US2252364A - Phase corrector for multiplex telegraph systems - Google Patents

Description

Aug. 12, 1941. G. R. CLARK PHASE CORRECTOR FOR MULTIPLE TELEGRAPH SYSTEMS Filed Jan. 31, 1940 2 Sheets-Sheet l l mm mmu m L w M m M W R m C T wbmmyufi m r m mg X l A 1.... w I. 7 +L l J E A @2 0? m +1 u 03383 u Wk 5 g vvvvvv IAAAA Au 12, 1941. e. R. CLARK PHASE CORRECTOR FOR MULTIPLEX TELEGRAPH SYSTEMS Filed Jan. 31, 1940 2 Sheets-Sheet 2 fi 6 SEEM mmwmwmu G 6 QmmEm mmumwmu K M R L 0 T E VT m Ba 5% 58 8% b8 ATTORNEY tator.

Patented Aug. 12, 1941 UNITED STATES PATENT OFFECE PHASE OORRECTOR FOR MULTEPLEX TELEGRAPH SYSTEMS Gilbert R. Clark, Brooklyn, N. Y., assignor to Radio Corporation ofAmerica, a, corporation of Delaware Application January 31, 1940, Serial No. 316,458

9 Claims. V (01. 1'78-53) This invention relates to a phase corrector for a multiplex telegraph system. In order to maintain a receiving commutator in phase synchronism with the speed of transmission of telegraph signals, it is necessary that the signals shall produce a corrective action whenever the speed of the receiving commutator departs from which, for example, represents the moment of passage of the commutator brush over the cenby Richard E. Mathes and assigned to the assignee of this application. In the Mathes disclosure the times of occurrence of two local circuit closures are compared one with the front edge of a marking impulse, and the other with the back edge of such a marking impulse. Two distinct time intervals are, therefore, derived which may be differentiated from one another in such manner as to produce the necessary phase correction of the distributor.

In the instant application certain of the teachings of Mathes, as above explained, are adopted. However, I have found certain advantages to be derived froln using a specific circuit arrangement as will be hereinafter set forth in detail.

It is an object of my invention to provide a circuit arrangement for multiplex distributor phase correction such that the phase correcting factor may be made a function of the ratio between-two measured time intervals,,0ne of which elapses between the front edge of a marking signal and a local impulse near the .normal center of a time allotment for dot unit impulse reception. The other time interval starts from a moment near said center of the unit signal time allotment and terminates with theback edge of the marking signal.

It is another object of my invention to provide circuit arrangements for the performance of phase correction in a multiplex receiving system such that accurate and dependable control r to of the speed and phase of the distributor may be insured.

My invention will now be described in detail, I

reference being made to the accompanying drawings in which,

Figure 1 shows diagrammatically a preferred circuit arrangement including elements in combination for carrying out the invention; and

Figs. 2 to 8 inclusive are plots of electrical impulses which occur in various parts of the circuit arrangement. These plots are all drawn to the same horizontal time scale in order to show the correlation of essential operations in response to signals.

By means of the circuit arrangements herein shown I am enabled to gain considerable advantage over the use of previously known phase correcting apparatus. The old method of making the correcting factor dependent upon the location of the front edge alone, or the back edge alone, of the signal in relation to the phase of the distributor, became undependable when the signal itself was distorted. With'front edge correction the elongated mark elements may overlap into following space elements enough to adversely affect the multiplex signal distribution; inasmuch as the commutator brush is maintained at a fixed point with respect to the front edge of the mark signal.

My method is somewhat similar to that of the aforesaid Mathes application, in thatI produce a correction factor which is derived from the correlation of both front and back edges of the marking signal with the center of the marking signal elements as defined by the commutator. My system provides further advantages, however, since the circuit constants are more stable, and a very heavy signal with relatively short space elements can be used for multiplex.

Briefly my method is one which applies phase correction to advance or retard the distributor brushes in accordance with a ratio derived from two time interval measurements. One of these time intervals is a function of that part of a marking signal of dot unit length which extends from its front edge to a predetermined position of brush passage over a commutator segment. .The other of these time intervals is a function of that part of the marking signal of dct unit length which extends from a predetermined position of brush passage over a commutator segment to the back edge of the signal. The two predetermined positions of brush passage should be equally spaced from the center of the segment.

In the following description of a preferred embodiment of my invention all elements referenced V1, V2, V3, etc., will be understood to be discharge tubes, whether of the vacuum tube, or of the gasfilled ionization discharge type. The gas-filled tubes, however, are conventionally designated in Fig. 1 by means of a black spot in the discharge zone.

Furthermore, the several resistors are referenced R1, R2, R3, etc., while the several capacitors are referenced C1, C2, C3, etc. Specific reference to each of the tubes, resistors, and capacitors Will, for the sake of brevity, be frequently made by their identifying symbols alone.

The time interval which starts wi th the front edge of a marking impulse and terminates with a commutator brush passage over a predetermined portion of a segment near the center thereof is represented in the circuit arrangement by a charge stored on condenser C13. Correspondingly the time interval which starts with the passage of the commutator over a predetermined portion of, a

- segment near thecenter thereof, and which terminates with the back edge of a marking signal, is represented by a charge stored on the condenser C14. These charges are held untilaspace occurs following a marking impulses. At that time both of thetubes V14 and V15 are rendered conductive through their control grids, and the impedances of each tube are determined by the amount of each charge respectively stored in C13 and C14; Consequently more or less current will flow in V14 than in V15.

A differential relay S is provided'with windings such that current from the B+ terminal of the power supply source flow more or less through each of the anode circuits V14 and V15, in consequence of which the armature I2 is moved either upwardly or downwardly in order to drive the correction motor l3 in the proper direction for correcting the phaseof the main motor I4.

The correction motor l3 as shown in Fig. '1 is provided with a pinion gear l5 meshing with a gear l6 mounted on the frame and field of the motor M. The correction motor I3 is required, therefore, to operate only momentarily while the relay S makes one contact or another, and the slight movement applied to the frame and field of the motor I4 is sufficient to restore the proper phase of the distributor brushes which are rotated by saidmotor 14.

In accordance with the circuit arrangement shown in Fig. 1, if C13 has a higher potential than C14, as when the receiving commutator is running slow, then the armature I2 is deflected upwards and the correction motor advances the phase of the distributor. If, on the other hand, C14 has the higher potential, the reverse action takes place by deflecting the armature l2 downwardly so as tooperate the correction motor IS in the opposite direction and thus to retard the phase of the distributor brushes.

I will now explain how the incoming signals and locally derived impulses from the commutator are made effective to store charges on the condensers C13 and C14 so that the proper control ofthe relays S may be obtained.

The received multiplex signal, consisting of a tone for mark and no tone for space, is passed through transformer T1 and rectified by the full wave rectifier tube V1. A diagram of the signal input before rectification is shown in Fig. 2.

The rectified signal voltage appears across resistor R1 in the polarity indicated, withthepositive and connected to ground.

The negative signal voltage is applied to the control grids of two electron discharge tubes V14 and V15, thereby biasing these two tubes to cutoff whenever the signal is on mark.

Likewise, the negative signal voltage, after being filtered by the network consisting of capacitors C1, C2 and resistor R2 is also applied to the grid of discharge tube V2, biasing this tube to cut-off when the signal is on mark. In the curve of Fig. 3 the control of conductive times in tubes V2, V14 and V15 is plotted on the same time scale as that of Fig. 2.

A voltage divider consisting of resistor sections R3, R2, R1 is inserted between the plate supply source 3+ and ground. Since R3 and R2 have a high value compared to R1, this circuit has little effect on the rectified signal voltage developed across R1. Furthermore, since R3 is large compared to R2, the rectified signal voltage across R1 is applied practically undiminished to the grid of tube V2.

The grid of V2 is prevented from ever becoming appreciably positive, due to the high resistance of R3.

The network keeps the grid of V2 at zero bias or slightly positive until the signal voltage becomes strong enough to overcome this positive threshold bias. Thus, weak line noises are prevented from causing false signal start pulses in the output of tube V2.

The current in V2 is cut off sharply when the mark starts, and. starts flowing again when mark ends. A steep wave front voltage change appears between the anode of V2 and ground. This change is in the positive direction when mark starts, and in the negative direction when mark ends.

Condenser C3 has a low capacity, so that, when mark starts, it quickly charges through R4 and R5 to the higher plate potential of V2. This temporary charging current causes a positive pulse toappear across R5 when mark starts, and a negative pulse when mark ends. Fig. 4 shows a .curve (in full line) of the momentary positive and negative pulses across R5.

The positive front edge pulse is designated M and is carried by the conductor similarly referenced in Figs. 1 and 4.

In the same way network C4, Re derives a positive front edge pulse and a negative back edge pulse, which are applied through grid R2 to the grid of V2. The negative back edge pulse is inverted by tube V: so as to appear as a positive back edge pulse between the plate of V3 and ground.

This positive back edge pulse passes through the output coupling C5, R9 and is designated P in Figs. 1 and 4.

In addition to the distributor segments which are employed for channel distribution of the multiplex signals, certain segments and a brush, or two adjustable brushes and one segment, are employed purely for phase correcting functions. This brush and segment combination is known as a kicker, and its function is to deliver .short, sharp impulses into one or more local circuits such as will be hereinafter described. In the drawings two points of circuit closures are indicated, one of which is under control of kicker contacts designated kicker A; the other circuit closure is provided by contacts designated kicker B. .It will be understood that these circuit closures are accomplished at certain instants of the commutator cycle equally removed from the center ofa brush passage over the center of a given channel segment.

Closure of kicker A produces a surge impulse across resistor R10 from the power supply terminal 3+ to ground. Prior to this closure, however, condenser Cs has received a charge, since it is in circuit with resistors R9 and R10 between the source terminals C and 13+. Closure of kicker A thus allows Cev to discharge, and to impress a surge impulse across R9.

At the end of the kicker A closure C6 is again connected only through R10 to the supply source 3+, and the recharging current produces a positive pulse N or N across Be. This positive pulse is applied through R111 and R19 to the control grids of two gaseous discharge tubes V6 and V7.

Kicker B has a short closure just following the center of every multiplex signal element and impresses a positive pulse on the conductor so labeled.

In this case, however, it is necessary to suppress pulse 0 during signal space elements, since the back edge period applies only to the back end of the mark elements. The space between mark elements is used forthe differentiating action between V14'andV15, as explained previously.

Pulse 0 is'suppressedduring space by the expedientof obtaining the positive supply source for this pulse from the plate of tube V2. When the signal is on space, tube V2 is conductive and the anode potential is very low. Hence the pulses O are very weak.

During'mark, however, tube V2 is cut off, and its anode potential is the same as the plate supply source. The pulses 0 derived from the opening of kicker B are now strong.

Decoupling network R11, Cr, R12 is added in order to prevent kicker B from interfering w1th the proper operation of V2. If any such interference existed, it would produce false M-pulses, and also (through the full amplification of V2), false P-pulses.

Condenser C7 acts as a positive supply voltage reservoir for kicker B and is isolated from the plate of V2 by R11 so as not to affect the wave 4 front of the voltage changes between the plate of V2 and ground. I

Incidentally, this same decoupling network prevents the steep wave front voltage changes across V2 from initiating false O-pulses prior to the closure of kicker B.

We now have four positive pulses M, N,- O and P for eachsin'gle marking element. The M- and N-pulses define the front edge integration period. The O-pulse when immediately followed by a P-pulse defines the back edge integration.

The operation of the front edge integration circuit (consisting of V4, V6, V8, V10, V12, C13 and associated parts) will now be described.

The operation of the back edge circuit is identical except for the pulse sequence.

V4 and V6 are grid-controlled thyratrons. They are inter-related by means of a famihar circuit R16, R20 and C9. As is well known, if Vs has been ionized and then V4 is caused to be ionized, C9 reduces the anode potential of V6 to a negative value long enough for V6 to de-ionize. Although the anode potential soon rises again above the point where it will sustain iomzation, nevertheless Vs will remain out if it has sufficient negative'grid bias.

When the signal is on space, the state of th1s circuit is always as follows: V6 is ionized and V4 is out. This will be called the idle state.

When the signal goes on mark, a positive pulse M is generated by V2 and is applied through R14 to the grid of V4. This pulse ignites V4, thereby extinguishing Vs.

The circuit remains in this new state, the active state, until the kicker A closure takes place. At the break of the kicker closure a positive pulse N is generated and applied through R12 to the grid of Va. V6 now ignites and V4 goes out, thereby restoring the circuit to its original idle state.

The active period (during which V4 is ionized) therefore coincides with the front edge period, and its duration is less than half the length of a dot unit signal, as shown by comparison 01R Figs. 2 and 6.

It will be observed that when V4 is ignited by pulse M, the anode of V6 is made negative by the existing charge on C9. This charge normally persists long enough to permit V6 to de-ionize. It is purposely made still longer to permit a certain necessary wipe-off action to be completed with respect to C13.

The wipe-01f is effected in the following way: During the idle state the grid of Va is at zero bias or slightly positive, due to the high resistance voltage divider network R22, R24 between the anode of V6 and ground. Therefore, V8 is conductive, and its anode potential is low.v

When the active period begins and the anode of V6 receives a mometary negative surge impulse from condenser C9, Va is sharply cut off because its grid becomes negative also. The anode potential of V8 immediately rises toward the full voltage of the plate supply source.

C11 quickly charges to this new potential through R32 and R34, and in so doing creates a strong, sharp positive pulse across R34 which is applied through R36 to the grid of V12.

This positive pulse is created at the instant that V4 is ignited, and cannot occur at. any other time, since the anode of V6 never becomes negative at any other time. The pulse is strong enough to ignite V12 even if the anode voltage of V12 (which is supplied by C13) is very low.

Of course, once V12 is ignited, it immediately discharges C13 through R38 down to the point, say, 15 volts. at which point there is not enough potential to sustain ionization and V12goes out.

All this must happen at the commencement of the active period and before C9 charges enough to permit the anode of V6 to become positive again. While the anode of V6 is negative, diode V10 is non-conductive and isolates C13, R38 and V12 from the rest of the integrating circuit.

When the anode of V6 does become positive its potential continues to rise toward the full voltage of the B+ supply source, since the tube is non-conductive. The high potential at the anode of V6 begins to charge C13 through R211 and V10. The resistance and condenser values are such that C13 only partially charges within the duration of the active period (as in a saw tooth generator) so that the charge on C13 is substantially proportional to the duration of the period.

The end of the active period occurs when pulse N ignites V6 and restores the circuit to the idle state. When this happens the anode of V6 drops to, say, 15 volts. Since C13 is at some higher potentia1, diode V10 prevents C12 from discharging down to, say, 15 volts. C13, therefore, holds its charge, except for leakage and the screen currentdrain of V14 duringspace, until the next front edge period begins.

The circuit arrangement shown in the lower part of Fig. 1 will be seen to have certain points of similarity with that of the upper part. Thus, tubes. V5 and V; are inter-related in the same way as tubes V4 and V6. Likewise V9 is the counterpart of V8 and the control circuits for all of these tubes are similar in the lower portion of the diagram to those in the upper portion thereof. In view of this similarity'of circuit arrangements it willbe suflicient t-o briefly summarize the operation of tubes V5, V7, V9 and V13 in the following paragraphs. I

Kickers A and B, as stated above, make circuit closures periodically at predetermined points of brushpassage over each commutator segment. Kicker B is effective, however, only during marking periods, that is, when V2 is non-conductive. Fig. 5 shows the action of kicker B to deliver O-pul'ses during. mark intervals only. These pulses are impressed upon the grid V5 to ignite thesame at. the commencement of the back edge integration. This period is terminated with the back edge of the signal as denoted by the P-pulse which is impressed through R19 upon the grid of V1. When V7 is ignited a surge impulse from C10 reduces the anode voltage on V5 to the point of extinction. 1

This action, as explained in the foregoing paragraph, takes place without intervening actions, provided the marking impulse of the signal is of one dotunit length.

If the signal continues on mark at the end of a dot unitlength, it is necessary to ignite V2 in response toan N -pulse derived from the operation ofkicker A. Thus, V5 becomes extinguished either at the endof a true marking impulse or at predetermined points of circuit closure by kicker A when. the marking impulse is prolonged. During. these prolongations of the marking impulse the reversing operations of V5 and V7 take place but are ineffective, as shown by the shaded square wavesof Fig. 7-. The back edge integration which is efiective in all. cases is initiated by an O-pulseigniting V5 and extinguishing V7. When the anode potential V7 rises to that of the supply source B-lcondenser C14- charges through the rectifier V11. Momentarily, before this charge begins, the action of tubes V9- and V13 is brought about in the same manner explained above in respect to V8 and V12, for the purpose of igniting V13 so as to discharge C14 to a fixed value, say, 15 volts. Thereafter the rectified current flowing across V11. increases the charge on C14 proportionately, to the. duration of the extinction time of V1. In the back-edge integration circuit, resistors R111, R21, R31, R23, R25, R27, R29, R33, R35, R37, R39 and R41. are counterparts respectively of resistors R16, R20,.R30,. R22,. R24, R26, R28, R32, R34, R36, R33 and R40 which are shown in the front edge integration circuit. Furthermore, the functions of C11 and. 012 are identical.

Since there is no true back edge integration period tobe measured until an O-pulse is immediately followed by a P-pulse within an interval of one dot unit length, all other occurrences of the O-pulse which take place during a prolonged marking signal must be rendered ineffective by an N'-pulse from kicker A. During these prolonged. marking signals, therefore, C14 may become discharged repeatedly. But only on the last period, that is, at the back edge of the marking signal, does the stored charge remain on C14 to be rendered effective when V15 becomes conductive at the commencement of a spacing signal. A comparison of ignition inter- Vals of V5 (which correspond to extinction intervals of V7)- with the time intervals tobe differentiated by relay S, as shown in Fig. 8, willshow that, regardless of the length of the marking signal, the front edge integration period is always compared with a back edge integration period. The differentiation is shown by the symbol I to represent that the time intervals to be compared are more or less close approximations to equal time intervals, and that they are'equal when no correction factor is to be applied. In other words, when the back edge period is finally ended by the P-pulse, then the charge on C14 is left to be impressed on V15 at the commencement of a spacing signal.

During the lapse of a spacing signal tubes V2, V14 and V15- are rendered conductive. V2 when conductive renders the action ofkicker B ineffective and also maintains the idle state of the integration circuits, both for the front edge and for the back edge integration.

Fig. 5 shows two time graphs of pulses delivered by kickers A and B, respectively. As shown in the upper of the two graphs, an N-pulse occurs on the first opening of kicker A whichv occurs in a given mark period. If the mark is prolonged for more than a dot length, then kicker A delivers N'-pulsesuntil a space occurs. During space periods kicker A delivers pulses marked X, but their effects are annulled by the fact that tubes V6 and V7 are already discharging. The operation of kicker Bis shown in the lower of the two graphs in Fig. 5. O-pulses occur periodically upon each opening of kicker B, except during space-periods. These pulses are suppressed,.however, as shown at Y, due to the conductive state of V2.

Throughout the duration of the space signal the control grids of V14 and V15 are; unblocked and at zero bias, so that the charges on C1: and C14 respectively may control the screen grids of these tubes. The current flow through these tubes is substantially proportional to the screen grid voltage.

The anode currents from the supply source B-I- pass through the windings of relay S in opposi- When there is asufiicient tion toone another. difference in the two currents, the armature of relay S is pulled from its neutral position, thereby causing the correction motor to turn in the required direction.

A smoothing-out of anode currents of V14 and V15 is attained by means of condensers C15 and C16 respectively in combination with R44 and R45. These elements cooperate in preventing a chattering of the relay armature at the signal element repetition rate. The operation of the relay S to control the direction of rotation of the correction motor l3 and thereby to correct the phase of the main motor [4 has already been explained.

I claim:

1. Apparatus for maintaining a distributor in synchronous phase with the arrival moments of the mean centers of a train of marking impulses received over a circuit, said apparatus comprising a pair of condensers, a unilateral charging circuit connected across the electrodes of each condenser, a discharging circuit in shunt with each charging circuit and including a grid-controlled gaseous discharge tube, means operated by said distributor for producing two surge impulses, the first preceding and the second following by equal time intervals the centering of a distributor brush on a distributor segment, means for accumulating a charge on the first of said condensers which isproportional to the time lapse between the arrival of the front edge of a marking signal and the first said surge impulse, means for accumulating a charge on the second of said condensers which is proportional to the time lapse between the second said surge impulse and the arrival'of the back edge of said marking signal, means operative only during space-signal intervals for comparing the charges respectively stored on said condensers, means dependent on the operation of said comparing means for adjusting the phase of said distributor, and means for controlling each of said gaseous discharge tubes at the commencement of the charging pe.. riod of its associated condenser, thereby to reduce each condenser charge to a predetermined value.

2. Apparatus according to claim 1 and including a pair of gaseous discharge tubes for controlling the time intervals of charging the first said condenser, one of said tubes being subject to ionization in response to the arrival of the front edge of a marking impulse, the other of said tubes being subject to ionization in response to the occurrence of the first said surge impulse, and each of said tubes being subject to de-ionization control by the other said tube.

3. Apparatus according to claim 1 and including a pair of gaseous discharge tubes for controlling the time intervals of charging the second said condenser, one of said tubes being subject to ionization in response to the occurrence of the second said urge impulse, the other of said tubes being subject to ionization in response to the arrival of the back edge of said marking impulse, or alternatively under control of a recurring first surge impulse, whichever event occurs first, and each of said tubes being subject to de-ionization control by the other said tube.

4. Apparatus for maintaining a distributor in synchronous phase with the arrival moments of the mean centers of a train of marking impulses received over a circuit, said apparatus comprising a pair of condensers, means operative upon arrival of the front edge of one of said marking impulses and including a grid-controlled gaseous discharge tube for dissipating a residual charge on one of said condensers down to a fixed voltage, means in circuit with a diode rectifier tube for thereafter building up a new charge on that condenser substantially proportional to and limited by the time interval from said front edge arrival until the occurrence of a periodic circuit closure, means for producing a second periodic circuit closure, means for maintaining the first and second circuit closures one in advance of, and the other equally retarded with respect to a brush passage over a given segment center of said distributor, means operative in response to one of said second circuit closures for dissipating a residual charge on the second of said condensers down to the same fixed voltage, means for thereafter building up a new charge on the second said condenser substantially proportional to and limited by the minimum time interval from one of said second circuit closures to the arrival of the back edge of said marking impulse, means operative during a subsequent space interval for differentiating between the voltages built up on the two said condensers, and means responsive to the differentiating means for accelerating or retarding said distributor.

5. Apparatus according to claim 4 and including a grid-controlled gaseous discharge tube in themeans for dissipating aresidual charge on the second of said condensers.

6. Apparatus according to claim 4 and including a diode rectifier tube in circuit with the means for building up anew charge on the second said condenser,

7. A'synchronizing system for telegraph apparatus comprising a first condenser arranged to store a charge commencing with the arrival of the front edge of a marking signal and terminating with a local circuit closure, a second condenser arranged to store a charge commencing with a second local circuit closure and terminating with the arrival of the back edge of said marking signal, a rotary distributor member normally in phase with the mean centers of a train of marking signals, means operable by said rotary member for producing said circuit closures, the first immediately before and the second immediately after said rotary member passes across a mid-position for distribution of a signal element, means including a unilaterally conductive device in circuit with each of said condenser for impressing said charges thereon, means including a grid-controlled gaseous discharge tube in circuit with each condenser for dissipating its charge down to a fixed voltage value prior to the impress of a new charge thereon, and means for difierentiating between the charges on the two said condensers thereby to accelerate or to retard said rotary distributor member.

8. A synchronizing system for telegraph apparatus comprising a first pair of grid-controlled gaseous discharge tubes each arranged to be rendered conductive by a positive impulse applied to its grid, means including a capacitor connected across the anodes of said tubes for extinguishing the discharge in each said tube when the other said tube is rendered conductive, means operable on arrival of the front edge of a marking signal for producing one of said positive impulses applicable to the first of said tubes, a rotary distributor member to be maintained in synchronism with the incoming signals, means operable by said rotary member for applying a second positive impulse to the second tube of the pair, a second pair of grid-controlled gaseous discharge tubes each arranged like those of the first pair and further arranged to be" controlled, one tube by a positive impulse derived from the operation of said rotary member, and the other tube by a positive impulse derived at the instant of arrival of the back edge of said marking signal, capacitive storage means operable by the two said pairs of gaseous discharge tubes, a differential relay for determining the relation between two charges accumulated by said storage means, and phase corrector mechanism operable upon said rotary distributor member in response to the action of said relay.

9. Apparatus for maintaining a distributor in synchronous phase with the arrival moments of the mean centers of a train of marking impulses received over a circuit, said apparatus comprising a pair of condensers, a unilateral charging circuit connected across the electrodes of each condenser, a unilateral discharging circuit in shunt with each charging circuit, means operated by said distributor for producing two surge impulses normally equally disposed about the center of each one of said marking impulses, means for accumulating a charge on the first of said condensers which is proportional to the time lapse between the arrival of the front edge of the marking signal and the first said surge impulse, means for --accumulating a charge on the second of said condensers which is iproportional to the time lapse between the second said surge impulse and the arrival ofthe back edge of the marking signal, means efiective at the outset of .each charging period for completing the appropriate one of said discharging circuits thereby to permit the charge on each condenser to be built up from a predetermined reference level, means for comparing the charges respectively stored on said condensers, and means to utilize the result of said comparison to adjust the phase of said distributor.

GILBERT R. CLARK.

Images