary reversal of the equilibrium state of the circuit, the latter triggering over as before to position No. 2, but as condenser S is unable to maintain the steady state of the circuit in this new position (it is only able to transmit surges) the circuit over action is substantially independent of the wave-form of the input pulse; it is proportional, on the other hand, to the resistance-capacity time-constant of the trigger circuit, taking into account the effective internal resistance of the tube plate circuit itself. As soon as the input pulse ceases, the circuit triggers back to position No. 1, the speed of this restoring process being again only dependent on the circuit and not on the wave-form of the input pulse.

Input E. M. F. of peak voltage less than a predetermined value has no efiect on the trigger circuit, this marginal sensitivity being dependent chiefly on the average negative voltage on the grids of A, the higher this voltage, the less voltage required to operate the trigger, up to a point where double stability is no longer obtained. It is thus clear that the circuit of Figure 2 conforms to the three requirements as stated above: as arepeater, it is (a) inoperative by E. M. F.

below a certain peak voltage; (b) it gives a peak output that is constant when operated at all; and (c) it gives an output pulse having a steepness of immediately afterwards suddenly restores itself to position No. 1, and this last action takes place after a time dependent on the circuit constants and not on the'duration of the input pulse.

The circuit of Figure 3, therefore, gives the effect desired for double pulse operation; the

duration of the output pulse is controlled by appropriate adjustment of condenser S together with the other circuit constants.

Another and, in general, preferred form of repeater circuit utilising a pulse generator syn-' chronised by incoming pulses will now be described. It is very suitable for double pulse working-or for any system in which sharp pulses of constant wave-form are employed. It will be referred to here as the synchronous type of repeater circuit. It relies on the well-known property of an oscillator e. g. of relaxation oscillator to lock itself into step with asmall E. M. F. applied to its'grid, when this E. M. F. has a frequency of the same order as that ofthe naturalfrequency of the relaxation oscillator.

In order to explain clearly the working of this kind of device, one may consider any type of circuit having two more or less symmetrical positions of stable equilibrium. If such a device is initially adjusted at exactly its neutral equilibrium point, midway between the two stable positions, and if it is then left entirely undisturbed by noise-it will theoretically remain. at this neutralpoint indefinitely. In practice, of

course, the intrinsic circuit noise will give a small bias in one direction or the other, resulting, finally, ina swing to the corresponding stable position; but if at the start an impressed D. C. pulse v be applied in the direction of either stable positiona: pulse substantially greater than the .noise--then the rate of swing-over will be increased. Ifthe circuit be quenched after time t and if the swing" wave-form-be truly expon'entialthe peak voltage reachedby this swing before quenching will be: ce where k is the'increment (i. e. negative decrement) of the swing-over action'; an effective amplification of the pulse by the. factor e is thus obtained, a

factor which, as in the normal superaction circult may reach very high values. One simple form'of symmetrical double-stability relay circult suitable for theabove purpose is shown in Fig. 311. It comprises essentially a double triode and two resistances R1 and R2.

If this circuit be quenched".to its neutralequilibrium (i. e. both plate currents'equal) at regular intervals, it will carry outthe above function of pulse amplification. There is, however, a variation of the cir-- cult of Fig. 3a which (for the purposes of I synchronous repeatering) does not require an externally applied quenching oscillatorand this is simply the conventional multi-vibrator circuit. In the multi-vibrator, (which is the circuit of Fig. 3a, with the direct connections between plates and opposite grids replaced by connections through condensers), the two stable equilibrium positions are changed to two quasi-stable conditions-(i. e. two positions in which the conditions are relatively stable for a limited time) and the circuit automatically triggers over from one of and synchronised oscillator in a nearly ideal manner for the property of synchronisability by very small impressed voltage pulses to be displayed. I I

Fig. 6 shows the multi-vibrator typeof synchronous repeater circuit as applied to a double pulse cable system without carrier.

It has the advantages of extreme simplicity of construction, and capability of two-way operation (as is explained below); at the same time it conforms to the three requirements carried out by the trigger circuit of Fig. 3. A double triode B is associated as shown with a multivibrator circuit of the usual design. Double pulse signals of positive sign in the cable A, from either direction, are stepped up by transformer C to the effective grid impedance of the tube, and coupled to the multi-vibrator by condensers E and F. When the input pulses arriving at the coupling point D are of sufiicient amplitude (2 volts or less), and if the multi-vibrator is adjusted so that its natural frequency "is slightly lower than thatof the incoming pulses, the local oscillator will lock itself into step with these these two positions to the other, and vice versa.

Just before each triggering, the circuit condition may be described as in quasi-neutral equilibrium; and the arrival of an impressed D. C. pulse during this period of the triggering cycle results in the subsequent triggering taking place earlieror later-than would otherwise be the case (according to the sign of the pulse). If the circuit were artificially quenched at a definite time interval after the arrival of the pulse, the peak volts represented by the swing-over at the moment of quenching (reckoned from the voltage at the neutral equilibrium point) would be linearly proportional (within certain limits) to the peak value of the impressed pulse-and would thus represent this pulse greatly amplified, exactly as in the case of the relay circuit of Fig. 3a. In the usual multi-vibrator circuit, then, with the additionof quenching-there is an intrinsic amplifying action on the super-reaction principle; If, now, it is desired to-use this action to give amplification of periodic signals before their application to lock into step a synchronous repeater, there should be added such a free oscillator after the multivibrator circuit. But in the latter there is such a free oscillator already present; and as modulation takes place not on the amplitude but on the phase of the signal pulses, there is not required anyartificial quenching and there is no interest in obtaining, for the synchronising process, an

pulses, in frequencyand phase.

Alternatively, if that form of double pulse system is used in which one of each pair of pulses is resupplied at the receiver, the natural frequency of the multi-vibrator may be adjusted to be slightly less than half the pulse frequency as originally produced at the transmitter, the local oscillator then looking into step with the first (or second), pulse of each of the original pairs of pulses. 1

In either case, the adjustment is such that exact locking remains during the time modulation of the incoming pulses by the speech or other wave-form. to be conveyed.

The outputvoltage at point D due to the multi-vibrator action has, of course, a rectangu- 'lar wave-form of the same nature as that of single pulse modulation. To transform each cycle of the multi-vibrator into a sharp pulse as I, required for the double pulse system, the small r condenser F and the resistance G is added, of such values that the wave-form of the voltage at point H is substantially the derived curve of' the wave-form at D. Also to prevent the negative pulses that would otherwise occur at H due to the second half of each multi-vibrator cycle, rectifier K is bridged across resistance G in such a sense as to give a substantial short-circuit to these negative pulses. This correcting network is amplified replica of the signal amplitudes, it is of the same type as that used in the double pulse transmitter described in French Patent No. 833,929 filed on June 18, 1937, and its first addition No. 49,159 filed on July 5, 1937. The multi-vibrator action causes, therefore, at point H a sharp positive peak a small fraction of a pulse cycle after each input pulse from the cable; and these local pulses, in a practical case, may have a level of 40 db. or more above that of the input pulses i. e. an effective amplification of the order of40 db. or more is obtained. From point D the amplified, local pulse is applied back to they cable through the same transformer C as is used for the coupling of the input pulses.

The manner in which the back action of the repeater onto the preceding repeater in the chain is prevented (which otherwise would give rise to local singing conditions) is explained later in the present specification.

A simple form of synchronous repeater suitable for use on double pulse modulation (or other system comprising sharp pulses of constant amplitude and wave-form), these pulses modutrains;

(b) A demodulator to the pulse frequency;

-A free oscillator of peaky waveform locked into step by the amplified output of (b); and

(d) A sourceof high frequency of the original frequency (or nearly so), modulated by the output of the locked oscillator (c).

The design of such a repeater will be discussed in detail below. It will be shown that a suitably adjusted super-reaction circuit, with self-quenching provides a satisfactory solution, and provides it by theuse of one vacuum tube only.

A suitable 2-way repeater circuit is shown, in its simplest form, in Fig. 7. A triode A-of a 'design slntable for oscillation at the carrier frequency used-is connected as an oscillator by means of inductance C and condenser 13. The high frequency grid connection is coupled through the blocking condenser D; and the grid leak is formed by the two resistances E and Fin series.

The high frequency oscillations are made to start and strip, in trains by the action of the grid current in charging negatively the condenser G-in the well-known manner. The function of E is to prevent the high frequency from being shorted to ground by G, and may therefore be replaced if'desired by a high frequency choke. If the repeater is to be used for operation in free space, the oscillating circuit is coupled, as shown, to a suitable'antenna H by means of coil J; if for cable operation, J is connected to this cable-impedances being suitably matched.

As is well known, such a squegging" oscillator will act as a self quenched super-reaction receiver; in a manner similar to that described above, the circuit will therefore greatly amplify incoming signals in the neighbourhood of its natural frequency. The circuit also acts as a demodulator to the amplified signals; if the latter arrive in pulses, therefore, a strong component at this pulse frequency will result. The requirements pointed out above under (a) and (b) are therefore fulfilled.

Respecting requirement (0) it will be seen that the provision of a free oscillator of peaky waveform is also already carried out in the "squeg oscillation of the, circuit; and if this latter is adjusted to be of nearly the same frequency as that of the signal pulses-the free squegwill be locked into step with the latter (within certain limits of amplitude) as is desired. The frequency of the high frequency generated by the circuit B-C has already been'adjusted to be approximately that of the incoming carrier; as, therefore, at each squcg cycle a train of the local high frequency is produced, requirement (:1) is also fulfilled. V

The result, therefore, is the desired synchro. no repeater-obtained in this case by using the normal sinusoidal type of super-reaction instead of the aperiodic described above.

At this stage,-it may be of value to refer to a recent experimental result obtained on a repeater ofthedeSl8no1Figure7. Thecarrier frequency was about 100 'megacycles and the tainable, great interest is now centred on the possibilities of very high frequencies-of the order of 10,000 megacycles or even higher-these carrier frequencies being, in most cases, conveyed by means of some form ofdielectric guide.

In the former art no practical means existed for the repeatering of such frequencies-r-one of the obstacles that has so far proved most serious when plann ng a transmission system on these The synchronous repeater, however-of the type illustrated in Figure 7is capable of adaptation'to these much higher frequencies; it is, therefore, the only practicable solution of the problem that is so far known. Figure 8 shows a simple form of such'a repeater for frequencies of the order of 10,000 megacycles. A is a solid anode" type of diode .magnetron tube oscillating at approximately the carrier frequency of the incoming pulses (double pulse system). In series with the high tension supply B is placed a resistance Cshunted by a condenser D. With suitable values of C, of 'D and of the tube supply voltages and constants, it has been found by the applicant that as soon as high frequency waves start being generated a change in plate current occurs which causes a building up of an additional charge on condenser D that eventually alters the plate voltage sufficiently to stop the oscillations-in fact, a series of short wave trains of carrier frequency occurs exactly as in the circuitof Figure 7. The circuit of Figure 8 may,

therefore, be used as a repeater for these very high carrier frequencies, in the same manner as explained in connection with Figure 7, the waves to and from the magnetron being coupled by a a tance be 30 km., and the pulse frequency be 8 kilocycles; then the time ,of travel (assuming free space) is millisecond. Let repeater B, Figure 11, initially lag being repeater A in phase by exactly millisecond: it will then receive synchronizing signals at the moments of maximum sensitivity as a super-reaction receiver,

and will lock into step, being controlled by A.

, The pulse generated by B, however, will arrive at "A" 0.2 millisecond after the latter has generated the pulse that locked B in the first instance, and 0.25 millisecond in advance of the third pulse from A. This interval of 0.05 millisecond is great enough to prevent A, at this moment, from being sufflclently near to the point of sensitivity as a receiver to b affected appreciably by this pulse from B. The result, therefore, will be that A controls B, but that B does not control A, it is a one-way action.

If, however, the initial conditions are such that Alagsfl millisecsbehindBinphasathc situation is exactly reversed' B will thencon.

The amount of the wave-front of the speech .If B lags by 1 5 millisec. behind A-and. Cby

the same amount behind B-sigrials will pass from A to C through the repeatering action of B;

if, on the other hand, B lags by this amount be-" hind C, and A by the same amount behind B, signals are capable of being passed in the direction C- -A, through the repeater B. Eachrepeater is thus capable of acting in either direction, according to which of its stable equilibrium positions is in operation. But, it will control in only one direction at a timehence there is no possibilty of local echoes round the repeater sections. Further, whichever stable position it is in initially, it will remain in until, artifically disturbed by some means.

Another feature of the present invention relates to means suitable for changing all the repeaters from one stable position to the other as desired. Among the possible solutions for singlechannel working ('e. g. for ultra short wave radio link) is the following method:

Assume (as an example of the possibilities only) that a form of Vodas Control is required, symmetrically from each end of the link, that gives control to that end from which speech is originated (assuming that it is from one end only at a time), and thereafter. keeps the control at this end, even after cessation of the speech, until the direction of the incoming speech is reversed. For this purpose the transmitter terminal equipment at each end would comprise (essentially) simply a single tube in a circuit (see Figure 9) almost exactly similar to the re- .peaters (Figure 7).

As a transmitter, incoming speech from the line is applied to an auxiliary grid through condenser K shunted by the metal rectifier L and resistance M, in such a manner that th average potential (positive) obtained by this auxiliary grid follows approximately the peak value of the incoming speech. The effect of this speech will thus be always to accelerate the squeg wave trains, andnever to retard them in phase (1. e. the average frequency will be increased).

Let us assume that initially the link is passing signals in the direction EDCB-A, Figure 11. On the cessation of speech in this direction, speech then is originated at A. The speech advances (periodically at audio frequency), the phases of the pulses from A; the result is: (a) an almost immediate loss of control of B onto A-and (b) as soon as the pulse from A arrives at B slightly in advance of that from C onto B,

- the control of B passes from C into the hands of A-and remains there as long as speech is artortiontime is thus of the order of 1 millisec. (neglecting other-possible sources of initial distortion due tothe receiver-causes which can be riving at A--as this speech always advances the pulses from A, and never retards them.

Immediately afterwards, as soonas the pulse from B arrives at C slightly in advance of that from D onto C, the control of C--similarly passes from D into the hands of B; and this process continues until the whole control is in the direction A-BCD etc., the reverse of the initial condition.

used up. in this changing-over of the controls before the latter process is completed (and hence the extentof the initial syllable distortion) is a function of the ratio of the number of repeaterssections to the number of pulses per second. For example, with 100 repeater sections a minimum of'100 pulses would be required for the control change-over to be completed; and if a pulse frequency of 100 kilocycles, is used, theinitial dis,-

largely eliminated by suitable design).

When'speech in the direction A -B--C--DE ceases, the control remains in this direction until speech is originated at the opposite end, when the control changes over once more, in a manner exactly as described.

It is thus evident that the single-tube synchronous repeaters described will automatically reverse .their direction of operation, according to the direction of the speech. 1

Any of the above-mentioned types of repeater circuit (with the exception of the synchronous type on'2-way operation) mayfof course, be used on multi-channel working-either. to

repeat the pulse wave-form representing the err-'- velope of the whole group of channels combined together according to frequency difference" methods (the methods usual in carrier systems), or to repeat each pulse in turn of each channel or channel-group separated in time by the distributor method.

To accomplish this result, no essential change in repeater design is required; it is sufficient that the pulses given out by the repeaters shall be sharp enough in wave-form to avoid appreciable cross-talk between channels.

One simple form of terminal apparatus, suitablefor l2-channel operation on the distributor principle using double .pulse modulation, is illustrated in Figure 10. (Only two channels are shown equipped.)

The circuits shown are particularly suitable when a number of such groups of channels are operated simultaneously, as part of the equipment--the synchronisers A and B-can in that case be used in common for all the groups. An advantage of the design is that the transmitting and receiving terminals are identical in all respects; and any given channel independently of all the others, may be caused to change from the transmitting to the receiving position immediately-either by means of a single switch tion approximately rectangular, the whole of this portion lying just within the first twelfth of the 10,000 cycle period (see portion 1 of curve a). The negative half-cycle of this applied E. M. F. consists of a sharp negative pulse 2 occurring half a period after .the beginning of the first portion 1.

The circuits constants ofthe tube are adjusted so that the start of positive half cycle of. the multi-vibrator oscillation must always synchronise with some part of the section 1. In the absence of speech from line 1in which case the grid biases on the two grids of C are equal-the multi-vibrator natural frequency is adjusted so that stable locking takes place at such a phase that thepositive half cycle commences at point (3b) at the centre point of the portion 1. As, however, the voltage represented by 1 is practically constant throughout its duration, a slight difierence in potential difference between the two grids of C sufllces to cause locking to take place at a difierent phase, giving 'rise to the start of the positive half cycle at any point between 4 and 5 according to the strength and sign of this potential difference between the grids-a potential diflerence at audio frequency which is obtained as shown from the incoming speech from line I, as shown, by means of the transformer F. (This part of the circuit is the same as that used in the pulse modulation transmitters described in French Patent No. 833,929 filed on June 18, 1937, and its addition French'Patent No. 49,149 filed on July 5, 1937).

The start of the multi-vibrator half cycle, however, can never occur outside the limits given by points 4 and 5, on account of the strong locking action due to the E. M. F. of curve a.

The multi-vibrator constants are adjusted so that the end of the above positive half cycle is determined by the arrival of the negative pulse 2 from the synchroniser A. This moment is thus fixed and independent of the incoming speech.

As in the double pulse modulation transmitter already referred to, the rectangular waveform of curve b is then transformed into a sharp pulse G (curve 0) by means of the small condenser G in series with resistance H. The two dotted pulses each side of i on curve c show the extreme limits of variation, during speech modulation, of the sharp voltage pulse that is then present across the resistance H. Transformer N at the end of the corresponding curve, in thecase of channels other than No. I). Point 5 of curve b will now represent the starting time of themulti-vibrator half cycle in the absence of signals or speech.

The synchroniser B at the receiver-0f the same design and adjustment as A at the transmitter-is locked into step with A, either by conveying some of the output of A to B by a separate channel, or by other appropriate means. The exact value of the distributor frequency (referred to above as 10 kilocycles) and/or the propagation time of the cable (if necessary by adding a delay network) are modified so that this propagation time is an integral number of half periods of the distributor frequency. If a pulse from channel I at the transmitter synchroin series with blocking condenser K (large enough to prevent distortion of the pulse wave-form) is thenused to step down the pulse into the impedance of the cable, asshown. As in the double pulse transmitter, theundesired negative pulse due to the end of the multi-vibrator positive half cycle at 1 is eliminated by shunting resistance H by the rectifier L in such a manner as to eliminate this unwanted pulse without aifectlng the desired pulse 6.

The circuit of tube D, associated with line 2, is exactly similar, except that the applied E. M. F. from the synchroniser A is now taken from another output branch of the latter in such a manshows the output pulse given by this second multi-vibrator to the cable.

The remaining gap of 10/12 of a period of the 10,000 cycle frequency is filled up by the remaining ten channels in turn-no channel over-lapping in time with any other, thus avoiding inter-channel interference.

As already stated, and as shown in Figure 10,

the apparatus at the far end of cable-the end nises at the receiver with a channel I? pulse at this receiver end, the same result then will also take place in the reverse direction.

Let us consider the action of the pulse 6 from transmitter channel I on itsarrival at the receiver. Owing to the particular adjustment de--' scribed above in the synchronism between B and A, pulse 6 will arrive at the centre point in "time" of an applied E. M. F. from synchroniser B-an E. M. F. of the same wave-form and phase as the portion I of curve a. The arrival of the pulse causes the multi-vibrator of receiver channel l to start its positive half cycle earlier than would otherwise have been the caseit is- 'ceiver multi-vibrator No. 1 will minim in step with it at its new values, thus following the "time modulation at the transmitter. The result will be curve I) exactly the same as that of the transmitter multi-vibrator at the distant end of the cable.

As the pulse frequency components are filtered from transformer M by the small by-pass condensers shown, (or by a more complicated filter if necessary), the grid windingoi' M will have throughit a current proportional to the average area of curve b--i. e. proportional to the original speech wave-form at the transmitter. This speech therefore, will be given to the output line from M.

In exactly similar manner, multi-vibrator 2 of the receiver will have its starting time modulated by the pulses from channel 2 of the transmitter-and similarly for the remaining ten channels-each receiver multi-vibrator being entirely unafiected by pulsesfrom channels other than its own, asthese other pulses occur-always at times of insensitivity to'their locking .action.

To operate any channel in the reverse direction, it sufllces to raise the natm-al frequency of the circuit of tube L (for example) to givenormal locking at point 3 instead of at 5, and to lower that of the circuit of the corresponding tube (e. g. tube 0) to give normal looking at point instead of at point 3. Tube L will then control C, instead of the original reverse effect. The design of the synchronisers A and B- common to all the groups of 12 channels-and having a twelve-phase output of the wave-form described abovemay be carried out by any method already known to the art, for example, by a method similar to the pyramid of rectangular waves described in British Patents Nos. 344,444 and 363,403. I It will be seen the invention is not limited to the arrangements described by way of examples, but-on the contrary the invention is capable of many applications without going beyond the scope of the appended claims. For instance, Fig. 5 shows a transmission system somewhat analogous to the so-called 4-wire system in which two way lin'es l1 and I: are connected by suitable.

coupling network (hybrid coils H1 and Hz) to two uni-directional lines on which are. bridged repeaters R1. R2 R1 R: according to features of the present invention.

We claim:

1. A repeater system for repeating along a transmission path, signal modulated carrier waves wherein the signals are represented by time displaced impulses on said carrier waves, comprising a relaxation oscillator, means for biasing said oscillator normally to non-oscillatory condition in the absence of input impulses and means responsive to the received impulses for triggering said oscillator into oscillating condition, thus producing upon arrival of each impulse a corresponding single impulse of substantially constant amplitude level and having a constant wave front depending on the characteristics of the oscillator, and means for bridging said oscillator across said transmission path for repeating signals conveyed in either direction along said path.

3. A repeater system as claimed in claim 1 for aperiodic pulses further comprising a plurality of repeaters placed in succession in said transmission path, each said repeater having means for locking its oscillator in step with the received impulses, said repeaters being spaced at such distance with regard to the average pulse frequency that the successive repeaters are locked successively in phase by an incoming signal on the first repeater and follow these signals until these signals are stopped and the system is made available for a signal in the opposite direction, being therefore free from echos.

4. A repeater system according to claim 1 further comprising means responsive to signals arriving over said transmission path in one direction for conditioning said repeater for operation in the said direction.

5. A repeater system according to claim 1 wherein said oscillator is used as terminal equipment for transmission or receiving signals, iurther comprising means controlled by transmission and reception of signals therethrough for alternatively conditioning said oscillator for transmission or reception.

6. A repeater system according to claim 1, further comprising means responsive to operation of said oscillator for terminating said oscillations after a predetermined period.

7. A repeater system according to claim 1 wherein said oscillator comprises a dual stability system, and means responsive to received signals for synchronizing operation of said'oscillator.

8. A signaling system as in claim 1, wherein a plurality of tandem connected repeaters are provided, characterised in this that each said repeater comprises a vacuum tube, and that each said vacuum tube is duplicated by another tube in parallel whereby if one tube fails the second tube alone gives sufilcient gain to operate the next repeater in the series, the output of each said repeater being a simple relay device, whereby its operation is unchanged by the failure of the first tube in question.

' EDMOND MAURICE DELORAINE.

ALE-C 'HARIEY,REEVES.

Images