Nos. 7 to 12 KARL FREDRIK WINCRANTZ. Managing Director of ...

VOL. 7 1930Nos. 7 to 12KARL FREDRIK WINCRANTZ.Managing Directorof Telefonaktiebolaget L. M. Ericsson1922—1930.ENGLISHEDITION

THE L. M. ERICSSON REVIEWENGLISH EDITION.JOURNAL OFTELEFONAKTIEBOLAGET L. M. ERICSSON, STOCKHOLM.Responsible publisher: HEMMING JOHANSSONEditor: WOLDEMAR BRUMMER.Issued quarterly. / / / / / / / / / / / / / / / / / Yearly subscription rate: 7/ -All communications and subscriptions to be forwarded to the Editor.Karl FredrikWincrantz.Karl Fredrik Wincrantz, born in Stockholm in1874, graduated from the University of Technologyin 1897, but already in 1893, while still a youngundergraduate, he was engaged as an assistant inthe Royal Board of Telegraphs, where he remaineduntil 1900, when he entered the service of theStockholms Allmanna Telefonaktiebolag as its AssistantManager under Mr. H. T. Cedergren. During1898—1906 he also served in the Royal PatentOffice, where he was some time Secretary, sometime Chief Engineer.On the death of Mr. Cedergren in 1909, Wincrantzwas appointed his successor as ManagingDirector of the A.-B. Stockholms Telefon, formedthe year before, which owned and ran the largeprivate telephone net of Stockholm and neighbourhood.He led this firm until 1918, when itwas taken over by the Swedish Government.The Aktiebolaget Stockholms Telefon was thenre-organized as the Allmanna IndustriaktiebolagetH. T. Cedergren, which firm took over the StockholmTelephon Cable Works at Alvsjo and workshopsin Stockholm. Wincrantz was at the sametime appointed Manager of the new firm.In 1922, the Industri A.-B. Cedergren was amalgamatedwith Telefonaktiebolaget L. M. Ericsson,and with Mr. H. Johansson Wincrantz became aco-Manager of the enlarged company. In 1925, thecompany general meeting having resolved thatthere would be only one managing director, Wincrantzwas appointed sole Managing Director, andretired from that position on September 3rd ofthis year.- 98 -

The Svenska Radioaktiebolaget Audio-frequency Generator, withcontinuously variable frequency Adjustment.By TorbernLaurent.I. Introduction.The present paper is a description of theSvenska Radioaktiebolaget Audio-FrequencyGenerator, Model TFG 529, intended for audiofrequencymeasurements. The following pointsare satisfied by this generator:1) The frequency must be continuously variablefrom 200 to 10,000 cycles, with possibilityto increase the range by a variable auxiliarycondenser for the 50—200 cycles range.2) The audio-frequency current must be practicallysinusoidal.3) The frequency adjustment dials must allowa sufficiently accurate setting of the frequencyfor any and every occurring measurement.4) The frequency must be practically unaffectedby the properties of the valves, and of voltagevariations in the D. C. supply of the audiofrequencygenerator.5) The frequency must be practically unaffectedby the output utilized.6) Utilized output must be practically thesame, whatever frequency the apparatus is setfor.7) The audio-frequency generator must becapable of a minimum effect of 200 m\V.8) The audio-frequency generator must notgive rise to static or magnetic fields which mayhave a disturbing effect in the neighbourhood.0) The audio-frequency generator must notcause disturbing currents in the D. C. batterysupply leads connected to the generator.Points 1, 2, and 7 are intended to make anyusual audio-frequency measurements possible. Ifpoints 3, 4, and 5 are complied with, frequencymeasurements will be superfluous. A calibratingtable for the frequency adjustment dialsof the audio-frequency generator, made up onceand for all, will enable the generator to be setfor any frequency desired. Compliance withpoint 6 facilitates measurements with constanteffect at different frequencies, and provides acertain protection for thermo-couples possiblyused in the measuring devices.Points 8 and 9 are introduced to allow themeasuring devices to be placed in the immediatevicinity of the audio-frequency generator,as is usually done when a test board is used, andalso to allow the generator to be run by directcurrent from the same batteries as other appliances,e. g. telephone repeaters.In order to satisfy all these demands by comparativelysimple means, the design of the audiofrequencygenerator has been based on an entirelynew principle, whereby many important simplificationsand improvements have beenachieved.To obtain a suitable design, it has been foundexpedient to introduce a few extra valves. Butas measuring devices are not used in large numbers,nor continuously, the increase in the numberof valves cannot be considered an economicdrawback.2. The principle of the Audio-FrequencyGenerator.The new fundamental principle applied in theSvenska Radioaktiebolaget Audio-FrequencyGenerator is the use of a serial resonance circuitas a tuning circuit, instead of the parallel resonancecircuit previously exclusively employed,which endows the audio-frequency generatorwith particularly valuable properties as regardspoints 1, 2, and 6, while at the same timeseveral important advantages are gained fromthe point of view of manufacturing.Fig. 1 is a diagram of the principle of theaudio-frequency generator. The tuning circuitconsists of a variable condenser C. in series withan inductance coil L, which by means of a switch99

Fig. 1.may be taken into use wholly or partly. Whenonly a portion of the induction coil L is used,a resistance r„ r v or r, is also introduced intothe tuning circuit.The potentiometers P x and P., are in serieswith the tuning circuit. From the former thereaction voltage is taken, and from the latterthe grid voltage for an amplifier (4.,, T u A vA„ T 2 ), from the output transformer of whichthe audio-frequency current is supplied. In thisgenerator two oscillating valves are used, 4, andA t , which are connected in cascade by meansof the resistance coupling M.A special problem has been the arrangementof the D. C. feed to the valve 4,, which on accountof point 9 must prevent access of theaudio-frequencies to the anode battery, while atthe same time the arrangement must not act asa load on the valve A 2 , either as regards capacitycr inductivity. Experience had proved thatwhile a D. C. feed through a resistance, with thecondenser connexion to the cathode necessitatedfor passing the alternating currents, requires anexcessively large condenser, a D. C. feed by chokecoil requires this coil to possess excessively highself-induction.By the device shown in fig. 1, however, theproblem is solved in a satisfactory manner. Thisdevice consists of two resistances R1, and R2.,two large condensers K l and K 2 . and two largechoke coils D 1 and D 2 . The impedance formed bythe coil D.,, connected in parallel to the condenserA',, is negligible in comparison to the impedanceof the coil D,. The condenser A'„ with acapacity K v and the coil D,, with a self-inductionD, and a loss resistance R D as well as the resistancesR, and R.,, with the resistance valuesR l and R., respectively, are so dimensioned thatFor all frequencies within the normal workingrange of the generator, the impedance betweenanode and cathode of this device will be equal toan ohmic resistance Jt x .The resistance coupling M with its choke coils(not shown in the figure) is also dimensioned sothat the phase angle between the EMF in theanode circuit of the valve A x and the grid voltageof valve A 2 for practical purposes may be disregardedfor the generator's frequency range.According to the above, the reaction voltagewill thus be in phase with the original oscillations,which is an essential condition for ensuringfrequency stability as desired by point 4.By using two oscillating valves, the phase ofthe reaction voltage is changed 180° so that theconditions necessary for oscillation are obtained.The natural frequency of the generator is thusdetermined solelv bv the current-resonance of100

the condenser C and the coil L. The condenserC, made up of mica condensers and one air condenser,has a loss resistance of negligible magnitude,and the coil L, which has no iron core,has a loss resistance practically independent ofthe frequency, and equal to its D. C. resistance.By current-resonance, the impedance of the tuningcircuit will therefore be practically equal tothe D. C. resistance of the utilized part of thecoil with a resistance r v r„, or r 3 added, sufficientto make it equal the resistance of thewhole coil. This impedance is therefore a pureresistance, the size of which is independent ofthe size of the condenser C and of the portionof the coil L utilized. Any voltage and currentin the oscillating system will thus be independentof the frequency setting, that is to say, theamount of reaction and, apart from a small distortionin the amplifier, the amount of outputwill be independent of the frequency setting. Bypoint 5, the output should be independent of thefrequency setting, but the fact that the amountof reaction is also independent of this settingis of great value, particularly for obtaining ahigh frequency stability and satisfying the demandsof point 2.The design of the tuning circuit is interesting.The loss resistance of the induction coil L should,on account of points 2 and 4, be of approximatelythe same magnitude as the resistance P, andthe potentiometer resistances P, and P 2 . Thereason for this is that the frequency stabilityand the ability of the oscillating circuit to filterharmonics increase with the reactance of thecoil compared to the ohmic resistance in the circuitpassed by the oscillating currents. As theloss resistance of the coil must be comparativelysmall to prevent the self-induction from beingexcessively large, the resistance R i must besmall, which is feasible by using two oscillatingvalves, with consequent powerful excess of amplification,which may be suitably consumed bythis resistance. For the same reason, the potentiometerresistances P, and P., should besmall, which is all to the good, as the potentiometersare designed as rheostats. However, thedesign requires the induction coil L to be madecomparatively large, whereby the condenser Cbecomes comparatively small. This also is entirelybeneficial, as the impedance of the inductioncoil is purely a matter of winding, whichhardly affects the cost of the coil, while the costof a condenser is practically proportional to itscapacity.3. Apparatus design.Fig. 2 shows a front view of the Audio-FrequencyGenerator, and fig. 3 shows the arrangementsat the back of the panel.I 1851 Fig. 2.The manufacture of the induction coil is rathera troublesome problem. Iron must beavoided in the coil for the sake of the frequencystability; to secure frequency stability, and withdue regard to the task of the oscillating circuitto filter away harmonics, the loss resistance ofthe coil must be comparatively small in proportionto its self-induction; to avoid any possibledisturbance of the neighbourhood (point 8), thecoil must not give rise to any external magneticfield. The problem is solved by using an aircoilenclosed in a 3 mm thick copper casing,visible in fig. 3 as a cylindrical box marked"CUB 240". For the highest frequencies anothersmall coil, marked "CUB 1 a" is used. Fig. 3also indicates how the other component partsof the Audio-Frequency Generator are arranged.On the front of the Audio-Frequency Generatorpanel (fig. 2) the terminal connexions of theD. C. voltages are shown on the left, and on theright the terminals for the audio-frequency current.At the bottom a jack-strip is seen, to whicha I). C. instrument may be plugged for checkingthe anode- and grid currents. The four lowerdials belong to a decade condenser with a variableair-condenser for fine adjustments, and- 101 -

the induction coils are connected as desired bythe two keys located above. As will be readilyunderstood from the above, these dials and keysare used for setting the frequency. The filamentvoltage, indicated by the voltmeter locatedbelow the valves, is adjusted by the dial to theleft of the keys; the dial on the right controlsthe audio-frequency power output, and the reactionis adjusted by the dial above the keys.A 6-volt filament battery, a 120-voIt anodebattery, and a 24-volt grid battery are requiredfor running the audio-frequency generator.The audio-frequency generator works the bestwhen the I). C. voltages are correctly adjustedand the reaction is kept near the singing point.The audio-frequency generator shown in figs.2 and 3 can be provided with a can cover andfitted on a test rack instead of in a box.Fig. 3.The Use of Personal Telephone calls in Sweden, and in TrafficBetween Sweden and other Countries.The long distance telephone traffic of Stockholmhas increased considerably during 1929.In 1928 the number of trunk calls emanating fromStockholm was 4.663.033. The correspondingnumber of calls in 1929 was 4.889.418, showingan increase in the number of outgoing calls by4.9 per cent.The number of incoming trunk calls in Stockholmis about 10 per cent, larger than that ofoutgoing calls, and the incoming traffic during1929 may therefore be estimated at 5.380.000calls. In considering these figures from the pointcf view of comparative frequency, however, itshould be remembered that the Stockholm freedistrict includes 180 Local Exchanges outside thetown proper, involving about 32.000 subscribers'stations within a radius of 40 kilometres (24miles), and that up to 200.000 free calls — shortdistance or district calls — are put through oncertain week-days, or about 52.000.000 annually.Similar calls abroad are almost invariably chargedfor and included in the number of trunk calls recorded.Personal calls continue to increase in number,and at a more rapid rate than the total traffic.Personal calls originated in Stockholm thus increasedfrom 1927 to 1928 by 7 per cent., andfrom 1928 to 1929 by 9.4 per cent.In 1924, 29 per cent, of all calls originating inStockholm were pre-advised.In 1927 the proportion had increased to 39 per cent.» 1928 • » » » , 41 »> 1929 » > > > > 43 > >102

showing a continuous large increase in personalcalls. This indicates that the Swedish public fullyappreciates the advantages of the personal call.To the administration, these personal calls — inspite of the low pre-advice fee — mean an increaseof income which, as far as Stockholm isconcerned, covers 50 per cent, of the total costof the supervision and service staff at the TrunkExchange.The transmission of pre-advices on the trunklines naturally takes up some time, but this isvery short, amounting to an average of at theoutside 15 seconds each time. Out of the 477trunk lines serving Stockholm, 228 of which areexclusively employed for outgoing calls, only about3 lines would therefore be occupied for this purposeif all the advices from Stockholm could beassumed to be transmitted one after another onthese lines, irrespective of the destination of thecall.The following table indicates how the personalcalls emanating from Stockholm are distributedamong the several Rate Districts:In comparison to the foreign traffic of 1928,this is an increase of 54 per cent, in the numberof calls, and of 46 per cent, in the number ofpre-advised calls. Personal calls from Stockholmhave increased by 53 per cent., while the increasein personal calls to Stockholm is 36 per cent.As it may be of interest to compare the use ofpersonal calls in the traffic with the differentcountries, a table is given below showing theStockholm traffic during 1929 on some of theprincipal international trunk lines.The traffic to the neighbouring countries ofNorway and Denmark shows the highest percentageof personal calls, and their incidence isabout the same in both directions.74 per cent, of the calls to, and 70 per cent,from, Norway, were pre-advised.The corresponding figures for the Danish trafficare 65 and 61 respectively. The high percentageof personal calls is accounted for by thefact that this form of call is quite familiar tothe traffic with these countries since many years,and the percentage is approximately the same asfor the longer trunk lines of Sweden.But for other countries the differences arelarger:As already stated, the personal calls emanatingfrom Stockholm during 1929 amounted to 43 percent, of the total number of calls.The increasing use of personal calls with increasingdistance is very apparent in this table.If we now pass on to the telephone trafficbetween Stockholm and abroad, we find a considerabledevelopment during 1929. 317.766 internationalcalls were put through, 153.344 ofwhich were pre-advised as personal calls or ascalls to a specified extension line.This shows the percentage of personal callsfrom Stockholm to these countries to be considerablyhigher than in the opposite direction.As the fees in international traffic are manytimes larger than in inland traffic, one would expectthe percentage of personal calls from abroadto be larger than has actually been the case.When the fee is high, it is of course more importantto find the person to whom one wishesto speak, or to obtain the information desired,than when the fee is lower, so that the heavyexpense of the call may not be wasted. Consideringalso how the call period is computed, a personalcall is worth while. In a call to a certainstation, the taxed call period is reckoned from the103

Personal calls allowed only from Oct. 1929.mcment when the two stations are connected, eachhaving previously replied to the call signal. Theinterval between this and the moment when theperson enquired for by the caller arrives at thetelephone apparatus, if he is net pre-advised, mayfrequently cost quite a lot of money.The reason why the personal call is so muchmore rarely used in traffic to Sweden, however,may be found partly in the fact that this formof call is new in most other countries, and thatconsequently the subscribers there are not awareof its existence and value, and partly in that theinternational regulations for taxing personal callswere originally unfavourable in so far that inmany cases the caller had to pay for a call eventhough he did not find the person to whom thepre-advise was directed. From October 1st 1929,however, these regulations are improved so thatno call fee is payable in case the person preadvisedis not at hand.Gradually, as knowledge of the personal calland its advantage to the subscriber becomes moreuniversally spread abroad, the frequency withwhich it is employed will certainly increase. Theexpansion during 1929, in comparison to 1928,is already considerable.In the directionfrom Germany lit Sweden, these calls have increasedduring 1929 by 18 per cent.,from Holland to Sweden, these calls have increasedduring 1929 by 60 per cent.,and from Switzerland to Sweden, these calls have increasedduring 1929 by 197 per cent.Stockholm, February 14th 1930.A. Lignell- 104 -

The Hallsberg Electric Interlocking Signal Plant.By Herman Holmqvist, Signed Engineer.Progress in the field of electric interlockingsignalling deviees has heen rapid in recentyears, and the system applied in the Malmo plant(1925) led to further improvements. The firststep was the Hasselhohn interlocking plant ( 1926),where the mechanical locking system was whollydiscarded, and the levers only connected electrically.As this plant has now heen in use for sometrafficked junction. The plans were therefore revised,and the result may be said to be a compromisebetween the old system and the new whichis of great interest and has worked well in practice.The Hallsberg railway station is a junction oftwo important railway lines, the electrified Stockholm—Gothenburgline and the steam Krylbo—Mjolby line, which latter has double tracks fromFig. 1.years, and has proved perfectly satisfactory fromthe point of view of security and reliability, thetime has been considered ripe to apply the systemto other large installations planned, primarilythen at Lund, Gothenburg, and Stockholm C. Inthe mean time, however, plans for another largeplant, Hallsberg, had been completed in the beginningof 192.S, with the intention of using theelder mechanical locking devices, i. e. allowingonly two train routes to he laid by each lever.But the advantages of the Malum system, permittingtrains from any of the lines to enter or leaveall tracks, were considered desirable at this heavilyOrebro to Hallsberg. On account of the topographicalconditions the Krylbo—Mjolhy line is nota through line, but the tracks from Kuinla andfrom Asbro both enter at the east end of thestation yard. This implies a change round or exchangeof the engines of all the trains at Hallsberg.and the Palsboda track must necessarilybe crossed either coining in or going out. Inaddition there are through carriages to and fromOrebro in a majority of the trains on the StockholmGothenburg line, involving a hurried shuntingof passenger coaches from one train to another.Train traffic goes on all day and night.1 105

Fig. 2.

with groups of express trainsbetween 1—2 and 3—t a. m.Further, a number of goodstrains, usually very long, passthe station on their way toand from the shunting yard,situated immediately to thewest of the passenger station.This is a centre for the StateRailway long distance goodstraffic to and from Gothenburg,Malmo (Xassjo), Stockholm,and Krylbo (Norrland).As far as possible, the more importantgoods trains are madeup and sorted out in this marshallingyard, and the work ismostly done at night.The above indicates thatsafety devices at this placemust fill a very real need, andsuch have therefore long been planned.The electric interlocking plant now erected atHallsberg is principally concerned with the passengerstation only. The attached diagram (fig. 1)shows the disposition of the tracks in the junction,4 passenger tracks and 2 goods tracks.Between the passenger tracks there are two platforms,connected mutually and to the stationbuilding by subways. The signal cabin is situatedFig. 4.107Fig. 3.on the cuter platform between the covered instairways to the passenger subway, and is intendedto be run by the train dispatcher alone,which for the present, however, does not seemfeasible without assistance when large train groupsare passing. In the signal cabin (fig. 2) is anilluminated track plan, as well as shelves for relays,instrument panel, etc. Space being ratherlimited on account of the siting of the cabin, thecuter wall towards track No. IVhas been doubled, forming akind cf bay (fig. 3). The outerwall consists of hinged shutters,and on the inner wall theterminals of most of the cablesfrom the yard, safety devices,signal transformers, crossinggatesrelays, and so on aremounted.All the tracks are providedwith track circuits extendingbeyond the distant signals,whereby the arrival of trainsis signalled. This is essential,as no less than five pairs oflevel crossing gates are controlledelectrically from thecabin. All main signals areelectric light signals, and thedistant signals are gas light

signals. On account of the curving track, thehome signal from Asbro has a repeat signal 150metres further out, showing only green lights, butextinguished when the home signals indicates stop.A number of dwarf signals are also put up in thestation yard, normally showing the neutral signal— two lights at an angle of 45° to the right(fig. 4). The dwarf signals are in this instancenot used for the guidance of ordinary shunting,lint are designated for the laying of starting routesand for the passage of goods trains to and fromthe shunting yard. The centrally controlled pointsare as a rule connected in pairs on the same lever,all of which are provided with point locking devicesin connexion with track circuits. The pointscan also be changed over locally from localoperating contacts in the yard. Certain points andscotch-blocks are locked by electric locking devices,and others are locked by control locks, the keys ofwhich are kept in locks, electrically connected tothe respective coupling circuits, in the signal cabin.A smaller, mechanical, switching stand is put upin the shunting yard for safeguarding the start ofthe goods trains towards Ostansjo, which have tocross the track leading to the shunting yard.A detailed study of the signal interlocking gearshows that this is comparatively roomy, withspace enough for 72 levers numbered in sequence.Signals and points are given the same numbersas their respective levers. The dwarf signals aredesignated by the letter D, and the main signalsby the letter H, for instance D 61 and H 69/70,The former signifies that this dwarf signal hasbeen set for neutral when the lever 61 is to theleft, and the latter that the main signal is controlledby the levers 69 and 70. The points, aswe have said above, being generally interconnected,both points are given the same number,e. g. 18, the one furthest westwards being called18 a, and the other one 18 b.The interlocking plant is of the Ericsson standarddesign, but with an unusually broad tracklocking gear, providing space for no less than40 road rulers (fig. 2).To lay the points for an incoming train, twolevers have generally to be moved, an inner one,choosing the right track, and an outer one actuatingthe light signal. Either lever locks the pointlevers in its group mechanically, and checks thestop position of the dwarf signals required. Theincoming tracks are indicated to the engine driverby different main signal combinations, but notby the dwarf signals as is the case for the outgoingtracks. The start signals show a green anda red light only, but each outgoing route is indicatedby two dwarf signals also, each controllingits section of the track. Three levers musttherefore be moved to lay an outgoing route.The centre dwarf signal is first set, and movingthe lever sets this in the 45° go-position. Whenthe inner dwarf has also been set to 45°, theouter start signal lever is moved, which sets thestart signal to clear line and the two dwarf signalsto the 90° go-position. On the power cable gallowsacross the platform repeat signals withgreen lights are fixed, announcing to the stationmaster when the line is clear for start, when thenumber of lights enables him to check which ofthe three lines is indicated. If the train is so longthat it reaches beyond the first dwarfsignal, thego-signal may still he given from the start signal,although the dwarf signal will only show 45°.No clear line signal can be given for eitherincoming or outgoing tracks until the level crossinggates are down. The normal procedure istherefore first to set the signal levers for theproper route, which does not give a clear linesignal. When the train enters the outermosttrack circuit, or for the outgoing line when thetrain is nearly due to start, the gate lever is turned.When the gates are right down, which, includingthe cautionary ringing signal, will takeabout 40 to 50 sees., the clear line signal automaticallyappears in the respective signals. Thewiring connexions are such that when the signalhas once shown clear line, the gate lever mayagain be turned to its normal position, whichdoes not affect the gates until the last truck axlehas passed the level crossing, when the gates areautomatically raised.The abovementioned division of the train routesinto several sections, each with its signal lever,makes it possible to combine these levers so thateach line entering the junction may be routedto or from any track, wherever the track systemwill permit this. Track V and VI may also beentered or left by two routes, an inner one bypoints No. 39 and an other one by points No. 35.This is of great importance in avoiding, as muchas possible, blocking the level crossing in thewestern part of the station yard by the long goodstrains.— 108 —

R 1724 Fig. 5.The new L. M. Ericsson designs of operatingmachinery, incorporating a point lock (fig. 5),control the points electrically. The motors aredriven by 130 volt D. C. By discarding the hooklocking device (fig. 4) lubrication and cleaningof the points is facilitated, especially in wintertime when ice and snow often interfere with theworking of the hook locks. Rollers are furtherprovided underneath the point tongues, whichwill carry the tongues during switching and makelubrication of the points quite unnecessary.The arrangements for local operation of thepoints are of new design in so far as, to allowthis, the switch lever must not only be placed inan intermediate position, but a relay placed ina small box close to the switching lever mustalso be dropped. This is done by means of atumbler switch above the interlocking gear (fig. 2,the far part of the cabin). A control lamp inthis box is also lit when the relay is attracted.When the relay drops, the motor circuit fromthe levers is broken and the scotch-block is disconnectedfrom the track circuit. Permission forlocal switching may thus be given and retractedeven if the train is standing on the track circuitthrough the points concerned, which is a greatadvantage when the trains are so long that thereis not room enough for them in the loops. Adim light at the local operating controls in theyard announces in the now usual manner thatthe points may be operated locally.The loose key used for local switching mustbe taken out after each switching operation andinserted the opposite way in the key-hole. Thisis designed so that when the key is inserted thehandle must point the way in which the tongueshave to move. When the points are connectedin pairs, the points furthest from the local leveralways move the first, which makes it easy tosee when the switching is completed, as therewill be no current for the near points until thefar ones have closed. As an additional check,a switching light is always provided at the farpqints.As mentioned above, five pairs of level cross-Fig. 6. Fig. 7.— 109 -

ing gates, the furthest of which is no less than950 metres distant towards Palsboda, are controlledfrom the signal cabin. No shunting takesplace either at this crossing or at the one on theAsbro line, and consequently no proper gatesignals are given for the trains. The droppingof the gates here is controlled solely by mainsignals. As the motor traffic at these level crossingsis very heavy, light-signals are installedwhich show a red light to the road when theinterlocking gear by means of a control lock,provided with contacts on the local lever. Thismakes the main signal independent of the positionof the gates. When the gates are set forlocal control, a dim light on the local lever contactis lit. When the gates are set for centralcontrol this is shown by a light in a control lampclose to the switch in the cabin.The illuminated track plan in the signal cabin(tig. 2) is of a neat and practical design, andFig. 8.cautionary ringing signal logins. A special devicein the gate lifting machinery disconnects thesesignals while the bar is still lifting, before it isright up. Similar light-signals are installed atthe level crossing in the western part of the stationyard. The gates are raised and lowered byelectric winches of the usual type, placed nearthe gates and joined to them by steel wire ropes(fig. (>). Contact devices on the gates provide acheck that the gates are actually down. Switchesin the signal cabin (fig. 2 to the right) make andbreak the current for the bar winches. If thecurrent should fail, the winches may be operatedby means of a crank handle. The gates at thecrossings where shunting occurs may also beclosed or opened on the spot. For this purposea special switch is turned in the cabin, and theguard at the gates must then drop a relay in theindicates the position of both track relays, signals,and level crossing gates. The track plan lampsare 3 and 6 volt telephone lamps behind varicolouredlenses. White lights on the track circuitsindicate that these are free of vehicles.When the lamps in a dwarf signal in the planare extinguished, that signal shows stop. Neutralis marked by a white light, the 45° position bya yellow light and the 90° position by a greenlight. The main signals show red and greenlights. When the distant signal shows a greenlight, this is also shown on the plan. When thelight is changed to white, the lamp in the planis turned out. The level crossing gates in theplan are marked by lamps showing a red lightwhen the bars are up and green when they aredown. There is also a plain check light for thelevel crossing signals towards the roads. The— 110 —

signal levers that must be moved for a certaincombination of the various train routes are plainlynoted on the track plan. Normal points positionscan also be read from the plan. A trainroute may thus be laid without recourse to lockingschedules, solely by observing the positionof the points in the plan and the number of thelevers for the route in question.A special connexion has been used for the trackcircuit, with a condenser connected in front ofD. C. primary cells, and have repeating relays inthe cabin.A machine room for the supply of current isarranged in one of the station buildings, wherethree-phase 220 volt A. C. is provided by therailway light-supply (fig. 9). A reserve of 2X220volt D. C. is also available from local supplylines. D. C, 130 volt for the point-driving andlevel crossing gate motors, and 30 volt for controlcurrent, is obtained from two Westinghouse me-Fig. 9.the feed transformer. This gives a practicallyconstant secondary current, which is advantageousfor the shunt values of the track circuits. Thetrack relays are of two-phase type and placedin the interlocking machine, with relay transformersin cast iron boxes along the track ( fig. 4).Feed transformers and condensers are placed inwooden cases in the station yard (fig. 7). Trackcircuits outside the main signals are fed fromtal rectifiers (to the left in figs. 8 and 9). In caseof A. C. failure, this may also he obtained froma D. C. driven rotary converter I to the right infig. 8).A plant of this size naturally requires a largecapitaloutlay, but the operating economies effectedby the reduction of staff made possibleby the installation of these safety devices providegood interest on the initial expenditure.111

On Cross-talk between Telephone Lines.By M. Vos.Introduction.The present theoretical research has for itspurpose to calculate the cross-talk betweentwo parallel homogeneous double wire lines, e. g.two homogeneous overhead lines located on thesame poles. The immediate reason for makingthis research was two questions which le ComiteConsultatif International des CommunicationsTelephoniques a Grande Distance (C. C. I.),das Nebensprechen in Kombinierten Fernsprechkreisen»,E. T. Z., 1920, page 188, also publishedby J. Springer, Berlin, 1919; F. Breisig, "Oberdas Nebensprechen in Fernsprechkreisen",E. T. Z., vol. 42, 1921, page 992; and K. Kupfmuller,"Cber das Nebensprechen in mehrfachenFernsprechkabeln und seine Verminderung",Arch. f. Elektr. vol. 12, 1923, especially Part III"Theorie des Nebensprechens in langen homogenenLeitungen", page 173.Fig. 1.submitted to the 3rd Commission of Assessors2 at its plenary session in Berlin in June1929. The questions concerned were 6 b and16 b, which in translation read as follows:1) What value should he prescribed for crosstalkbetween any two carrier channels in thesame transposition system.2) Choice of a system of uniform transpositionsin short sections for aerial line networksin various countries.Cross-talk between homogeneous telephonelines has previously been theoretically dealt withby several investigators. Amongst these mayhere onlv be mentioned: R. Lichtenstein, "Cber1 A lecture delivered at Ihe Meeting of the Swedish F.Iectro-Kngineers Association, on March 7th. 193.2 Peals with questions relating to Transmission and Maintenance.As the method of calculation used by theauthor differs from those used by earlier investigators,and certain important, not generallyknown results have been attained, a publicationof the present research may be justified.Most earlier investigators have concentratedon the calculation of the so-called near-endcross-talk, which in ordinary audio-frequencytelephony is of the greatest interest. In carriertelephony with different carrier frequencies forthe two directions of talk, far-end cross-talk is ofthe greatest interest.Fig. 1 shows diagrammatically the ordinaryarrangement of two such carrier current channelswhich operate with the same carrier frequencies,one for each direction, on two separatedouble wire lines on the same poles between the112

terminals A and B. 5, and S,' are two transmittersat A, which operate with the same carrierfrequency fa S., and S,' are two transmitters atB, which operate with the same carrier frequency/„; the frequencies generated at A by S, and S/are received at B by the respective receivers M,and M1' the frequencies generated at B by S 2and S 2 ' are received at A by the respective receiversif, and A/,'. Consequently the transmissionfrom A to B occurs for both channels withthe same carrier frequency f v and from B to Awith the same carrier frequency f 2 . In order thatturally unable to suppress the frequencies whichare generated by the transmitter at the otherterminal, and induced into the channel throughan electro-magnetic coupling, i. e. the filters areunable to prevent far-end cross-talk. One ofthe main results of the present research is, however,that under certain conditions also far-endcross-talk can be made to disappear, though anelectro-magnetic coupling exists between thelines. The practical consequence of this is thatwe may be able to save the expense of a specialsystem of transpositions in short sections forthe transmitters S, and S,' do not affect thereceivers M 2 and M./ respectively, belonging tothe same channel and located at the same terminal,there are inserted before the receivers twofilters F, and F,' respectively, which only passthe carrier frequency /„ and the side-band intendedfor these receivers. Likewise the filtersF, and F s ' are inserted at B before the receiversM 1 and 3// to prevent the transmitters S 2 andS/ respectively from influencing these receivers.An induction between the lines 1 and 2 cannotcause near-end cross-talk between subscriber 1and 2 at A, because a tension of frequency /,induced in line 2 is suppressed by the filter F'...and a tension of the same frequency /, inducedin line 1 is suppressed by the filter F 2 . The conditionsat the terminal B are quite analogous.With such an arrangement as shown in fig. 1near-end cross-talk between two high-frequencychannels is, consequently, not to be feared, althoughthe same frequencies are used on separatelines on the same poles.Fifr 2.- 113On the other hand, the filters inserted are nacarrieroperation, and be satisfied instead withthe existing transposition or twisting systems.Formulating the Problem.The problem the author has first placed beforehimself is as follows: (cf. fig. 2).Two parallel homogeneous double wire lines1 and 2 of the same length / are given. Betweenthese lines there exists a magnetic couplingwhich is constant for the whole length of thelines. A point on the line is characterised bythe distance .r from one terminal, which we callthe "near-end". The other terminal, which issituated at the distance x = / from the nearendwe call the "far-end".At the near-end an alternating current generatorwith a sinusoidal tension E, independentof the load, is connected to line 1. Line 2 isterminated in an impedance B,„ at the near-endand in an impedance R n at the far-end. Line 1is at the far-end terminated in an impedance B„.Line 1 may also he denominated as the induc-

ing. and line 2 as the induced line. Our taskis then, to calculate the tensions over the terminatingimpedances W,„ and /f L ,,. With a givengenerator tension E these tensions V. 2tl and V%indicate respectively the magnitude of the nearamifar-end cross-talk between the two lines.Method ofCalculation.As the two double wire lines together forma system of four parallel lines, the problemshould properly be dealt with by the aid of Maxwell'sequations for the electro-magnetic field ofsuch a system.The electro-magnetic field of a system of parallelconductors has previously been dealt withby several investigators. Lord Rayleigh 1 wasthe first to work out such a theory on the basisof Maxwell's equations. Afterwards, Max Abraham-in his text-book "Theorie der Elektrizitat"has further developed the same. The stricttheory for the transmission of electro-magneticwaves along a system of lines has up to thepresent day been carried through only for a fewspecial line arrangements. A summary accountis given by Max Abraham in "Enzyklopedie d.Mathm. Wissen", Vol. V, Art. 18, par 12, entitled"Induktionswirkungen von Wanderwellenin Nachbarleitungen".K. W. Wagner 3 has shown that also for non-Staticnary phenomena on a system of parallellines it is permissible to calculate as if the fieltldistribution were stationary, provided one canneglect the ohmic resistance of the conductorsand the leakage in the surrounding medium.These assumptions purport that the electro-magneticfield is entirely transversal, i. e. that thelines of force run in planes that are at rightangles to the longitudinal direction of the systemof lines. In reality there is always a componentin the longitudinal direction due to the resistanceof the wires, but a simple calculation goesto prove that for ordinary overhead long distancelines this component is so small in proportionto the transversal one, that in practice we mayalways reckon as if the field were merely transversaland stationarv.i Urd Rayleigh, Phil. Mag. Series 5, Vol. II. 1897. p. 199.- M. Abraham Theorie der Elrrtrizitat" Vol. 1, 5, 191H, par. 72,p. 293.K. \V. Warner "Induktionswirkungen von Wanderwellen inNacbbarleitnngen" E. T. 2. Jahrg. 35, 1914, p. 639.Wagner 1 considers a system of n parallel cylindricalconductors with arbitrary section. Theground is denominated by 9 and the conductorsby 1, 2, 3 n. By the use of magnitudeswell defined for stationary electro-magnetic phenomena,e. g. magnetic induction coefficientsand electrical potential coefficients, as well asknown theses in respect of the transmission ofelectro-magnetic waves over a system of parallellines, Wagner arrives at the result that the tensionV, between a conductor »- and earth at acertain point of the system can be expressed asa linear function of the currents in all n conductorsat the same points viz.:J,, is the current in the conductor »-.W„, is the mutual surge impedance betweenthe conductors « and y.W„ is the surge impedance for the very conductorv-Wagner further shows that W, and W„ aregiven by the following relations:andwhere: L uy is the mutual magnetic inductioncoefficient between the loops formed bythe conductor u and earth and the conductorv and earth.L rr is the self-induction coefficient ofthe loop formed by the conductor andearth.c is the speed of light in space andf is the dielectric constant of the mediumsurrounding the line system;the permeability of the medium is assumed= 1.As Abraham- had already shown, the generalsolution applying to the differential equationsfor such a system of lines is:andIn these equations qv is the charge per lengthunit on the conductor r, t is the time variable,1 loc. cit.2 Abraham loc. cit. p.114

X is the distance to the point referred to from afixed point, e. g. the near-end of the line system;/ and g are arbitrary functions. The equations(4) and (5) represent two electro-magnetic waveswhich are transmitted with the same speedone of form f, in a positive direction,the other of form g, in a negative direction.Equation (4) is the expression for the chargewaves, and equation (5) for the current waves.The functions /", and ;/, are arbitrary, so thatwaves of any form whatsoever may occur insuch a line system, but they are determinedby the limiting conditions.where r'„, = r',.„ is equal to the distance l>etweenthe conductor u and the mirrorimage of the conductor »>, or vice versa.On the basis of equation (2) there is then:andso is also:In the same way the self-induction coeffientof the loop formed by the conductor r and theearth is:Fig. 3.Calculation of the Surge Impedances\V„, and \V„,.As the surge impedances \V„,, and W„ for asystem of parallel conductors are directly proportionalto the corresponding magnetic inductioncoefficients for the same system, we mayfirst calculate these.For two conductors ;< and y, fig. 3, placedupon a height h„ respectively h, above ground0 and at a distance r„, from one another, themutual magnetic induction coefficient, is, as weknow:where h, is the height of the conductor v aboveground and o, the radius of the conductorV-On the basis of equation (8) there is then:Calculation of the Surge Impedancesof Two Parallel HomogeneousWire Lines:of a SystemDoubleWe consider a system of two parallel homogeneousdouble wire lines of which fig. 4 showsan arbitrary cross-section. The line I is formedby the conductors 1 and 2 at an average height115

HI above ground 0; V and 2' are the mirrorimages of 1 and 2, when the ground is consideredas the mirror surface. The line II is formedby the conductors 3 and 4 at an average heightHII above ground; 3' and 4' are their mirrorimages.The use of the equation (1) on all four conductorsthen gives us the following system ofequations:andThe system of equations (10) by this becomes:ror the tensions between the branches wetherefore get:Fig. 4.In these equations V„„ V.,,,, V.„, and V 411 aretensions between the respective conductors 1, 2,3, 4 and earth; J v J.,, J. u J i currents in the respectiveconductors; W tv W T2 , W,.„ IV 44 surgeimpedances between the conductors.In a double wire line the current in one branchis equal to, but of an opposite direction from,the current in the other branch, so that we areable to assume:if we put:the equation (13) passes into:11(5

Through W r . — W llt (equation 7a) is also:The equation (15) thus passes into:Z n and Z.,., are the surge impedances for thedouble wire lines 1 respectively 2. Z,, = Z 21are the mutual surge impedances between thedouble wire lines.For such a double wire line h > 30' = 9.15 m.;r 12 = 12" = 30.5 cm.; 2 o = 0.104" = 2.64 mm.;Q = 0.052".As the distance between the conductors issmall compared with their height above ground,r u ' ^ 2 h, as we have assumed here before, andthe formula (17) holds good. With f = 1 andC = 3 - 10 1 " cm/sec. the surge impedance is thencalculated to:Fig. 5.Calculation of the Surge Impedance Z lt for aBalanced Double Wire Line.or, since 1 e. m. e. is equal to 10 -9ohm:For a balanced double wire line (1—2) is:andwith the result that according to (9):Calculation of the Mutual Surye Impedance Z r ,between two Double Wire Lines (1—2) andAccording to equation (7) is:Since according to (7a) W nto W„ we get:is always equaland then accordingto (14)so that:orSince according to (14):As in most caseswe get:By way of example we may calculate the surgeimpedance for a double wire line, which is verymuch used by the Bell Company in U. S. A. 11Ct.: "Carrier systems on Long Distance Telephone Lines".B. S. T. Jl. Vol. 7, July 1929. p. 815.If the distance between the lines is short comparedwith their height above ground, we can.without any very great error, put:so that (18) is reduced to:117

tion) and by V1'', the tension for a retrogressivewave, the total tension between the branches is:By way of example we may calculate the mutualsurge impedance between two double wirelines of the same design as in the previousexample, erected beside each other on the samecrcss-arms as those practised by the Bell Companyin U. S. A. (cf. fig. 5). 1The height above ground is, as before, h > 30= 9.15 m, i. e. great in comparison with thedistance between the conductors, so that formula(19) may be employed. The distance betweenthe conductors in every double wire lineis: r 12 = r M = 12". As the distance betweenthe insulator pins of the same cross-arm isIn the same way it holds good for double wireline (2):If we define a current as positive when it flowsin the direction of increasing x-values, and if thecurrents in the progressive and retrogressivewave on line 1 are I1' and I" respectively, thetotal current is:in the same way it holds good for double line 2:Fig. 6.throughout 12", r„, = 12"; r ]4 = 36"; r„ = 24"and r. 4 = 24" and with f = 1,The use of the equation (15 a) on the progressivewaves on the lines 1 and 2 then resultsin:Calculation of Cross-talk between two HomogeneousParallel Double Wire Lines when theElectro-magnetic Coupling is Constant all alongthe Line.We will now consider a system of two homogeneousparallel double wire lines with a constantelectro-magnetic coupling between the linesas per fig. 6. If we denominate the tension betweenthe branches of line 1 by V,' for a progressivewave (transmission in a positive direciCS. B. S. T. JI. Vol. 7, July 192«. P. 581.In these equations Z 1X and Z 22 , as before, denominatethe wave resistances of lines 1 and 2respectively, and Z ]2 the mutual surge impedancebetween the lines. For the retrogressive waveswe get:If we assume that the tension caused by thegenerator V 10 is sinusoidal and of the angularvelocity a>, undamped waves are transmitted onthe line system both in a positive and negativedirection with a speed of— 118 -

) line 2:where a is the wave-length constant for the linesystem. If A 1 ' is the amplitude of the voltagewave at the beginning of the line 1 (x = 0) thevoltage at a point x on the same line is:In the same way we obtain for the voltage atpoint x on the line 2 for the progressive wave:For the retrogressive waves we get:But the voltage V 20 at the beginning of theline 2 must also be equal to the voltage dropwhich the current I20 causes in the impedanceR 20 , i. e.:2) for x = I.a) line 1.or according to the aboveIf B t ' and B x " are the amplitudes of the progressiveand retrogressive current waves at thebeginning of line 1, and B.,' and B." the equivalentcurrent amplitudes for line 2, we get thefollowing system of equations:The voltage drop over the impedance R^ bywhich the line is terminated must also be equalto V n so that:By combining (29), (30) and (31) we get:If the electro-magnetic coupling between thelines is so loose that the reaction of the currentin line 2 on the voltage on line 1 can be neglected,(22) and (23) are reduced to:b) line 2.andWe assume in the following that this is thecase. By the combination of (22 a) and (23 a)with (24) and (25) we get:Here, too, V a at the terminal of the line 2must be equal to the voltage drop which thecurrent J 2/ causes in the impedance 7? a , whichterminates the line 2, i. e.By a combination of (33), (34) and (35) weobtain:If we consider the limiting conditions at bothterminals of the line system (0 and /) for bothdouble wire lines, we get:1) for x = 0a) line 1If in the following we put e +jal = m ande -jal = n, (32) and (36) respectively are:By inserting in (27), (28), (32 a) and (36 a)for A/, A" 2 , ,4/' and A" their values extractedfrom (36) we get the following system of equations:119

B.,' is calculated from (38 a) in a known mannerto:or after reduction:For B." we obtain in the same way:In the form of a determinant the system ofequations (37) becomes:The near-end cross-talk then becomes:We assume now at first that the lines are terminatedin impedances which are equal to thesurge impedance of the respective lines, so that:The determinant (38) then becomes:Since:andThe near-end cross-talk tension across the terminatingimpedance R, 0 = Z.,. z , according to(28), therefore is:If we put:we obtain by reduction:For the purpose of calculating the currentI20 in the terminating impedance R 20 = Z 22 atthe near-end of the induced line 2, which currentgives rise to near-end cross-talk, we needonly calculate B1' and B2'' I20 is then:The far-end cross-talk current !», i. e. thecurrent in the terminating impedance R% = Z?>of the induced lines is calculated according to(34) to:In view of (40) and (41) we get:we get:120

or also:I. e.: If the inducing line 1 at the far-end isterminated in an impedance R-i equal to thesurge impedance Z tl of the inducing line, andthe induced line 2 at the near end is terminatedin an impedance B.,,, equal to the surge impedanceZ.,., of the induced line, the far-end crosstalkbetween the lines disappears, no matterwhat the frequency and the length of lines maybe. In this case is does not matter in what impedancethe induced line is terminated at thefar-end, since current and tension there are equalto zero.If this conclusion is correct, it must also provethat B! and B," are independent of the impedanceR»i in which the induced line is terminated,and that it is unnecessary to assume thatB 2; is equal to the wave resistance Z a of theline.In order to control this we shall now onlyassume that R u = Z u and R n = Z22.The determinant (38) then becomes:We put:As will be seen, (39) is identical with (30),no matter whether R 3l — Zu or not.For B," we obtain in this case:orThe calculation of the determinant results in:As will be seen, also (50) is identical with(41) and this proves that far-end cross-talk disappears,no matter in what impedance the inducedline is terminated at the distant end.For the sake of completeness we calculatealso /,„, i. e. the current fed into the inducingline by the generator V,„.Since I 10 , = B1' + B1" we compute B,' andB1" from (38 a) to:The calculation of the determinant results in:In the preceeding case B.,' hecomes:and121The current fed into the inducing line by thegenerator then becomes:

The reciprocal value is:This result was to be expected.With the aid of B/ and B," we are also easilyable to calculate the current J u and the tensionV\t at the near-end of the inducing line:According to (30):which, in view of (51) and (52) becomes:The near-end cross-talk attenuation is then:The cross-talk is greatest, i. e. b becomes aminimum, when cos 2al = — 1; thenandFrom these formulas it is obvious that onaccount of the terminating impedance R u havingbeen selected equal to Z„ no reflected waveis developed at the near-end of the inducing line.For the amplitude of the voltage waves onboth lines and in both directions we get withregard to (26), (40), (41), (51) and (52):By way of example we may with the aid offormula (64) calculate the minimum cross-talkattenuation between two double wire lines whichare very much used by the Bell Company inU. S. A. 1 , and for which we have previously inthis article (cf. pp. 8 and 9) calculated theirsurge impedance as well as the mutual surge impedance.With the values for the surge impedances computedon pages 8 and 9 viz:the minimum attenuation for these non-transposedlines becomes:Calculation of the near-end cross-talkbetween the lines.According to (44) is:attenuationThe author has also with the aid of formula(54) made certain calculations concerning thecross-talk attenuation between two twisted pairsof the type customary in Sweden, and which arelocated beside each other on the same crossarms.The minimum cross-talk attenuation wtas calculatedat 6.06 Neper for two twisted pairs whichrotate in the same direction and in the samephase.Sincewe getComparisons with the Results Obtained byKiipfmiiller '- by another Method.Kiipfmuller's equation (45) for near-end crosstalkbecomes with our denominations:The absolute value ofis:122In this equation K is Kupfmiiller's "electromagneticcoupling" hetween the lines at point XI1 Cf. B. S. T. Jl. Vol. 7. July 1928, pp. 581 and 592.2 Kupfmuller. Arrh. (. F.lektr., Vol. 12. 1923, p. ISO.

φ is the angular velocity and y, respectivelyy> 2 the propagation constants for the lines 1and 2.where C is the capacity of the double wire lineper unit length. A comparison with (43) thenshows that:If we assume the electro-magnetic coupling A'lo he constant and equal to K 0 for all x's, theintegration gives:If in accordance with our previous assumptionswe put:we get:Calculation of the Cross-talk between two HomogeneousLines when the Coupling Z 12 isa Function of x.We consider now an arrangement accordinglo fig. 7. The inducing line, with the surge impedanceZ 11 , is assumed as very short, its lengthis denominated by dx; at x, the beginning of theline referred to, an alternating current generatorFig 7A comparison with our equation (43) showsthat the two equations (43) and (67) lead tothe same result if:orNowis the velocity of propagation ofthe waves on the line system, for which reason(69) may he written:Kupfmiiller shows that equation (67) can alsobe written in the following form:with sinusoidal tension Vj, is connected to theline; at the other end x + dx the line is terminatedin an impedance equal to the surge impedanceZ 11The induced line with the surge impedanceZ22 is much longer; its length is /. At x = 0Ihe induced line is terminated in an impedanceequal to the surge impedance Z.,.,. At the otherterminal x = I the line is terminated in an arbitraryimpedance. Between the lines there isan electro-magnetic coupling which between xand x + dx is constant and equal to Z tx .The arrangement according to fig. 7 is thenperfectly equivalent to an arrangement as perfig. 6, in which R u = Z n and R„ = Z„. Thepart of the induced line to the left of x, whenlooked at from x, is equivalent to an impedance7,... The impedance of the part to the right ofx + dx, on the other hand, may be arbitrary, as123

we have previously seen in the case of an arrangementas per fig. 6.With regard to the equivalence of the arrangementsand equation (44) the tension at point xof the induced line is:The tension hetvveen the branches of the inducedline at x + dx, on the other hand, accordingto (46) is:As may also bewritten:The arrangements according to fig. 8 is equivalentto the arrangement as per fig. 7. Thepart of the inducing line added to the left of xhas no inducing effect upon the line 2, and thetension V r , set up by the generator V 10 at thepoint x is assumed to be equal to \\ x in fig. 7.Xor has the part of the inducing line attachedto the right of x + dx any effect upon line 2,and its impedance, looked at from point x + dx,is, as in fig. 7, Z 1VUnder the circumstances we are able towrite:and in view of (76) and (77):Fig. 8.The tension dV 2x at point x sets up a tensionat point 0 of the induced line:As the tension dV 2(x+dx) = 0. the tension ofthe induced line at the far end x = I also becomeszero, so that we are able to write:For a line arrangement according to fig. 8,in Which the electro-magnetic coupling is notlimited to a single point on the line, but wherethe coupling can be expressed as a variable functionof at, in view of the principle of superpositionholding good in this case, we get:This holds good irrespective of in what waythe line is terminated at the terminal /, nor doesit matter whether the part of the line between(at + dx) and / has the wave resistance Z ri ornot, or whether it is homogeneous or not.We now pass on to a consideration of an arrangementaccording to fig. 8. in which an electro-magneticcoupling Z, 2 between the lines onlyexists between the points x and x -f- dx.With this we arrive at the main result of thisresearch, viz:That the far-end cross-talk between two homogeneouslines of which the inducing line at thefar end and the induced line at the near end areterminated in their own impedances, always disappears,no matter how the electro-magnetic124

coupling may b2 distributed along the linesystem.If we assume that Z v , in equation (81) is constant,i. e. independent of x, we get:As will be seen, (83) is identical with (44),which ought to be the case if our calculationsare right. Likewise (82) is identical with (46).In a future work the author intends to showthat similar results are obtained for the crosstalkbetween two parallel homogeneous telephonelines whose line resistance is low- comparedwith the surge impedance. The waves arein that case damped. If the propagation constanty = p + j a , ( a is the wave-length constant;ft is the attenuation constant) is equal forthe two lines, and if the lines are terminated aswe have previously stated, the equation (44) forthe near-end cross-talk can be written:The far-end cross-talk tension is, as before,practically:By way of summary we may say that the farendcross-talk between two parallel homogeneoustelephone double wire lines practically disappearsif the far end of the inducing line and thenear end of the induced line are terminated intheir own impedances. The distribution of theelectro-magnetic coupling along the line systemis, in this case, without any influence.From a physical point of view we may givethe following explanation of the cross-talk whichoccurs in lines that are not terminated in theaforesaid way.If the induced, but not the inducing, line isterminated in its own impedance, there occursa reflected wave at the far end of the latter. Thisreflected wave may be considered as being awave emanating from a generator at the far end,i. e. it causes near-end cross-talk at the far-endof the induced line. The far-end cross-talk ob-served is, therefore, in this case in reality a nearendcross-talk due to the reflected wave, and itsintensity is directly proportional to the amplitudeof the reflected wave. When the inducingline, but not the near-end of the induced line, isterminated in its own impedance, far-end crosstalkis due to reflexions in the near-end of theinduced line of the same wave, which causesthe near-end cross-talk. As has previously beendemonstrated the wave emanating from the generatorproduces at a coupling point betweenthe lines a wave in the induced line, which isenly transmitted in the direction of the near-endof the line system.. This wave gives rise to nearendcross-talk, but since we have assumed thatthe induced line at this point is not terminatedin its own impedance, a part of the wave is reflectedand travels along to the far end, whereit also causes far-end cross-talk.If neither line is terminated in its own impedancefar-end cross-talk is caused by superpositionof the phenomena described for the twoprevious cases.The Practical Importance of the Results of theResearch.It seems to be obvious from this research thatin carrier current telephone systems whichoperate with different frequency bands for thetwo directions of conversation, we might be ableto considerably diminish far-end cross-talk,which is the only one of any importance in thisinstance between the different carrier channelson the same poles, by dimensioning the equipmentsat the terminal and intermediate amplifyingstations in such a way that their input Impedancefor all important frequencies coincidesas much as possible with the characteristic impedanceof the lines connected to them. As thereare no essential technical difficulties in bringingabout good agreement between the said impedances,it seems as if this method, which directlyattacks the cause of cross-talk, could diminishthe demands which must he made upona transposing or twisting system for lines fortransmitting the high frequencies used in carriercurrent telephony. We might, possibly, even besatisfied with the systems of transposition andtwisting which are now customary for low frequencytelephony and thus effect a great savingin the line costs.— 125 —

Practical Points About Automatic Fire Alarm.By Harold Ekman, Supervising Engineer of the Stockholm Fire Brigade.The stupendous advances made in technicsduring the past twenty or thirty years hashad a revolutionizing effect on individuals andthe community as a whole. What seemed out ofKin. t a. Diagram of an Automatic Fire Alarm installation.(Eacfc fire-alarm section comprises a certain part of the premises.)Fig. 1 b. Diagram of an Automatic Fire Alarm installation.(The black dots represent the automatic thermo-contacts.)reach yesterday may to-day be made available toall and sundry, and as a result of the abundanceof technical aids and the necessities created bythem we make ever greater demands on the functioningof the organs which supply our needs.— 126 —The wave of rationalization rolling over theworld is receiving added stimulus from the economicpressure resulting from depression inbusiness and severe competition. But of whatavail is a good organizationand well-planned operation,based on what seems to bedependable calculations, ifthe whole venture can beupset by the ravages offire. The material objectsone can insure but not thepresent and future prospectsof the business.Economical fire protectionis not only a nationalgain, it is a good investmentfor the individualbusiness man who has realizedin time the benefitsaccruing from it.An efficient fire protectionorganization must embraceboth arrangementsfor preventing the inceptionand spreading of thefire, and fire extinguishingdevices suitable for the localconditions. One featureof such an organizationwhich is of primary importanceis that absolutelydependable means shouldbe provided for alarmingand directing the fire-crewto the spot. It should bepossible for the fire-crewto find the right placequickly, because any loss of time means thatthe fire will get a further start, and mayattain such a size that the available fire extinguishingarrangements will be insufficient tocope with it, no matter how efficient the organi-

Fig. 2. An Automatic Fire Alarm Station Apparatus for 6 Sections,including Charging Board for Storage Batteries. The diagram showssignal lamps and various illuminated signs which, together withbells, automatically register fire alarm and faults of every kind,and also control that every switch is correctly thrown.zaticn happens to be. I wish at this point toemphasize the immense importance of discoveringthe fire in good time, and the fact that thereis an intimate connection between the stage atwhich the fire is discovered, and alarm given, andthe losses which the fire will cause in the end.The fire grows in practically all cases from aninsignificant start, is developed gradually withsteadily increased power of expansion, and thecurve for the spread of the fire rises violently.Minutes are valuable, especially at the earlieststage of the fire, and the results that can be accomplishedat that time with a very simple fireextinguishingoutfit may at the later stage bebeyond the capacity of even the best fire fightingorganization.Experience teaches us a number of interestinglessons on this score and gives valuable informa-Fig. 3. Central Fire Alarm Board for a large station (311 Sections).tion which is collected in the statistics of firelosses and the journals of the fire brigades.In practically all cases of fire where some personhas been present at the inception, or has arrivedon the spot shortly after the outbreak, theperson has himself or with aid called to his assistancebeen able to put out the fire or in anyevent confine the fire to a limited area. Veryfew large conflagrations have for this reasontaken place during working hours, except inattics, warehouses or other places of that kindleft without regular attendance. In a community,factory or other working premises peopleare all the time in watchful movement, and thefire, as already intimated, is in all these casesimmediately discovered and, according to statistics,is usually put out with a hand fire-pump,chemical fire-extinguisher or the like, and a seriousconflagration avoided.It is of interest to examine the journals of theStockholm fire brigade to see what they have to127

Fig. 4. Central Fire Alarm Board for a large station (open).Fig. 5. Thertno-contact.closed.Fig. 6. Thermo contact after giving alarm.l28

Fig. 7. Thermo.-contact filled in Wood-drying Kiln.Fig. 10. Alarm Bell for A.Ccurrent.Fig. 9 Alarm Bell of powerful type,with bell of 250 mm diam.Fig. 8. Tbermo-contact fitted underneath stage floor.— 129 —

Fig. 11. Motor-generator for automatic connection to Fire AlarmBoard. To be used il several A.C. alarm bells are automatically togive fire alarm.relate on the subject of calls to put out firesduring night hours or after working hours or,generally speaking, concerning large fires.This is a very interesting chapter. Most largefires have occurred in premises which have beenunguarded during the inception and early stageof the fire, and the fire brigade has in these casesoften been alarmed by outsiders after the firehas bursted windows and shutters, or even burntthrough floors and walls. All the largest fires inStockholm during recent years, among themSvenska Teatern, Galarvarvet, Hasselbacken, Alhambra,Ostermalmsteatern, Separator, Tattersall,have started between 11.45 p. m. and 3.30Fig. 13. Same Fire Alarm Box as Fig. 12, but with the door open.a. m. and in consequence have not been discovereduntil a far advanced stage. The recentfire in the paper storage of the Svanstrom firmat Herkulesgatan also belongs to this category.The fire raged undisturbed in the large basementstorey, where it stored up an enormousamount of energy and choking gases, which madeit difficult to localize and attack the heart of thefire. In the last mentioned case the delayedalarm caused losses amounting to several millionkroner.From what has been said in the foregoing oneis justified in the cause of fire protection to insiston the enormous importance of the quickestpossible means to give notification of the out-Fig. 14. Branch Fire AlarmBox which may be connectedto automatic fire alarm station.The box is to bo connectedin a separate loop orin an existing thermo-enntat Isection.Fig. 12. Fire Alarm Box with telephone, intended as main firealarm box for transmitting automatic alarm to F'ire Brigade.Telephone can also be used for manual operation.Fig. 15, Fire Station Apparatus forFire Brigade, with two telegraphinstruments.— 130 -

R 1591Fig. 16. Fire Station Apparatus of the Fire Brigade in the town ot Jonkdping.break of fire, and demand that the most up-todatedevices which modern technics have createdshould be utilized for preventing such abnormallylarge fires.With such devices I have in mind the automaticfire alarm system, where the alarm isput into function by the fire itself at the veryoutbreak. This system signifies that one nolonger leaves it to chance to determine the momentfor giving alarm, but the work of puttingout the fire can be instituted at once, and,furthermore, by means of a special alarm boardit is possible to determine where on the premisesthe source of the fire is located.The automatic fire alarm system is based onelectric circuits connected to fire alarm contacts(thermo-contacts) distributed over the premisesthat are to be protected and which are susceptihleto heat and connected by electric circuitsto a local alarm board. Should one or more ofthese contacts be heated owing to a rise of thetemperature in that part of the premises, theyautomatically begin to function and close a certaincircuit by means of the relays and otherreceiving devices on the alarm board. As a resultof such changes, signal bells or sirens areput into operation, or the alarm may be transmittedthrough fire alarm boxes direct to thefire brigade. The alarm board of the receiving3.131 —set indicates inter alia from whichplace on the premises the alarmhas been given. For facilitating thisplace indication the thermo-contactsare grouped into a numbera separate, locally delimited, firealarmsections, which each have adesignation on the alarm board.The alarm board is preferably tobe set up at the main entrance orother frequented place on the premises.A controlled storage batterysupplies the current requiredto operate the system.The value of such an automaticalarm installation can not be overrated,provided it complies in alltechnical respects with the demandsfor reliability in operation:otherwise it may do more harmthan good. It is necessary to keepin mind that an automatic firealarm system, without any human interventionand under the very worst practical conditions,must infallibly register and transmit it may beonly a single important fire alarm and thatmany years after the system has been installed,at a time when the interest in, and the attendanceand care of, the system may not be so much aliveas when it was new.The decision to instal an automatic fire alarmsystem of this kind means both for the insuredand the insurance companies the transfer of animportant responsibility onto the system, whichis then made practically the sole protectoragainst fire.The technics relating to electric signalling havemade rapid strides during the last decade ortwo; what was considered satisfactory andgenerally speaking good enough sometime agomust now be discarded in view of recent progressand the increased demands made on the reliabilityand effectivity of such devices.The alarm board is the most vital part ofthe whole receiving outfit. This apparatus shallconstantly control not only the circuit thermocontacts,battery and, in some cases, main firealarm box for transmitting the alarm to the firebrigade, but also in as far as possible its ownfunctions. That is to say, it must consist interalia of a larger or smaller number of relays, i. e.

R 1290 Fig. 17. Fire Station Apparatus of the Kalarina Fire Brigade in the City of Stockholm.Fig. 18 Telegraph Rogm of the Ostermalm Fire Brigade in the City of Stockholm.132 —

Fig. 20 a. "Svenska Tcatern" lire in Stockholm 29.6.1925. Alarm turned in 3.13 a. rSlage ceiling has collapsed but the ceiling joists over auditorium are still intact.Fig.20 b. "Svenska Teatern" fire in Stockholm 29.6.1925. Ceiling; over auditorium has collapsed133

Fig. 19.small sensitive electrical magnets which are actuatedand balanced by weak electrical currentsand provided with variously connected electricalshort circuits and circuit breakers. The numberof such relays must be made as small aspossible, at the same time as theabsolutely necessary organs SOT thereliable operation of the fire alarmmust be insisted upon. Furthermore,the design and constructionof these relays must be the bestpossible, and all the relays of thefire alarm system must automaticallycontrol the current. The mutualcombination of the relays isalso of great importance.The electric contacts of the alarmboard included in the fire alarmcircuit shall also be as few andreliable in operation as possible;thus preferably circuit-breakingcontacts or such as are closed when the systemis at rest, which consequently automaticallycontrol the current and break the circuit whengiving alarm. A working contact, i. e. a contactwhich under normal conditions is open, is un-Fig. 21 a. Tallersall" fire in Stockholm 18.3.1913. Alarm turned in 1.31 a. m.- 134 -

A further demonstration of the risk of leaving the fire to itself.135Door covered by sheetiron on both sides which has resisted the fire Iron door deformed by fire.Effects of fire on unprotected steel STRUCTURESFig. 21 b. "TATTERSALL" fire in Stockholm 18.3.1913

eliable, as its contact surfaces are liable to becomeoxidized or covered with an insulatinglayer of dust, oil or the like which, with theweak currents employed, may not let throughthe current when wanted.The arrangements for restoring the system aftera fire alarm has been given, or in case ofline faults etc., should be simple and definite,and the connections properly guided, so as tonuke faulty connections impossible. After readjustmentthe alarm board shall plainly andpermanently indicate the fault which has madethe readjustment necessary.The automatic fire alarm system devised andintroduced by Telefonaktiebolaget L. ML Ericssonhas been designed in the smallest detail withthe view to comply with the above demands forreliability. With same it has been found possibleto produce the necessary functions withoutmaking the system dangerously complicated.Required functions in connection with givingfire alarm have been amply provided for, andthe system besides automatically controls all thefaults which might prevent alarm being givenat the proper time. All parts of the system arethus constantly automatically controlled bymeans of closed current, so that the faults ofany kind whatever are immediately indicated assoon as that appear.The main features of the system have conducedto establish even greater reliability, whichis carried to a point always regarded as thehighest desideratum, namely the elimination ofthe effects of dangerous line failures, so thatthese, even if they should occur at the very momentthe alarm is to be given, cannot preventthe alarm from being transmitted. This timemoment,and the current impulses generated inconnection with same, may be very critical,especially if the circuit is very extensive and thusexposed to strains of different kinds in thecourse of years. The risk of line faults andether failures in a system of any kind is ofcourse bound to increase in order as the systembecomes older.Even the sensitive thermo-contacts have beendesigned in a special way in order to obtain necessarycontrol. At the same time the degreeof safety for producing the alarm impulses hasbeen quadrupled for a single thermo-contact, inaddition to which the different thermo-contactsin a certain section collaborate in giving thealarm.Space does not permit a more detailed accountof the special arrangements for detecting faults,giving alarm etc. The foregoing is merely intendedto give an idea of the importance of automaticfire alarm, and the technical demandssuch a system must comply with.—136 —

New Interlocking and Signalling Plant at Lund.By Ivar Larsson, Sitjiud Enyineer in the State Railways.The railway station at Lund was one of thefirst in Sweden to be equipped with interlockingand signalling installations. In the years1899—1901 there was erected a plant with twomechanical interlocking machines for the southpart of the station, and in 1901—1902 there waserected an interlocking machine in the northernpart, so that a complete plant was obtained thatwas equipped with those devices and arrangementsfor safeguarding trains and securing the work ofthe station service which were then consideredi>erfect. Signalling technics have, however, prosidiarycompany, the "Signalbolaget", received ancrder to supply the material and erect a completeelectric installation.This new installation embraces only one interlockingmachine. This has been located at thenorthern end of the station. The switch movementsthere are more numerous and more complicatedthan in the south part of the station, andit was in consideration of this circumstance thatthe place was selected.The cabin is on three floors. The ground-floorcontains the power plant for the interlocking Bin-Part of the station yard. The signal cabin on the right.gressed incessantly, and, when the station was rebuiltin the year 1927, the old plant was quiteantiquated. It was provisionally adapted to thenew station, and there was, of course, no impossibilityof completing the same, but withoutdoubt a new structure was the only rational thingto be able to satisfy up-to-date requirements ofsafety and labour saving. For this reason a perfectlynew interlocking and signalling safety plantwas erected, following those modern lines andprinciples which have been the foundation of theplants erected by the State Railways in recentyears.Interior of Cabin.The L. M. Ericsson Company, through its subchine,a workshop for repairs and the heating installationfor the cabin. The first floor is reservedaltogether for relays and other auxiliaryinstruments, while the top floor is the switch-roomproper. From the same a good view is obtainedof the northern part of the station, but no viewcan be had of the central and southern parts ofthe station.The switch apparatus is manufactured with onlyelectric interlocking register: all locking and dependenciesare. therefore, carried out electrically.Previously a similar interlocking installation hadbeen erected by L. If. Ericsson on behalf of theState Railways, viz. at Hasslcholm (described in- 137 -

General plan of tracks, signits, points, and track circuits.this Review in 1927, Xos. 1—3). The machinesupplied for Lund is constructed on the same linesas that at Hassleholm, but the design had beenaltered by making use of the experiences gainedat Hassleholm. The whole machine is higher, sothat the levers are approximately on a level withthe elbow of an average-sized man. By this meansthe advantage of a more natural grip of the leversis gained. These levers are arranged in a row thesame as in L. If. Ericsson's older interlockingmachines. The shape of the handle on the leverhas been altered so as to present similarities to asmall door-handle. It has been possible to effectsuch grouping on the apparatus that every otherlever is a signal-lever and every other an interlockinglever. The signal-lever is in a normal positionwith its handle straight up. It can 1M?switched over to right or left (each movement70 2 ), thus being used for operating different signals.The switch-lever has in its normal positionthe handle obliquely downwards to the right, andon switching over it is turned 140° to the left.Particular care has been devoted to the developmentof the design in so far that both the lever'shorizontal shaft, which is rigidly connected withthe handle, and the co-operating vertical contactshaftget a perfectly steady and exact motion, sothat make and break occurs with all desirableprecision. Control windows above the lever havebeen utilized in the following way. The switchmanis informed by a white or red signal-plate ifthe switch occupies a position coincident with thelever, and closes completely, or if such is not thecase. Besides, a blue pointer in the window of thelever indicates that the switch is blocked for changing.This pointer disappears when the switch isfree. A white or blue signal-plate in the windowsabove the signal-lever indicates if the lever is disengagedfor changing over or locked.The interlocking machine is in no way connectedwith any release instruments or the like operatedby the train dispatcher. The signalman must,therefore, himself lay the tracks and display the"clear" signal. This arrangement is consideredsuitable here, because safety measures have beenadopted for automatic control that the tracks arefree from vehicles when a start-signal can be displayed.This system is preferable when the trafficin a station is in any way extensive. As, therefore,the operator on his own responsibility lays thetracks, it is important for him to be able to controlin a simple manner that the track is clear to orfrom an intended track when the start-signal isgiven. This has been done in such a way that theswitch installation is equipped with small auxiliaryinstruments in the shape of a push button switch,one for each track. On a sign-board l>elongingto each switch it is indicated which incoming oroutgoing signal is meant, as well as the numberof the track to or from which a track is to be laid.Hefore the signal lever for a main signal becomesdisengaged for switching over, this push-buttonswitch must tie changed over. If the signal leverthen becomes free for switching over, the operatoris sure that the points or switches are set for thevery track intended, i. e. he obtains a simple andeasily grasped control instead of having to examinewith meticulous care the position of the levers,in order to prevent shifting switches.On enamelled signs above the levers there aregiven the necessary directions for the positionwhich other levers must occupy to enable thelever to be operated, and by this means all thetrack-tables are set out on the very interlockingmachine.The installation is made for altogether 80 levers,and was at the beginning fitted with 29 signallevers,31 switch-levers, 3 interlocking levers, 1bcom-lever, and 16 spare places. Its length is6.6 m. It is lacquered the ordinary deep-greencolour, and its appearance is smart and attractive.For the State Railway lines connected to it,double track line to Uppakra and Stangby, theblocking of the line is arranged by means of a3-field block instrument located at each gable ofthe machine.The illuminated track plan is set up separatelybehind the interlocking machine. On this planare set out the signal images of the main anddwarf signals, as well as the condition in the tracksections provided with track line. All instructionson the track plan take place by means of lampscountersunk behind separate index windows. Anoccupied track is marked by a lighted lamp in thetrack window, and a free track with an extinguishedlamp, an arrangement which on fairlylarge track-plans gives a better survey than if themore customary opposite method were employed.All tracks (19 home and 19 outward tracks)are locked by track interlocking, and the release138139

is effected automatically by the train. For emergencyrelease there are separate, sealed keys. Thetrack for the home roads are not locked until thetrain has reached the track-line immediately outsidethe home signals, i. e. it is approximately1400—1200 m. outside the appurtenant advancesignal. If an expected train has not reached thispoint, the signal can be taken up and the road releasedwithout having to break the seal and usingthe emergency key.With the exception of the interlocking group inthe north, the track area of the interlocking installationis equipped with complete track lines.These have been excluded in the north on accountof the expense, because the direct survey from theIncoming and outgoing signals at north end of yard.interlocking apparatus has been considered to constitutesufficient safety measures.The devices for actuating the switches are of anew design, and they are made with interior mechanicallocking, for which reason the hook- andlink-locks hitherto employed for locking the pointcan be dispensed with. It is hoped by this meansto counteract, inter alia, to some extent the difficultiesarising in case of a fall of snow. Theswitches are, in so far as circumstances permit,coupled in pairs to the same lever, and it happensin some instances that three motor actuating devices,two for switches and one for scotch block,are connected to the same lever. Both parallelcoupling and serial coupling have been employed.The entire number of switch and scotch blockactuating mechanisms is 49. There are no mechanismsfor local manipulation. Four interlockingmechanisms for merely locally operated switchesand scotch blocks are laid down.The entire station area connected with the interlockinginstallation is equipped with a completesystem of dwarf-signals. These are used not onlyfor interlocking motions but also for train movements.All switch roads are interlocked by thesesignals, and for the southern and central parts ofthe stations switches in front of switch-vehiclesare locked when the appurtenant dwarf-signal hasbeen passed, even if the signal in question is takento stop. As to the northern part of the station,which can be surveyed direct from the interlockingplant, no such locking has been employed, because,as has been stated here before, no tracklinesnecessary for this purpose have been putdown. The dwarf-signals are of the standard typeused by the State Railways, optical signals beingmade by two white lights which form a horizontalconnecting line at an angle of 45° or vertical. Forthe outward tracks the dwarf-signals have alsoto serve as outward signals. The circumstancethat a track is switched off is shown by the verticalposition of the dwarf-signals. By this means onehas obtained safe signalling with the aim that thelocomotive driver will be able to check that thisoutward track is set and clear, without having toresort to outward signals for each track. We are,therefore, restricted to one outward signal for eachline, and also an inner outward signal of maintype near the main tracks in view of the trainspassing at a high speed.All main signals are light signals. Track signallingon the main home signals is done with oneto three green lights. Besides, the home signalshows on signalling for a main line train whetherthere be a free passage or not. In the former casea white intermittent light is visible, in the lattercase a green intermittent light, of course also inboht cases a fixed green light. Tracks can be setto and from all seven pairs of rails for the StateRailway lines, whereas the trains of the crossingprivate railway can be signalled in and out ontracks 6 and 7 north and south, as well as in andout on track 7 in the south.On platform 1 near the station building are locatedrepeaters for the home and outward signals,so that the train dispatcher can check their interlocking.To facilitate the work of the train dispatcherin his clearing, there are also arranged— 140 —

at certain places separate repeat signals for themost commonly used outward tracks.One detail of the installation that may be speciallymentioned are the arrangements for the levelcrossings, which exist south of the station building,about 650 m. from the interlocking installation.The street is shut off by means of electricallyoperated booms, which are operated fromthe interlocking plant. The booms are interlockedby the home and outward signals, so that there isno need to risk the booms being forgotten andthe trains are yet signalled along over an openroad. When the booms are lowered and a signalis set for a train to go ahead, the boom lever canat any moment whatsoever be returned to a positioncorresponding to raised booms. The boomsremain nevertheless in a dropped position, but theboom-motor obtains automatically a current whenthe train has passed the level crossing. For thepurpose of warning street traffic (this place isvery busy occasionally) there have been set upspecial signals which by a powerful red light,visible even in daylight, show that the booms arebeing, or have been, lowered. These signals, oneoutside each room, show stop already whenthe warning bell on the booms begin to ring,i. e. a good while before the booms themselvesbegin to drop. When the booms rise the signalsgo out as soon as the booms have been raisedsufficiently for allowing vehicular traffic to pass.These extra signals have proved to be of greatbenefit; no difficulties whatsoever have occurredto have the road traffic stopped and the crossingcleared when the booms are lowered, though theyare operated from an interlocking cabin from whichthe street traffic cannot be observed. It is obviouslyso that when the way-farer knows that adrop-boom is operated mechanically so as to dropunrelentlessly, it inspires a great deal more respectthan a crossing-keeper close by, who, in caseof need, can stop the motion; in the latter casemany a way-farer is tempted to hum' on to thetrack, although the lowering of the boom hasstarted, in order to escape having to stand waitingbefore dropped booms.The interlocking installation is operated bothwith direct and alternating current. The City ofLund Electric Works supplies both kinds of current.3-phase alternating current, 50 periods, istaken from its alternating current net, with a highvoltage transformer erected within the stationarea, and on the secondary side 3X'30/100 55volt are drawn in the interlocking installation'sown transformer. The lamp voltages are 55 in thedwarf-signals and 12 volt in the main signals; thelatter voltage is obtained from local transformersset up in cabins close to the signals. The tracklinesare fed with alternating current, c:a 2 volttension between the rails, and this voltage is steppedup to about 4.5 volt for feeding the track-phaseof the track-coils located in the interlocking cabin;the local phase is fed with 110 volt, the interlockingcontrol coils (the SS-coils) operate with twophases, both with 110 volt. Direct current (12volt) for operating currents for signal and releasecoils as well as block magnets is obtained from aPower station.metal rectifier (copper oxide rectifier). The interlockingmotors, made for 120 volt direct current,are fed direct from the city's direct current net.A converter is also set up in the power-room ofthe installation as a reserve. The latter is operatedwith direct current. It has been considered thatsince the city has very reliable arrangements forthe supply of direct current, a separate reservebatten' can be dispensed with. It has, therefore,been possible to carry out the entire power plantcomparatively simple. The energy consumption ofthe installation is about 32000 kwh alternatingcurrent and about 450 kwh direct current perannum.As has been stated in the preceding, the entireoperation is now confined to a single interlockingmachine. This concentration affects, as a matterof course, the signalling staff beneficially. In— 141 —

the old plant, with its three interlocking machines,11 men were on interlocking duty with the sametraffic conditions as now. The new interlockinginstallation has hitherto been served by a staff ofaltogether 6 men, but it is quite possible that thisnumber can be reduced when the staff have becomesufficiently accustomed and expert. Thesaving in personnel is thus at least 5 men. Theinstallation has cost 280000 Swedish kronor. Althoughthe motive for such installations in thefirst instance must be considered as a measure forsafeguarding the trains and bringing about sufficientspeed and rapidity in operating the station,the saving in personnel already effected in thiscase constitutes a very good contribution towardspaying interest and amortisation on the initial investment.The installation was fitted up in the summer of1929, the work taking about 3 months, which mustbe looked upon as very smart. Electrically operatedspecial drilling machines and welding machinesare, as much as possible, used for the mechanicalwork of fitting up, and this work was done bythe State Railways themselves. When working ata very busy station it is of great advantage for therate at which the work is done, if electric currentis available and can be used for operating portablemachine tools.Interior of relav room.H 1690View of station yard with level crossing. Station building on the right.142

On Impedance and Impedance Measurements as well as a Descriptionof the Impedance Measuring set Manufactured bySvenska Radioaktiebolaget.By TorbernLaurent.I. Introduction.For telephone engineers who are not inclusivelyengaged in electrical measurements andthe mathematical dealing with the results ofmeasurements, there arises often the need ofhrief descriptions of various methods of measurementand calculation as a support or refresherfor the memory. Such descriptions may oftenalso be of value as a medium of instruction forless qualified personnel. In the present paperbrief descriptions of some devices generally usedin line impedance measurements as well as ofthe mathematical dealing with such impedancemeasurements have been summarized. The calculationsare, furthermore, in certain instancesillustrated by figures which, as is well knownfrom experience, many a time may make it easierto understand the methods.Special importance has been attached to thesuitability of the methods for practical purposes,and it has been shown how the measurements,and especially the calculations, may be simplified.The same views also form the basis for theconstruction of a new impedance measuring set,described in the last part of the paper, which ismanufactured only by Svenska Radioaktiebolaget.The principles of the instrument have previouslybeen published by the author in "TekniskaMeddelanden fran Kungl. Telegrafstyrelsen"(Technical Papers from the Telegraph Office)No. 4, 1925, and in "Elektrische NachrichtenTechnik", Volume 5, Part 5, 1928.The resistances r form a ratio 1:1, G is anaudio frequency oscillator supplying simple harmoniccurrent of the angular frequency φ, HTFa telephone receiver, N a variahle non-eapacitativeand non-inductive resistance, C a variablecondenser practically free of losses, and O, respectivelyO, switches. The measurement is effectedby varying the resistance N and the capacityC until a sound minimum is obtained inthe telephone receiver HTF.After setting the sound minimum the followingrelations hold good between the impedanceZ, expressed in amplitude Z and phase angleφ or resistance R and reactance X, as well as resistanceN and capacity C:For bridge-coupling according to figure 1 applies:the plus sign applies to switch position 2 andthe minus sign to switch position 1.For bridge coupling according to figure 2 applies:2. Impedance Measuring by Means of Resistanceand Capacity.Figures 1 and 2 show the bridge devices inmeasuring an impedance Z with a small respectivelylarge phase-angle.The plus sign applies to switch position 1 andthe minus sign to switch position 2.143

Fig. 2.Fig. 1.Sonic drawbacks or difficulties with thesemeasuring devices for practical measurementsin telephone technics may be pointed out.Mostly the bridge-coupling according to figure1 is made use of, and no matter whether theimpedance is to be expressed in \Z\ and φ or Rand X, comparatively difficult calculations arenecessary. The bridge-coupling according to figure2 is as a rule not resorted to unless thephase angle of the impedance is so great thatmeasuring with bridge-coupling according to figure1 is impossible because the resistance Ncannot be made indefinitely great. It thereforevery often happens that we obtain series ofmeasurements made with different bridge-couplings,and, therefore, the uniformity in the calculationsis lost. Besides, it is often tiresome tohave to change a measuring method in a measur-ing series, since the impedance of the objectmeasured occasionally is dependent upon thecurrent intensity.The capacity C consists as a rule of mica condensers,which by means of decades are interconnectedin various ways in such a mannerthat, together with a variable air condenser, theyform a continuously variable capacity. The capacityvalue should naturally be capable of beingdirectly read from the decade settings, withoutany conversions and corrections. Thiscalls for a certain adjustment of the mica condensers,which is particularly troublesome andexpensive.3. Impedance Measuring by Means of Resistanceand Self-induction.Another measuring device is based upon theuse of a variable self-induction. In fig. 3, whichshows the measuring device, the denominations144

, N, G, Z and HTF possess the same significanceas in figures 1 and 2. L is the variable selfinductionwith a loss-resistance a, which is balancedin the bridge by the equally great resistanceb.Measuring is carried out by varying the resistanceA' and the self-induction L until asound-minimum is obtained in the telephone receiverHTF. in which casethe plus sign applies to switch position 1 andthe minus sign to switch position 2.In measuring with this arrangement the impedanceas a rule is expressed by R and -Y, liecausethe relation between N and L, as well asR and X, is obviously remarkably simple. Forthis reason impedance measurement by self-inductionis generally preferred to impedancemeasurement by condenser.The variable self-induction consists as a ruleof a variometer with an empirically graduatedscale. The practical disabilities or drawbackswith such a variometer are as follows:The variometer is large and cumbersome.The accuracy of measurement is adventuredby mechanical changes in the rotor's suspensionin the stator.The variometer's open field does not permit theproximity of iron. E. g. the variometer cannotbe mounted on an iron rack or covered by aniron casing or the like.The variometer is subject to disturbances bymagnetic induction.These drawbacks or disabilities can be diminishedby certain special measures, but neverentirely eliminated.4. Calculating the Constants of a Telephone linefrom Open and Closed Circuit ImpedancesMeasured.By impedance measurements from one terminalof a telephone line it is possible to fully determinethe electrical characteristics of the line.Let us assume at first that the characteristicsof the line are of interest to us for a certain angularfrequency φ only. We measure then theopen circuit impedance Z t and the closed circuitimpedance Zt between the branches of the lineat one terminal for this angular frequency, thebranches at the other terminal being open orshort-circuited.In accordance with a method of calculationwhich is deduced in the work "TelefonledningarsElektriska Egenskaper" (Electrical Qualities ofTelephone lines) by H. Pleijel, we shall calculatefrom these impedances the different electricalqualities of the line.Assume that the amplitudes and phase anglesof the two impedances are determined bymeasurements for the angular frequency OJ, viz.Fig. 3.145

The characteristic impedance Z. of the line iscalculated according to the equationsFor determining the propagationof the line we calculate firstconstantsBy calculation we getby which we obtain the attenuation bs of theline according to the equationorwhereWe now calculate the characteristics of theline according to equation (4)and the wave-length constant a fromorwhereand A' is a whole number which grows by leapswith increased frequency and an increasedlength of line.It is of a certain interest that the same calculationresult is obtained if we putorAccording to equation (5) we calculateAccording to equation (6) we may calculateFor determining in the above manner comparativelylarge line attenuations the calculationsmust be carried out with great accuracy.from which value the attenuation may be calculated:Example 1.For a copper line with a length ofS =20.90 km.impedance measurements have been made withthe angular frequency φ = 5000 radians sec.from one terminal of the line, the followingresult being obtained:According to equation (7) we calculateFor such fairly low frequency and such fairlyshort line as in this example, the constant Kprobably is 0,146

ErgoIf the line is homogeneous the so-called primaryline constants, listed below, may be calculatedfrom the so-called secondary line constantsdetermined by the above calculations. The primaryline constants are:r = the resistance in ohms/km. double line.a = the leakage 1/ohm km. (= Siemens/km.):L = the self-induction in Henry/km. double line.C = the capacity in Farad/km.The relation between primary and secondaryconstants is:If we take out the primary constants fromthis system of equation we obtainorAccording to the equations (9) we obtainr = 5 ohms/km. double line.L — 2.43 m. H/km. double line.a = 1.05 • 10~ 6C = 0.00516 ,uF/km.mhos/km.These values are reasonable, which proves thatwe have selected a proper value for the constantK.An arbitrary apparatus composed of resistances,condensers and coils, with two input andtwo output terminals e. g. a transformer or anelectrical filter may be looked upon as a telephoneline with a certain characteristic impedanceand certain propagation constants,which are determined from open and closed circuitmeasurements in the same way as for atelephone line. Measurements carried out fromthe input and output terminals as a rule givedifferent characteristic impedances but, providedany possible ironcore coils in the apparatusare only lightly magnetised, the same propagationconstants are always obtained.5. Line Impedance as a Function of the MeasuringFrequency.Example 2.For the line in example 1 has been calculatedby whichThe characteristic impedance and the openand closed circuit impedances of a line as afunction of the measuring frequency are discussedin a paper "Matningar a Gotlandskabeln"(Measurements Made on the Gotland Cable), byJ. Skoglund in "Tekniska Meddelanden franKungl. Telegrafstyrelsen" (Technical Papersfrom the Telegraph Office) Xo. 7, 1921. Figure 44. 147

Fig. 4.is taken from this paper, and it renders in curveformmeasured impedances as a function of themeasuring frequency on a 16 km. long loadedcable.As we see, both the open and closed circuitimpedance vary periodically with rising frequency.This is the result of the superpositionof the waves coming from the measuring endof the line, and those reflected at the far end,and returning to the measuring end. The samething is apparent from the equations of the openand closed circuit impedances.148

We are, therefore, in the first section, whereA = 0.At any rate, the constant A can clearly be determinedby measuring the open or closed circuitimpedance as a function of the frequency,and fix the section concerned from their periodicalvariations.The wave-length constant grows almost proportionallyto the measuring frequency and, consequently,the functions cos 2 a s and sin 2asvary periodically between the values + 1 and— 1. The attenuation fi, on the other hand,grows comparatively slowly and the functionscosh 2 ps and sinh 2 fis are not periodical. Theopen and closed circuit impedances, therefore,van.- periodically between the values6. Approximative Equations for CalculatingAttenuation.It has been mentioned in chapter 4 that incalculating comparatively large line attenuationsthe calculations stated must be carried out withgreat care and accuracy. This entails a fairlytroublesome mathematical labour, which may,however, be saved by making use of the followingapproximative equation:As will be seen from fig. 4, the curves may bedivided into sections between which the value schanges its sign. As will be seen, the value ofthe constant A depends upon in which sectionthe measuring frequency falls, and if this canbe judged, we are also able to determine thevalue of A' for the said frequency.The signs for the sections are as follows:1st section f positive, Z t > Z k , φ k >φt, K = 02nd » a negative, Z,\ < |ZJ, φ k > φt, K = 13rd • e positive, \Z,[

Fig. 5.If we measure the impedance from one terminalof a line terminated in its own impedancehut suffering from an inhomogeneity at a pointat the distance s from the measuring end, weohtain an impedance with the amplitudeFigure 5 shows curves for such impedances.If we plot curves which pass through the maximumand minimum points of the impedances(see the curves drawn with dashes), we obtaincurves with the equationswhere p and q are magnitudes that are determinedhy the nature of the inhomogeneity. Thisequation has the same character as the equations(10), and the impedance y therefore, variesperiodically.and the relation150

Fig. 6.can be calculated from these for an arbitraryfrequency.In calculating (ps + p) from the equations(14) and (15) we obtainobtainingoi in radian/secIf the inhomogeneity consists of a break onthe line or a short circuit between the branchesp = 0. The impedance curve consists in suchcase of an open or closed circuit impedance fora length s of the line, and for this length theattenuation bs can obviously be calculated bymeans of the equation (16).Example 4.Figure 6 shows the amplitude curve of ameasured closed circuit impedance on a 64.4km. long loaded cable, whose attenuation weare to calculate.According to the curves for Z im „ and Z^minwe calculate firstFrom the value obtained for n 0 - we thencalculate the attenuationobtainingφ in radian/sec (is in neper (i in neper/km.3000 0.73 0.01135000 0.76 0.01187000 0.79 0.0123The value bs + p can in the same way be calculatedfor an arbitrary inhomogeneity accordingto the equation (16). In the next chapterwill be shown, how the distance s to the inhomo-151

geneity can be calculated from the amplitudecurve of the impedance. With a knowledge alsoof the attenuation per km. line j}, which, e. g.,can be determined from the amplitude curvefor the closed circuit impedance in accordancewith example 4, the attenuation to the inhomogeneitycan be calculated and by thisor of an inhomogeneity in the self-inductionof the line, we obtain that character for theamplitude curve of the impedance, which is apparentfrom curve B in fig. 5, i. e. the curvefor |yl max and y min diverges with rising frequency.For such inhomogeneityIf the inhomogeneity consists in a change ofthe characteristic impedance from Z to Z l at thepoint of inhomogeneity and the impedances arepractically purely ohmic and independent offrequency, we obtain that character of the amplitudecurse of the impedance which is apparentfrom the curve A in fig. 5, i. e. the curves ' y | mMand ] y ' min are running parallell. For such inhomogeneityis practicallyExample 5.The inhomogeneity represented by periodicalvariations of the amplitude curve A lies at thedistance a 4- 64.4 km. from the measuring end.The attenuation per kilometer is the same asfor the line in example 4. We are then able tocalculate p according to equation (16 a).m2n 0 fis + p p3000i 095v^T= 0.720 1.23 0.5215205000 U^= 0.720 1.25 0.49lbOO7 000 v ^ = 0.735 1.28 0.4916boergo p is almost constant. The inhomogeneityis actually a change in the characteristic impedancefrom Z = 1350 ohm to Z 1 = 2900 ohm,brought about artificially. The latter characteristicimpedance is represented by a resistanceR = 2900 ohm. According to equation (17):which practically coincides with the p-valucscalculated from the amplitude curve A.If the inhomogeneity consists of an increaseor decrease of the capacity per km. of the line,wherefor the surplus or deficit δc in capacityandfor the surplus or deficit δL in self-induction.Example 6.The inhomogeneity represented by periodicalvariations in the amplitude curve B lies at thedistance S = 64.4 km. from the measuring end.The attenuation per km. is the same as for theline in example 4. We are then able to calculatep according to equation (16 a).a n 02δs + p p3000 j^jj = 0.869 1.68 0.955000 ^ ^ = 0.789 1.41 0.65lo20'000 !S? = 0.723 1.26 0.471 / 30The inhomogeneity consists in reality in a riseof the capacity per km. of the line, broughtabout artificially, by additionFrom equation (18) follows:o) K p3000 1300 • 3000 • 0.089 • 10~ 6 = 0.347 0.895000 1350 • 5000 • 0.089 • 10" 6 = 0.600 0.627000 1400 • 7000 • 0.089 • 10 « = 0.872 0.46which practically coincides with the /^-valuescalculated from the amplitude curve B.For a line with two capacitative inhomogeneitiesthe amplitude curve of the impedance assumesthe character shown by curve C in figure5. By introduction of the "inhomogeneity impedance",lines with several minor inhomoge-152

ncities can very easily be dealt with (see thepapers cited).In plotting amplitude curves for investigatinga line's homogeneity, the line must be terminatedso as to give as little reflexions as possible. Thetermination of the line should, therefore, havean impedance as nearly equal to the line's characteristicimpedance as possihle. As a suitabletermination may, by way of example, be connecteda balancing network made for the line.Example 7.From the amplitude curve in figure 6 of theclosed circuit impedance of a 64.4 km. long loadedcable line we obtain 6 maximum values betweenthe angular frequencies ω = 3450 andω = 8050. The distances between the consecutivemaximum and minimum values appearto be almost equal. We may then assume thatthe function m is constant within the said frequencylimits and according to equation (20)we get8. Localising Inhomogeneities from AmplitudeCurves for Measured Line Impedance.As has already been stated, we are able todetermine the location of the inhomogeneityfrom measured impedance curves. Since an arbitraryfault on the line as a rule entails an inhomogeneityon the line, we are able to localiseline-faults of an arbitrary nature by means ofimpedance measurements. The theoretical basisfor localising inhomogeneities has been dealtwith in the papers cited in the previous chapter.If ω t and ω2 are the frequencies correspondingto two consecutive maximum or minimum valuesof the amplitude curve, it holds good that thedistance to the inhomogeneityExample 8.The amplitude curve C in figure 5, which isplotted for a line of the same nature as inexample 7, shows two inhomogeneities. In calculatingthe distance to the inhomogeneity whichcorresponds to the shorter frequency distancesbetween the amplitude values we obtainaccording to example 7where ip' is a function. If this function is known,the distance to the inhomogeneity can consequentlybe calculated. From the equation (19)it is evident that the distance to the inhomogeneityis greater the shorter the frequency distancebetween the consecutive maximum or minimumvalues of the impedance curve.In order to determine the function 1/ we plotan open or closed circuit impedance curse fora line of the nature in question and with aknown length » x . The frequency distancesw 2 — w, between the maximum or minimumvalues of the impedance are calculated from thecurve, and by thisby which according to equation 19:or(The actual distance was 64.4 km.)In calculating the distance to the inhomogeneitywhich corresponds to the greater frequencydistances between the amplitude values we obtainand153(The actual distance was 32.2 km.)In the case of loaded cables, it is impossibleto determine the distance to the inhomogeneity

with any greater accuracy than one coil section,i. e. approximately 2 km.9. Localising Cross-talk by Means of ImpedanceMeasurement.In accordance with a method published in thepaper "Mctod for uppmiktning av laget av enoverhorning mellan tvenne ledningar" ("Methodfor Measuring the Position of Cross-talk betweentwo Lines"), by E. Fridh, in "Tekniska Meddelandenfran Kungl. Telegrafstyrelsen", No. 3,1926, localisation of cross-talk can be accomplishedby means of impedance measurements.Figure 7 shows diagramatically the measuringdevice: An impedance curve for line I is plottedwith the measuring bridge (R, HTF, X). Theaudio frequency oscillator G, which supplies themeasuring current to the potentiometer P alsosends out a disturbing current on line II. Atthe cross-talk point this current causes a disturbanceon line I, which appears as an inhomogeneityin the impedance curve. In order to allowthe periodical variations in the impedance curve,which correspond to the disturbance, to becomemore prominent, impedance measurements arecarried out both with and without any interferingcurrent on line II.The difference between these two measurementsis altogether due to the disturbance, andcan be made great relatively to the impedanceby making the measuring current small in relationto the interfering current by means of thepotentiometer P.The calculations are carried out in the sameway as in localising inhomogeneities. In thiscase a value for S x («, oij) must be used, whichis the mean value between the S t (w, — ojj) valuesfor the lines I and II, e. g. calculated from closedcircuit impedance curves plotted for each line.10. Impedance Adaptation.At junctions between two telephone lines ofdifferent kinds or at junctions between a telephoneline and a subscriber's apparatus, forexample, a greater or smaller fraction of the telephonecurrents is generally reflected. This reflexionis always detrimental to the transmissionof speech, the result being additional attenuationand distortion of the speech transmitted. Thisholds good particularly in respect of long telephonelines equipped with amplifiers. As a ruleit is, therefore, a problem in telephone engineeringalways to arrange the lines and apparatusin such a way that the reflexions are reducedas much as possible.154

The relation between the magnitude of the reflectedcurrent and that reaching the point ofreflexion is:where Z x and Z, are the characteristic impedancesof two interconnected lines or Z x thecharacteristic impedance of a line which is connectedto an apparatus of an arbitrary kind withthe input impedance Z 2 .We assume that the difference in phase betweenthe impedances Z, and Z, is q\ the relationbetween the impedances may then be representedby the vector diagram as per figure 8.forThe reflexion disappears completely when alsoif> = 0.We must, therefore, endeavour to get theamplitudes of the two impedances as equal aspossible and also the difference in phase angleas small as possible.The question remains, however, how the impedancesof lines and instruments may be altered,so that impedance matching is attained at thepoints of interconnection. As a rule neither linesnor instruments are altered directly, as thismostly encounters insurmountable practical difficulties.The way, therefore, is to introduceapparatus which are capable of effecting moreor less good impedance matching in such cases.We have such an apparatus in the transformer.Figure 9 shows diagrammatically impedancematching between the impedances Z x and Z 2 byFig. 8.According to the cosinus theorem:We put this expression =the denominationV'y and introduceand obtain thenIn order to make the reflexions as small aspossible, we must make y as small as possible.If y is derived in respect of )) we obtain:means of a transformer with a ratio of turns of1 :n from the terminals I, 2 to the terminals 3, 4.We assume that the transformer is an ideal one,i. e. that the shunt impedance between the linebranches and the series impedance in the linebranches caused by the transformer, may be ignored.As a matter of fact, we can design andbuild transformers in respect of which this holdsgood approximately, or in which the relationbetween the series impedance and the shuntimpedance is adapted in such a manner that thesame result is attained.If we measure the impedances from the terminals,1, 2 towards the transformer we shallthen obtain an impedancefrom which it will be seen that the minimumreflexion isand if we perform the same measurement fromthe terminals 3, 4 we obtain155

n is, however, a scalar magnitude from whichit follows thatAccording to the preceding we obtain a minimumof reflexions at the terminals 1, 2 and3, 4 ifbe connected with an unloaded undergroundcable. By loading the cable it is, however, possibleto alter its characteristic impedance insuch a way that it coincides better with that ofthe overhead line, and the addition of suitableloading coils may, therefore, occasionally be anappropriate means for bringing about good impedanceadaption. (See the paper "Pupiniseringav inledningskablar" ("Loading Terminal Cables"),by A. Holmgren, in "Tekniska meddelandenfran Kungl. Telegrafstyrelsen", No. 6 b—7,1924.)and this gives the condition for the minimum ofreflexions, viz.We are obviously unable to eliminate by meretransforming those reflexions which are causedFig. 10.by a difference in the phase angle between theimpedances Z, and Z.,. Nevertheless, telephonelines as a rule have small phase angles, makingthe differences in phase angles also small. Sufficientlygood impedance matching is thereforeobtained by mere transforming.Figure 10 shows repeating coils, manufacturedby the Svenska Radioaktiebolagct, which, interalia, are used for impedance matching.If the characteristic impedances of two connectedlines do not have the same impedancefrequencycurve, the impedance can be matchedby repeating coils for a single frequency only.Such a case occurs when an overhead line is to11. Line Balancing.In line balancing we are faced by the problemof designing by means of resistances, condensersand possibly, self-induction coils, an impedancenet whose impedance for all important frequencieswithin the voice range is as far as possibleequal to the input impedance of the telephoneline. This line balance and the line itself areconnected to the same hybrid coil in a telephonerepeater.From the preceding it is obvious that the inputimpedance of a telephone line is determinedboth by the wave outgoing on the line and thewaves reflected at the inhomogeneities of theline and at the other terminal. The outgoingwave may be balanced with a line-balance whoseimpedance is, as far as ever possible, equal tothe characteristic impedance of the line. When,for example, an overhead line is led into the stationby a terminal cable, near-by reflexions mayoccur. These reflexions may be balanced byconnecting before the line-balance an artificialline with, as far as possible, the same electricalcharacteristics as those of the terminal cable.On account of the attenuation of the line,which the reflected currents have to pass, distantreflexions are less noticeable in the input impedanceof the line than near-by ones. Such reflexionscannot without very great practical difficultiesbe balanced. Since from all points ofview it is better to try directly to reduce theline reflexions to a minimum, efforts are nevermade to balance reflexions from distant points.Often the reflexions are also variable, due tothe fact that at the far end of the line other linesand instruments with different electrical characteristicsare alternatelv connected, and even156

the line itself may by reason of climatic changesbe subject to periodcial changes.In order to calculate a line-balance whichreproduces the characteristic impedance of theline, we must know the primary line constants(e. g. obtained from open and closed circuitimpedance measurements) and also, in the caseof loaded line, the self-induction of the loadingcoils and the coil spacing.The characteristic impedance of an overheadline can with fairly good approximation be re-are the primary line constants, including theself-induction and resistance of the coils, andS is the coil spacing and d the length of cablefrom the beginning of the line to the first loadingcoil in km., we obtain the following dimensioningformulae:When the line-balance is assembled, its impedanceshould be measured and compared withthe measured input impedance of the line. Ifthe coincidence is poor, either the line-balanceor the line is defective. The former is testedby checking the elements constituting the linebalance, and the latter by DC or AC linemeasurements.The relation between the detrimental currentreturning to the amplifier and that sent out onthe line is determined by the expressionFig. 11.produced by a resistance R in series with a condenserK (see figure 11), whereandL, C and r are the primary line constants.A fairly short unloaded terminal cable beforethe overhead line may with good approximationbe reproduced by a T-device (see fig. 11) withtwo resistances g'2 and a capacity rj, dimensionedin such way, that Q = total resistanceand t) = the total capacity of the cable.The characteristic impedance of a loaded linemay, for example, be reproduced by a Hoyt-balancesuch as shown in fig. 12. If L, C and rwhere Z L is the input impedance of the line andZ B the impedance of the line balance.Assume for a moment that the input impedanceof the line equals Z B . The balance isthen complete and no currents return into theamplifier. As to reflexions, the amplifier thenhas an apparent interior resistance = Z B .Suppose now that the input impedance of theline is Z L . In this case a reflexion occurs betweenthe amplifier and the line, determined bythe above term, and the reflected current is justthat one returning into the amplifier.The balancing problem can, therefore, be dealtwith as a reflexion problem, where it is a questionof diminishing the reflexions between theline and the line balance. According to chapterFig. 12.10 we are then able immediately to state thatZ L shall, as far as possible, be equal to Z B \and that the difference in phase angle betweenthem shall be as small as possible. We assume157

that ZL is almost equal to Z B and that thesaid difference in phase angle, which we denominateq, is small.We are then able to writeThe amplitude of the input impedance of theline or the impedance of the line-balance thusappears as denominator in equation (25), andwe are consequently forced to calculate the amplitudein those instances where the impedancesZi and Z B are expressed in resistance and reactance.If we express equation (25) in amplitudes\Zj\ and \Z B and,difference in phase angleq, (expressed in arc measure) the equation assumesthe following simple appearance:We are thus able to judge the conformity betweena line and its line-balance by two terms,one merely containing the amplitude relation\Znand the other merely the difference in phase1**1angle ^.12. How Should an Impedance be Expressed inthe Most Appropriate Manner?From chapters 4 and 6 we find that by calculatingthe constants of a line the simplest wayis to have the impedance expressed in amplitudeand phase angle; from chapter 7, that amplitudecurves can be used for calculating the attenuationof the line and the magnitude of inhomogeneities;from chapter 10, that we arechiefly interested in the impedance amplitudes,when it is a question of impedance matching;and from chapter 11, that the impedances arel>est expressed in amplitude and phase angle injudging the conformity between a line and itsline-balance.There exists, therefore, an actual need to obtainmeasured impedance expressed in amplitudeand phase angle, which in no instance can beasserted about impedance expressed in reactanceand resistance. By impedance measurements ishere not meant measurements of resistance, capacityand self-induction.The fact that in spite of this we generally expressimpedance in reactance and resistance issimply due to the trouble and difficulty in conversingthe bridge readings of the impedancemeasurement bridges now generally used (seechapters 2 and 3) into impedance expressed inamplitude and phase angle, while resistance andreactance can be easily obtained.This is, of course, decisive in such cases wherethe impedance not necessarily has to be expressedin amplitude and phase angle, e. g. inlocalising inhomogeneities and cross-talk, in accordancewith chapters 8 and 9, in which casethe calculations can as well be made from theperiodical variations in resistance and reactance.A natural consequence is also that an endeavourhas been made to get along with impedance expressedin resistance and reactance even in suchcases where impedance expressed in amplitudeand phase angle would have been more valuable.The problem of measuring directly with asimple bridge the amplitude of the impedanceas well as a measure of its phase angle is, however,now solved both theoretically and technicallyby the Impedance Meter Type IM 329manufactured by Svenska Radioaktiebolaget.This improved measuring method makes itpossible without inconvenience to express an impedancedirectly in amplitude and phase angle,which, consequently, in future should be normallyemployed.13. The Principles of the Svenska Radioaktiebolaget'sImpedance Measuring Set.Fig. 13 illustrates the principles of the measuringbridge. The various elements are given thefollowing denominations, which, simultaneously,indicate their electrical magnitudes:R two equal ratio resistances, A T a variableresistance, X the unknown impedance, G anaudio frequency oscillator with the output voltageV and the angular velocity' to, ML p an aircore transformer with the primary self-inductionL p and a variable mutual inductance M, K'an impedance and HTF a telephone receiver.Measuring is done by setting the resistance Nand the mutual impedance M for sound minimumin the telephone receiver HTF. The currentin the receiver is in such a case practically equalto zero. A current I., passes then through theresistance N and the impedance X. a current /,158

The resistance X, therefore, becomes equal tothe amplitude |A'| of the impedance and thephase angle ^ can be calculated from the valuesfor ω, M and A.The mutual impedance can be made both negativeand positive (e. g. by inversion of the secondarywinding of the transformer and the settingof the bridge can, therefore, be made forboth negative and positive phase angles tp.The calculation of the phase angle can beFig. 13.through the two resistances R, and a current l tthrough the impedance A" and the primarywinding of the transformer ML pIf we assume that the impedance A" is of sucha nature that the current l t is in phase with thecurrent l x the induced E.M.F. ω 3// 3 in the secondarywinding of the transformer will be inphase quadrature to the current /j and to thepotential drop /?/,. The vector diagram shownin fig. 14 is then obtained, q: representing thephase angle of the impedance X.From this vector diagram follows immediatelythatandIf A" is the impedance value of the impedanceA connected in series with the primary impedanceof the transformer, which is mainlydue to the inductive resistance of the self-inductionL,„ we obtain the relationby which the above equations becomeFig. 14.further simplified if the impedance A (whichis obviously pure ohmic) for all measuring frequenciesis set on a value that is proportionalto ω.If, therefore,where F is a constant, the equations (27) assumethe appearancewhere the phase angle is determined merely bythe mutual inductance M.14. The Practical Design of the Svenska Radioaktiebolaget'sImpedance Measuring Set.Fig. 15 shows the exterior appearance of theSvenska Radioaktiebolaget's Impedance MeasuringSet, and fig. 16 the arrangements below thepanel. Fig. 17 shows the markings and designationson the front of the panel, and fig. 18the circuit diagram.159

Fig. 15.Fig. 16.160

The resistance N consists of a decade resistancevariable by means of four decades between0 and 11110 ohm with steps of 1 ohm.If it is desired to measure impedances beyond11110 ohm or with greater accuracy than 1 ohm.the short circuit plate between the terminalsmarked "N" should be removed and an additionalresistance connected between said terminals.This resistance is then connected in serieswith the decade resistance N.The design and construction of the transformer(ML,, in fig. 13), whose appearance willbe seen from fig. 10, is very interesting. Theprimary is a non-capacitive winding on a ringmade of insulating material, giving a homogeneoustoroidal-shaped magnetic field. The secondarywinding is without inter-winding capacityarranged outside the primary and is providedwith a series of terminals which are connectedto the contacts of the four decade switchesmarked '. tg ^- (see fig. 17).The mutual inductance may, therefore, bevaried by connecting to the telephone receivercircuit a variable fraction of the secondarywinding. The secondary is inversed in relationto the receiver by means of the switch marked*qj*, reversing the sign of the mutual inductance.The air core transformer, which replaces thedecade condenser in the bridge according tofig. 1 and 2 or the variometer in the bridge accordingto fig. 3, possesses the following advantages.1) The air core transformer can be mademuch smaller than both the decade condenserand the variometer.2) The air core transformer can without anydifficulty be adjusted with the desired precision.3) In contradistinction to the variometer theair core transformer does not send out any exteriormagnetic field, thus obviating magneticdisturbances on near-by instruments.4) For the same reason the air core transformeris insusceptible to iron masses in the vicinity,which may endanger measurements madewith a variometer.Fig. 17.161

The air core transformer can, therefore, withoutany disadvantages be mounted on an ironrack or covered with an iron casing. This willbe the case when the instrument is mounted ontest-racks.5) For the same reason the impedancemeasuring set can be protected against exteriormagnetic disturbances, because the instrumentcan he covered with a protective iron casing.6) The adjustment of the air core transformeris not endangered by mechanical wearand tear, as, for example, is the case with variometerswhere the bearings are gradually worn.In chapter 13 is stated that the impedance A"(see fig. 13) must be of such a nature that thecurrent /., is in phase with the current /,. Thisis done by adjusting the handles marked "ffor the frequency / which is to be used for themeasurement. The switches "/" are both providedwith a frequency scale.The magnitude of the impedance A (see equation28) can by means of the handle marked"f'" be varied for 10 different values, selectedwith regard to 10 of the most important measuringfrequencies, viz. 400, 500, 600, 800, 1000,1200, 1400, 2000. 2800 and 3600 cycles/sec. andthe positions of the handle marked "f" aremarked with these frequencies. If we exchangeFig. 18.— 162 —the short circuit plate between the terminalsA' for a variable resistance, an arbitrary valuecan be added to the impedance A'.The dimensioning of the impedance A is suchthat a correction for the slight error which occursthrough capacity in the air core transformeris considered.The unknown impedance is connected to theterminals marked "X"; an audio frequency oscillatorto the terminals markedU GEN"; and atelephone receiver to the terminals markedU HTF".In order to obtain more accurate adjustmentof the handles marked u f the switch marked"AAA. MATS" is placed in the position "AAA"(calibration), a sound minimum thus being obtainedin the telephone receiver if the saidhandles are properly adjusted. By this meanswe obtain simultaneously a check on the measuringfrequency. In measuring, the said switchmust be in position U MATX" (measuring).15. Measurements.After every alteration in the measuring frequencyf the handles u f" are adjusted for thisfrequency. As has already been pointed out, adjustmentof the said handles can be made more

safely l>y means of adjustment for a soundminimumwith the switch "KAL. MATN* in position*KAL* (calibration), inasmuch as we obtainat the same time a check on the measuringfrequency.The switch "KAL. MATX" is put into position"MATff and the measuring is done by adjusting. . • f w"a sound-niinimum with the decades —— tg —and "A". The switch "m", whose positions arealso marked with the sign of the phase angle ofthe object to be measured, may have to be reversedto enable sound-minimum adjustment.From the adjustment of the decades "AT" weread directly the amplitude of the measured impedance,and from the adjustments of the de-T 9* r

Fig. 1.Fig. 2. Fig. 3.

Electric Interlocking Plant at Vanneboda Station.By G.Pervall.At the end of the year 1927 tenders were in-.. vited for an interlocking plant for VannebodaStation, on the Grangesberg—OxelosundRailway. The station, whose track system isshown on the sketch, fig. 1, is a junction for thelarge ore shipments from the various ore fieldsin the Bergslagen District to the port of Oxelosund,and serves as a junction for passengertraffic to and from the surrounding part ofBergslagen. The tender concerned a mechanicalplant, but for purposes of comparison tendersfor an electric plant were also invited, becauseon account of the size and traffic conditions ofthe station it was impossible to decide withoutany further ado which type of plant would bemost advantageous. The railway has alreadya large number of mechanically operated plants,which have been quite satisfactory ami withwhich the staff are quite conversant. The stationswhich have been equipped with such installationsare, however, smaller than Vanneboda.On the other hand, there did not existany electrically operated plant, and, consequentlyno staff capable of running such.In order to get an economically practicablemechanical inlerloeking plant, a locking machineof detached crank type and with signals of thesemaphore type was also a sine qua non forVanneboda. The scheme for the electrical interlockingplant also necessitated its being capableof being housed in a low building, suitably locatedbetween the platform tracks and in sucha way that the train dispatcher would be personallyable to look after it in the course of hisduties, whereas for shunting purposes the centrallyoperated points would be manipulatedby local devices. To facilitate inspection of thepermanent way tracks were to be put down atboth ends of the station for checking whetherthe line was open and setting the signals againstthe trains. The signals in this instance wereto be made in the shape of daylight signals withthe lamps of the main signals normally fed fromthe existing 127-volt alternatic current electriclight net. The motor current battery was toserve as an emergency power source in case ofa breakdown in the supply of alternating current.Fig. 2 and 3 represent sketches on the samescale, showing the two frames suggested. Thelength of the electric interlocking machine (fig.3), a normal 24-lever frame, is 1950 mm., andthe length of the crank apparatus (fig. 2) is5040 mm. As will be seen from the sketch, theelectric interlocking machine contains 7 spareplaces for future enlargement, whereas the possibilityof spares in the crank apparatus islimited to 2 cranks and one track lever.On comparing the two types of plant the railwaymanagement, after the costs for certainwork which the railway itself was to carry out,e. g. laying down of line-drums etc. for the mechanical,and the erection of a cabin for the electricalplant, that the initial costs of the twoplants were on the whole equal. It was found,however, that the train dispatcher himself wasnot to manipulate the mechanical interlockingplant, but a separate operator or operators would— 164 —

F 1584 Fig. 4. Interlocking Machine.be required for this purpose, enchancing the runningor working expenses of the mechanical plantas compared with the electrical. The railwaymanagement consequently decided to have theelectrical plant on account of its being the moreeconomical. The choice proved its merits alreadyfrom the beginning inasmuch as a desideratumof laying down further tracks couldeasily be accomplished, which would have beenimpossible if a mechanical plant had been erected,because the local capacity of the crank-apparatusdesign in this respect was already fullyutilized.From the machine, shown in fig. 4 are operatedthose points which are to occupy different positionsfor the most frequently occurring tracksand are locked by the aid of electrical locking devices,fig. 5 the points and switcheswhich are operated locally and requirelocking for the varioustracks. The locking devices areprovided with point contacts, theyare integral parts of the apparatus,and are connected with lock-magnetson appurtenant lever in sucha way that the latter cannot be putover unless the points are in alcckable and proper position.Out of the points 23, 24 and26 located immediately outside theinterlocking plant, as well as thescotch blocks Sp IV and Sp V thetwo scotch blocks and point 23are provided with point contactswhich are connected with a relayequipped with an optical signal inthe machine, enabling control ofthe proper position when a trackdepending upon the said pointand scotch block is to be put over.Points 24 and 26, which are facingpoints for the said track, arelocked by key-locks which cooperatenot only reciprocally butalso with a key-lock on top of theinterlocking machine, this keylockbeing provided with contactsfor obtaining the necessary electricaldependence between thepoints and the signal lever correspondingto the track.As has been stated in the preceding, the mainsignals are erected as electric daylight signalsand are controlled by means of signal controllamps placed in separate housings on top of theinterlocking machine (see fig. 3 and 4). Toprevent the lamp in the light signal from becomingextinguished, the control lamp is providedwith a shunt-resistance. Fig. 6 showsthe h ome signals B 1/2 and C 1/2/3, which aremade with masts of reinforced concrete. Automaticbells with the use of insulated tracks havealready previously been arranged at the twolevel crossings at the outer ends of the station.These tracks have also been utilized in the plantfor the purpose of being able to control togetherwith tracks specially laid down for the same,that the parts of the tracks at the outer ends ofR 15S3 Fig. 5. Electrical Locking Device with Locally Operated Scotch Block.— 165 —

R 1588 Fig. 6. Home Signals B 1/2 and C 1/2/3.the station are clear of vehicles. All track linescan be controlled by the track relays, Scy, Scietc., provided with optical signals, these relaysbeing housed in the casing on top of the machine.The relays are under normal conditions currrentless,but are supplied with current via the pedalcontact which is an integral part of the machine,when the tracks have to be controlled. The necessarydependence is obtained by contacts onthe track relays, so that no signal can be set forclear if a track-line on the corresponding trackis occupied by vehicles. The track-lines laiddown through the centrally operated pointsare also used for locking a respective point-leverto prevent its change over while any vehicle ispassing the point or is in it.The direct current necessary for the plant issupplied by two Xife accumulators, a motor current— and a control current battery which arecharged by means of mercury rectifiers from theexisting alternating current net.The plant has now been in operation sincethe spring of 1928, and has all the time provedto fully come up to the desiderata of the buyerin so far as reliability, convenient and easyoperation both in dispatching trains and shuntingare concerned, as well as low charges foroperation and maintenance are concerned.Fig. 7.Cabin.— 166 —

Porcelain Insulators and Insulator Porcelain.Observations and Views on the Causes of Insulator Failure.By Sten Velander. Professor at the Royal Technical University, Stockholm, Sweden.Report to the World Engineering Congress at Tokyo 1929.The subject discussed in this paper has beendealt with earlier by the author; first, in thetreatise mentioned on the next page and publishedas No. 90 among the "'IngeniorsvetenskapsakademiensHandlingar" (Monographs of TheRoyal Swedish Institute for Scientific-IndustrialResearch), and second, in a report to the Paris Congressof 1929. At that stage, however, the extensiveinvestigations given here in condensed formof the destructive temperature differences, whicheven moderate leakage currents are able to produce,had not been completed. These researchesalso show how the thermal stresses arise and howlittlethey have to do with the different coefficientsof expansion for porcelain, iron and cement.They strongly support a number of explanationspreviously advanced more as hypothesesof insulator failures. However, they mustbe viewed against the background of earlier researcheson operating experiences, etc., which latter. have therefore been reviewed in this report despitethe fact that they were published earlier inthe papers mentioned.Prcelain is of great importance as an insulatingmaterial within all branches of electricalengineering. Only if the porcelain insulatorsare reliable and durable at all points can ourelectric plants and especially our power transmissionsystems give fully first class service.In other fields research has very largely siftedand marshalled the phenomena concerned andsolved the most important problems, but in thecase of insulator porcelain as well as many otherinsulating materials much still remains beforecalculation, design and manufacture have theproblem so well in hand as they have withinother branches of engineering.This drawback has especially made itself feltin electric power transmissions, where the deteriorationof insulators has reduced and stillseriously threatens the reliabilitv of the service.Innumerable theories concerning insulator deteriorationhave been advanced, tested, and rejected.The porcelain insulators on our power linesare exposed to many different stresses, bothelectrical and mechanical. The electric forcesare derived, first, from the continuously operatingworking voltage and, second, from impactstresses set up by excess voltages. The mechanicalforces are, first, in the main continuousstresses from wires or cables, second, oscillatingforces transmitted through the conductors, third,internal stresses due to faults in manufactureand erection, fourth, thermal stresses due toexpansion of the porcelain, cement, iron, etc.resulting from temperature rises and above allunequal temperature distribution in the insulator.The magnitudes of these different forces arenot directly determinable, not even approximatively.For this reason as well as in view ofthe complicated shape of the insulators and theproperties of the porcelain, it is at present impossibleto establish on a mathematical basis thestresses occuring in the insulator. We mustcontent ourselves with the fact that in manycases these forces and their resultant stressesbecome so intense that they, at least gradually,bring about a deterioration of the insulator.Insulator statistics obtained from variouscountries and distributing systems as well asthe existent literature on operating experienceswith porcelain insulators have been analysed inorder to ascertain by an indirect method whetherand under what conditions the different kindsof forces are capable of causing insulator failure.These investigations were commenced some 7years ago. For particulars of this analysis the— 167 -

eader is referred to the description of theauthor's insulator researches published in 1929under the title of "Porzellanisolatoren und Isolatorenporzellan"in the "IngeniorsvetenskapsakademiensHandlingar" (Monographs of theRoyal Swedish Institute for Scientific-IndustrialResearch, Stockholm). The method adopted atthis investigation was to compare the relativenumber of insulator failures under conditionsthat had been different with respect to the magnitudeand frequency of the forces capable ofproducing electrical or mechanical stresses inthe insulators.This siudy of operating experiences has giventhe following result:The electrical forces play no noteworthy primarypart in insulator deterioration. The initialfault is practically always a crack produced bymechanical or thermal stresses. Electric punctureis a secondary phenomenon.The forces with which wires and cables actupon the insulators are a distinctly contributorycause of insulator failure. The continuous stressprobably contributes its share, though the effectof the vibrations is most clearly discernible.The internal forces, strains in the porcelain,swelling of the cement, extra stresses from theerection, and suchlike, appear to be without appreciableeffect where due care in manufactureand erection has been practised.The thermal stresses arising from unequaltemperature within the insulator appear to playa very important part as a destroyer of insulators.In all coast-regions, where the atmosphereand winds are more or less saliferous,semi-conducting deposits settle on the insulators.As a consequence there arise powerfulleakage currents, which with for instance 50 kVpin-type insulators may develop an effect of amagnitude of .5 k\V per insulator, at least ifthere are no salt ribs. The generation of heatthen principally takes place under the lowestpetticoat, where the salt deposit is least, i. e.the resistance greatest. Heating tests, whichimitate the actual conditions under which heatingoccurs through leakage currents, have beenperformed on two different types of 50 kV insulators,one consisting of three pieces joinedtogether with hemp (v. Fig. 1) and the othermade in one piece. The salt ribs on the insulatorin Fig. 1 do not appreciably alter thedistribution of temperature, having only the effectthat under otherwise equal conditions theyincrease the leakage resistance and thus reducethe leakage effect. With a heating effect of 200watts placed under the lowest petticoat, differencesof temperature amounting to over 100° Cwere obtained within the insulator, as can beseen from Fig. 1. The other tests gave correspondingresults. With a leakage effect of .5 kW,the temperature differences to be reckoned withwould then be at least 200° C for such insulators.In cemented insulators still larger differencesof temperature may doubtless arise. In this wayconsiderable thermal expansions and stresses areproduced. Poor service from insulators in districtsalong salt-water coasts is in fact an experiencecommon to all parts of the world. Ina similar way the pronounced variations of temperaturein alpine regions appear to contributeto insulator deterioration.It has not been possible to establish any effectfrom the height of the temperature in itself, andthus there is no support for the hypothesis thatascribes insulator deterioration to the differentcoefficients of thermal expansion for the porcelainand cement. Nor has it proved possible tofind any operating experiences which confirmthe theory that the slow chemical alteration andinduration of the cement is a cause of insulatorfailure. On the other hand, as shown by anotherinvestigator, the cement has the propertyunder otherwise similar conditions of powerfullyincreasing the differences of temperaturewithin assembled cemented insulators. On eachside of the cement joint there are formed capilliary,highly heat-insulating, layers of air. Theeffect of the cement in insulator deterioration isthus of a different character from that till nowassumed, but is also quite certainly considerablyless than hitherto thought. The same influence isdoubtless exercised by all kinds of joints. Forinstance, the previously-mentioned heating experimentswith an insulator composed of threeparts joined together by hemp showed that thehemp joints absorbed 25—30 % of the total differencein the temperature between the outerand inner surfaces of the insulator (v. Fig. 1).The slow chemical change which occurs in thecement has however been the hitherto acceptedcause of the so-called ageing of the insulators,i. e. that insulator deterioration does not begin— 168 —

at once, but first after some years,when it begins to manifest itself asa cracking process in the insulators.In order to shed some light on thisageing phenomenon I have analyseda very large volume of insulator statistics,principally from Swedishtransmission lines. The insulatorson the Swedish plants in questionhave been under such control thatthey have practically always been replacedbefore the cracking has proceededso far as to give rise to electricalpuncture. My investigations aswell as others have shown that thecracks do not arise all at once butcommence as fine probably submicroscopiccracks, often on the inner sideof the porcelain shells and thenwander very slowly through the porcelain.It may take months for thecrack to penetrate the 15 to 20mm. shell and years before the porcelainis completely cracked. Such anevolution of the crack proves that itcannot be produced either by internalstrains or by occasional stressesreaching up to the breaking limit.The investigation further showedthat, as above mentioned, the numberof insulator failures is not evenlydistributed all years. The first fewyears are practically speaking freefrom such failures. After that thenumber of cracked and destroyed insulatorsgrows for every year. Withinsulators of good and uniform qualityunder similar conditions a veryclear and regulated variation in thefrequency of insulator failures is obtained.If for a group of such insulatorsa curve is set up showingthe frequency of insulators with differentlengths of life, the maximumfrequency is obtained at a certainservice age, the frequency then diminishingfor both shorter and longerlives. Fig. 2 shows a frequency curve of thisdescription. As will be seen from the figure,the frequency curve for the life of the insulatorsR 1748 Fig. 1. Heating Test on Porcelain Insulator.At the points denoted by o> on the surface and within the porcelain the temperaturemeasured is given in ° C under a heating effect of 200 watts. Isothermshave been drawn in on the basis of these temperature. Surrounding temperature 20° C169 —corresponds with unusually great exactness tothe normal frequency curve founded on thetheory of probability.

R mm Fig. 2. Above, Frequency Diagram. Below DistributionCurve for Life-times of Pin Type Insulators Type --lion a South-Swedish Overhead Line.Number of insulators: 867; average life: 9.5 years. The broken line indicatestotal number of replaced insulators; the unbroken line total number replacedon account of cracks — on account of a general change of insulators all havenot been in service until they cracked.The life-time has been counted as extending to time of discovery of a visiblecrack. The occurrence of cracks was controlled by half-yearly inspections. Thefrequency diagram indicates the number of insulators (in percentages of thetotal number) falling within each half-yearly interval of the life-time. Thedistribution curve is the cumulative curve of the frequency diagram.The small circles and dot-and-dash curves give the frequency curve calculatedby the equationand the corresponding distribution ram (probability curves) obtained by integration.In the equationu — frequency, here in " for the half-year;N — the total number, here 100 %;y — the average deviation; calculated in the same way as the radius of intertiafor the surface between the frequency curve and the axis of the abscissaewith reference to the line of symmetry of this surface;m — the average life-time, the abscissa for the line of symmetry of the frequencycurve;i — the time, the length of life.170Thus, around a certain normal life-time insulatorswith a shorter or longer life distributethemselves in a manner which conformably tothe theory of probability corresponds to a naturaldispersion, dependent on chance, aroundthe normal value. For insulators put into serviceat the same time the frequency curve for insulatorfailure coincides with the frequencycurve for the life-times or service age of theinsulators and may then be used for theanalysis. If the insulators are put intoservice at different times, however, thefailure frequency must be converted intolife-time frequency in order to get accurateresults. For many different sortsof insulators and for different powertransmission lines in Sweden as wellas in other countries such frequencycurves or, as the case may be, distributioncurves (i. e. the integral curve forthe frequency curve) have been set upand their correspondence with the probabilitycurve ascertained. This is readilyachieved by seeing if the distributioncurve drawn on so-called "probabilitypaper"turns out to be a straight line.When analysing insulator statistics,however, different sorts of insulatorsmust not be confused, nor must statisticsfrom lines with dissimilar externalconditions be compared, for then theresult will be a distribution curve consistingof several different components,each showing a different normal lifetime.Such an indiscriminate collectionand treatment of the primary materialleads to erroneous and misleading conclusions. In those cases believed to showa constant failure percentage per annum,i. e. no normal life-time for theinsulators, just such very heterogeneousfailure statistics would appear to be involved.Judging from everything, therefore,we have to proceed from the assumptionthat a given sort of insulator undergiven external conditions has a normallife, around the normal value of whichthere is a dispersion corresponding tothe theory of probability. Nor is thisfact reconcilable with the hypothesis ofan occasional high stress attaining to thebreaking limit and therefore cracking the insulator.The hypothesis of the subsequent indurationof the cement as cause is also incompatible withthe progress of insulator deterioration, for insulatorshave considerably different lengths of

R 1751 Fig. 3. Above, Distribution Curves for Life-times of PinType Insulators of Types A1, A2 and D on a South-Swedish Overhead Line.Broken lines indicate total number of replaced insulators and unbroken lines thosereplaced on account of faults (cracks). The small circles and dot-and-dash linesindicate the probable course of actual distribution curves. For A, compare Fig. 2.The curves marked St are for insulators on straining towers, those marked Su forinsulators on suspension masts (flexible).Below, Stress-Strain Diagram obtained from Bending Tests on porcelainrods sawn out from the different insulator types.The diagrams marked g are for glazed porcelain, those marked u for un glazed.The coefficient of linear expansion established by tests is given for .4, and D.life under varying climatic conditions, underheavy or light conductors, etc., and the cementcannot harden faster if the insulator supportsa heavy cable than if it carries a light wire, etc.The course and development of insulatorcracks as well as the frequency curve for thelife-times of the insulators admit, however, ofa very natural and plausible explanation if oneassumes that insulator cracks are a fatigue phenomenon.The possibility of fatigue originatingcracks in porcelain has been contested by many,and this for various reasons.It has been considered that the fatigue-crackpresupposes a recrystallisation, i. e. a physicalchange in the material, a change whose nonoccurencehas been ascertained by microscopicand electrical tests on insulators that have beenin service a long time. The fatigue-crack doesnot however imply any change whatever of thematerial, but at the most strained spot in thematerial there arises, in consequence of varyingforces acting for longer or shorter periods, amolecular crack that slowly makes its way furtherinto the material and finally becomes visible.For such a gradual disturbance of the cohesivebond between the molecules considerably smallerstresses are required than those necessary toproduce immediate fracture.It has further been denied that porcelain,glass, and other silicates could exhibit fatiguecracksbecause the limit of proportionality andthe ultimate strength coincide; that is to say, nopermanent set appears in these materials, it beingconsidered as established that for metals thefatigue limit lies at about the limit of proportionality.However, a closer examination showsthat in certain cases the fatigue limit for metalsfalls considerably below the limit of propor-171

R 1750 fig. 4. Comparison between Stress-Strain Diagrams.To the left for a strong and weak insulator porcelain. To the right for steeland cast iron.tionality. Further is to notice that even if glassand porcelain do not show any permanent deformationsunder short-time loads, this does notimply anything respecting the conditions underloads acting for months or years. At least forglass, moreover, it has been established thatsmall loads applied during a long period alsoproduce a permanent set.It is thus not straightway evident cither thatthe fatigue limit for all materials is bound upwith the limit of proportionality or that rapidlyrepeated loads produce fatigue cracks in glassand porcelain because they generally do so inmetals A few fatigue tests on glass and porcelainperformed in the course of a few dayscannot therefore be assigned any conclusivevalue. To ascertain the behaviour of insulatorporcelain under varying stresses extensive longtimetests are required. Such have been commencedby me. It would however be desirableif parallel experiments could be undertaken atother centres, as such investigations absorb verylong time.So far, then, it seems that nothing concerninginsulator failure has been advanced that isin conflict with the results and conclusionsreached by me on the strength of my analysisof operating experiences and insulator failures,172viz. that crack-formation is thedominating feature in insulatordeterioration and that the crackis a fatigue fracture produced bythose heterogenous and varyingforces and stresses which withmore or less intensity act uponevery insulator in service.If within the engineeringworld a construction is mechanicallylacking in durability, thismust be remedied in the firstplace by altering the dimensionsin order to keep down the stresses,and in certain cases by areduction of the forces at play.Such measures are also utilisedin the case of insulators, and asthey are comparatively wellknown I will on this point merelyrefer the reader to my detailedtreatise. When improvements ofthis nature in shape and dimensionshave not the desired effect, it is usualwithin the majority of other technical fields togo in for higher quality, to try a stronger material.Should for instance cast iron fall shortof what is required, steel is used, in some casesalloy steel.Unfortunately this aspect of the problem hasbeen entirely neglected where insulators areconcerned. In all but a few minor details insulatorshave been manufactured and purchased,delivered and tested, as if all porcelainwere of equal quality from a mechanical pointof view. When deliveries of iron and machinemanufacture come into question, the quality andproperties of the material are most carefully controlledand tested. In most insulator deliveriesit has not as yet been the practice to troubleabout the mechanical properties of the porcelainused.While engaged on my analysis of insulatorstatistics I found that different deliveries of insulatorshad markedly different lengths of lifein spite of shape and dimensions as well as externalconditions being similar. I then wentin for a purely mechanical bending test on prismaticaltest-pieces sawn out from the insulators.These tests showed that insulators made of weakporcelain, under otherwise similar conditions,

always had a relatively short life, while insulatorsof mechanically strong porcelain had a relativelylong life. Fig. 3 gives some typical instancesof the results from these tests. The testalso revealed that the porcelain used in normaland according to the customary specificationsfully first class insulators may be subject tosuch extreme variations as are indicated inFig. 4. The porcelain in our insulators has thushad, and unless our views and specificationsare changed will presumably continue to have,mechanical properties that are aimlessly and capriciouslyallowed to vary within a latitude thatis relatively speaking almost as wide as thatover which all our varieties of iron and steelextend. Hence it is not to be wondered at thatoperating records and views concerning insulatorshave been so exceedingly at variance.The chief reason that mechanically weak porcelainhas found frequent use in insulators isobviously that nobody has given heed to thesignificance of the mechanical properties. Attentionhas been exclusively concentrated uponthe electrical qualities, and porcelain with highdielectric strength has been demanded, thoughthis is as a rule accompanied by inferior mechanicalqualities. Moreover, weak and brittlepcrcelain is generally easier to manufacture.Although the investigations on the reliabilityand durability of insulators with reference to themechanical properties of the porcelain used arenot yet so great in number, they point so unmistakablyin one direction that it can be assertedwith a high degree of certainty that the use ofa porcelain with high mechanical strength andductility ought to yield insulators of very highreliability and considerable durability even undersevere climatic conditions. This is naturallyen condition that certain other features such assuitable shape, etc., which experience has taught,ought to receive attention, are not neglectedeither in the manufacture or in the erection.It is conceivable that a variety of porcelainmight be selected that is resistant to short-timestresses but relatively less resistant to fatiguestresses than another sort which happens to havea somewhat lower ultimate strength. This point.however, must be cleared up by a conscientiousstudy of the resistance of different sorts of porcelainto the various forms of stresses. Thiswill be a problem for ceramists, who must takeas an example the splendid advance attained bysystematical and conscientious research in the ,iron and steel industries during but a few decades.For the present, however, we must contentourselves with demanding a high ultimatestrength in the porcelain material used; concomitantlythere ought to be a high probabilityof also obtaining great fatigue strenght. In anycase the probability of securing such ought tobe considerably greater than if, as hitherto, nodemand is made with respect to the mechanicalproperties of the insulator porcelain.Summary.Reliable and durable insulators are of thegreatest technical and economical importance forelectric plants and overhead lines. Analyses ofoperating records and insulator statistics frommany countries and lines show that the primarycause of insulator failure is a crack producedby mechanical (and thermal) forces. The electricpuncture is a secondary phenomenon. On accountof the development and course of thecracks and of the time it takes for the cracksto arise, it may be concluded that in all probabilitythe cracking is due to a mechanical fatiguephenomenon. The view of some observersthat porcelain and glass do not exhibit fatiguefracture is not founded on conclusive investigations.Rest-arch is however in progresson this point. The porcelain used in insulatorsreveals exceedingly variable mechanical properlies.The variation is as great as l>etween ordinarycast iron and high-grade steel. Low resistanceand low ductility in the porcelain appearto involve poor service reliability ami ashort life for the insulators. Therefore, withdue attention to suitable shape, etc.. the insulatorsshould lie made of mechanically strong andtough porcelain. To ensure that our insulatorsshall stand on a level with the rest of the plantin quality and reliability, a conscientious andmethodical inquiry into the mechanical propertiesof the different sorts of porcelain underdifferent stresses is recommended.— 173 -

The Svenska Radioaktiebolaget Valve-testing set.By TorbernLaurent.1. General.The Valve Testing Set Model RPR 829 of theSvenska Radioaktiebolaget is an instrumentworking on entirely new principles, and is particularlysuitable for practical measurementsand tests of amplifying valves, e. g. in a telephonerepeater station, or in a valve retailingshop. This Valve Testing Set will assess the effectiveamplification of a valve in a way which,while expressing the working conditions objectively,is still perfectly explicit. Further, theinternal resistance of a valve, its grid voltagespace, and the magnitude of the non-linear distortionintroduced by the amplifier, can be determinedby the Valve Testing Set.This is very simple in design, and althoughthe effective amplifying capacity of a valve willbe directly measured, no audio-frequency generator,valve voltmeter, etc. is required.The measurements are very easily taken, andthe method allows great precision without anydifficulty.The Valve Testing Set is suitable for routinetests as well as for a more thorough investigationof various types of valves.The risk of measurement errors caused by anoverloaded valve, which is present when valvesare measured by any previously known methodof amplification measurement, is entirely eliminatedin this Valve Testing Set.In all probability the Svenska RadioaktiebolagetValve Testing Set will make valve tests ona much larger scale than hitherto economicallypossible, and valve consumers will thus alwaysbe assured of good quality amplification.2. The Principle.It is generally known that an amplifier valve,the anode- and grid circuit of which are coupled,will act as an oscillator if the conditions requiredfor oscillations are present. The amplifyingcapacity of the valve, and the amount of reaction,are also among the factors involved inthese conditions for oscillation. For a certainamplifying capacity of the valve, the reactionmust not fall short of a certain minimum valueif the conditions for oscillation are to be fulfilled,and under otherwise unchanged conditionsthis minimum will diminish as the amplifyingcapacity is increased. This minimum reactionvalue is thus a function of the amplifying capacityof the valve, and the Svenska RadioaktiebolagetValve Testing Set is based on this fact.Fig. 1 gives a diagram of the Valve Tester.R is the amplifier valve tested, receiving its gridbias, filament voltage, and anode voltage fromthe batteries B,, B 2 , and B 3 respectively. Theanode transformer T is provided with severalprimary winding taps (2, 4, 8, 16, 32, 64), representingdifferent anode load impedances, andthe anode is connected to the tap which will givean anode load impedance as closely correspondingto the normal working conditions of thevalve as possible. Connected to the secondaryside of the transformer is a condenser C which,together with the secondary self-induction ofthe transformer, forms an oscillating circuittuned to an easily audiable frequency, e. g.800 c/s. Parallel to the condenser C is the potentiometerP 2 , in series with a high resistancetelephone receiver H, shunted with a small resistancer. Ry means of the potentiometer P 2 ,the amount of reaction can be varied, and bylistening in the telephone receiver H one ascertainswhether the device oscillates or not.The grid bias of the valve may be varied bythe potentiometer P v above the grid battery B„and as a rule this is set for the grid bias withwhich the valve is meant to work.— 174 -

Fig. 1.When the D. C. voltages and the anode loadimpedance of the valve are adjusted as closelyas possible to the desired working conditions,the measurement is taken simply by setting thepotentiometer P, to the limit of the oscillatingcapacity of the system, which can be observedby listening in the telephone receiver H. Theamplifying capacity of the valve is then readdirectly on the scale for the setting of the potentiometerP 2 , which is graduated empirically.After this preliminary survey of the workingof the Valve Testing Set, some of the attendantconditions will be explained in more detail.To enable a reaction-coupled valve to oscillate,the reaction device must be such that the alternatingvoltages returned to the grid from theanode induce in the anode circuit of the valvean E.M.F. which is exactly in phase with theoriginal E.M.F. If, as is done in the Valve TestingSet, a tuned circuit is introduced in thereaction device, the reaction phase characteristicswill very largely depend on the frequency,when this is anywhere near the resonance frequencyof the tuned circuit. Somewhere nearthis resonance frequency of the tuned circuit,a frequency fulfilling the above phase conditioncan therefore generally be found, and the valvewill consequently oscillate with this frequency.If the valve, as in fig. 2, is regarded as a sourceof current, the E.M.F. and internal resistanceof which is E and R { respectively, loaded by atuned circuit consisting of a capacity C, an inductionL with a small loss resistance /, and thepotentiometer resistance R B , the resistance R Bwill cause a voltage drop V z , the relations ofwhich to the electromotive force E will be studiedbelow.The impedance of the tuned circuit is obviously(a) = the angular frequency)orHenceorFig. 2.We assume the imaginary component of thedenominator to be small in comparison to thereal component, and the amplitude relationshipmay then be written175

and the phase angle between the voltage V x andthe E.M.F. Eat the limit of oscillations for the valve (E = /.tV g ).The amplifying capacity of the valve can obviouslybe determined by means of the relations(5) and (6), when a and the anode impedanceare known.The anode impedance is assumed to be selectedas nearly equal to the internal resistance Riof the valve as possible. If we regard fig. 3,illustrating the same point as fig. 2, but in whichthe transformer T of the ratio n:l is put in (theresonance impedance of the tuned circuit is assumedto be included in the potentiometer resistanceR B ), we therefore getIf the phase changes caused by, for instance,the transformer T, which it is the task of theangle y to compensate, are small, the amplitudeVrelationship ~ will, according to equ. (3), beb.independent of the angular frequency w- Thiswill make the reaction independent of the phasecompensation, and in the computation of theamount of reaction the tuned circuit need onlyThis makes the primary voltage of the transformerA measure of the amplifying capacity of thevalve has been chosen which will as far as possibleexpress the working properties of the valve,particularly when used as an amplifying valvebe regarded as a shunting impedance =fCWhen the reaction is adjusted to the limitof the oscillating capacity of the valve, the gridvoltage V g , regenerated by the E.M.F. acting inthe anode circuit E, will be just enough to inducein the anode circuit an E.M.F. equal tothe original E. If the amount of reaction is definedaswhere V g is the voltage at the output end of thereaction device when this is disconnected fromthe valve grid, we may, in other words, saythatFig. 3.n I6J5 Fig. 4.for electric communication purposes, and this isdefined as the natural logarithm for the relationbetween the voltage V,. re-transformed to a fixedcharacteristic impedance selected in the ValveTesting Set to 1M? = 600 ohms, and the grid voltageV 0 . Consequently we get the amplifyingcapacity176

orIf the resistance H 0 of the potentiometer R Bcorresponds to an amplifying capacity b = 0,we getwhere V 0 is the voltage drop at the resistanceR 0 . Obviously the relation3. Description of the Apparatus.Fig. 4 shows the external appearance of theValve Testing Set, and fig. 5 the arrangementsat the back of the panel. Fig. 6 is a circuit diagramof the Valve Testing Set, where the symbolsR, T, C, P,, P, and r have the same significanceas in fig. 1.The batteries B t , B 2 , and B :1 are connected tothe terminals V g „ V k , and V m and by the voltmeterV the grid voltage (with two ranges), thefilament voltage, and the anode voltage, may beread by inserting the plug B in the jacks A,, A„A., and A t respectively. The telephone receiveris connected to the terminals HTF.The primary winding taps of the transformerT represent the following anode resistances:(The anode impedance being practically pureohmic)will then be obtained, and consequentlyAlthough the anode resistance cannot alwaysbe made to correspond exactly to the internalresistance of the valve, the amplifying capacityindicated by the reading on the potentiometerP, will nevertheless be practicallyR 1656 Fig. 5.The scale setting of the potentiometer whenthe potentiometer resistance is R0 represents azero-point on the scale. Potentiometer readingsabove and below this zero point will then representnegative and positive amplifications respectively.as will be shown in the next chapter.In fig. 4 we see the voltmeter V to the left atthe top, the socket for the valve to be tested tothe right at the top, and in the centre the potentiometerP.,, the setting of which is read on amovable scale visible through a window. Underneaththe voltmeter V is the potentiometerPj by which the grid bias is regulated, andunderneath the valve socket is the dial for adjustingthe anode resistance. At the bottom wesee the terminals, and in the centre the plug B,which is lifted and plugged into the jacks abovewhen the D.C. voltages are measured.4. Theories for the Impedance adaption in theAnode Circuit.According to the above, only certain values,differentiated by multiples of 2. can be selectedfor the anode resistance. We will now examinethe significance of this limitation of the number177

Fig. 6.of possible anode resistances. If we regard thevalve as a source of current with an E.M.F. ofa V g , an internal resistance Ri, and loaded bythe anode resistance Ry as shown in fig. 7, thevoltage drop caused by the resistance R y will beThe Valve Testing Set will therefore measurean amplifying capacityorFig. 7.This voltage, re-transformed to a characteristicimpedance of 600 ohms, will beThe first term of this expression is obviouslythe actual amplifying capacity tested, and thesecond term represents the error in measurementoccasioned bv the internal resistance R t of the- 178

valve not being the same as the anode resistanceIty.When R y = /?, we haveBut, on the other hand, if R y ± ft,,As a function of R y ,value when R y = R,.We assumeb has thus a maximumwhere R and 2R are two /^-values which mayhe set on the scale, and introduce the symbolWhen using ft y = ft, and R y ~2R respectivelyfor measurements, we thus get the respectivevaluesft 2 —b v By computing the difference ft 2 —ft, fromthe scale values ft, and ft, observed, the amplifyingcapacity sought may thus be determinedwith the assistance of these curves, and the internalresistance of the valve may also be computedif the resistance ft is known.We see, however, that if the highest of thetwo values b 1 and ft, is accepted (in other words,the highest scale value obtained when testingwith different anode resistances) as being equalto the amplifying capacity b 0 , the error cannever exceed .015 neper, i. e. 1.5 per cent. Inpractice, an error of that description is usuallyof no importance. The determination of the internalvalve resistance cannot of course be madewith any very great precision, but then we arenot interested in the internal resistance of thevalve except in so far as this affects its effectiveamplifying capacity, and this influence is ratherinsignificant in the neighbourhood of the bestimpedance-adaptation.5. Amplification Tests with different anoderesistances.The above shows that the amplifying capacityof an amplifying valve, defined asOn the above assumption, ft, and ft, will consequentlybe the two largest scale values obtainedwhen using different anode resistancesfor the tests, and this offers an opportunity toascertain the two anode resistances for whichthe above assumption is valid.The equations (11) contain only two unknownquantities, viz. b 0 and ft,, which may obviouslybe computed from the values of b1 ft,, and ft.Fig. 8 gives curves computed from the equations(11) for the measurement errors b 0 — ft,, as wellas b0— b2, and the relation ft, : ft as functions ofand its internal resistance may be determinedwith the Valve Testing Set by making a seriesof tests with different anode resistances. In amore thorough investigation of a valve withthe Valve Testing Set, such a test series is thereforecompleted.Fig. 9 shows actual amplification tests ofsome different types of Marconi valves madewith different anode resistances. The roundrings mark the scale values observed, and thecrosses the internal resistances of the respectivevalves given by the maker.This shows distinctly what we have alreadypointed out, namely that the amplificationreaches a maximum value when the anode resistanceis equal to the internal resistance ofthe valve. We further note that the maximumamplifications recorded on the scale only differslightly from the amplifying capacities b 0 tobe determined. For a more exact determinationof the amplifying capacity b„ and internal re-— 179 —

Fig. 8.180

-10Fig. 9.sistance W, of a valve, the correction curves offig. 8 are used, in which case the symbols b v b.,etc. refer to the plotting in fig. 8 of b as a functionof Ry,6. Amplification Tests with different Grid Biases.To determine by means of the Valve TestingSet the grid voltage space of a valve and themagnitude of the non-linear distortions arisingin the valve, a series of amplification tests aremade with different grid biases. The fulldrawncurve in fig. 10 gives the result of such a testscries on a Marconi valve. We note that theamplification increases with decreased grid bias,for the reason that the slope of the valve characteristicis increased.In this connexion we will comment upon acircumstance which may he of some interest.When the amplification of an amplifier is mea-— 181 —

Fig. 10,sured in the usual manner with an audio-frequencygenerator, artificial line, and valve voltmeter,the valve is made to work during the testwith an audio-frequency voltage of a certainamplitude on the grid. The amplification arrivedat is then determined hy a kind of averagevalue for the slope of the valve characteristicwithin the grid voltage range utilized by theaudio-frequency voltages. When measuring theamplification with the Valve Testing Set however,the conditions are different. If the reactionis slowly increased until the amount isleached when the valve hegins to oscillate, thisamount of reaction must be characteristic of thecondition of the valve before oscillation had begun.The amplification measured will thereforebe determined by the slope of the valve characteristicat that point of the curve which correspondsto the grid bias selected.We will introduce the designation "point amplification"for the amplification measured bymeans of the Valve Testing Set, as this amplificationcharacterizes conditions in one particularpoint of the valve characteristic.The point amplification must obviously be independentof the grid bias within the range ofgrid voltages employed if the amplification obtainedis to be free from distortion or, in otherwords, within the said range of grid voltages thetulldrawn curve b = f(V 0 ) of fig. 10 must bea straight line parallel to the axis of abscissas.If this is not so, the amplifier will introduce— 182 —

non-linear distortion, the magnitude of whichwill depend on the amount of deflection of thecurve b = f(V g0 ) from the mean value of thepoint amplifications within the range of gridvoltages employed. The dotted curve of fig. 10shows the mean value of point amplificationswithin ranges of grid voltages distributed equallyon either side of the —10-volt grid bias insuch a way that the average amplification of arange of grid voltages will correspond to theordinate at the limit voltages of that range. Ifwe consider that when amplifying audiable currents,the range of grid voltages nearest to thegrid bias will be used the most, the dotted curveshows that the point amplification of the gridbias in question may also be regarded as ameasure of the average amplification. The advantageof defining the amplifying capacity ofa valve as point amplification for the grid biasin question is that the amplification capacitywill be explicitly defined. Deviations in thepoint amplification at other grid biases whencompared to the grid bias employed will thenbe regarded as sources of distortion.It must be pointed out that when measuringamplifications by the usual method previouslymentioned, the amplifying capacity ascertainedwill partly depend on the properties of the indicatorappliance used (e. g. a valve voltmeter).As a matter of fact, different types of indicatorappliances will treat the disturbing voltagescaused by distortion in the amplifier in varyingways. Valve voltmeters, based on rectificationof only positive or negative half-cycles of themeasured voltages, do not usually give the sametest result when the negative and when the positivehalf-ccyles are rectified, and the questionthen immediately arises which is to be consideredcorrect. There is a great risk of getting the6. 183Fig. 11.measurement results all wrong with these amplificationtest devices, if from one cause or anotherthe valve is overloaded, which risk is nonexistentwhen the Valve Testing Set is used.7. The Effective Amplification of an Amplifier.Fig. 1 shows an amplifier connected withoutreflection losses on the input and output side tolines with the characteristic impedances Z, and/., respectively. The grid transformer, whichhas a ratio of 1 :n, introduces into the device anamplification ofwhere ti g represents an effective attenuation inthe transformer. If Z 1 is = 600 ohms, we getIf the attenuation in the anode transformer is£ a , and the amplifying capacity of the valvethe amplifier will bethe effective amplification ofIf the amplifier consists of several valves connectedin cascade, the amplifying capacity ofeach valve will enter as a term of the effectiveamplification of the amplifier, and each couplingbetween the valves will be represented byits own term. If all the terms entering in theeffective amplification are added together, exceptthe amplifying capacity of the valves, anamplification exponent, characteristic of theamplifying device, is obtained.It this amplification exponent be determinedonce and for all. the effective amplification ofthe amplifier can be determined by adding tothis the amplifying capacities of the severalvalves, measured by the Valve Testing Set.

The characteristic impedance 600 ohms whichenters into the definition of the amplifying capacityof a valve, has been chosen because thatcharacteristic impedance is frequently used intelephone technics, particularly for measurements.8. Valve Testing.The Valve Testing Set is primarily intendedfor routine tests of valves. The anode resistanceis then adjusted in as close correspondenceas possible to the specified internal resistanceof the valve. It is of course also possible to producea valve tester for certain types of valves,so designed that anode resistances identical withthe internal resistances of the valve types inquestion may be employed. The valves are testedwith normal filament voltage, anode voltage,and grid bias, and the amplifying capacity ispassed as sufficient if it reaches a certain specifiedvalue. Tests should also be made withFig. 12.— 184 —say 75 per cent higher and lower grid biases,when the valve will be passed if the amplificationsmeasured do not deviate from those atnormal grid bias by more than a certain specifiedamount.If, for some particular calculations, it is desiredto graduate the reaction potentiometer scaleto any other amplification units thanthis can of course easily be done. The reactionpotentiometer may for instance be graded inwhere S is the slope, and these quantities maythen be directly measured.Fig. 12 illustrates the relation between the expressionsfor the amplificationand

On the Calculation of Delays in an Automatic Telephone System.By Stig Ekelof.When in an automatic telephone system it isa question of calculating the number ofswitches required for carrying a certain amountof traffic, we as a rule start from the conditionof allowing a certain "loss", in other words,we estimate the number of switches in such away that, for instance, 2 %o of the calls are"lost" through all switches being engaged. Inthis case it is necessary that all subscribers whodo not at once find a disengaged switch, immediatelyreplace their receiver and do not renewtheir call until a short time has elapsed. As abasis for calculating on these lines serve, byway of example, the well-known lossformulaeof Erlang.Sometimes it may, however, be of interest tocalculate with u delays" instead. We premisethen that a subscriber who cannot get throughimmediately waits with the receiver off, untilhe can get a free line finder. In this case wewant to be able to calculate the total numberof calls which are delayed or the number delayedmore than a certain period.In the ensuing pages we shall give an accountof the calculation of accompanying charts fordelay calculation worked out by the author.The foundation for these consists of certain delayformulaededuced by Prof. Pleijel. It should bepointed out that Prof. Pleijel in his deductionalso took into consideration the otherwise generallyneglected fact, that the number of callsdiminishes in the proportion in which the subscribersare engaged.Deduction of Prof. Pleijels Formulae.We assume a group of l subscribers who fortheir calls have available x linefinders. Asregards the registers it is assumed that thedevices and arrangements are such that evervline-finder always has at its disposition a disengagedregister.We now consider a certain time interval bthe "busy hour", and assume that every subscriberduring the same makes e calls. Everycall is premissed to occupy the linefinder duringthe constant time interval a.The system is, furthermore, imagined to beof such a nature that a subscriber who callsat a moment when all Iinefinders are engaged,will wait until a linefinder is free. If thereare several subscribers waiting simultaneously,we assume that they will get a linefinder inthe same order in which they have called.Our first problem then will be to investigatehow often it happens that m subscribers are engagedby reason of these calls with concomitantconversations.For solving the same we make use or Erlang'smethod of "statistical equilibrium*. We considera very large number P of groups of / subscribersaccording to the preceding. At a certain momentof time we have different conditions inthe different groups with regard to the numberof subscribers who are engaged. The entiresystem should nevertheless be in a statisticalequilibrium, so that the number of groups P m ,which have m subscribers engaged (we say:the number of groups of the order m) shouldbe equal at every moment. The condition forsuch an equilibrium to exist is obviously thatduring every time element as many groups ofthe order m pass into such of order m + 1 byadditional calls as the number of groups whichpass from order m + 1 to m on account offinished calls.We introduce "the traffic intensity* y, bywhich is meant the number of calls which onan average originate during the time of a conversationa.We denote further by r the number of callswhich are made during the time interval a in185

a group P 0 , i. e. in a group where no subscribersare engaged. If we take a as time unit,the number of calls in such a group duringthe time interval dt becomes equal to zdt andin all groups P 0 we obtain altogether P 0 zdtcalls.In a group of order m we have only /— msubscribers who are able to make calls. Thetotal number of calls within these P m groupsduring dt is then obviouslyorWe now insert the probability S,„ of havingat a certain arbitrary moment m subscribersengaged in conversation.ObviouslyIt is here assumed that none of the / subscribersare engaged by incoming calls. Inreality we must as a rule premiss that the incomingtraffic is equally large as the outgoing,so that on an average y subscribers are constantlyengaged by calls from other subscribers.Consequently, on an average at the mostl—y subscribers are able to make calls, i. e.we kan include the effect of arriving conversationsby assuming that the group has only/ — y subscribers. The difference for large I is,however, very slight, for which reason, in orderto make the numerical calculations more convenient,we ignore this correction. Should itbe found in a certain case that this is not permissible,we have only everywhere in the formulaeto substitute l—y for l.We now select the time interval dt so smallthat we are able to ignore all those cases wheretwo or more calls occur or get lost in a groupduring this period. The expression P„,—-then gives us the number of groups which onaccount of new calls are passing from ordern to m + 1.For calculating the number of groups thatpass from order m + 1 to order m we assumefirst that in + 1 < I , so that all engaged subscribersarc talking. The IB + 1 conversationsgoing on will clearly all be finished after theinterval 1. If we assume that the conversationsfinish continously, (m + 1) dt are concluded inthe period dt, so that the number concludedwithin all P m t groups iszdtand consequentlyfrom whichwheredenominates the binomial coefficientFor m = x we get speciallyWe assume now instead m > x and putm = x + r. All linefinders are then engaged,so that x conversations are on while r subscribersare waiting. In the time interval dt thenxdt conversations disappear. The condition forequilibrium becomes in that caseor if we here put the probabilityfrom whichi. e.equilibrium consequently de­The statisticalmandsfor R1 = S1186

By the use of the previously deduced expressionfor S y we then geti. e.If we also here substitute S m for Rm we getIf we employ the notationwhether m'—x,we can thus writeandorFor determining S 0 we make use of the factthat the sum of the probabilities for all possiblecases is equal to 1. The highest possible numberof engaged subscribers in a group is clearly/. We therefore get the relationas well asIf these values are inserted for r and S„ inthe expressions (1) and (2) for S m and H m , weobtain ultimatelyIn order to enable the system to operate properly,the number of delayed subscribers must,as we are aware, be very small, so that 1'H mis very small in proportion to 1. We are thenable without serious error to substitute S m forR m in the above relation and then getandOur next problem will be to express r in y.Because y is the number of calls which on anaverage fall during the time interval 1 withina group, the number of calls in this period inall P groups together becomes P . y, so thatWe now pass on to the actual problem, viz. thecalculation of how many calls hare to wait forlinefinders longer than a certain period of time.A call originating in a group with m < x engagedsubscribers will get a linefinder at once.If a call comes in a group where x subscribersare engaged it will on an average have to waitfor the period — = the period which on anaverage elapses before a linefinder becomes idle,reckoned from a certain moment when all areengaged. If x + 1 subscribers are engaged, thecall will be delayedand if generallyx + r — 1 subscribers are engaged the delaywill beThe number of calls for the time interval

The total number of calls in the period dtis, however, Pffdt, so that the number of calls2R— 1which are delayed for a time— , calculatedin proportion to the entire number of callsbecomesThe number of calls delayed for a timethus isFor r 0 =1 we get especially the totalof delayed subscribersnumberIt is also interesting to know the average delayT per call. The number of calls in all groupswith r — 1 delayed subscribers was xP x + rdt forthe period dt. Each one of these calls has towait for the time so that the totaldelay isIf we sum up for all groups with r> 1 anddivide by the total number of calls Pydt, wegetWe uow introduce thenotationsand then haveFormulae for Numerical Computation.We pass on to deducing formulae suitablefor numerical computation. It obviously dependsupon summing up expressions of the formwhereandAccording lo the precedingThe order of magnitude of a. p and X is188

We, therefore, make an extremely slight errorif instead of the upper limit l— x in the sumswe put :x.For computing D and D l we shallthese sums in powers of a p.expandIt is then appropriate to introduce the notationit l y ~ l = the sum of all possible productsof v of the numbers 1, 2 ... s— 1.On inserting the expressions for ,T,» 'we seelhat the other a and b will be composed ofsums of the formWe shall, therefore, first calculate n n for somedifferent values of n.We getWe have thenwill have to satisfy te relationhenceFor D and D t we now getwherewithWe now getas well aswe immediately calculateto189

withForconsequentlyIf we introduce thenotationas before, we will havewe getAs a rule it is therefore sufficient to includeonly the first terms in the parentheses and putis here independent of ;/.Spec, we getFor high values of r (busy traffic) x grows,however, and may in such cases which arepossible rise to approximately 0,8. In suchcase the correction of the neglected terms becomesmuch greater, so that we should gel toosmall values for D and D t .This is, however, balanced by the formulafor R m at high z giving too large values. Forwith a busy traffic our H m are comparativelylarge and by substituting H m by S m in the expressionSo + S1 . . . + S x + R x +1 + .... + Ri = 1we get too great a value for S„ and consequentlyfor Rm.We therefore calculate withIf we have properly calculated a certain N r0according to the preceding formulae, it will beeasy to extract from this value the N r for otherr-values.First N R0 + 1 can be computed fromwhereis smallThusin proportion toFurthermorewhereWe now begin by computing X r , i. e.thenumber of calls delayed for a periodin proportion to the tolal number of calls.The equations (5) and (9'') give usand analogously190

generally thusExample: We supposel = 500 subscribersx = 45 linefindersIn this way the accompanying charts havebeen computed.Abscissa is « = 2r— 1, as ordinates have beenput N 1 N2, N 3 etc. for u = 1, 3, 5 etc.Approximation Formulae.We easily find simple approximation formulaefor the number of delayed calls S and theaverage delay T.According to the equations (5 1 ) and (9")We firsl calculatewhich is independent of y.FurthermoreThe equations (6) (9') and (9 11 )givefurthermorethusThe average delay per delayed call finally iswas, as we know (see page 157) the timewhich a call was delayed on an average if itwas the only one delayed. With this intervalas unit we get consequently191

Fig. 1. Delays at free selection of 20 linefinilers over a multiple of 500 subscribers.Example: Let us suppose a group of 500 subscribers with x = 20 linefinders. .M= 700 SM and the duration of conversationa = 2 minutes. We want to know the number of calls delayed > 6 seconds, F'irst we calculate e fromThe curve for 700 SM then gives us tor this ((-value P = 13 ‰

Fig. 2. Delays at free selection of 25 linefinders over a multiple of 500 subscribers.193

Fig. 3. Delays at free selection of 30 linefindera over a multiple of 500 subscribers.194

Fig. 4. Delays at free selection of 35 linefinders over a multiple of 500 subscribers.195

Fig. 5. Delays at free selection of 40 linefinders over a multiple of 500 subscribers.196

Fig. 6. Delays at free selection of 45 linefinders197over a multiple of 500 subscribers.

Fig. 7. Delays at free selection of 50 linefinders over a multiple of 500 subscribers.198

Fig. 8. Delays at free selection of 55 linefindersover a multiple of 500 subscribers.199

Fig. 9. Delays at free selection of 60 linefinders over a multiple of 500 subscribers.200

Working Reliability and Maintenance of the L. M. EricssonAutomatic Telephone System.By Anders Lignell, Director of Telephones,Stockholm.By working reliability of a telephone systemwe understand the capacity of the systemto function correctly if properly used by thesubscriber.In the manual system, this reliability dependsnot only on the correct functioning of the technicalplant, but also — and not least — on theefficiency of the operators, while in the whollyautomatic plant the personal operator is eliminatedand the reliability depends entirely onto what extent the technical devices can betrusted to perform their various functions.Continuous reliability control is highly importantfor satisfactory service, assuming thatthe control is arranged to test the whole systemand that the examination includes a sufficientnumber of calls to make the control results representativeof the whole traffic.If that is the case, and if the control is appliedto actual traffic and not to test calls madeby the staff, a clear idea of the real quality ofthe service is obtained, as the latter method forseveral reasons will not give the desired completeand reliable information regarding thetraffic. Further, provision should be made forfaulty connexions to be locked during the controland the faults traced and remedied. Thecontrol will then simultaneously act as a generalsearch for and removal of faults in the plant.In an automatic system, where a remainingfault in any organ may disturb a great portionof the traffic, the importance of this is obvious.The efficiency of an automatic system is frequentlyjudged by the number of repaired faultsrelatively to the number of subscribers or calls.But the fact that a comparatively small numberof faults are discovered and put right is noguarantee that no faults remain in the system,nor that the service is satisfactory. The onlymeans of ascertaining this is by continuous reliabilitycontrol, which will give an actual pictureof the service efficiency. The traffic controlpanels in Stockholm are therefore arranged:partly for continuous reliability control,partly „ tracing faults and assisting thesubscribers (when any directoris occupied too long, or whenthere is a fault caused by thesubscriber or by the equipment,a red lamp is lighted),and partly „ individual control, i. e. controlof all traffic on a certain subscriber'sline.The accompanying Table I gives the result ofcontinuous reliability control in 1929 of our oldestautomatic exchange "Norra Vasa" — a10 000 line exchange with 7 300 subscribers —which at the end of 1929 had been working for6 years.The "Total" column shows that out of 71 629controlled calls from subscribers — an averageof 5 969 month, evenly distributed between alltraffic routes —93.53 per cent, have been faultless,3.28 „ „ of the faults have been causedby the subscriber,0.02 „ „ „ „ „ have been causedby the operator ina manual exchange,0.17 „ „ „ „ „ have been causedby the technicalequipment (subscribers'stationsor lines, or exchangeequipment).Out of the faults caused by the technicalequipment — 0.17 per cent — the origin of 57,or 0.08 per cent., have been located; 40 of these.— 201 —

TABLE I.or 0.05 per cent., have been located in the automaticsystem, and 17, or 0.03 per cent., in otherexchanges, direct lines, or subscribers' stations.If we assume that the same percentage of the64 not located causes of faults can be referredto equipment outside the automatic plant, whichis a safe assumption, as it is considerably moredifficult to locate faults outside the exchangethan inside it, only 0.107 per cent, of the causesof faults can be blamed on the automatic equipment.The Table also shows very nearly the samepercentages during each of the several months,which proves that the reliability has been fairlyconstant throughout the year.The average of the faults committed by subscribersduring the year is 3.28 per cent.This high percentage is explained by the factthat the change to the automatic system in thewhole telephone area has not yet proceeded farenough for the majority of the subscribers tobecome familiar with the proper way of dialling,and by faulty memorizing by the calling partywhen translating the manual numbers.— 202 —

TABLEII.The results of the control during January andFebruary 1930 are shown in Table II below.It will be noted out of 9 825 controlled calls— 4 900 per month —97.40 per cent, have been faultless,2.50 „ „ of the faults have been causedby the subscriber,0.10 „ „ „ „ „ have been causedby the technicalequipment.Out of the faults caused by the technical equipment— 0.10 per cent., of 9 825 controlled calls— the causes in 5 instances — 0.05 per cent. —have been located in the automatic system, in1 instance — 0.01 per cent. — in a manual exchange,and in 4 instances — 0.04 per cent. —time has been too short for the cause to be located.A maximum of 0.09 per cent of the faultscan thus be charged to the automatic system,but this figure is too high, as some of the 4 unlocatedfaults certainly should be referred tocauses outside the automatic plant. Losses frominsufficient exchange facilities do not occur inthe L. M. Ericsson system, as any such lack willonly prolong the time for obtaining a connexion.The control may therefore be said to havegiven the automatic system an excellent characterfor reliability.Faults caused by the calling party are alsoreduced from 3.28 to 2.50 per cent.The results of the control designed to assistsubscribers and simultaneously to trace faults ifrequired are shown in the attached Table III.The red lamp is lit immediately if the callerdials a combination which is not on the selectorregister, and after 24 seconds if the director isheld too long (handset removed from the cradlerest), if an insufficient number of figures aredialled, or if there is a technical fault.During the year 19 736 calls on red lampshave been examined. In 19 361 of these, or 98.1per cent., the fault has been the subscriber's, whohas then received assistance and requisite advice.Out of the 375 remaining instances, or 1.9per cent, of these red lamp signals,149 have been faults in own exchange,31 „ „ „ „ other exchanges,23 „ „ „ „ subscribers' lines, and20 „ „ „ „ subscribers' stations, atotal of 223 faultswhich have been traced and remedied, whiletime has not allowed 152 of the faults to betraced. 189 causes of faults in the automaticsystem (40 by continuous control and 149through red lamp signals) have thus been removedby the control, which is 35 per cent, ofall faults remedied in the automatic system duringthe year. The importance of this control forthe working reliability of the exchange is obvious.Regarding maintenance costs, it might be mentionedthat in the Kungsholmen automatic Exchange,opened in June 1928 and equipped for15 000 subscribers with an average of 10 640— 203 -

TABLE III.connected lines during 1929, the year's maintenancecosts have been Kr. 4:02 per subscriber.This cost includes all work and material,cleaning of premises and equipment, also cost ofnight attendance as well as staff holidays andsick leave. The maintenance staff has consistedof nine male repairers, 6 of whom are detailedto the selector rooms and 3 to the cross-connexionfield and the faults department, and two femaleassistants.The average traffic is 6 calls per day andsubscriber, and 0.7 calls per subscriber in TheBusy Hour.The third kind of control — individual control— is used for faultfinding subscribers, and isessential to facilitate the straightening out ofcomplaints regarding recording of calls or theservice generally, which cannot be satisfactorilychecked in any other way.April 15th 1930.— 204 —

Lead Covered Rubber Cable Installations.By EinarThe jubilee celebrated in 1929 in memory ofthe invention by Edison of the incandescentelectric lamp was at the same time a jubilee ofthe popular use of electricity. It is true thatelectricity had been utilized as a source of lightquite a number of years before, but it was notuntil the invention of the handy little Edisonglow lamp that electric light became a domesticsource of light. By this invention electricity becameby and by the property of everybody whichat present we can consider it to be.During the first decades of this developmentelectricity was, however, to costly a merchandisefor the ordinary man. It is true that electicitygeneration plants were erected in a great numberof cities, but owing to the high costs onlyprominent dwelling-houses, hotels and publicbuildings were provided with electric light.For industrial purposes electric light came intouse at rather an early date, since it had beenfound out that the output of labour increasedwith improved illumination.In the solid dwelling-houses and other technicallycomparable places where the first installationsof electric light were introduced, theinstallation conditions were rather favourableand in addition the voltage applied amounted asa rule only to 110 volts. The strain on the installationmaterial was therefore unimportantand as a natural consequence hereof the materialwas, in many respects, of rather a frail construction.It took so long a time before electric light hadbecome in a proper sense generally spread that atradition, rather deeply rooted, as to the constructionof electrical installations had had timeto develop before that. Unfortunately, this traditionwas not based on profound studies of theproblem but was sooner the result of experiencegained in the working places. The manufacturersthemselves did not pay much attention toOlsson.the installation problem and the specific installationmaterial was constructed chiefly from thepoint of view of cheapness. This point of viewbecame more and more prominent in the courseof time, and a considerable portion of the installationmaterial sold during the years 1910—1920was appreciably inferior to the material in themarket during the previous ten years.When electricity came into use in industrialplaces it became necessary to undertake a numberof reconstructions and it was now that theheavy armoured conduit material and the singlecoreconductor installations were wrought out.However, both these systems suffered from certaindeficiencies, but as long as electricity wasused chiefly in thickly settled communities or inindustries having installers of their own, thedifficulties due to unsuitable construction of theinstallation material were fairly well overcome.That no vigourous measures were taken, may beexplained by the fact that electricity was so newthat those busy in this sphere knew nothing elsebut that it should be troublesome. Besides, theintroduction of electric light always involved sogreat an improvement that the difficulties encounterednow and then were not considered tobe of any great importance.The difficulties began in earnest when, withthe invention of the three phase alternating current,it became possible to distribute electricityalso in rural areas. To this contributed aswell that the local conditions in the offices of afarm are of such a kind as to expose the installationmaterial to very hard stress of chemical,electrical and mechanical nature.In Sweden the electrification fever in therural districts during the time of the world-wardisclosed in sharp colours illuminated by firesand marked by deaths and accidents the defectsof the materials used, and forced on the reconstructionwork which finally resulted in the lead— 205 -

Light installation with lead covered rubber cable material incow-stables.covered rubber cable system coming every davmore and more into use.It was in the first place the fire-insurancecompanies who drew the attention of the manufacturersand other technicians to the subject,and an intensive work commenced with aview to repair the deficiencies. Now the installationproblem was studied in detailfor the first time, and soon enough itbecame evident that the accidents occurredcould not very well be regardedas the result of the concurrenceof a number of unlucky circumstancesbut that rather a combination oflucky circumstances was required inorder that no accidents should happen.In other words, the installationmaterial used was unsuitable andthe installation methods did not givesafe results.If we make a short technical reviewof the way in which older installationsystems, single wire conductorson porcelain insulators andarmoured conduit installations, actagainst stress of different nature, we obtain theproofs of the unsuitability of these systems.It is perfectly clear that the mechanicalstrength of an installation with thin single conductorson porcelain insulators is very imperfect.The insulators situated at a distancefrom one another of about one meter give a badhold to the wires which are easily torn away.The insulators themselves are easily brokenwhich also applies to the fittings and the switches.From an electrical point of view thestrength of a plant is measured by its insulation,that is, its resistance to earth, and experiencehas shown that this insulation is, in the case ofsingle conductor installations, to a high degreedepending on the humidity in the place. If theinsulation measurements are executed in wintertimewhen the localities are warm and damp itis often impossible to obtain any measurablevalue by the aid of an ordinary insulation measuringinstrument. One is therefore quite justifiedin saying that in the case of single conductorinstallations a more or less strong leakage ofelectricity always takes place in damp places.This circumstance is also reflected in the prescriptions,where no definite value of insulationis stipulated for single conductor installations.It is, above all, the joints insulated by handwhich cause the bad insulation. Such a jointinsulated with rubber and insulation tape excludesthe moisture as long as the tapes are new,but as soon as the rubber and the tapes havedried thev absorbe moisture. Since the wholeHusum sulphate mill, entirely equipped with lead covered rubber cable.— 206 —

Husum sulphate mill, hall of paper machines.braiding of the conductor is generally soakedwith moisture the outer part of the conductorbecomes alive, electricity leaks out and there isdanger of life present.Most of those who have been working in placeswhere the electrical installations are executed bymeans of single-core conductors know wellenough how easy it is to get a "shock" from theconductors.It is those leak currents which make a singleconductor plant so little resistant from a chemicalpoint of view. As an example may be mentioneda single conductor installation mountedon the damp hay-loft above a horse-stable. Thisinstallation had been in service onlya few months, but during this periodthe wires had worn off a couple oftimes, and from this reason been soheated that they had broken. Glowingparticles had fallen down intothe hay stored on the loft. The heavymoisture on the loft had come fromsome ventilation ducts leading fromthe stable to the loft. Thanks to thismoisture the conductors were destroyedbut at the same time the haybecame so damp that it was not seton fire by the glowing pieces of conductorsfalling down into it.It is to the electrolysis always occurringin damp places that the destructionof material is due. Electrolysisalways presents itself where twometals of different nature are connectedwith one another by means ofa fluid acting as electrolyte. This isgenerally the case with all the waterof condensation in a damp locality,which absorbes or dissolves in it anumber of corrosive gasses.From what has been said above weobtain the following conditions forinstallations in damp places; firstly,that all live parts should be hermeticallyenclosed in an effective manner,secondly, that all combination of differentmetals should be avoided or,where this is not possible, that themetals should be well insulated fromone another.An armoured conduit installationis mechanically strong as far as concerns theconductors placed in the conduit. On the otherhand the fittings are more fragile. This obtainsespecially with regard to rigid tube pendants.The ordinary armoured conduit consists as arule of rather thin stuff and this applies alsoto the conduit boxes. A piece of tube screwedinto the cover of a box and thus rigidly fixedis easily broken even in the case of a relativelysmall stress at the lower end of the tube wherethe fittings are.From electrical standpoint an armoured conduitinstallation is solid as long as the temperatureis the same all over the various parts ofLead covered rubber cable installation in cow-stables.207

the tube system. Since this, however, is neverthe case the electric strength is generally ratherproblematic. The prescriptions require an insulationof at least 220.000 ohms between anytwo successive fuses or for each group of lamps,and as long as the installation does really showthis insulation it is also reliable. But ratherlittle is required to lower the insulation so muchthat danger arises. The worst enemy of tubeinstallations is condensation of moisture withinthe tubes. There are still people who believethat an armoured tube installation is a closedsystem, and it is possible to so construct an installationof this kind that it becomes hermeticallyclosed. This is done by sectionalizing thesystem into a number of small parts by meansof filling in compound and by making tight allthe tube joints by the aid of tow prepared withread lead. However, if only a rubber packingin a conduit box. is dried, or the sealing of aswitch becomes untight, this is enough to givethe external air free access to the tube system.As a rule the external air is supposed to haveaccess to the tubes and it is endeavoured to reduceby drainage the risks thereby involved.Condensation of moisture in the tubes alwaystakes place when the tube passes from a hotroom to a colder one. As soon as the humidityof the air exceeds the point of saturation correspondingto the existing temperature, part ofit comes out in the form of water of condensation.This water of condensation attacks theinsulation of the conductors and makes thebraiding rot away on the wires and the tubesrust internally. Current leakage takes place aswell at joints dried through in the junction boxesas at points where the insulation is damaged.Most of the fires caused by electricity are dueto defective tube installations.The result obtained from researches as to thenumerous accidents in the rural areas was disheartening,and it became clear to all partiesconcerned that the installation problem ought tobe considered in full if a lasting result was to beexpected.Upon the application of a number of personsinterested in this question the Sieverts Kabelverktook up the problem in 1922. Quite naturallyit was in the first place the conductors whichwere made the object of the interest of the factory,and by utilizing the experience gainedFig. 1. Lead covered rubber cables:Bare cableUnarmoured cableArmoured cablefrom the manufacture of lead cables, the constructionof lead covered rubber cables shownin tigure 1 came out as the result of a numberof experiments. This cable consists of tinnedcopper conductors insulated by means of vulcanizedrubber. In the case of multi-core cablestwo or more such conductors are twisted togetherand further insulated with vulcanizedrubber so that a circular cross section is obtained.A lead sheath is pressed around therubber and the bare lead covered rubber cableshown at the top of figure 1 is finished. Conductorsof this kind with uncovered metal sheathmay only be utilized in places where they areprotected from both chemical and mechanicalinjurs'. The lead sheath of the cable shown inthe middle of figure 1 is protected from chemicalaction by means of a serving of asphalt compoundover which are wrapped several carefullyimpregnated cellulose tapes. Outermost is placeda braiding, likewise impregnated. Conductorsof this construction can be installed anywherewhere they are not exposed to serious mechanicalinjury. The third cable shown in figure 1is the armoured lead covered rubber cable thathas been most in use up to now. It has proovedto bear exceedingly well any kind of strain ofmechanical as well as of chemical and electricalnature. The armouring consists of two leadplatedand asphalted iron tapes.The outer braiding of the lead covered rubbercable is impregnated either with a kind of asphaltcompound or with red lead and linseedoil.The first-mentioned impregnation is moreresistant from the chemical point of view and itshould be employed in all places where it is notdesired to have the cable painted in other thanblack colour. But where that is the case the redlead impregnation is used. Places in which the- 208 —

ed lead impregnated cable is not resistant arethose where the reaction of the condensed humidityis basic. For linseed-oil is an organicgrease which is easily changed into soap andwashed away when attacked by basic solutions.As an example of places where the red lead impregnationshould not be employed may be mentionedcattle-stables of any kind and laundries.To secure a reliable installation it is notsufficient to possess excellent conducting materialsbut it is also necessary that fittings andswitching devices should be just as reliable. Tosolve this problem was thus the next step in thework. As mentioned before the investigation ofthe older installation systems had clearlv mani-Fig. 2. Complete lamp fitting with switch and shade.fested that in order to obtain a resistant installationin damp places all live parts ought to behermetically enclosed. This was also made acondition not to be dispensed with when workingout the constructions. Simplicity of constructionand resistance against all kinds ofstrains were also requirements which it was endeavouredto fulfil as far as possible. The resultis indicated in figure 2 showing a complete lampfitting with conduit box, switch and lamp holderwith protective globe and shade.The junction box is made of lead-plated ironplate, japanned cast iron, or bakelite. In themetal boxes there is a connecting block of bakelitecarrying two, three or four connectionsockets. The connections are performed byloopingthe conductors around the sockets andfixing them with nuts. The switch is connectedto the junction box by plugging three contactplugs on its upper side into the sockets of theconnecting block.The switch is a press button switch manoeuvredby means of an external and an internallever arm, both joined to a shaft placed in aspecial bearing. The cover of the switch and itsinner conducting parts are made entirely ofbakelite. On the off side of the switch, countedfrom the junction box, there are two sockets forthe connection of the lamp holder.To the external manoeuvre arm of the switchis fixed a strap equipped at its other end with apulling ring. This pulling strap can if necessarybe led through a guide pulley or through glassrings to the place from where it is desired tohandle the switch. It is, however, preferable tohave the plant arranged in such a way that thisprocedure is not necessary.The lamp holder is also of bakelite, and bymeansof contact plugs it is either, like theswitch, directly connected to contact sockets inthe junction box or to a switch fixed to thejunction box. The lamp socket is provided withnormal Edison thread but this thread is not conducting.Contact with the lamp is effected bymeansof a spring contact and of an elastic ring.Contact is not made until the lamp is almost entirelyscrewed into the socket. Neither the lampholder nor the switch are provided with anyscrewingdevices, but are both held fast to thejunction box by means of screws from theholding ring fixing the glass globe to the lampholder.Between the different parts of the fittingsare inserted packings of first-rate rubber.Thanks to the broad and even surfaces betweenthe various parts the stress on these packings isnot hard and their life-time is therefore considerable.The holding ring also holds in place the shadesbelonging to the fittings. If necessary a protectiveguard may be screwed to the holding ring.The protective globe consists either of ordinarytransparent glass or of glass with opalescentcasing. The former construction is usedin rooms where no very accurate work is to becarried out and the latter in places where thelamp glare may be dangerous or diminish thelabour output.The reflectors of the fittings are of two kinds.One shades the light so as to produce a lightcone with a top angle of 120° whereas the other— 209 -

Fig. 3. Connector with rubber-lead packing.one gives a light cone with a top angle of 180°,that is, the light is shaded in such a way thatthe upper limiting surface becomes a plane goingthrough the luminous body of the lamp. Thereflectors are coded according to the top angleof the light and are called 120° and 180° reflectorsrespectively. The first-mentioned one ismore economic and should be preferred in roomsthe height of which exceeds 3.5 meters.The most important detail of the constructionis the jointing at the point where the conductorenters the junction box, and much constructionand experimental work has been devoted to thisdetail, a work that finally resulted in the leadcovered rubber joint shown in figure 3. It consistsof a rubber bushing both ends of which arecovered with lead. Fig. 3 shows how the closingup of the joint is carried out. When the packingis compressed a metallic connection between thelead sheath and the junction boxes is obtainedat the same time as the joint is tightened. TheFig. 4. Junction box with three outlets.lead sleeving on the rubber packing is hard pressedto both the lead sheath and the metal socketof the box. At the same time as the lead groundsthe metal sheath of the conductor it protects theFig. 5. Complete lamp fitting without switch but with 180° reflector.rubber from attacks of the air, which prevents,if not entirely, yet to a high degree, the rubberfrom aging. Experience gained from deliveredinstallations has shown that the rubber inpackings being in service since more than 6 yearsis still just as soft and elastic as it was wheninstalled.The parts described above, viz. junction box,switch and lamp holder with protecting guardand reflectors can be combined in many differentways.If a box is equipped with a cover as indicatedin figure 4 we obtain a connection- or junctionbox. There are junction boxes with one, two,three, four or five outlets, and of connectionblocks there are threekinds, viz. blockswith two, three andfour connecting sockets.By combiningthese two parts wecan obtain 15 differentjunction boxes.If in addition thepackings are alsovaried an almost infinitenumber of combinationsis obtainedwhich should be ableto satisfy almost anyimaginable requirements.In the boxesone-, two-, three-, andfour core conductorsup to 6 mm 2 can beFig. 6. Pendant with leadcovered rubber conductor.210

Fig. 7. Connecting cover for pendant with lead covered rubberconductors.connected. For the connecting in of conductorsthicker than 4X6 mm 2 other boxes with biggerconnectors are used.To each one of the abovementioned types ofjunction boxes can be screwed a lamp holderwith or without reflector. In figure 5 is showna combination of junction box, lamp holder withprotective glass and 180° reflector. By fittingthe switch in between, further combinations areobtained.To complete the system a pendant constructionhas been worked out. This pendant shown infigure 6 consists at the top of an ordinaryjunction box covered by a pendant cover of theconstruction indicated in figure 7. To this coverbelongs a connector differring from the onedescribed in the preceeding only by its somewhatgreater length. In the pendant cover there isfurther a bakelite block carrying connecting pinsin the case of the upper cover, and connectingsockets in the case of the lower cover, which isotherwise almost identical with the upper one.The pendant cable is armoured with iron wireand from figure 7 is seen how the connection isbrought about. The armouring wires are bentaround a ring which is placed between thepacking and the screwing sleeve with a washeron each side. When the screwing sleeve is pulledtight the mechanical fastening of the armouringwires is brought about as well as the closing upof the joint and the grounding. At the bottomis fitted a lamp holder which can be equippedwith reflectors or protective guards as wanted. Ifa switch is to be provided for such a pendant itshould be mounted between the upper pendantcover and the junction box.Pendants always have some advantages overrigidly mounted fittings. So, for instance, thelamps last longer because they are less exposedFig. 8. Junction box for connecting lead covered rubber cable tosingle core conductor.to vibrations. In addition, when using pendantsit is easier to place the source of light at thepoint where it gives the best effect. The mostimportant advantage is, however, that this kindof fitting is not by far so much exposed to chemicalaction. The walls of a room are generallycolder than the air in the room, and the risk ofcondensation is at highest along these. It maybe true that the absolute percentage of humidityin the air close by a pendant fitting at a distanceof one meter from the ceiling, is the same as inthe air near the ceiling, but since no coolingdown takes place no condensation of water canoccur on the fitting. It is therefore oftenpossible to employ pendants without protectiveglobes even in relatively damp places. In suchcases lamps up to 200 watts can be used for thefitting shown in figure 6.To the system beloag further devices for hermeticalenclosing of lamps up to 200 watts. Itdoes not pay to close in lamps of bigger sizehermetically since the costs of the coolingarrangements required will be too high. As isknown the life-time of a lamp is much reducedif it is not kept well cooled.When reconstructing older installations constructedwith single core conductors, it is oftennecessary to have the possibility of combiningFig. 9. Simplified fitting for rubber-lead conductors.— 211 -

in an easy way this system with the lead coveredrubber cable system. For this purpose are usedjunction boxes with connectors for single coreconductors as shown in figure 8. This figureshows a so-called Y-box to which two singlecore conductors are joined. The connectors forthese conductors consist of a porcelain bushingand a rubber ring. In other respects it is identicalto the box shown in figure 3. The rubberbushing shall close tightly to the rubber insulationwhich necessitates the removal of thebraiding.If the rubber-lead conducting material is tobe joined to an armoured tube installation thisis done simply by screwing the tube directly orvia a reduction nipple into the junction box, thethread in the conductor inlet being of the normalsize for the standard 19.2 mm armouredconduit (OP-tube).In places which are not damp a simplifiedform of fitting as shown in figure 9 may be employed.It consists of a standard table lampholder mounted in a junction box. The box isthen covered with a protective globe and theclosed fitting thus formed can be equipped withprotective guards and shades of various kinds.In dry places even if there is danger of firethese fittings are as reliable as those of the moreexpensive construction described previously.They are fire-proof only as long as the protectiveglobe is unbroken, and this holds true ofboth constructions. In a damp room the simplifiedfitting is reliable as long as the protectiveFig. 10. Supporting board with cable clip.globe is unbroken but if the globe is broken —an eventuality always to be reckoned with —the cable will easily be destroyed when the dampair gets access to the cable ends in the box. Thisdanger does not exist in dry rooms where thesimplified fitting may therefore be advantageouslyutilized.In order that the advantages of the rubberleadconducting system shall be fully acknowledgedit is necessary to have it installed in asatisfactory way both technically and economically.It has already been pointed out thatthe worst enemy of this material in damp placesis electrolysis which should therefore be neutralizedto the greatest possible extent. This isdone by means of insulation. The fittings andconductors are insulated from the wall or theceiling by being mounted on a board impregnatedwith some good wood-impregnating stuff.This board is mounted at some distance fromthe wall boarding. The distance varies betweenFig. 11. Wall inlet with supporting cleats.10 and 50 mm according to the proportion ofhumidity in the place where the installation iscarried out. On this board (see fig. 10) the conductorsare mounted with double clips of impregnatedwood or porcelain. The latter materialis better but somewhat more expensive.The fittings and the clips are fastened by meansof galvanized screws and the supporting boardsby means of galvanized nails.Within breweries the use of supporting woodenboards is not desired. The fact is that the woodcan get mouldy, and mould is one of the worstthings met with in a brewery where only suchfungi are desired as are necessary for the fermentationof the beer. The conductors aretherefore fastened immediately to the wall, tothe ceiling, or to the iron framework with porcelainclips.In slightly damp places the supporting boardcan be dispensed with and cables and fittingsbe installed in the same way as in breweries.But in this case wood clips can be used withoutany risk.In dry places the cable is installed in the same— 212 —

Fig. 12. Junction box for switch.way as an ordinary conductor in a soft metaltube (kuhlo-conductor) with a sheath of leadedor galvanized iron. But very often it is moreconvenient to make use of a supporting board.So for instance the installation will be easier andcheaper if a supporting board is used in the casecable. These cleats should have such dimensionsthat the bend of the cable does not exceed theone corresponding to a bending radius of fivetimes the diameter of the cable.Ceiling inlets are carried out in the same waywith the only difference that the cable shouldbe equipped with a mechanical guard up to aheight of 1.5 m above the upper surface of theframework. This guard can consist either otan iron tube fixed to the cable with compoundor of a properly adapted wood duct.When peeling the cable great care should betaken that the lead sheath is not damaged andalso that it is well cleaned up at the point encompassedby the packing. This is necessary toFig. 13. Electrical installation project for big farm.where the conductors are to be mounted underneatha concrete framework across the beams.Wall and ceiling inlets should be handled verycarefully. The cable is here specially protectedby means of a tube fastened to it with compund,before installing. The tube should be longenough to protrude at least 10 mm outside thewall, and the space between the tube and thewall should be carefully tightened, in a woodenwall with tow and oakum, in a stone wall withconcrete or plaster. Figure 11 shows a wall inletwith a lead tube fastened to the cable. Inthe same figure is also shown how supportingcleats should be applied at outward bends of theobtain an effective contact between the leadsheath and the junction box.Figure 12 shows the connections in a junctionbox intended for a switch. The connecting blockin a box of this kind must be equipped withthree connection sockets in order to permit theoutgoing conductor to be connected either beforeor behind the switch. The connectingsocket in the middle is surounded by a red ringand to this one the neutral wire should bejoined. One of the conductors in the cable isred coloured and by joining it to the neutral atthe distribution point and to the red sockets ofall the junction boxes the correct connection is— 213 -

obtained. On installation the lamp holder willbe automatically connected to the middle connectionsocket, and this whether a switch ismounted in between or not. The connections inthe box are so arranged that the switch as wellas the lamp holder can be joined to the junctionbox in one way only. This kind of connectionthus eliminates every possibility of confusion.If the connections in the junction box areTo derive full advantage of the rubber-leadcable system it is necessary to plan the installationscorrectly with application of all the possibilitiesof combination offered by the system.This can be illustrated in the best and simplestway by an example. Figure 13 indicates a bigfarm the offices of which are in part built together.The current is three phase alternatingcurrent of 500 volts. From the aerial line passingcarried out in the way shown in figure 12 theswitch will serve not only the lamp holder mounteddirectly on the switch but all the lamps behindit as well. If, on the other hand, the outgoingconductors are connected to the samesockets as the incoming conductors the lampslying behind will be entirely independent of theswitch. If the outgoing conductor is a threecoreconductor some of the subsequent lampsmay be manoeuvred by means of the switchfixed to the junction box and others be left independentof it.The junction boxes can of course be utilizedfor one or more junctions at the same time asthey hold the switch and the lamp holder. Thejunctions can be connected before or behind theswitch.Fig. 14. Light installation in the offices of a big farm.— 214by the farm an underground cable has been laidto a small transformer station erected in theyard between the buildings. From this transformerstation underground cables go to all thebuildings. For power purposes current of 500volts is used but for the illumination there is atransformer 500/220 volts. The lighting conductorsto the various buildings also consist ofunderground cables. In this way all live bareparts on the ground of the farm are avoided andat the same time the best possible reliability ofservice is secured.Figure 14 shows the light plant in the offices.It consists entirely of lead covered rubber cable.Since it is important from the point of view ofsaving current and with respect to the risk offire that the conductors in the buildings should

not be alive any longer than is absolutely necessary,every group of lamps is equipped withan outdoor group switch which also lights anoutdoor lamp. By looking out into the farmyardthe owner can therefore immediately seewhether the conductors within the buildings arealive or not. All the switches appartaining to theplant ar? placed near the lamps. Thus all specialswitch conductors with the dangers theyinvolve are avoided, and simultaneously thenumber of junction boxes is reduced.The question of costs is often decisive for thechoice of installation system. The cheapestsystem is generally preferred. But even herethe old paradox applies that "he who pays dearbuys cheap". It is not the initial material costsand the installation costs that should be comparedto one another, but it is the sum of amortizationand maintenance costs that gives thereal comparative values.Figure 15 shows graphically the costs of installedconductors for three installation systems,Fig. 15. Graphical tabic indicating installation costs with different conductor systems.If in planning an installation one hesitateswhether to install a junction box or to put ina few meters more of cable it may be mentioned,by way of information, that the installation costsfor a junction box correspond to that of fivemeters of cable. Thus, if it is possible to savemore than five meters of cable it will be cheaperto install an additional junction box but in othercases not.As is obvious from figure 14 the length perlamp of the conductors is relatively small in thecase of a well planned installation. Since thereliability is always depending upon the numberof junction points and the length of the conductors,the rubber-lead conductor system offers,in this respect as well, important advantagesover older installation svstems.viz., single-core conductors on insulators fordamp places, armoured conduit system and leadcovered rubber cable installations. From thetable is seen that the difference between thecosts of installation of the conductors is rathersmall for the three systems. The rubber-leadconductor is somewhat cheaper than the singlecoreconductor but somewhat more expensivethan the armoured conduit. This is made stillclearer by figure 10 showing a table of the costsof construction for a number of installations.Compared with the single core conductor therubber-lead conductor system is cheaper throughout,but it is somewhat more expensive thanarmoured conduit installations. The table showsthat this additional cost in the two cases is 5and 9 per cent respectively. However, if we8. 215

Fig. 16. Examples of cable costs for a number of installations.compare the sums for the various systems itwill be seen that the rubber-lead conductorsystem in the cases given is cheaper by 11 percent on the average.In the case of a complete installation therubber-lead system will, on the contrary, generallybe higher in price, and the examples givenin figure 17 show an average additional cost of22 per cent. This additional cost falls entirelyon the fittings. The fittings are constructed tocomply with requirements quite other and morerigorous than does the ordinary installation material.and their price must therefore be higher.In order to reply to the question whether it willpay to install this more expensive material it isnecessary to know the relative life-time of thevarious systems. Unfortunately it is not yetpossible to answer to this question in full, hutexperience has furnished certain figures whichmay be guiding. In the case, for instance, of avery complicated installation where the singlecoreconductor system had to be replaced everythree months the rubber-lead conductor installationwas still after six years in a satisfactoryshape. The construction of the system excludesany condensation whatever in the conductingcables and if it should occur in the fittings theyought to have been exceptionally badly mounted.The fact that all live parts are hermeticallyFig. 17. Total installation costs for installations indicated in figure IS.enclosed in a carefully grounded metal coveringalmost excludes leakage. The result of all thisis that the rubber-lead conductor installationsare the least dangerous and the most fire-proofof all the systems. The electrical strength andthe high insulation of this system place it in aspecial class. For an installation with leadcovered ruhber cable it is stipulated a minimuminsulation of 5 megohms, or more than 22 timesas high as that prescribed for armoured conduitinstallations. To this comes that there are nodifficulties whatever to reach this value, and asa rule the insulation is many times higher.In addition the rubber-lead conductor systemhas been throughout worked out with a view toneutralize chemical action. As has been mentionedabove this purpose has also been fulfilled.This installation system is the only one whichit is allowed to use in damp places where thereis at the same time a risk of fire. According tothe actual prescriptions it can and may be installedin any place where electrical installationis at all allowed. It makes possible unificationand standardization of installation technics, oneinstallation system for all kinds of places.In order to make it possible for owners ofelectric plants to have their installations modernizedwithout undue loss of time, a close collaborationof all parties busy in the sphere of instal-216 -

lation technics is required. Above all it is ofgreat importance that all consulting engineersmake themselves thoroughly acquainted withthe properties of the rubber-lead conducting materials.To facilitate this work the consultingfirm Kraftkontrollen at Malmo has, at therequest of Sieverts Cable Works (Sieverts Kabelverk)prepared a paper on installations inplaces exposed to moisture and to risk of fire.This paper from which the abovementionedfigures 15, 16 and 17 are taken, shows clearlythat the rubber-lead conductor system is alwayspreferable both from technical and economicalstandpoint as soon as it is the question of installationin industrial premises or in the offices ofa farm.For a contractor it is much less risk to workwith first-rate material than to handle materialnear the limit of what is permitted. When it isto be tendered for an installation which is intendedto be constructed throughout of first-ratematerial the contractor runs no risk of having toreplace or reconstruct part of it. He can determinehis costs entirely in advance and almostall points of uncertainty are eliminated. In thesame way as a first-rate tailor gets better paymentfor a suit than does a second-rate one, acontractor who always deliveres a first-rate jobcan be well paid.CONTENTS: Karl Fredrik Wincrantz. — The Svenska Radioakticholaget Audio-frequency Generator, with continuouslyvariable frequency Adjustment. — The Use of Personal Telephone calls in Sweden, and in Traffic betweenSweden and other Countries. — The Hallsberg Electric Interlocking Signal Plant. — On Cross-talk between TelephoneLines. — Practical Points about Automatic Fire Alarm. — New Interlocking and Signalling Plant at I.und. —On Impedance and Impedance Measurements as well as a Description of the Impedance Measuring set manufacturedby Svenska Kadioaktiebolaget. — Electric Interlocking Plant at Vanneboda Station. — Porcelain Insulators and InsulatorPorcelain. — The Svenska Kadioaktiebolaget Valve-testing set. — On the Calculation of Delays in an AutomaticTelephone System. — Working Reliability and Maintenance of the L. M. Ericsson Automatic Telephone System.— Lead Covered Rubber Cable Installations.217

Vol. VII. 1930.CONTENTS.Biographical data. Xos. Page.Karl Fredrik Wincrantz 7 to 12 98FireProtection.Practical Points about Automatic Fire Alarm 7 to 12 126Metering Devices.Impedance and Impedance Measurements as well as a Description of the ImpedanceMeasuring Set Manufactured by Svenska Radioaktiebolaget 7 to 12 143Instrument for Grouping Fifteen-minutes Loads in order of their Magnitude ... 1 to 6 60Localisation of Line Faults with the Resistance and Capacity Bridge constructedby the Svenska Radioaktiebolaget 1 to 6 69New Swedish Electricity Meters 1 to 6 86The Svenska Radioaktiebolaget Valve-testing Set 7 to 12 174Miscellaneous.Use of Electricity in Modern Life 1 to 6 48PlantHighLowConstruction.Tension.Lead Covered Rubber Cable Installations 7 to 12 20,>Porcelain Insulators and Insulator Porcelain 7 to 12 167Static Condensers for the Improvement of the Effect Factor in A. C. nets 1 to 6 33Tension.New Swedish Carrier Current Telephone and Telegraph Systems on TelephoneLines 1 to 6 23Railway Signalling.Electric Interlocking Plant at Vanneboda Station 7 to 12 164Hallsberg Electric Interlocking Signal Plant 7 to 12 105Interlocking and Signal Plant at Lund 7 to 12 137Telephony.Calculation of Delays in an Automatic Telephone System 7 to 12 185Organized Service for Information as to Subscribers' Numbers in Large TelephoneExchanges 1 to 6 5Suburban Telephone Traffic 1 to 6 16Use of Personal Telephone Calls in Sweden, and in Traffic between Swedenand other Countries 7 to 12 102Working Reliability and Maintenance of the L. M. Ericsson Automatic TelephoneSystem 7 to 12 201Theoretical.Cross-talk between Telephone Lines 7 to 12 112Wireless.The Svenska Radioaktiebolaget Audio-Frequency Generator, with ContinouslyVariable Frequency Adjustment 7 to 12 99Kurt Lindbtrg, Boktntckeriakliebolaq, Stockholm 1930.