US2405280A

US2405280A - Signal correction circuit

Info

Publication number: US2405280A
Application number: US585526A
Authority: US
Inventors: Alda V Bedford; Karl R Wendt
Original assignee: RCA Corp
Current assignee: RCA Corp
Priority date: 1945-03-29
Filing date: 1945-03-29
Publication date: 1946-08-06
Anticipated expiration: 1963-08-06

Description

Aug. 6, 19%.

A. V. BEDFORD ET AL SIGNAL CORRECTION CIRCUIT Filed March 29, 1945 5 Sheets-Sheet 1 All ll Ann.

IN VEN TORS Huy/7 l/.EDFRD Kem l?. WEA/77 By @QM Aug. 6, 1946.

A. V. BEDFORD ET AL S IGNAL CORRECTION CIRCUIT @fram/5y Aug. C, 1946.

A. V. BEDFORD ET AL SIGNAL CORRECTION CIRCUIT Filed March 29, 1945 5 Sheets-Sheet 3 IA .Y

wmf/Heime #fz ,a Te/ssieip Hyp) INVENTCRS HLM? MEDFa/w By QM Aug 5, 1945; A. v. BEDFORD ETAL 2,405,280

S IGNAL CORRECTION CIRCUIT Filed Mal-n 29, 1945 5 sheets-sheet 4 A. v. BEDFORD ET Al. 2,405,280

Aug. 6, 1946.

l SIGNAL CORRECTION CIRCUIT 5 Sheets-Sheet 5 Filed March 29, 1945 INVENTORS irme/vif Patented Aug. 6, 1946 NT OFFICE SEGNALI CGR-RECTION CIRCUIT Application lillarch 29, 1945, Serial No. 585,526

(Cl. 17g- 1.5)

l1 Claims.

The present invention relates to wave transmission systems and more particularly to an irnproved method of and means for improving the low-frequency fidelity of received communication signals by means of an improved D.C. insertion circuit in, a synchronized communication system..

The invention, by way of example, will be described hereinafter as an improvement in a secret telecommunication system of the general type described in the copending U. applica tion of Alda V. Bedford, Serial No. 36,630, filed May 20, 194C/l. Said copending application discloses a system wherein, for example, a speech signal comprising a complex wave S is modied by means of a coding signal comprising a comm plex wave K in a manner whereby the instantaneous ordinates of the resulting coded signals are the product SK of the corresponding instantaneous ordinates of the speech signal and the cod" ing signal. The resulting unintelligible coded signals are transmitted by any conventional means to receiver wherein thev coded signals are combined with decoding signals generated in the receiver and having instantaneous ordinates corresponding to the reciprocals of the corresponding instantaneous ordinates of the coding signal component of the transmitted signalY The final signals. therefore, are derived from the product of the transmitted signal SK and the decoding signal The coding and decoding signal generators at the transmitter and receiver, respectively, are synchronized by a unique system, wherein synchronizing pulse signals, each. comprising a first signal pulse immediately followed by a second signal pulse of opposite polarity, are superimposed upon the coded signals SK at predetermined intervals. At the receiver the reversal in polarity between the two synchronizing pulses is employed to synchronize the decoding wave generator.

The instant invention comprises an improved method of and means for improving the lowfrequency delity of received coded signals by means of a D.C. insertion network combined with a novel split/channel filter system disclosed in the copending application of Alda V. Bedford. Serial No. 583,343, filed March 17, 1945. In the instant system, the coded signal is corrected for spurl-ous D.-C. and low-frequency components by deriving control signals corresponding to the instants when the coding signa] component K crosses its A.C axis. Since the coded signals have instantaneous ordinates which are the product of the corresponding instantaneous ordinates of the communication signal S and the coding signal K, the control signals corresponding to the times when the coding signal K crosses its A.C.

. proved D.-C.

axis may be employed to control a D.C. insertion circuit, for correcting the product signal SK to cross its A.C. axis at the same instant. The split-channel filter system is utilized to remove spurious high-frequency signal components which are introduced by the D.C. insertion network.

Among the objects of the invention are to provide an improved method of and means for improving the low-frequency fidelity of received communication signals in a synchronized communication system. Another object of the invention is to provide an improved method of and means for inserting D.C. and low-frequency signal components which have been lost during transmission of a communication signal. A further object of the invention is to provide an iminsertion circuit in combination with a synchronized secret telecommunication system. An additional object is to provide an improved secret telecommunication system including a novel D.C. insertion circuit combined with a split-channel filter network. Another object is to provide an improved secret telecommunication system including a receiver having, in combination, a novel D.C. insertion network synchronized with the telecommunication signals and a unique split-channel lter system for improving the low-frequency fidelity of received sighals,

The invention will be described in greater detail by reference to the accompanying drawings of which Figure l is a block schematic diagram showing the basic elements of a complete telecommunication system including the instant invention, Figure 2 is a partially schematic, more detailed, block diagram of a complete telecommunication system including the instant inven tion, Figures 3 and 4 are groups of graphs which are explanatory of the operational characteristics of the circuit of Figure 2, Figure 5 is a schematic circuit diagram of a signal multiplier network forming one of the components of the circuits of Figures i and 2, Figure 6 is a schematic circuit diagram of a signal reciprocal network forming another component of the circuits of Figures 1 and 2, Figure 7 is a schematic circuit diagram of the D.C. insertion and split-channel filter networks comprising the novel circuit elements of the instant system, and Figure 8 is a group of graphs explaining the operation of the circuit of Figure 7. Similar reference characters are applied to similar elements throughout the drawings.

BAsIc SYSTEM Coding transmitter Referring to Figure l, the circuit to be described includes transmit-receive switches 49, 5l and el which control the circuit for transmis sion of coded signals when the switches are in the positions T1,T2, and T3, and alternately control the system for receiving and decoding such signals when the switches are in the positions D1, D2 and D3, respectively. In the transmitting condition, a complex coding wave generated by the code wave generator 2 and code wave synchronizing network 2 is applied to one input circuit of a wave multiplier circuit 59. Speech signals, derived from a microphone 55, are applied to a second input circuit of the wave multiplier 59 whereby complex coded. signals SK, having instantaneous ordinates which are the product of the instantaneous ordinates of the speech signal S and decoding signal K are applied to a signal mixer circuit 63. Synchronizing signals, each comprising a pair of pulses of opposite polarity occurring at regular spaced time intervals, are superimposed upon the coded signals SK in the signal mixer circuit 63. Peak values of the mixed coded and synchronizing signals from the mixer circuit 63 are limited in a limiter circuit IS and are applied to modulate the conventional radio transmitter 8l connected to an antenna 83.

BASIC SYSTEM Decoding receiver When the circuit is employed for the reception of coded signals, the signals received on the receiving antenna 84 are applied to a conventional receiver 85 which detects the coded signals SK and the synchronizing pulses from the signal carrier. pulses thence are applied to a novel D.C. clamp circuit IBO, comprising a D.C. insertion network and a split-channel lter system which will be described in detail hereinafter. The D.C. clamp circuit I!) is controlled by square Wave pulses derived from the coding wave generator 2 through limiter circuits of a reciprocal wave network II. The square wave control pulses applied to the D.C. clamp circuit Iliii are characteristic in time of the intersections of the coding `wave K with its A.C. axis. The D.C. clamp circuit corrects D.C. and low-frequency spurious components of the received signal at each instant that the coding wave K crosses its A.C. axis, whereby the received coded and synchronizing signals are corrected in amplitude at closely spaced time intervals. Y

The corrected signals derived from the D.C. clamp circuit Iill4 are applied to a differentiating circuit 8'! which selects the synchronizing pulses and controls the code wave synchronizing network 2' for synchronously pulsing the code wave generator 2. The corrected signals also are connected to a synchronizing pulse blanking circuit 93 which removes the synchronizing pulses from the coded signals and applied the blanked, corrected SK signal to the wave multiplier circuit 59. The coding wave K derived from the code wave generator 2 is applied to the reciprocal circuit II which converts the instantaneous signal ordinates to a value The reciprocal signals alsoare applied to the wave multiplier 59 whereby the product of signals sKXIl or s' is derived and applied to a reproducer ID3. The

received decoded signal is designated S' because of inherent distortion in transmission and reception. The details of the various circuits described heretofore and the Voperational characteristics thereof will be described in greater de- Vtail by reference to subsequent figures of the drawings.

The coded signals and synchronizing l COMPLETE SYSTEM Coding wave generator Referring to Figure 2, the coding wave generator employed for both transmitting and receiving coded speech signals comprises a conventional free-running multivibrator circuit I which generates pulses at a rate, forl example, of one hundred pulses per second. A typical multivibrator of this type, the frequency of which may be controlled by recurrent applied control pulses, is described in U. S. Patent 2,266,526, granted to E. L. C. White on December 16, 1941. It should be understood that pulses of either polarity may be applied in any known manner to key the multivibrator, and that similarly output pulses of either polarity may be derived therefrom. The generated pulses are applied to the input of a conventional delay network 2 comprising a plurality of

series inductors

3, 5, 6, 9, II and a plurality of shunt capacitors 4, S, 8, Ill, I2, I4. The remote terminals of the resultant pulse delay network 2 are terminated by a resistor I 3 matching the surge impedance of the network. It should be understood that the delay network 2 may include a relatively large number of filter sections as indicated by the dash lines interconnecting the iilter sections l', 8 and 9, Ii), and that equalizers and booster amplifiers may be inserted in the delay network at desired points to maintain pulse amplitude relations at optimum values.

Pulses applied by the multivibrator I to the input of the delay network 2 provide similar pulses at the junction of each of the succeeding

series inductors

3, 5, 6, 9, II wherein each succeeding pulse is delayed a predetermined amount with respect to pulses occurring at other prior network terminals. A complex coding wave thus may be obtained in response to each pulse applied to the delay network by combining in either polarity differently delayed pulses derived from a plurality of such predetermined points along the delay network.

Separate isolating resistors l5, Il, I9, 2|, 23, 25 each have one terminal connected to different points along the delay network, and have their remaining terminals connected to separate movable contacts of a plurality of singleepole doublethrow switches 21, 2Q, 3i, 33, 35, 31. The corresponding iixed contacts of the several switches are connected together to provide two lines 35i, 1 I, which are terminated through resistors 43, 45, respectively, to ground. The remaining terminal of the line 39 is connected through a coupling resistor 41 to one iixed contact T1 of a first transmit-receive single-pole, double-throw switch 49. The remaining terminal of the second line 4I is connected through a polarity-reversing amplier 5I and a second coupling resistor 53 to said rst fixed contact T1 of the first transmit-receive switch 49. Thus each of the pulses per second, dervied from the multivibrator I and applied to the input of the delay network 2, provides a plurality of pulses of either polarity occurring at predetermined intervals during each one-hundredth second period, as determined by the points of connection to the delay network and the arrangement of the switches 21, Z9, 3l, 33, 35, 3l. Therefore, a very complex coding wave may be applied to the first iixed contact T1 of the first transmit-receive switch @97, merely by selecting the desired arrangement of the pulse selecting switches. It should be understood that the total delay provided by the pulse delay network should be at least slightly less than the pulse period of the multivibrator l in order that only one pulse may be traveling along the delay network at any predetermined instant.

In the typical secret telecommunication system of the general type described in applicants1 copending applicati-on identiied he etofore, the coding signal generator includes a delay network having 80 sections and a plurality of seamen `al switches which may be preset to any desired code and selectively actuated by a clock mechanism to change the code continuously or at predetermined desired intervals. Identical coding signal generators are employed both the tra.. mitter and receiver in such a secret telecommunication system. By means of simple transmit-receive switches the coding signal either is combined with the speech signal for transmitting a coded wave, or l reciprocal values` of the coding signal are derived from a reciprocal circuit responsive to the coding signal generator and are combined with the received coded signal to decode said received. signal. Much oi the decoding apparatus including the generator for the code signal is identical to the coding apparatus. Hence, by means oi' the simple transmit-receive switches, the various elements of the apparatus may be employed at different times for dual purposes, in a single unit for either transmitting or receiving the coded signals.

Coding transmitter Referring to Figure 2, the system may be employed as a coding transmitter by switching the movable contacts of each of single-pole` double-throw transmit-receive switches included therein to engage the liked contacts T1, T2, T3, T4, T5, Te corresponding to the transmit condition. Signals derived, for example, from a microphone 55, which may be fed through a speech amplifier, not shown, are applied through a second transmit-receive switch El to one input cir of a wave multiplier 5, which will be described ence to Figure 5 of tl A.,

generator described heretofore, are through the *first switch lid., to a `second input circuit of said wave multiplier whereby coded signals SK having instantaneous ordinates corresponding to the products of the corresponding 1nstantaneous ordinates of the speech signal S and the coding signal K are applied through a third transmit-receive switch tl to one input circuit of a first mixer circuit 63, which may comprise any conventional network wherein applied signals are combined algebraically.

Transmitter synchronizing pulse operator ative square wave pulse c is applied through a fifth transmit-receive switch T3 to a second input circuit of the rst mixer circuit 63, and is applied through a sixth transmit-receive switch T5 to key a third multivibrator il which generates a positive square wave pulse indicated by the graph d of Figure 3. It will be understood that the positive square wave pulse d will be initiated at the termination of the negative square wave pulse c in a manner well known in the multivibrator art. The positive square wave pulse d is applied to a third input circuit of the mixer circuit 93 whereby the coded signal SK, the negative square wave pulse c and the positive square wave pulse d are combine-:l to provide a communication signal includin-g-the coded/wave SK andY the synchronizing signal comprising a negative sanare wave pulse immediately followed by a positive square wave pulse. It should be understood that, if desired. the synchronizing signal may comprise a positive pulse followed by a negative pulse since multivibrators may be keyed by, and can provide, pulses of either polarity, providing proper connections thereto are provided in a manner known in the art. rhe combined coded signal and synchronizing signal derived from the mixer 63 will have a wave form, for example, of the type illustrated in graph f of Figure 3, including the pulses I, I, shown in dash lines.

A pulse derived from the third multivibrator "il also is applied to key the rst multivi rat-or l to generate a positive square wave pulse e, illustrated in Figure 3, which is applied to the input l the delay network to initiate a succeeding pulse which will be progressively delayed along the delay network. Since the rst multivibrator i is keyed by the pulse from the third multivibrator ll immediately preceding the time for the generation of a normal pulse by said nrst multivibrator, it will be seen that the coding wave generator will be self-running, and will be main tained at a substantially constant treuren since the pulse rate therethrough will be sub-etantially dependent upon the time of the successive pulses applied to the delay network 2. if for any reason the first multivibrator i is not properly keyed by the third multivibrator fr". the irst multivibrator will merely generate a pulse e which will be applied to the delay network 2 at a slightly later interval. The slightly delayed pulse upon reaching the seventy-ninth tap of the delay network therefore will key the second and third multivibrators in the manner described heretofore and provide a new set of synchronizing pulses which will actuate the rst multivibrator l in 'synchro-nism thereafter.

The coded signals SK combined with the syn-- chronic-ing pulses c and d are applied to second limiter 'iii whereby the high amplitude portions l of the synchronizing signal are clipped to a maximum level 1I indicated by the dash lines in grap-h f of Figure 3. The thus limited combined coded and synchronizing signals are applied as a communication signal to a conventional radio transmitter 3l which includes a transmitting antenna 83.

Coding signal rocciosi In order to convert the circuit thus described to operate as a coded signal receiver, the movable contacts of each of the transmit-receive switches "i3 and are switched to the corresponding fixed contacts D1, D2, Da, D4, D5, D5, corresponding to thA re eive condition. The combined coded signal and synchronizing 7 signals transmitted from the transmitter 8| are smeared and phase-shifted somewhat due to non-linearity in transmission to resemble the solid portion a: of the graph f of Figure 3, and as received by means of a conventional radio receiver 85 are applied, through the equalizer network 86 and D.C. clamp circuit lei), described hereinafter, to a conventional wave differentiating network 81 which may be of any type well known in the art. For example, a wave may be differentiated by applying it to a network comprising a small series capacitor and a shunt resistor. The transmitted signal f of Figure 3 after being differentiated at the receiver resembles the graph g of Figure 4 wherein a relatively large -pulse Q occurs at an instant correspondingto the reversal in polarity between the received synchronizing negative and positive pulses and wherein low-frequency components are substantially removed from the pulse Q. It should be understood that instead of differentiating the received signal, it may be treated in any other known manner to derive a control pulse in response to the reversal in polarity of the negative Y and positive synchronizing pulses.

The receiver first multivibrator l being free running, as described heretofore, the delay network 2 will provide recurrent pulses at its seventy-eighth tap which will be limited by means of a third limiter 8S to provide limited pulses represented by the graph h, of Figure 4. The thus limited pulses h are applied `to key a fourth multivibrator 9| which generates a relatively long blanking pulse illustrated in graph' i of Figure 4. The long blanking pulse iis applied to a blanking circuit 93 which blanks out portions of the received signal, as will be explained in greater detail hereinafter.

Receiver synchronizing circuits Similarly, each of the recurrent pulses derived from the eightieth tap of the delay network 2 are applied to a fourth limiter 95 which clips th'e upper portion of the applied pulse as explained heretofore with respect to pulse b, to provide a short pulse illustrated by graph 7' of Figure 4, The limited pulsey is applied through the fourth transmit-receive switch 69 to key the second multivibrator Il to provide a relatively long positive square wave pulse lc. It will be noted that the positive pulse 7c is of relatively longer duration than the negative pulse c previously described as generated by the second multivibrator 'il when said multivibrator is employed in the transmitting circuit. The different pulse polarity and duration may be accomplished in any well known manner by changes provided in the multivibrator circuit constants and the connections thereto, when the multivibrator is switched from th'e transmitting to the receiving condition.

The positive square wave pulse 1c derived from the second multivibrator l! is applied throughthe fifth transmit-receive switch 'i3 to a second mixer circuit 91, to which also is applied the differentiated wave y derived from the differentiating circuit 8l. The thus mixed signals illustrated by graph l of Figure 4 include a pulse peak Z which corresponds in time to the occurrence of the large positive pulse Q 0f the differentiated received wave g. As explained heretofore, the pulse Q corresponds to the reversal in polarity of the received synchronizing negative and positive pulses. The wave l derived from the second mixer circuit 97 is applied to a fth limiter 99 which' clips the mixed signal at a level z to pro- 8 vide in its output circuit a short somewhat triangular pulse, illustrated by graph m of Figure 4.

The triangular pulse m is applied throughthe sixth transmit-receive switch 15 to key the third multivibrator ll to provide a positive pulse represented by graph n of Figure 4 which is applied to key th'e first multivibrator l as described heretofore with respect to the pulse d in the transmitting network. It should be understood that, if desired for extremely precise synchronism, the pulse m may be changed from triangular to square Wave shape by clipping at a low-level and then by amplifying the clipped lower portion of the pulse in a manner known in the art. The pulse n therefore causes th'e first multivibrator vI to generate a positive pulse-o which iseappliedY to the delay network 2 in the same manner as described heretofore with respect to the positiv pulse o of the transmitting network.

As explained heretofore with respect to the operation of the multivibrator circuits in the transmitting condition, if the circuit falls out of synchronism, the various multivibrators will provide pulses at somewhat increased time intervals until such time as a synchronizing pulse occurs at a proper instant to pull all of the multivibrators back into synchronism. Since pulses are derived from the delay network 2 at intervals of the order of .01 second, it is apparent that the various circuits will fall into synchronism in a relatively short time which seldom will exceed one full second.

Due to phase distortion in the transmission or radio circuits interconnecting the transmitter and receiver units, it is possible that the effective time of occurrence of the received synchronizing pulses will vary in different receivers with respect to the received coded speech. To correct for such variations, the circuit constants of the third multivibrator Tl may, in any known manner, be altered in the receiving condition so that the width of the pulse n may be varied to provide keying of the first multivibrator l at the precise desired instant. The manner of varying .the circuit constants of multivibrators to provide pulses of desired polarity and duration in response to predetermined applied keying pulses is well known in the art.

Signal decoding system The received signals derived from the radio receiver are applied, through the phase equalizer network 86 and the D.C. clamp circuit |00, to the input of the blanking circuit 93 which interrupts the received coded signals during the occurrences of the recurrent blanking pulses i, whereby the transmitted positive and negative synchronizing pulses may be removed from the received coded signal. This condition obtains when the coding signal generator of the receiver is in synchronism with the transmitter coding signal generator, since the fourth multivibrator 9! is responsive to pulses derived from the seventy-eighth tap on the delay network 2. Blanking circuits are Well known in the art. They may comprise, for example, a push-pull amplifier for the signal, arranged so that the blanking pulses i are superimposed on the grid-cathode circuits so that both tubes are simultaneously driven to cut-off during the blanking period. The thus blanked received signals comprise the transmitted signal components SK which are applied through the second transmit-receive switch 51 to one of the input circuits of the wave multiplier 59.

9 Similarly, the coding signals K generated by the receiver coding generator are applied to the input circuit of a reciprocal circuit IGI, which will be described in detail hereinafter by reference to Figure 6 of the drawings. Signals derived from the reciprocal circuit il!! will have instantaneous ordinates corresponding to the reciprocal values of the instantaneous ordinates of the synchronized coding wave K generated in the receiver. rEhe reciprocal wave and SK applied thereto, the output signals applied through the third transmit-receive switch SI to a reproducer E93 will be substantially characteristic of the original speech modulation signals S. The signals applied to the reproducer H53 have been indicated as S since some phase distortion is inherent in the various circuits described and especially in many radio transmission circuits. It should be understood that the signals S derived from the third transmit-receive switch 6l may be applied to actuate any other desired type of utilization apparatus, n

not shown,

Signal 'mult/plier Figure shows a 'typical wave multiplier circuit forming a portion of both the coding wave transmitter and receiver circuits described heretofore with reference to Figures 1 and 2 cf the drawings. This multiplier circuit is described and claimed in the copending U. S. application of Alda V. Bedford, Serial No. 517,967, led January l2, 1944, and assigned to the same assignee as the instant application. The circuit utilizes the property of well known electrical devices which provide an instantaneous output voltage which is proportional to the square of the instantaneous input voltage over a reasonable voltage range in a single polarity. Such circuits or devices will be referred to as squaring circuits, and will be designated as V where referred to hereinafter.

l'n the preferred form of the multiplying circuit, the waves S and K, to be multiplied, are added together with four dilerent polarity combinations and squared in four different signal channels. Then the four squared signals are added together with suitable polarities to obtain the product SK in the output circuit of the multiplier network, as will be l lustrated by the following equations:

rality of small copper oxide rectiers V1, V2, Vs, VA, known commercially as Varistors Because of the particular variable resistance characteristics of the Varistor, the current therethrough is substantially proportional to the square of the applied voltage over a reasonable range of applied voltage of a single polarity. The multiplier network 5S is shown as including a rst triode thermionic tubo i having its grid electrode connected to the movable contact of the first transmit-receive" switch 49, whereby signals characteristic of either the coding wave K or the reciprocal thereof may be appiied to the tube grid cathode circuit. A second thermionic tube H3 has its grid electrode connected to the movable Contact of the second transmit-receive switch 51, whereby either the speech signals S or the blanked, received signals SK may be applied to the tube grid-cathode circuit. The operation of the circuit will be explained hereinafter with the switches 49 and 57 in the transmitting position whereby the signals K and S, respectively, are applied to the grid-cathode circuits of the tubes IH and H3. Push-pull output signals are derived from each of the tubes by means of connections to the corresponding tube 'anode and cathode circuits as indicated in the drawings.

In order that the desired sum voltages be obtained, the signals S and K are applied to a network of resistors in the following manner; Signals S andK respectively traverse resistors Rs and R7 to provide a signal proportional to (S-i-K) at point (S-i-K); the signals S and -K respectively traverse resistors R5 and R12 to provide signal (1S-K); the signals -S and -K respectively traverse resistors R10 and R11 to provide signal (-S-K) and the signals -S and K traverse respectively resistors R9 and Ra to prosived signal (-S-l-K). Thus, at each of the four designated junction points, a sum of voltage is obtained as indicated in the circuit diagram. The network also includes resistors R12 and R15 leading respectively from points (S-K) and (-S-l-K) to ground, and resistors R11 and R15 leading respectively from points (S-l-K) and (-S-K) to the positive terminal of the source of bias voltage which is applied through a voltage-reducing resistor R13. An SOOO-ohm resistance has been found satisfactory for the resistors R13, R14, R15, and R16, while 100,000-0hm resistance has been selected as the value of resistors R5, R6, R7, Re, Re, Rio, R11 and R12.

The sum voltages at the four points of the network are applied with bias voltage A and -A to the four Varistors V1, V2, V3, and V1 respectively, all of which control the current through the common load resistor R17 to provide thereacross the product output voltage SK. The output across R17 is proportional to the sum of all the voltages which would have been generated if each It will be understood that the term A in the 7" Varistor had supplied current to a separate reabove equations is the D.C. bias added to the A.C. waves to cause all of the signal amplitude variations to have the same polarity with respect to the squaring devices.

The squaring circuit illustrated employs a plusistor, as indicated by the foregoing squaring equations. It is to be noted that the Varistors V2 and V4 are connected with opposite polarities from the Varistors V1 and V3, so that the D.C. bias voltage must be different, By reference, re-

11 Y spectively, to the third and fourth equations it will be seen that the values (-S-i-K-A) and (S-K-Al are each preceded by another minus sign and included in brackets before squaring to indicate properly mathematically the effect of the reversed connection on these two Varistors. These ve equations show that, ideally, only the desired voltage SK is produced across the output resistor R17.

For compensating for small dissimilarities in the Varistors and other circuit elements, it has been found desirable to provide variable res'mtors R1 and R3 connected as voltage dividers in the anode circuits of the tubes and H3, respectively for adjusting the relative amplitudes of S and -K AWhile in the foregoing the term multiplying circuiv has been used to dene the circuit, it will be seen that the circuit actually is a sort of modulator which is completely balanced in the sense that only the side band frequencies are produced, -while the input frequencies and the harmonics 4thereof are suppressed.

The output signals SK derived from across the output resistor R17 are applied to the movable contact of the third transmit-receive switch 6|, whereby they may be selectively applied to either the reproducer |63 or to the rst mixel` 63, depending upon the desired operation of the circuitsof Figures 1 and 2.

Signal reciprocal circuit ,The reciprocal circuit shown in Figure 6 of the drawings is described and claimed in the copending application of Carl A. Meneley, Serial No. 484,304, iiled IApril 23, 1943, and assignedto the same lassignee as the instant application. In this circuit instantaneous reciprocal values of an applied coding wave K are obtained by means of an electrical network in which the wave K is clipped on both its positive cycle and on its negative cycle to produce a substantially rectangular wave, and

' inwhich the wave K and the -rectangular wave are added together with one of them reversed in polarity, preferably after the peaks of the posi- -tive and negative cycles of the wave K have been squashed or attened somewhat. 4The circuit includes no appreciable capacitive or inductive reactances (the blocking capacitors in the circuit presenting negligible impedance) and, therefore, provides the reciprocal of substantially any applied signal waveform regardless of its frequency components.

Referring to Figure 6, the graph G represents a typical coding wave K which is applied to the input terminals |2| of the circuit. The graph J represents the reciprocal Wave which is the sum of the flattened coding wave K, represented by the graph H, of reversed polarity, and of the rectangular wave shown in graph I. The squashed or flattened wave H may be obtained by passing the wave G through a circuit that changes its resistance `with a change in `applied voltage produced kby clipping the positive and negative cycles of the flattened wave H at the voltage levels U and L respectively, for example, close to the A.-C. axis of the signal, and then by amplifying the clipped signal.

The wave K applied tothe input terminals |2| may, ifA desired, be amplified by meansof an 'amplifier tube |23 to provide a peak-topeak The rectangular Wave I maybe 4 12 amplitude, for example, of the order of G0 volts. The amplied K wave then is applied through a blocking capacitor |5| and a resistor |52 to a copper oxide rectifier unit |53 which functions as a non-linear resistor having the property of decreasing in resistance as the applied Voltage increases. The resistor |52 is of high enough resistance so that the driving source for the nonlinear resistance unit |53 is of high impedance whereby there is only a slight variation in the current flow through the unit |53. The unit 53 may consist of a pair of copper oxide rectiers |53a and |53b connected to conduct current in opposite directions.

rI'he voltage appearing across the non-linear unit |53 is the voltage wave H, which is the wave K having a flattened wave form. This voltage, which is amplied by a cathode biased vacuum tube |54, appears across an anode resistor |56 and a portion of the anode resistor LiL-|51 of a second amplifier tube |58.

The rectangular wave I is produced, in this particular example, by applying the output of the tube |56 through a blocking capacitor |59 and a high impedance resistor |5|to a pair of diodes |62 and |53, which are connected to conduct in opposite directions. Resistors |54 and |56, of comparatively low resistance, are connected in series with the diodes |52 and |53, respectively. A biasing voltage drop for opposing current flow through diode |63 is produced across the resistor 55 by connecting a source of voltage (not shown) thereacross, a resistor |51 being in series with the voltage source. The diodes |52 and |63 clip the applied wave H symmetrically about its A.C. axis, because a voltage which causes current iiow through the diode |52 and resistor |64 is built up across the capacitor |59 by the positive cycle pulses owlng through the diode |63. Thus, the diodes |62 and |63 become conducting on alternatecycles when the signal voltage exceeds the D.C. voltage drop across the resistors |64 and |65, respectively, The resulting rectangular wave I is amplified and reversed in polarity by the tube |58. The wave I and the flattened wave H add in the portion of the anode resistor WiL-|51 that is common to the tubes |54 and |58 to produce the desired reciprocal wave 1 /K shown in graph J.

If the wave H is flattened correctly, and if the Waves H and I are added with the correct relative amplitudes, the resulting signal will be substantially a true reciprocal of the wave K. The only substantial departure from a true reciprocal signal Will be where the wave K crosses the A.C. axis. Here the reciprocal value is infinity whereas the maximum amplitude of the wave l /K necessarily has a finite limit. The waves H and I may be mixed with the correct relative amplitudes by adjusting a variable tap |1| on the anode resistor |51-|5'|. The correct shaping of the flattened wave H may be obtained by selecting a non-linear resistor unit |53 having a suitable voltage-resistance characteristic and by ad- "justing the value of the variable resistor |52.

As stated heretofore, the above-described reciprocal circuit is purely resistive so that its opform. There are various ways of determining when the circuit has been adjusted to give substantially a true reciprocal. One way is to conneet the reciprocal circuit into the signalling systems of Figs. 1 or 2 and, while transmitting speech or music, adjust the resistor |52 and the variable tap at the receiver until the speech or music has a minimum of distortion.

It should be understood that oppositely-connected diodes may be substituted for the copper oxide rectifiers |530 and |5317, described hereto.. fore. When properly biased, the two diodes should be operated along the lower knee of their operating characteristic and in the proper region to shape the wave K in the desired manner to provide the iiattened Wave H.

It will be understood that the device not limited to the particular circuits illustrated since the waves H and I may be derived from the wave K in various other ways, and since the two waves may be combined by means of a variety of other circuits.

The square wave signal I also is applied through a suitable coupling capacitor l'lf to output terminals |15, which are connected to the D.C. clamp circuit |09.

D.-C. clamp circuit Referring to Figures 7 and 8, the received coded Wave SK including the synchronizing pulses is a product of the speech wave S and the code wave K, with the synchronizing pulses superimposed thereon at regularly recurring intervals. Since the coded wave SK is a product and includes a multiplication factor comprising the coding wave K, the coded wave SK should have zero ordinates at the same times that the coding wave K has zero ordinates. If the low-frequency components of the coded wave SK are attenuated or shifted in phase, those portions of the wave SK occurring at times corresponding to zero ordinates of the coding wave K will be displaced from their zero values.

The D.C. clamp circuit l iid, comprising a component of the receiver network, operates to displace the distorted received coded Wave SK upward or downward as required to have zero ordinate values at each instant when the locally generated coding wave K has zero ordinate values. This tends to resto-re the low-frequency components of the communication signal including its D.-C. components. By the same action, thev D.-C. clamp circuit also tends to remove any spurious low-frequency components, such as come from the power supply, surges from switching in the power supplies, etc. Another possible source of low-frequency surges is high pulses of radio noise, which may cause a momentary flow of grid current in some amplifier tube of the systern. Such grid current iiow would charge the associated coupling capacitor, and the charges would leak oii relatively slowly through the associated grid leak. A spurious signal comprising relatively long duration, low-frequency surges would be generated by these narrow noise pulses. Since the low-frequency noise surges are converted to objectionable spurious frequencies by multiplication with the decoding wave l/K, the use of the D.C. clamp circuit effectively decreases the eiTect of this type of interference. Also by improving the overall low-frequency fidelity, the clamp circuit improves the quality of the transmitted speech.

In the circuit of Fig. 7, the square coding Wave p derived from the terminals |75 of the reciprocal circuit IGI (see Fig. 6) are applied to a phase inverter tube Ill to derive the inverted wave q, shown in the graph q of Fig. 8. (Wave p is another View of the wave I of Fig. 6.) 'Ihe opposite polarity square waves p and q are effectively limited, and then added through a pair of diodes |19 and IBI to derive the triangular wave shown in the graph r of Fig. 8. The triangular wave 1' is differentiated by means of a small series capacitor |83 and a shunt resistor |85 to derive the pulsed wave shown in the graph s of Fig. 8.

The synchronizing signal blanking pulses shown in graph i of Fig. 3 or in graph t of Fig. 8 are applied to the grid of a blanking tube |81 to blank out all signals occurring during a short Vand including the synchronizing pulses. VThe blanked pulse signals s are applied to limiter tubes |89 and I9! which clip the pulses of the signals s at the levels |93 and |95, whereby pulses co1'- responding to the graph u are applied in opposite polarties to a pair of oppositely-connected diodes |91 and |99. The pulses u and u commence shortly before the instant at which the coding wave K crosses its A.-C. axis, and they end at precisely the instant at which the wave K reaches a zero value, as indicated by the two vertical dot-dash lines in Fig. 8. Since the diodes are connected in opposition and are pulsed by the pulses u and -u, the common connection of the cathode of the diode |91 and the anode of the diode |99 is effectively brought to ground potential at the termination of each of said pulses, as indicated by the hypothetical switch 2li l shown in dash lines.

The received coded and synchronizing signals derived from the receiver 8d, are applied through a coupling capacitor 293 to the grid of a iirst filter amplifier 2 Similarly the signals from the receiver 5 ar applied through a second coupling capacitor 29 to the grid of a second lter ampliner Due to the D.-C. setting action of the pulsed diodes |91 and |99 connected to the grid of nrst filter amplier 2&5, the received communication signal, shown in graph w as including spurious low-frequency components, as indicated by the broken line 2li, is corrected as shown in graph .r so that its ordinates are zero at each occurrence o one of the control pulses shown in graphs u and o. Graph c is similar to graph u with the exception that it is compressed to show more control pulses. Graphs o, w, a: and y are drawn to the same time scale.

rEhe abrupt D.C. setting action of the circuit thus described tends to provide spurious highfrequency components in the corrected communication signal, at the grid of the first filter amplifier although spurious low-frequency componente have :been effectively removed. By means of the unique split-channel complementary filter system described in copending application Serial No, 583,343, filed March 17, 1945, mentioned heretofore, the signal is further corrected to provide the fully compensated signal shown in the graph y which is applied to the terminals of the synchronizing blanking circuit 93 and the diierentiator 3l of the circuits of Figs. 1 and 2.

rlhe series capacitor 297 and shunt resistor 2| l in the grid circuit of the second lter ampliiier 299 provide a high-pass filter therefor, whereby substantially only the high-frequency components w of the received signal w are applied to the succeeding circuits 8l and S3. Also because of the low-pass filter provided by the series resistor 2|3 and shunt capacitor 2l5 in the'cathode circuit of the first filter amplifier 205, substantially only the complementary low-frequency components of the corrected received signal are applied to the succeeding circuits 8l and 93. This eifect is obtained by selecting the parameters of the filters 261, 2H and 2l3, 2li to be exactly complementary in that each have 70 percent response and 45 phase shift at some selected mean frequency.

Thus, the invention disclosed comprises a novel secret telecommunication system including a D.-C. clamp circuit for improving the low-frequency delity of received signals in a synchroized communication system. The D.-C'. clamp circuit is controlled by the coding wave to displace the received signal to its A.-C. axis at each instant that the coding signal, which is a multiplication factor thereof, crosses its A.C. axis.

We claim as our invention:

l. In a system employing a complex signal having a signal amplitude multiplication factor component and a spurious amplitude component, a circuit for removing said spurious signal component comprising means for deriving said multiplication component signal, a switching device, means responsive to reversals in polarity of said derived multiplication component signal for selectively actuating said switching device, and means responsive to said actuation of said switching device and operable upon said complex signal for removing said spurious component of said complex signal.

2. In a communication system employing a complex signal having a signal amplitude multiplication factor component and a spurious amplitude component, a circuit for removing said spurious signal component comprising means for generating a signal which is substantially a replica of said signal multiplication component, means responsive to said replica multiplication signal component for deriving control potentials, and means for combining said complex signal and said control potentials to remove said spurious signal component of said complex signal.

3. In a communication system employing a complex signal having a signal amplitude multiplication factor component and a spurious amplitude component, a circuit for removing said spurious signal component comprising means for generating a signal which is substantially a replica of said signal multiplication component, means including signal differentiating means responsive to said replica multiplication signal component for deriving control potentials, and means for combining said complex signal and said control potentials to remove said spurious' signal component of said complex signal.

4. In a communication system employing a complex signal having a signal amplitude multiplication factor component and a spurious amplitude component, a circuit for removing said spurious signal component comprising means for generating a signal which is substantially a replica of said signal multiplication component, means for limiting said replica signal to convert said signal to substantially square waveform, means including signal differentiating means responsive to said limited signal component for deriving control potentials, and means for combining said complex signal and said control potentials to remove said spurious signal component of said complex signal.

5. In a communication system employing a complex signal having a signal amplitude multiplication factor component and a spurious amplitude component, a circuit for removing said spurious signal component comprising means for generating a signal which is substantially a replica of said signal multiplication component, means for converting said replica signal to substantially square waveform, means responsive to said square waveform signal component forderiving control potentials, a network responsive to said complex signal and means for applying said control potentials to said network to remove said spurious signal component of said complex signal.

6. In Ya communication system employing a complex signal having a signal amplitude multiplication factor component and a spurious amplitude component, Ya circuit for removing said spurious signal component comprising means for generating a signal which is substantially a replica of said signal multiplication component, means for converting said replica signal to substantially square waveform, means including signal differentiating means responsive to said square waveform signal component for deriving Y control potentials, a network including a capacitor responsive to said complex signal, a selectively bi-directional network forV discharging said capacitor, and means for applying said control potenti-als to selectively actuate said bi-direc- .tional network to remove said spurious signal component of said complex signal.

'7. In a communication system including a complex signal having a signal amplitude multiplication factor component and a spurious amplitude component, a circuit for removing said spurious signal component comprising means for generating a signal which is substantially a replica of said signal multiplication component,v

limiting means for converting said replica signal to substantially square waveformJ means including signal differentiating means responsive to reversals in polarity in said square Waveform signal component for deriving control pulses, a network including a capacitor responsive to said complex signal, a selectively 'bi-directional network for discharging said capacitor, and means for applying said control pulses to said bi-direc tional network to selectively provide discharging paths for charges on said capacitor through said bi-directional network to remove said spurious signal component of said complex signal.A

` 8. Apparatus of the type described in claim 2 characterized in that said control signal introduces distortion in said combined signals, said apparatus including separate complementary filter means for said complex signal and said combined signals and means including said filter means for substantially removing said distortion from said combined signals.

9.,Apparatus of the type described in claim 7 characterized in that said selective discharging of said capacitor introduces distortion in said complex signal with said spurious signal removed, said apparatus including complementary lter means respectively for saidV original complex signal and said complexrsignal with said spurious component removed, and means including said filter means for substantially removing said distortion from said iiltered signals.

10. VIhe method of removing a spurious signal component from a complex signal which also includes an amplitude multiplication factor component comprising the steps of deriving said multiplication component signal, deriving control potentials in response to reversals in polarity of terized in that said amplitude changing of said said derived multiplication component signal, and complex signal introduces additional distortion selectively changing said complex signal ampliof said complex signal, and including the step tude to a predetermined value in response to said of filtering out said additional distortion therecontrol potentials to remove said spurious signal 6 from.

component from said complex signal, ALDA V. BEDFORD.

11. The method described in claim 10 charac- KARL R. WENDT.