470,495

PATENT SPECIFICATION

Application Date: Nov 14, 1935. No 31591/35

Complete Specification Left: Sept 22, 1936.

Complete Specification Accepted: Aug 16, 1937.

-------------------------

PROVISIONAL SPECIFICATION

Improvements in or relating to Multiplex Signalling Systems

I, ALAN DOWER BLUMLEIN, of 32 Audley Road, Ealing, London; W5, a British Subject, do hereby declare the nature of this invention to be as follows:-

The present invention relates to multiple signalling systems.

Multiplex systems are known in which a plurality of telegraph channels are obtained through a single circuit. The circuit may be a land-line or a radio link for example. In these systems it is usual to employ mechanical distributors to connect the circuit in turn to each telegraph instrument at the transmitting and receiving point simultaneously. The whole cycle of connection of the line to all the receiving instruments on the circuit must occupy a time less than the time occupied by one signal dot.

The present invention deals with systems in which telephone frequency circuits may be connected in multiplex manner through a circuit such as a radio link or high frequency cable intended to cover a wide range of frequencies.

In such systems, on account of the comparatively wide band of frequencies required for each channel, mechanical distributors are unsuitable.

It is an object of the present invention to provide improved distributors capable of high speed operation.

According to the present invention there is provided a system of multiplex telephony or telegraphy, said system comprising means from feeding signals from a plurality of channels through a single circuit, the arrangement being such that, in operation, there is transmitted through said circuit a train of elementary signals interspersed with synchronising signals having an amplitude outside the amplitude range of said train of signals, each elementary signal being representative of the signal in one of said channels, and means being provided for utilising said synchronising signals to control switching devices for connecting said channels in turn to said circuit.

The wave transmitted through the circuit may comprise a plurality of groups of signals, each group comprising a plurality of trains of elementary signals, and each elementary signal being representative of the signal in one of the channels. Each train then comprises the signals from a set of channels and successive trains comprise the signals from different sets of channels. Each group of signals includes an elementary signal from every channel, the trains within a group are separated from one another by synchronising signals and the groups are separated from one another by further synchronising signals of different duration or different amplitude or of both different amplitude from the synchronising signals separating the trains within a group.

In one arrangement according to the invention the switching means comprise a plurality of relays which may be thermionic in character, the relays being adapted for connecting the channels in turn to the line and being arranged to be operated successively by means of controlling signals derived from delay networks.

In another arrangement according to the invention, the switching means comprise a cathode ray tube in which an electron beam is caused to scan a screen having a number of targets associated with the incoming or outgoing channels, the tube thereby serving as a distributor.

Further, according to the invention there are provided means for feeding a part of the signal from one channel into a succeeding channel at the receiving end of the circuit for the purpose of neutralising or reducing cross-talk, which may be introduced due to distortion of signal wave-form.

In order that the invention may be more readily understood, several embodiments thereof will now be described with reference to the accompanying diagrammatic drawings wherein.

Fig 1 shows the wave form of a multiplex signal which is obtained in another arrangement according to the invention,

Fig 2 illustrates the wave form of a multiplex signal which is obtained in another arrangement according to the invention,

Fig 3 shows one way in which thermionic valves may be controlled in a multiplex distributor,

Figs 4 and 5 show distributor circuits for use at the sending and receiving ends respectively of a multiplex link,

Figs 6 and 7 illustrate modifications of Fig 5,

Fig 8 shows a circuit for reducing or neutralising cross-talk between channels,

Figs 9 and 10 show arrangements employing cathode ray tubes at the sending end and receiving end respectively and Fig 11 shows a construction of signal plate suitable for use in the tubes of Figs 9 and 10.

It is assumed that it is desired to transmit signals from a plurality of channels (telegraph, telephone or other signalling channels) through a single circuit which may comprise a radio link or a cable capable of handling signals covering a wide range of frequencies. The circuit is associated at its two ends with distributors which are adapted to operate synchronously to connect the transmitting and receiving channels successively to the circuit. Thus if there are n channels, the distributors first connect the first channel to the circuit for a short period of time thus allowing a signal to pass in this channel, the distributors then switch over to the second channel to allow a signal to pass in the second channel and then to the third, fourth, etc., up to the nth channel. The distributors then switch back to the first channel and the process is repeated continuously. The frequency at which the distributors make complete cycles must be at least as high as the minimum frequency which causes no deleterious effect on the signals being transmitted. Thus if the channels are telegraphic, the frequency at which the distributors complete a cycle of change must be greater than the reciprocal of the time occupied by one signal dot.

Fig 1 shows the wave form of a signal in the single connecting circuit in a system according to the invention. The wave comprises a series of uniformly spaced synchronising pulses of which two are indicated by the references O and O1 in the figure. The zero line of the signals is shown by the line x____x. The signal from the first channel is indicated at 1, that from the second channel at 2, that from the third channel at 3, and so on the remaining channels. After the synchronising signal O1 the channels are again connected in turn to the circuit and the portions of the composite signal due to the signals in the first, second and third channels are indicated at 1 2 and 3 respectively. It will be seen that, during the interval between the corresponding points in these consecutive cycles of the distributors, the signal value in channel 1 has become less positive, that in channel 2 has changed from a negative value to a positive value, whilst that in channel 3 has remained unchanged. By employing circuits of suitable time-constant in the channels at the receiving end the series of pulses 1,1 Ö. May be caused to produce a signal of a wave-form substantially corresponding to the wave-form of the signal in channel 1 at the sending end. Similarly for the other channels. If the synchronising pulses occur at a frequency of 10,000 per second, then signals embracing frequencies up to nearly 5,000 cycles per second can be transmitted through the circuit on each channel.

If it is assumed that the multiplex signal of Fig 1 can be transmitted with sufficient sharpness through a circuit capable of handling a frequency range extending to one half the number of pulses transmitted per second (the number of pulses per second is equal to the product of the frequency of the synchronising pulses and the number of channels) then a frequency range of 5,000 n cycles per second will be required for n channels each requiring the transmission of frequencies up to 5,000 cycles per second. It will be seen that this is exactly equal to the minimum frequency range required by a multiple carrier signalling system radiating n single side-band modulated carriers, each having a side-band width of 5,000 cycles per second, it being assumed that the channels are close-packed and are separated by infinitely sharp filters at the receiving end. In practice the multiplex signal requires a greater frequency band than that mentioned above but a corresponding multiple carrier system also requires a range greater than the theoretical minimum value mentioned above.

In the multiplex system considered above the synchronising pulses 0, 01 are employed for maintaining synchronism between the distributors at the two ends of the connecting circuit. Such an arrangement is quite satisfactory provided that the number of channels is not too great, for example, not greater than 20. If the number of channels (n) is much greater than 20 then errors may be introduced in dividing the intervals between successive synchronising pulses into n equal periods. These errors will not in general be of equal magnitude and in the same sense at the two ends and the result may be incorrect connections to the channels. For example, it might happen that when the 31st channel was connected to the circuit at the transmitting end, the 30th channel was connected to the circuit at the receiving end. Even if the error is insufficient to cause a signal pulse to be wholly fed to an incorrect channel, the error may be sufficient for a part of a pulse to be fed to an incorrect channel.

If a large number of channels is to be employed, for example 200, the signals are preferably divided up as shown in Fig 2. In this figure, rectangles L, M, N are employed to denote trains of signals. The train L contains signals from channels 1,2,3Ö.p, the train M contains signals from channels p+1, p+2, p+3 Ö..2p and so on. Thus if there are any q trains each deriving signals from different channels, a number of channels equal to the produce p.q may be represented in the composite signal. After the end of the qth train a repetition of the sequence begins. The trains L, M, N are separated from one another by synchronising pulses O similar to those of Fig 1. A sequence of q trains of signals which will be termed a group of signals is separated from the succeeding group by a group synchronising signal which differs from train synchronising signals O. Thus the group train L differs from train synchronising signals O in that the group signal has a longer duration than the train signal.

When the group signals are employed to operate a primary distributor and the train signals are used to operate a secondary distributor, it is possible to connect 400 channels in rotation without dividing any interval between synchronising signals into more than 20 parts. Thus if a group comprises 10 trains (q=10), each train representing 20 channels (p=20), then the 199th channel is selected by counting from group synchronising signal OO up to the 9th train and then counting up to the 19th signal in this train. If it is desired to transmit telephony involving frequencies up to about 5000 cycles per second, then the group synchronising signals must have a frequency of 10000 per second; in the example considered in which there are 10 trains per group the train synchronising frequency is 100,000 per second.

It will be observed that the group synchronising signal OO functions for group L as a train synchronising signal. If desired, a train synchronising signal of normal form may be inserted between the signal OO and the beginning of the group L. In another arrangement, either the leading or trailing edge of signal OO is timed to take the place of a train synchronising signal. In yet another arrangement, the group synchronising signal OO is broken into parts in the manner known for frame synchronising signals in television systems, one of these parts being used as a train synchronising signal. It will be seen that the group synchronising signals OO and the train synchronising signals O bear a close resemblance to the frame and line synchronising signals used in many television systems.

For multiplex telephony by the method just discussed, group and train frequencies are much higher than the frame and line frequencies commonly used in television systems. However, by using synchronising frequencies comparable with those employed in television, it is possible to realise high speed multiplex telegraphy. By employing somewhat lower synchronising frequencies multiplex telegraphy over a circuit with a band width of the same order as that of ordinary telephone circuits may be obtained.

It will be seen that for the successful handling of signals such as are shown in Figs 1 and 2, it is necessary for the transmission circuit to have a uniform frequency response over the required range phase distortion. The requirements of the transmission circuit are therefore the same for the above purpose as for a television link. If low frequency components are not transmitted (these components may not be fed to the transmission circuit, or may be lost either in the transmission circuit or at the receiving end thereof) they may be re-established at or before the distributor at the receiving end with reference to the peaks of the synchronising signals or with reference to some other recurrent fixed amplitude (for example the zero period shown at y in Fig 1) in a manner well know in television systems and generally referred to as D.C. reinsertion."

Fig 3 shows a circuit in which a number of valves is operated in turn from a delay network. The delay network consists of series inductances d and shunt condensers a, b, c and is terminated by a resistance R. The condensers c are equal to the condenser b, less an allowance for the input capacity of the valves 5,6. The condensers a are equal to half b, less an allowance for the input capacity of the valves 7,8. The ratio of the inductances d to the full size condensers b is made equal to R. The cut-off frequency of the filter so obtained is made well above the highest working frequency, so that phase errors and mismatching as cut-off frequency is approached do not prejudice its operation. Other forms of filters may be employed such as those having resistances shunted across the inductances to minimise phase or reflection errors, and similarly a more accurate termination than a plain resistance may be employed.

At the sending end the control grids of valves 5,6,7,8 are coupled to the separate channels and the anodes of these valves are connected in parallel and coupled to the single transmission circuit. When the arrangement of Fig 3 is used at the receiving end, the grids of the valves are connected in parallel and fed with the multiplex signal from the transmission circuit and anodes are coupled to the separate channels. The grids are normally biased beyond anode current cut-off but a suitable negative pulse on the cathode of a valve turns it on for a period equal to the length of the pulse.

If a pulse is fed in at the left hand end of the filter, it will operate each valve in turn, and if the duration of the pulse is approximately the time delay between successive valves, each valve will be switched on in turn and will be switched off as the succeeding valve is switched on. A series of valves such as shown in Fig 3 can therefore be used for passing signals to a transmission circuit or accepting them from a transmission circuit, the relative timing of the switching to the various channels being obtained from the delay network.

Fig 4 shows an assemblage of hexode valves adapted to pass signals from six telephone channels and a synchronising pulse in succession to a line 10. The screen connections and source of heating current and bias potential are not shown. Any suitable type of feed well known in the art may be employed to obtain the feed currents and biases for these valves or for other valves shown in other figures.

The telephone channels are brought in to transformers 11 which apply telephone frequency potentials to the inner control grids 12 of the hexode valves. These valves have high resistances in their cathode circuits, so that the average current of each valve is about the same. The outer control grids 13 of the valves are biased negatively so that no signals pass to the line, and the anodes are connected to the line 10 through a delay network D. This network is not shown in detail but it is represented by a box terminated by a resistance R. The other end of the delay network is connected to the line 10 leading to the receiver or to a radio transmitter through any desired amplification or matching means.

For transmitting a signal of the type shown in Fig 1, having a repetition frequency of 10,000 per second, the total delay from the point of connection of the first valve to the point of connection of the last valve is made just less than 1/10,000 second. Once every 1/10,000 second the pulse generator (not shown) applies a positive pulse to the outer control grid of all hexodes simultaneously. The extreme right hand hexode is not connected to a telephone channel, but has its inner grid connected through a condenser 14 to the outer grid so as to produce a large pulse of current in the line which represents the synchronising pulse O. Simultaneously, the other hexodes apply pulses to the delay network representative of the voltage at the instant in the telephone circuits, since the available current which can be passed to the anode by the outer control grid, depends on the potential of the inner control grid. These signals arrive simultaneously at the various tapping points along the delay network. They there divide, half of each signal flowing left to the terminating resistance R, and the other half flowing right to the line. Since the delays of the various valves to the line are different, the signals will arrive in succession at the line and produce a wave-form as shown in Fig 1, having, however, only six channels. The signals flowing to the left to resistance R will not be reflected (assuming the correct termination) and will therefore not reach the line.

It will be realised that in Fig 4 the signals have been applied to the delay network and the synchronising pulse has been applied simultaneously to all valves, whereas in the description given of Fig 3 the synchronising pulse was applied to the delay network so as to make each valve operate in turn. In the arrangement of Fig 4 the capacity of the various anodes can form part of the network, but the drive for the control signal must be sufficiently powerful to charge the outer control grids of all valves simultaneously. In the arrangement of Fig 4 the synchronising signal is transmitted ahead of the group of signals with which it is synchronised. By placing the synchronising valve at the other end of the delay network it can be arranged that the synchronising signal follows the group with which it is generated. With a reasonably steady pulsing frequency, the two arrangements are indistinguishable. For a slightly irregular pulsing frequency the receiver can be made to operate with either arrangement.

Fig 5 shows a receiving system for the signals generated in Fig 4. A delay network D is connected at the left to the incoming line 15 and is terminated on the right by resistance R. As before, six hexode valves are connected to this delay network. In this case the outer control grids 13 go to the network. A tapping is taken from the right hand end of the delay network to the control grid of a valve 16, which is arranged to operate as a self-biasing separator valve. It is arranged that the synchronising signal appears in the positive sense on the grid 17 of this valve so that it passes current at each synchronising signal, but as is more fully explained in Specification No. 422906, self-biases itself to be insulating for all other signal amplitudes. The output of this valve is taken to a further amplifier valve 18 which has sufficient output to charge the inner control grids 12 of all six hexodes in the required period. As an alternative, the coupling condenser between the two valves may be made very small, so that the synchronising signal is converted to a sharp negative followed by a sharp positive pulse may be utilised by taking the connection of the hexode grids 12 from the anode or cathode of valve 18. This or similar means may be employed to change the synchronising pulse into a suitable pulse for actuating the hexodes. The correct timing of the pulse actuating the hexode may be obtained by altering the position of tapping 19 along the delay network D. For example, if there is delay in passing the signals through valves 16 and 18, or if the tail end of the synchronising pulse is used to operate the hexodes, the tapping 19 may be moved along the delay network towards the left as far as is required to give the correct timing. The various hexodes receive appropriate pulses from the line, depending on the position of their tapping points. Every time the synchronising signal operates valves 16 and 18, the hexodes become conductive and pass a signal to the anode circuits in accordance with the signals coming along the line. The anode current therefore consists of pulses recurring 10000 times a second, carrying the telephone currents as a modulation. If desired, a filter may be put between the anodes and the telephone circuits to eliminate the pulsating carrier.

As shown in Fig 5, the synchronising signal used to operate the receiving train of valves is the synchronising signal preceding the group of impulses. By moving the tapping 19 to the other end of the delay network, the synchronising pulse following the group of impulses may be employed.

It will be noted that provision is made for using a sharp or short synchronising signal to operate the receiving hexodes. It may be advantageous to use a shorter pulse on receiving hexodes than is used on transmitting hexodes, so that the middle portion of the elementary signal only is used at the receiver. Such an arrangement gives less liability to cross-talk due to slight mis-timing. Alternatively, short signals may be used at the transmitter so as to leave a slight dead gap between the signals of successive elementary channels.

Fig 6 shows a block diagram of a six channel receiving unit for receiving signals of the type transmitted by the circuit of Fig 4, but employing delay networks in both the line and pulsing leads to the valves. The relays which may be hexode valves and are designated as rectangles H, have their outer control grids connected to a delay network D1 in the line circuit. The delay of this network is not as great as that of the network D of Fig 5, and may for example be of the order of 1/20,000 of a second. The synchronising pulses are collected from a tapping 19, as shown before, a little way back along the network, to allow for time delay in the separating mechanism S. The actuating pulses from the separating mechanism (which may be similar to valves 16 and 18 in Fig 5), are passed along another delay network D2 (terminated by a resistance R2) which has a delay equal to the difference between the total delay required and the delay in D1. The pulse is passed along this second delay network in the opposite direction and it will be seen that the signals are separated by the relay valve H, as though the total delay had been put either totally in the line, or totally in the pulsing circuit. This arrangement has the advantage that the valve capacities are included as part of the shunt condensers in the delay network, which calls for less energy from the separating mechanism S to actuate the six hexodes than is required in the arrangement of Fig 5. Such a reduction in power is of considerable advantage in the case of a system employing a greater number for example 20 hexodes. Similarly, the delay network D1 in the line circuit prevents the sum of the grid capacities of the hexodes being bridged directly across the line circuit.

Fig 7 shows a block diagram of a system for receiving a wave of the type shown in Fig 2. There are shown at S, separators employ the trailing edge relay valves H. The signals from the line 15 are passed to a number of delay networks D having a delay equal to the time period of one train of signals. The signals from the line 15 are also passed through a group pulse separator GS. This separates the group pulse (pulse OO in Fig 2) and passes a group pulse down the group delay network GD. This has a total delay equal to the total period of a group and has tappings along its length at time intervals equal to the time between successive trains.