Transoceanic Telephone Service ShorbWaveransmission

Transmission Features of Short-Wave Radio CircuitsBY RALPH BOWN1

Non-Member

TRUNK circuits between London and New York is good during daylight on the Atlantic. But in thewhich furnish telephone service between these two dawn and dusk period a frequency of about 14,000cities and also permit successful conversation by kilocycles (22 meters) is better. For the dark hours

means of toll wire extensions between the United States something like 9000 kilocycles (33 meters) gives bestand Europe more generally are being carried over both transmission and for midnight in winter an even lowerlong waves and short waves. It is the purpose of this frequency near 6000 kilocycles (50 meters) is advanta-paper to consider the transmission side of the new short- geous. Thus, in considering the short-wave range inwave circuits which the American Telephone and Tele- terms of communication circuits, we must shrink itsgraph Company and the British General Post Office have apparent width materially to take account of the severalmade available for this service. In doing this we shall frequencies required for continuous service.proceed from the more general considerations, relating At the present time the frequency spaces betweento wavelengths and communication channels, through a channels are much greater than the bands of frequenciesdiscussion of the principles governing the general design actually occupied by useful transmission. This elbowof the system, into a brief summary of practical per- room is to allow for the tendency of many stations notformance results. to stay accurately on their nominal frequencies but toThe frequency range so far developed for commercial wander about somewhat. But in spite of this allowance,

radio use is roughly 20 to 30 million cycles wide, extend- cases of interference are common and one of the activi-ing from about 10 kilocycles to perhaps 25,000 kilocycles ties which must be carried on in connection with aper second. There are two parts of this whole spectrum commercial system is the monitoring of interferingsuitable for transoceanic radiotelephony-the long-wave stations and the accurate measurement of transmittingrange which is relatively narrow, extending roughly frequencies to determine the cause of the conflict. Tofrom 40 kilocycles to 100 kilocycles, and the short-wave permit intensive development of the frequency spacerange which in its entirety is much broader, extending offered by Nature the greatest possible constancy andfrom about 6000 kilocycles to 25,000 kilocycles. accuracy of frequency maintenance in transmitting

It is evident that the long-wave region, including sets will be required.perhaps only 50 kilocycles, offers opportunity for The fact that channels have been assigned (withindevelopment of relatively few telephone channels, wide bands set aside for a particular service) with littleparticularly in view of the fact that it is in use by a regard to the geographical location of stations maynumber of telegraph stations. Also it must be borne result in neighboring channels having much strongerin mind that for telephony these waves are suitable signals than those in the channel being received. Whenfor only moderate distances of the order of 3000 miles this is so, a severe requirement is placed on the selec-and for routes in the temperate zones where static tivityofthereceivertopreventinterference.interference is moderate. The first transatlanticradiotelephone circuit opened in 1927 was a long-wave INTERCONNECTING WITH WIRE CIRCUIT EXTENSIONScircuit (58.5-61.5 kilocycles). In providing the next The skeleton of a radiotelephone circuit is in itsfew channels for the initial growth of the service the essentials very simple. It consists merely of a trans-opportunity to determine the utility of short waves mitter and a receiver at each end of the route and twowas embraced. oppositely directed, one-way radio channels betweenThe short-wave range is vastly wider in kilocycles them. These two independent channels must be ar-

but, nevertheless, has its limitations as to the number ranged at the terminals to connect with two-wireof communication facilities it affords. For a given telephone circuits in which messages in opposite direc-route of a few thousand miles a single frequency gives tions travel on the same wire path. The familiar hybridgood transmission for only a part of the day. For coil arrangement so common in telephone repeaters andexample, from the United States to Europe a frequency four-wire cable circuits might appear to solve thisof about 18,000 to 21,000 kilocycles (17 to 14 meters) problem,-were there not difficulties peculiar to the radio

Amercanchannels. In the short-wave case large vrariations in1. Department of Development and Research, Aeiattnainocri,hnrdoptswti hrTelephone and Telegraph Company, New York, N.Y.atnalnocrl h rdoptswt1 hr

Presented at the Winter Convention of the A. I. E. E., New York, intervals of time. These would tend to cause re-trans-N. Y., Jan. p7-31, 1930. mission of received signals at such amplitudes that

624

30-31

April 1930 BOWN: TRANSOCEANIC TELEPHONE SERVICE 625

severe echoes and even singing around the two ends of The function of the Receiving Delay circuit, thethe circuit would occur unless means were provided to Receiving Detector, and the relay RES is to protectprevent this. the Transmitting Detector and relays against operationTo overcome these fundamental transmission diffi- by echoes of received speech currents. Such echoes

culties, an automatic system of switches operated by the arise at irregularities in the two-wire portion of thevoice currents of the speakers has been developed.2 connection and are reflected back to the input of theThese devices cut off the radio path in one direction Transmitting Detector, where they are blocked by thewhile speech is traveling in the reverse direction and relay RES which has closed and which hangs on for aalso keep one direction blocked when no speech is being brief interval to allow for echoes which may be con-transmitted. The operation is so rapid that it is un- siderably delayed. The gain control potentiometersnoticed by the telephone users. Since this system shown just preceding the transmitting and receivingprevents the existence of singing and echo paths, it amplifiers are provided for the purpose of adjustingpermits the amplification to be varied at several points the amplification applied to outgoing and incomingalmost without regard to changes in other parts of the signals.system, and it is possible by manual adjustment to The relief from severe requirements on stability ofmaintain the volumes passing into the radio link at radio transmission and from varying speech load on therelatively constant values, irrespective of the lengths radio transmitters which this system provides permitsof the connected wire circuits and the talking habits much greater freedom in the design of the two radioof the subscribers. channels than would otherwise be possible.

Fig. 1 gives a schematic diagram of the United States THE RADIO CHANNELSend of one of the short-wave circuits showing the

One of the first questions which comes up in consider-______ _- i7 Fr>)m7Fnqarx lng the design of a radio system is the power which can

/ - ffi>s Qedccel 4 be sent out by the transmitter. The word "can" is used,I/ | HLJ- advisedly, rather than "should," since in the present

eWlXllS Q>Qil re art the desideratum usually is the greatest amount ofpower that is technically possible and economically

\\E--5 justifiable. There are few radio systems so dependablethat increased power would not improve transmissionresults. At very high frequencies the generation of

FIG. 1-CIRCUIT DIAGRAM ILLUSTRATING OPERATION OF lVOICE-PERATE SWITCINc, DVICE'Iarge powers iS attended by many technical difficultiesVOICE-OPERATED SWITCHING DEVICE

but fortunately the radiation of power can be carriedessential features of a voice-operated device which has out with much greater efficiency than is feasible atbeen used. This kind of apparatus is capable of taking lower frequencies. At 18,000 kilocycles (about 16many forms and is, of course, subject to change as meters) a single half-wave radiator or doublet is onlyimprovements are developed. The diagram illustrates about 25 ft long and it is possible to combine a numberhow one of these forms might be set up. This form of them, driven in phase by a common transmitter,employs electro-mechanical relays. The functioning into an antenna array which concentrates the radiatedof the apparatus illustrated is briefly as follows: the power in one geographical sector. In that direction therelay TES is normally open so that received signals effectiveness may be intensified 50 fold or more (17 db.)pass through to the subscriber. The relay SS is and waste radiation in other directions reduced ma-normally closed to short circuit the transmitting line. terially. Thus, one of the transmitters at Lawrenceville,When the United States subscriber speaks his voice New Jersey, used in the short-wave transatlanticcurrents go into both the Transmitting Detector and circuits when supplying 15 kw. radiates in the directionthe Transmitting Delay circuit. The Transmitting of its corresponding receiving station as effectively asDetector is a device which amplifies and rectifies the would a non-directive system of about 750 kw.voice currents to produce currents suitable for operating The transmitting antennas also give some directivitythe relays TES and SS which thereupon short circuit in the vertical plane, increasing the radiation sentthe receiving line and clear the short circuit from the toward the horizon and decreasing that sent at highertransmitting line, respectively. The delay circuit is an angles. It is not yet certain that vertical directivityartificial line through which the voice currents require is always advantageous and this effect has not beena few hundredths of a second to pass so that when they carried very far.emerge the path ahead of them has been cleared by the At the receiving station the radiated power hasrelay SS. When the subscriber has ceased speaking the dwindled to a small remnant which must be separatedrelays dropback to normal. from the static as far as possible and amplified to a

2 .For...dealddsrpino .ti ytmse"h volume suitable for use in the wire telephone plant.York-Londo:n Telephone Circuit" by S. B. Wright and H. C. Here again directive antenna arrays are of value. ASilent, Bell System Tech. Jl., Vol. VI, October, 1927, pp. 736-749. receiving antenna system sensitive only in a narrow

626 BOWN: TRANSOCEANIC TELEPHONE SERVICE Transactions A. I. E. E.

geographical sector, and that lying in the direction from which particularly hampers its use for telephony. Thiswhich the signal arrives, excludes radio noise from other is fading. Where fading is of the ordinary type, con-directions and thereby scores a gain of perhaps 40 fold sisting of waxing and waning of the entire trans-(16 db.) in the power to which the signal can be ampli- mitted band of frequencies, automatic gain controlfied without bringing noise above a given value. It also at the receiving station is of value and is employed in thescores against noise which arises in the tubes and circuits transoceanic circuits under discussion. The amplifica-used for amplification, since the combined action of the tion in the receiver is controlled by the strength of theseveral antennas of the array delivers more signal to the incoming carrier and is varied inversely with thisinitial amplifier stage where such noises originate. strength so as to result in substantially constant signal

Thus, it is evident that transmitter power, trans- output. Obviously this control can be effective only tomitting directivity, receiving directivity, and quiet the extent that the signal seldom falls low enough to bereceiving amplifiers are of aid in providing signal overwhelmed by radio noise.transmission held as far as possible above the radio When fading is of the selective type, that is, thenoise. In a well designed system the relative extents different frequencies in the transmitted band do notto which these aids are invoked will depend upon fade simultaneously, the automatic gain control systemeconomic considerations as well as upon the technical is handicapped by the fact that the carrier or controlpossibilities of the art. signal is no longer representative of the entire signal

There is one other type of noise than that provided band.by Nature which is of particular importance at short Selective fading is believed to result from thewaves,-electrical noise from the devices of man. One existence of more than one radio path or route by whichof the worst offenders is the ignition system of the signals travel from transmitter to receiver. These pathsautomobile. The short-wave transoceanic receiving are of different lengths and thus have different times ofstation at Neteong, New Jersey, is so loeated that transmission. Wave interference between the com-automobile roads are at some distance, particularly in ponents arriving over the various paths may causethe direction from which reception occurs. Service fading when the path lengths change even slightly.automobiles which produce interference cannot be If the path lengths differ by any considerable amount,allowed near the antenna systems unless their ignition for example, a few hundred miles, the wave interferencesystems have been shielded. Also, electrical switching is of such a character as to affect the frequencies acrossand control systems incidental to the power, telegraph, a band consecutively rather than simultaneously.and telephone wire systems at the station are shielded With the presence of selective fading there comes intoor segregated. being the necessity of guarding against rapid evenAt both the transmitting and receiving stations at though small variations in the transmitted frequency,

least three antenna systems are supplied for each since if such variations are present a peculiar kind ofcircuit, one antenna for each of the three frequencies quality distortion of the telephone signal results.normally employed. The design and arrangement of The varying load which speech modulation places onthese are dictated by the requirements flowing from the transmitter circuits tends to cause slight variationstheir uses. The purpose of the transmitting antenna is in the instantaneous equivalent frequency which areto concentrate as much power as possible in one direc- known as "frequency modulation" or "phase modula-tion. The purpose of the receiving antenna is to tion" depending on their character. To prevent thisincrease reception from the desired direction and to cut effect the control oscillator must be carefully guardeddown reception at all other angles. In the former the against reaction by shielding and balancing of circuitsforward-looking portion of the characteristic is of and the design must be such as to preclude variablegreatest importance, while in the latterithe rearward phase shifts due to modulation in subsequent circuitscharacteristics need greatest refinement, of the transmitter.It is apparent that if there are two paths of different

TRANSMISSION PERFORMANCE lengths, two components which arrive simultaneouslyIn short-wave telephone systems the width of the at the receiver may have left the transmitter several

sidebands is so small a percentage of the frequency of thousandths of a second apart. If the transmittertransmission that tuning characteristics of the antennas frequency has changed materially during this briefand high-frequency circuits are relatively broad and interval trouble may be expected. The trouble actuallyimpose little constriction on the transmission-frequency takes the form of a distortion of the speech as demodu-characteristic. A fiat speech band is easy to obtain over lated by thereceiving detector.3the range of approximately 250 to 3000 cycles employed Defects in short-wave transmission due to radiofor these commercial circuits. This relieves the short- noise, minor variations in attenuation, fading, andwave circuits from many of the problems of obtaining distortion are nearly always present to some extent and,sufficient band width which are troublesome in design- 3. Fo a dicsso oftihnmnnse"oeSuisi

ing long-wave systems. Radio Broadcast Transmission" by Bown, Martin, and Potter,Short-wave transmission is subject to one frailty I. 1R. E. Proc., Vol. 14, No. 1, p. 57.

April 1930 BOWN: TRANSOCEANIC TELEPHONE SERVICE 627

when any or all are severe, cause a certain amount of and South routes such as between North and Southlost service time. These interruptions are of relatively America, the interruptions associated with magneticshort duration and, furthermore, there is enough over- storms are less severe and of shorter duration.lap in the normal times of usefulness of the several The cycle of events which accompanied a particularlyfrequencies available, so that shifting to another severe magnetic storm4 in July, 1928, is shown graphi-frequency may give relief. There is, in addition, a kind cally in Fig. 2. The light dotted curve shows the varia-of interruption which from the standpoint of continuity tion in the horizontal component of the earth's field.

The heavy solid line follows the daily averages of theTren5mnision on 18.34-MC(16 Meters) Deal, N.J. to New 5avthgote,Engiand.

----o Transmissionon 60 llc.(5000 Meter5) Rocky Point L I to Cupar,Scotland short-wave received sig1al field. It iS apparentthatHorizontal Component ot Earth's Magnetic Field.,a p r n

oo - - -r M -e - - the disturbance took two days to rcaeh its peak and- Commercial Uncommercial No Datal

__ o New York to London

0 18~/ -- 720

DasBefore Storm Doys After.. l

oFI.2-EFFECT OF MAGNETIC DISTURBANCES ON RADIOIITRANSMISSIONI I II

of service is more serious. At times of disturbance of Lonon to New'Yrthe earth's magnetic field, known as "magnetic storms," 8short-wave radio transmission is generally subject tosuch high attenuation that signals become too weak to _use and sometimes too weak to be distinguishable.4These periods affect all the wavelengths in use and may zlast from a few hours to possibly as much as two orNon2three days in extreme cases. They are followed by a _ _recovery period of one to several days in which trans- < °mission may be subnormal. 8-

Severe static m~ay cause interruption to both long- land short-wave services at the same time but the short 6 I Xwaves are relatively less affected by it and are usually lllable to carry on under static conditions which preventl lllsatisfactory long-wave operation. On the other handIIsevere fading or the poor transmission accompanying June July August SepLa magnetic disturbance may interrupt short-wave FG -HR HWN RNMSINPROMNEO

service without affecting the long waves adversely,- SHORT-WAVE TRANSATLANTIC TELEPHONE CIRCUITin fact magnetic disturbances often improve long-wavetransmission in the daytime. The service interruptions the recovery to normal took nearly a week. The heavyon the two types of circuits are thus nearly unrelated to dotted line shows received field on long waves (60each other and have no definite tendency to occur kilocycles) and indicates that transmission wassimultaneously. This is the principal reason why both improved slightly at the same time the short waveslong-wave circuits and short-wave circuits appear were suffering high attenuation.essential to reliable radiotelephone service. The experience with transatlantic telephone serviceOn routes which are very long or which cross tropical on short waves covers a period of nearly three years,

areas which result in static sources facing the directive* * 1 1 ~~~~~~~~~~~~4.Data regarding other mnagnetic disturbances are given in areceiving antennas, long waves cannot as yet be suc- paper by C. N. Anderson, entitled "Notes on the Efect of Solar

cessfully employed and short waves alone are available. Disturba.nces on Tra,nsa,tla,ntic Radio Transmission," I. 1?. E.However, experience tends to indicate that on North Proc., Vol. 17, No. 9, September, 1929.

628 BOWN: TRANSOCEANIC TELEPHONE SERVICE Transactions A. I. E. E.

there having been available a one-way channel from course, not as severe as the disturbance illustrated inthe United States to England used as an emergency Fig. 2.facility for the first year and a half, a two-way circuit It is apparent that for these three summer monthsfor the next year, and two circuits since June, 1929. this new circuit gave a good account of itself andIt is only in this later period, however, that a circuit furnished commercial transmission for something likehas been available operating regularly with the amounts 80 per cent of the time that service was demanded of it.of transmitter power and antenna directivity which have In these same months the long-wave system sufferedbeen mentioned. its greatest difficulty from static, and we have con-The performance of the two one-way channels form- cretely illustrated the mutual support which the two

ing this circuit is charted in Fig. 3. The charts are types of facilities give each other.plotted between hours of the day and days in the year It should not be inferred from these data that theso that each unit block represents one hour of service short-wave transatlantic radio links furnish 80 per cent

of the time talking circuits as stable and noise free asgood wire lines. Under good conditions they do provide

43> ~facilities which compare favorably with good wirefacilities. On the other hand they may at times be

9X/. maintained in service and graded "commercial" undernAPters5rx conditions of noise or other transmission defects for

ReceMvng which wire lines would be turned down for correction,5tation

since the obviously undesirable alternative is to give no/°c\o OMrl¢0service at all until conditions have improved again.

The present development effort is largely directedtoward improvements which will insure not only agreater degree of reliability against interruptions butwhich also will improve the grade of service as a whole.

08un8oak ">,e52 In the foregoing little has been said about the sta-08m#:tions and plant since a description of these and the

operation of them are treated in two companion papersby Messrs. Cowan and Oswald. It may be well, how-

-\0 ever, to view the physical scene broadly as set forth onTwran5meittilerng N the accompanying map, Fig. 4.

5tatioNQj t , The geographical arrangement of the transmittingand receiving stations was governed among other things

(7Ot10 by transmission considerations. The two stations wereplaced about 50 miles apart because this is approxi-mately the distance for minimum signal and at a lesser

FIG. 4-MAP SHOWING TRANSMISSION CONSIDERATIONS AF- or greater distance the signals from the AmericanFECTING LOCATION OF STATIONS transmitter might be strong enough to offer some

interference to receiving the English or South Americantime. The solid black areas are time in which com- stations on adjacent channels. For the same reasonmercial operation could be carried on. The dotted they were placed at considerable distances from thestrips are uncommercial time. The blank areas are transmitters and receivers of other communicationfor time in which, for one reason or another, the circuit agencies. The Netcong receiving station lies to thewas not operating and no data were obtained. Perhaps north of the Lawrenceville transmitting station so asthe most outstanding feature of these charts is the not to be in paths of strong signals from the directivetendency of the lost time to fall in strips over a period of antennas which face northea'st toward England andtwo or three days. These strips coincide approximately southeast toward South America. This configurationfor both directions of transmission. The principal ones also places the transmitter outside the sensitive anglesare about July 10, 15 and August 2 and 17. These are of the directive receiving antennas.characteristic of the interruptions accompanying mag-netic disturbances of the kind which occur at irregular Discussionintervals of a few days to several weeks. They are, of For discussion of this paper see page 643.