US1922139A - Telegraph system for submarine cables - Google Patents

Description

Aug. 15, 1933. H. NYQUIST I 1,922,139

I TELEGRAPH SYSTEM FOR SUBMARINE GABLES 7 Filed Sept. 7. 1932 2 Sheets-Sheet 1 Condenser Cahle .D 'olew Circuit ending E7405 15' yual {jer' Cab Z9 Condenser Ill (demo Circuit Feceivz'rg End INVENTOR EWy uM/If BY M ATTORNEY Aug. 15, 1933. H. NYQUIST TELEGRAPH SYSTEM FOR SUBMARINE CABLES Filed Sepf. 7, 1932 2 Sheets-Sheet v 2 Oscillbyrqah ZZ; work with fzve cur-rent l l I .S e Equalgez fur f'-'- area, wit/run L n dotted q, I how/balmy. li i ikfl fiz'e ii fl I Circuit I Tr Z I l, 0 e90 Ci- CMJI ZZZ Receiuz'ny End Receiumy Device Recent/er fol F608 Mizynztudes L 1b .Feedflack Circuit Receiver for Three Magi/dazzles lNVENTOR EJV /quwt TI'ORNEY Patented Aug. 15, 1933 UNITED STA TES , 1,922,139 TELEGRAPH SYSTEM FOR sUBMARINE I CABLES Harry Nyquist, Millburn, N. J., assignor to American Telephoneand'Telegraph Company,

a Corporation of New York 1 Application September '7, 1932 Serial No. 632,011

17 Claims.

This invention relates to telegraph systems for submarine cables and similar transmission lines. Its purpose is to increase the speed of transmission of signals through such signaling circuits and to increase the reliability of operation. An-

other pu rpose is to construct a signaling system which is as free as possible from the characteristic interference normally present in such sys tems; A further purpose is to-obtain a form of received signaling .unit I which shall best lend itselfto such ends. Still a further object is to design a cable or transmissionline with associated apparatus which is best adapted for leading to the production of the preferred wave form at the receiving end. Still another purpose" is to design terminal receiving apparatus which shall be operated effectively and accurately by the; preferred wave form at the receiving end.

Mysystem relates to a cable which is unable to 20 transmit direct current or alternating current above a certain frequency 8 and in such a system, by myinvention, it is possibleto be as free from characteristic interference as one cares. For this reason, I can transmit as many different cur rent values as desired and so increase the speed of signaling or increasethe number of signaling channels. i a

In my system, if one sends as a signaling unit a square-top impulse covering one time unit of a duration related to the maximum transmission frequency for the cable, thecur-rent on the cable changes in form and at the receiver is spread over several time units in accordance with certain characteristics hereinafter set forth. In; this train of impulses at the receiving end, the main signal is followed by a trailing signal of substantially the same magnitudes but of oppositesign. In my system I provide a feed-back arrangement at the receiver which generates a local impulse 40 properly valued and properly timed to substantially neutralize the received trailing impulse. Also, on thiscable, in case a train of impulses of frequency s is transmitted, the received current resultingtherefrom approaches zero since the cable does not transmit the frequency s but, in spite of this, the feed-back arrangement acts to keep the receiving mechanism operating as long as no change occurs in the signals at the sendingend.

These and other features of my invention will be pointed out and explained in connection with the following specification and accompanying drawings, in which Figure 1 shows the sending end of a diplex circuit and Fig. 2 shows a corresponding diplex receiving end. Fig. .3 shows the preferred wave form which I designed my cable to produce. Fig. 4 is a portion of a record such as might be produced by the receiving apparatus of Fig. 2. Figs. 5,6 and I relate to thereceiving end of circuits using various numbers of current signaling magnitudes. One characterstic of submarine cables is that their cost is so great that it is economically desirable to go to great lengths to obtain as high a speed as possible even when this means greatly complicating the terminal apparatus. Another characteristic of such cables is that they are sub ject to interferences of very low frequency which usually necessitates suppressing the transmissionof direct current. Also such cables have a limiting frequency above which it is not possible to transmit messages which will be effective on the receiving apparatus. An outstanding problem then in submarine cable signaling is to transmit up tothe maximum frequency transmitted by the cable with the handicap of having notransmission of direct current. This specification gives an analysis of the requirements for and specifies a system which will transmit without distortion under these conditions. p

In the April 1928 Transactions of the American Institute of Electrical Engineers, page 617, I published an article entitled Certain Topics in Telegraph Transmission Theory to which reference will. be made from time to time in this specification. In that article, it was shown that the intelligence conveyed by a telegraph signal is fully contained in a band of frequencies extending from zero to a frequency s equal to the speed of signaling in dots per second, where the dot may be considered as made up of two time units, one for the marking direction and one for the, spacing. duration. It was shown that this is true regardless of the number of current magnitudes used in the signaling. It was also shown that the omission of any continuous range, however narrow, from the range just specified may, with certain signals, lead to the reception of an incorrect signal. Finally, it was shown. that, while a continuous range may not be omitted, individual frequencies may be omitted. In the latter case, the signal received must be supplemented by information derived from its previous history, as will be more apparent below.

I will, .first, consider'the specific problem of a telegraph circuit whose transmitted range does not extend beyond the frequency 8 and whose transmission is zero for direct current and for the frequency 5. Let itbe required to provide a telegraph system using four magnitudes or current values at the speed of s dots per second.

distinct signals which, as indicated m'a-y'be' continuous for a long time. These sixteen signals, while distinct, have this in common, that they are made up of no other frequency componentsthan direct current, 8 cycles, 28 cycles, cycles, etc. In other words, these signals. .containno frequencies transmitted by the telegraph circuit. The ultimate received current is zero in all these cases, although the beginnings and endings will be indicated by transients, the nature of which will depend on thesignal sent. In spite of this, the telegraph system" should be constructed to continue-to receive the. correct signal even after substantially no' current is being received.

While, as stated above, theultimate' current is thesa'me in the case of these sixteen signals, they differ in the manner in whichthey approach that finalvalue and the'teleg'raph' system must, ac cordingly, besuch as to receive an impression of P the right sort while the current'dies out andto continue receiving in accordance with that impressionas long as the current remainszer thatis until the transients due to the end of the't'raln of signals arrive. Todo this, the system must be capable of keeping itself going,-that"is,- it must have a feed-back property.

' Let-use consider a signal clement made" up 01 successive'impulses each of a durationequ'al to the marking or thespaci-ngtime for a dot, that is, of a duration s. I=mayassume" that the current in the middle or the first-time unit is on, inthe middle of the second 5' and soon and, after a small number'oi such time units the value hecomeszero. Then the system can be'made to receiveproperly provided the value at the middle of thefirst time unit-is used for reception and provided'thatthe values [3 etc. are made ineffectual by arranging matters so that'the reception ofa causes a neutralizing current 5 etc. to be fed backintothe circuit at the'proper instant. I will assume, tentatively, that the received current, due to a-unit impressed signal element in the firsttime'unit, is a, B and v inthe middle of the first,"second and third time'units, respectively, and'iszeroin the middle of all other time units.

In order that this should be possible in a system transmitting no direct current, it is necessary-that +fi+v'= Likewise-the assumption that the system is-to transmit no current of frequency s leadsto the requirement Thesetwo' equations may be rewritten as Consequently; a'sign'a'l element of value +1in'the middle ofthe firs'ttime unit, I in the middle of thethird and- O in-the middle of all'the others should be considered. 'Such an element is-indicat'ed inFig. 3' 'ofthe drawings. I have found that its properties'are' suchas to make it a particularly useful and preferred form of wave. As will be pointed out later in my invention, I so design the transmission system that a single flattopped impulse of a unit-timed duration will,,as it proceeds through the cable and associated networks, gradually shape itself so that when it reaches the receiver terminals, it will have produced a wave-shape of the form of Fig. 3.

Now, if I send a succession of impulses each capable, when considered alone, of producing at the receiver a wave of the shape of Fig. 3, and these impulses follow each other in successive time units; I will receive a wave whose beginning and ending-will be indicated by transients bearing *some-rel'ation tothe principal parts of the wave of Fig. 3, but between the beginning and ending transient conditions, the wave will have a steadystate value of zero. For example, if a second wa' e like Fig. 3 is superposed on that wave one time unit later, the second wave will attain its maximumpositive value +1- just when the first wave is-passing through zero in the middle of the second time interval, 1. e., the time intervalbetween the +1 and 1 values of the first wave. Hence'the +1 values of both waves will follow each other and neither will be suppressed. If a third wave like Fig. 3-is received in a third time interval, it will attain its +1 value just as the first wave reaches its 1 value and at the time the second wave is'passing throughzero, midway between its +1 and 1 values. The +1 element-of the third wave cancels the +1 element of the first-wave and the resultant of all three waves during this third interval is zero, and likewise; the +1 element of' a fourth wave will cancel the +1 element of the second wave while the first and third waves are passing through zero. Consequently, after the '+1-impulses of the first and second waves are received, a zero steady-state condition continues as long as the train" of impulses'is sent, but this steady-state condition isfollowed by a transient built upfrom thesuperposition of the last two waves like Fig. 3' which correspond to the last two impulses.

The considerations from now on will be divided into-two parts. The first will consist of specifying the transmission characteristics or signalshaping devices, or both; which lead to a received wave having the properties indicated and the second part will consist in specifying a telegraphsystemcapable of operating on such a received wave. The transmission characteristic of chiefinterest in this connection is one which We will call transfer admittance (Y) which is herewith defined as the ratio of the numerical value of the received wave to the sent wave, regardless of the nature of these values. That is Fr=YFsg where Fr and F5 represent "the received and the sent'waves, respectively.

Instead of working out the transmission characteristics directly, they may be obtained indirectly by making use of the results of my article referred to above. It is shown in appendix II-A (Equations 8 and 10) that the shape factor The required shape factor then for a wave such "that the current is +1 in the middle of the first timeunit and l in the middle of the third time unit and 0 in the middle of all other time units iv obtained by subtracting (2') from (1'). It is F(w) =2 (1- cos w/s-l-i sin w/s),

=1 4 (sin w/Zs) (cos w/2s1' sin (ll/2S),

' OiwSZqrS 21rs o If the sent signal element. is a rectangular wave coincident with the time unit .tovwhich it tains, I have, as shown in my article, for the shape factor at the sending end isi iel L The required transfer admittance is Y (Lu) i4 sin ((028) may be supplemented by other hops factors ouch This as shown in Fig. 2 at 22 and d in or; artic then leads to a choice of values for Y. When the required transmission Y has been determined above and wh computed or measured, the spec-.. cation of equalizers or signal shaping net-works to an'overall characteristic Y strai ht. r This invention does not concern itself with computationof equalizers, partly because r th e because the procedure of designing the co, varies in accordance with the given line aroc teristics. 'In general, the modifying of the to bring it to the characteristic Y can be accenplished by equalizers of the type now well connection with transmission circ But, in case it should be desired, these equalizers ma be supplemented by shaping the sent sign ls. This may be done in accordance with appendix VII of my'article referred to above or in accord ance with my United States Patent No. 1364.336 of March 12, 1929. The discussion in appep ia VII becomes applicable in this by pu fr (t)=+1 for 72:1, f1" (t)=:1 71:3 (t) :9 for all other valuesof h. The subse Equation 16 then finally becomes bit-heal part c.

Fig. 1 shows the sending end of the transmi sion system in schematic form. Thesystemis shown as a diplex type inwhichtwo systems of arrive;

t 'o magnitudes each are combined to forzna sin-- gle system of four magnitudes. The two contributing circuits ar of the synchronous printer type comprising a rotating brush passing over a numbe of segments circularly arranged, over which the signals to be impressed'on the Care is taken that the two circuits work synchronously with each other. The four voltages impressed may be l50, +50, +50, +150 and all the relays in Fig. l, well as in the following figures, are shown in marking position. A condenser is added to block low frequency interfer ence and, if desired, a Wave shaping device W of the kind described in my United States patent referred to above may beintroduced to supple ment the action of the equalizers.

shows the receiving end in schematic form, there being again a condenser for blocking low frequency interference.

a set of amplifiers and equalizers. After ampli There is also shown fication andequalization, the signal is passed to a distributor which is run inaccurate svnchronisin with the distributor at the sending end. The mechanism for maintaining the synehrcnism is not shown but will be taken up further below. The distributor is so arranged that only a small portion of the segments corresponding to the middle of the time unit fie-active. During this interval, the cable is connected to the grid of a vacuum tube. Until the next segment is reachedthis grid holds its potential and a sub stantially constant current flows in the plate circuit of the vacuum tube. Thus, the response of the receiver relays is independen't'of the intensity of the wave at other times than the middle of each time unit. It is necessary, thereiorejonly to see that the received wave has, the correct magnitude, it reaches the receiver, in the mid dle of the time units only. The current from the vacuum tube causes a set of marginal relays to operate and these, turn, cause the circuit to be separated into the two component parts, circuit I and circuit II, all. ina manner well known in the art. In addition to receiving the signal from the line, the'distributor also receives the feedback. wave which has been mentioned above and whi h will now be discussedmore fully.

When the first two signaleleinents of a message are being received, there is no diffi ulty apparent for the reason that the first signal element zero value at the middle of the second time element, as shown in 3 and therefore pro-,

arranged to impress a current of the samevalue l:

that received and so adjusted as to time delay, etc, that it arrives at the input of the receiver two time units after the which initiated it. This time delay is controlled by the time constants of the resistance and condenser elements shown in the feed-back circuit and by the time constant of the mechanically moving relay tongue. In the figure, there are shown two relaysin tandem for the purpose of introducing the approximately correct the delay may be obtained by varying the electricalv or the mechanical stiffness and inertia of 1 the various elements. From this, it will be seen that if there is received in the nth time unit a wave of the form of Fig. 3 which has Zero value delay and a precise adjustment of r:

has ruled on it four parallel at the middle of all time units except the and the third, then, in that nth time unit the received current is l and there will be an inter ing current of 1 from the line in the (n+2) th unit and a feed-back current +1 in the (n+2) th time unit. Thus characteri sti interference is eliminated and the receiver is for the next signal element. A little consideration will show that a series of dots are sent for an indefinite length of time, the system, by virtue of the feed-back arrangement, will con-- .tinue to record the receipt of dots so long the signaling conditions at the transmitting; are not changed.

It has been assumed that the receiving distributor runs in close synchronisin with the ing distributor. Approf. synchron srn. may be maintained by means of any suitable device such as a crystal oscillator. This, however, is insuflicient for my purposes. To obtain closer synchronism, cscillograph of the type velops and fixes the film maybe by operz-tting the switch record may be obtained of the h art, such as that shown L spend to the four currents ivl ceived under con to various sources of into received current does not co. pected value. One s 'nrce of lack of synchronisrn. By ref will be seen that the curve of 1' rent is very steep in the in time unit. This has a connection for, if the signal s receiving mechanism. of interference present l. nose as the received current on the unit and this co onent will be cated. On thoother h nd, it ill -.--t '1" the .receiv. a me ehind the incoming signal, there will be ilar component of interfen ceived current of the or plying this to 4, I ceived in 1 is posit ve and t. ceived in 2 is negative. This the receiving device is lagging; and should be advanced. The current received in 2 is the int ference in 3 positive. evidence that the receiving' device should tarded. Similarly, 3 and give evidence that i should be advanced. Denoting evidence that should be advanced by a and evidence 1 should be rearded by r, I have :tollo dence from units 1 to 11, inclusive, a, r, o, c, T, o, 1", T. On the Whole, then, there is no evidence that there is any adjustment re o. interference is largely due to other sci lack of synchronism.

If an adjust went is indicated of frequency. be made only when it is certain tha steady drift. Moreover, freon should be limited to one station, say the west one.

It will be obvious that the osc of Fig. 4 can be use for establ for adjustments other th se of s For instance, if the received current t numerically too large, the gain or? the receivii device shouldbe decreased. If it tends to be too bc cen that seven marginal relays and that the feed-back circuit conistead of two as in Fig. -ay arranged as 'ent fed back shall be the second determines a. branch determines a ttive current of hali added. The received curable into three circu'ts, I, II and III, iner which will be app ent from Pi ndition ""lGWil corresponds to the cur ent. For the next strongrelay nearest the plate is e is a'consequent reversal in J for the next current value, the plate is reversed as is Xt highest value, the next ted circuit III is reversed to shown.

be obvious how the system shown in c extended to obtain four circuits ine capable of keeping l6 magnitudes in general, n circuits from a line transmitting 2 distinct magnitudes, s positive integ'e l1lll11"8l' of magnitudes transmitted is simply this. I may assume in that a single receiving device making use 0. all th magnitudes employed. The comr or" such device need not concern us. It now-ever, possible to maize use of a feed-back circuit on the principles described as Will now made clear. For the purpose of explanation, l b suificient to take the case of five magnia; agnitudes may conveniently be he plate circi t as the five lowest s in Fig. 5.

out in Fig. 5 would function to pro proper feed-back current, the only peiently on spacing. That connections made to their spacing (Tl'iey are shown 5 on the marl:- ing' contacts.) resulting arrangement is n in 6. It will be understood that this e can be substituted for the upper dotted 4 Fig. 5. When this substitution is understood that the middle ach circuit will be reverse ition shown in 5.

It will be a parent that the feed-back circuits ..-.escribed can be made to give the proper changes in the magnitude of the current fed back. On th .e other-hand, it may not be obvious that the solute mine is correct. In the arrangement n.-wn in F Sjthe current fed back is -1/2, 1/2, +3 2, +5/2 or 4-7/2, Whereas the current rceived over the line would reasonably be expected to be symmetrical about zero, i. e., to be 2, -1, 0, +1 +2. The xplanation is that the constant difference of 1 units is taken care or p) Under these conditions,

ase, these relays may be removed and been'largely for thepurpose of expressing the inr of once for all in an initial adjustment. such as the adjustment of the grid potential.

A feed-back arrangement employing one branch for each marginal relay may be used-insteadof the arrangement shown. In the case where the number of magnitudes is three (which is practically an important case) this leads to the arrangement shown in Fig.- 7, which is somewhat simpler than that which would be obtained by the method illustrated in Fig. 6.

It .should be observed that while in all the cases given the magnitude of the main impulse arriving at the receiver is taken arbitrarily as unity-in the middle of its time unit, this isnot necessarily the maximum value of the impulse during its time unit;' also, while the magnitude of the trailing impulse is taken as minus unity in the middle of the second time unit thereafter, this is not necessarily the maximum value of this trailing impulse. 'Nor', is it necessary that the maximum value of these two impulses be the same, but only that at the middle of their re spective time units they shall be of unit magnitude and opposite in sign.

In using the term unit magnitudethroughout this specification it is to be understood that the unit is purely arbitrary and not necessarily equal or simply related to any of the units commonly accepted in the art. Also, in speaking of the nth time unit it is understood that the origin of time may be taken such that the middle of the time unit at the receiving end cor.- responds to that characteristic point of the main received impulse which is to be used for operating the receiving apparatus; Physically this middle of thetime unit may not necessarily be at the middle of the main received impulse, but the shiftingl of the time origin for purposes of describing this invention will nevertheless bring one to the desired point in that impulse.

Throughout the specification emphasis has been laid on the square topped wave. This has vention in a relatively simple mathematical form, and while the preferred form of transmitted signaling is a square topped wave, or substantially one, it is to be understood'that the invention is not so limited, but the elemental transmitted impulse may take on any desired form; Whatever this form may be the characteristics of the cable eration, I prefer to arrange the characteristics such that the main impulse is of unit magnitude in the middle of its time unit, and the magnitude' of the impulse two units later is equal but of opposite sign in the middle of its time unit,

and is zero in the middle of all other time units,

the invention is not limited in this'way. More I broadly, it may be stated that if mo, m1, m2, etc.,

are the magnitudes of the impulses in the middle of successive time units, then the conditions t be fulfilled are:

(1) mo+mz+m4+. .=0 and (2) m1+ms+m5+. .=0

-In general, this means that more than three time units have to be considered. If mo represents the magnitude of the main impulse in the middle of its time unit, then the feed-back arrangementmust besuch that impulses of 'm1,v '-m2, ma, etc., are fed back in the middle of the corresponding time units to neutralize the corresponding impulses. The simplest arrangement and the preferred one, is specifically that in which m2 is equal and opposite to me, and each of the other impulses is Zeroat the middle of its time unit.

vWhat is claimed is: i

1. In a cable signaling system the method of signaling which consists in transmitting over the cable an impulse element of a unit time duration receiving it as a train of impulses of magnitude mu, m1, m2, -etc.,= in the middle of the nth, (ml-nth, (n-|-'2 th, etc., time units, the magnitudes being so related that the algebraic sum of the magnitudes of even subscripts equals zero and the algebraic sum of the magnitudes'oi oddsubscripts equals and every magnitude previous to ma shall be equal to zero in the middle of its time unit.

2. In a cable signaling system the method of signaling which consists in transmitting over the cable animpul'se element of a unit time duration receiving it as a train of impulses of magnitude mo, m1, 1212, etc.,

signaling which consists in transmitting over the cable an impulse element of a unit time duration, receiving it as a train of unit magnitude in the middle of the nth time unit, of a substantially equal opposite magnitude in the middle of the n+2 th time'unit, and zero magnitudes in the middle of other time units. a

4. In a cable signaling system, the method of signaling which consists in transmitting over the cable an impulse element of a unit time duration, receiving it as a train of unit magnitude in the middle of the nth time unit; of a substantially equal opposite magnitude in the middle of the (n+2)th time unit, and zero magnitude in the middle of other time units, and introducing with the received wave a locally generated impulse in the (n+2) th time unit equal and opposite to that coming from the cable.

5. In a cable signaling system, a cable incapable of transmitting direct current and capable of transmitting alternating current only up to frequency s, the method of signaling with a speed of s dots per second which consists in transmitting impulses of duration modifying the transmission characteristics of the system so that the current value due to a single sent impulse consists of a unit magnitude in the middle of one timeunit, zero in the middle of the next time a. cable in tinuatien of the signals.

of s. dots p'er second which: consists in transmitting impulses of duration modifying the transmission characteristics. of the system so that the characteristic interference occurs chiefly in the ble of transmittingv direct current and capable of transmittingalternating current only up to frequencys, the method ofincreasing the number of current magnitudes which can I be used on the'cable which consists in concentrating the characteristic interference chiefly in a time .unit

subsequent to the arrival of the main impulse and neutralizing the characteristic interference by impulses generated at the receiverand triggered by the main impulse.

8. The combination of. claim 5 characterized by thefact thatithe neutralizing. impulse is timed.

to come in the time unit in: which the interference isconcentrated.

9.,In a cable signaling system for signaling over a cable which is adjusted by equalizers to give at the receiving end animpulse. of the same sign. as. the sent impulse followed by a trailing equal impulse of the opposite sign, the method which consists in sending from the receiving end into the receiver an impulse to neutralize the trailing impuls 10. In a subm rine signaling system comprisof transin direct current 1' alternai current abov irecuency the method. itaining. correct response to a sust'ined series of signals. of frequei s which die. down to zero current at'the' receiver wh ch consists in sending impulses from the receiver into th receiver until the termination ofithe zero current condition which results from the cen- '11. In a, cable system; a cable, a

transmitterand a receiver, means for impressing a sustained series of signals on the cable whose steady state value at, the receiver iszero, a feed ients at the beginning of the series, the feed-- back causing "the receiver to operate in accordance with the sent impulsesuntil the arrival of ialing system, a cable, a-

frequency s, the method of signaling with a speed a transient due toa change at the transmitter end.

13. In a cable signaling system, a cable and a transmitter adaptedto impress'thereon an im pulse, means to. modifythe transmission characteristic of the cable. such-that the received impulse consistsof an impulse of' the same sign as the transmitted impulse followed by asubstantia ly equal trailing impulse of the; opposite sign, means at the receiving end to impress on the receiving apparatus an impulse to. neutralize the trailing impulse.

14. In a cable signaling system, a cable and a transmitter adapted. to impress thereon an impulse, means to modify the transmission characteristic of the. cable. suchthat the received impulse consists of an impulse of the same sign as the transmitted impulse followed by a. substantially equal trailing impulse of the opposite sign, means at the receiving end. responsive to the first received impulse to impress on. the receiv' apparatus an impulse to neutralize th trailing impulse. I

15. In a cable signali11g,systcm,..a: cable and a transmitter adapted to impress thereon an impulse, means to modify. the transmission characteristics of i1 element is spread outv over and is-substan- J contained within three time units and of side +1 at the middle of the first. time unit, zero at the middle of the'second time unit,

and w l the middle of the third? time unit-,and.

means the receiving end to' impress on: the receiving apparatus an impulse to neutralize the portion in the third time unit.

16. in a cable signaling system; a cable,.equalizing networks attached thereto of such value where s is the maximum speed of signaling for the cable.

17. Ina cable signaling system. comprising a cable and a terminal station at each end thereof, a distributor at one endfor impressing: signals on the cable in accordance withamessage to be,

transmitted equalizers attachedtothe cable of such characteristics as to cause the impulse impressed at the sending end. to arrive at the receiving end as a main impulse followed by a trailing impulse of substantially equaland opposite magnitude, a distributor-at thereceiving end operating synchronously withtthe' transmitting distributcr, a feed-back relay circuit con.-

trclled by the received impulses toimpress on the receiver impulses equal and oppositeto: a trailing impulse, and means for adjusting thedelay.

of the feed-back circuit to cause the neutralizing impulse to be impressed at the time of arrival of the trailing impulse- Y HARRYNYQUIST...

the cable such that the received

Images