sign in sign up
<<
Home , Coaxial cable

NOVEMBER 1952 14.1

COA,XIAL CABLE AS A FOR CARRIER

by H. N. HANSEN *) and H. FEINER *). 621.395.44:621.315.212

------For many ycars carrier systems luwe been operated over two different types of cable, namely quadtled atul . During this period of rivai development, botli techniques have gradually improved and both have been successfully adopted by Telephone Aslministrations in various countries to build extensions of the European long-distance cable network. - This article reviews the history of the development period mentioned above and discusses the comparative advantages of tlu: two systems in a quoliuuiue way. Itfinally expresses the opinion that neither system has yet achieved a decisive advantage over tlie other in allfields of application.

Introduetion

The justification for carrier telephone systems lies Both the above systems were 'also adopted in in the economic advantage of providing several Europe. Between 1930 en 1940; coaxial carrier tele- speech channels over each physical telephone circuit phone systems were introduced in Great Britain, formed by a pair of line conductors. Before 'carrier and France 2) to work alongside the quad working was introduced, only one audio channel was , cables, on which 12-channel systems were installed. obtained from each physical circuit; with the result During and after the second world war, the Nether- that long-distance telephone sys~ems were composed lands Post, Telephone and Telegraph Service were of large multi-conductor cables whose cost formed able to show that on ca_refully constructed quäd a high proportion of that of the entire system. It cables, the number of channels per pair could be was, therefore, advantageous to reduce the cable considerably increased, namely from 12 or 24 up cost by increasing the number of channels per pair to 483). This .discovery had some reactions on the of conductors. development of the European cable network, At first, in the early 1920's, carrier systems were but interest in coaxial systems was nevertheless operated on open- lines and, somewhat later, maintained, and steadily increased. In Great on a few cable circuits. By about 1930 carrier tele- Britain a very considerable number of coaxial phone systems with 12 channels were introduced, systems has already been established, and the one such system being operated over each pair of P.T.T. in France is also engaged on similar activity. conductors of a paired or (somewhat later) "quad- It seems to us to be opportune to present in this ded" cable, the latter containing groups of four Review a brief survey of the rival development of conductors arranged in a star. At about the same coaxial and quad cables, and to attempt to assess time, in the experiments were being qualitatively their relative advantages and expected carried out with a type of cable which would admit fields of application. This will also provide a con- of several hundred channels per pair, viz. the venient introduetion to future publications des- coaxial cable, containing pairs of concentric con- cribing the work of our laboratories in the coaxial ductors. Thc first experimental multi-channel cable field. Two such papers have already ap- coaxial c~l'rier telephone system was installed in peared 4). 1929 at Phoenixville, Penn., by the American Telephone and Telegraph Corp. 1). About 20000 2) Verölf. a.d, Geb. d. Nachr. 6, issue of 28th Jan. miles of coaxial cable are now in operation in the 1937 (series of articles). A. H. Mumford, The -Birmingham coaxial cable U .S.A.; the transmission system is designed to provide system, The Post Office Electrical Engineers' Journal 30, a maximum of· 600 channels per coaxial tube. 206, 270, 1938; 31, 51, 132, 1938. R. Sueur, L'évolution de la technique des ligues a grande Although the above is only a fraction of the total distance depuis 15 ans, Aun. Tëlëcomm. 6,146-164., 1951. amount- of telephone cable in use, the coaxial cable 3) G. H. Bast, D. Goedhart and J. F. Schouten, A 48- channel carrier telephone system, Philips techno Rev. 9, has clearly proved its worth in a very short time. 161-170,1947. 4) J. F. Klinkha'mer, A through supergroup filter for carrier *) N.V. Philips Industry, Hilversum. telephone systems on coaxial cahleç, Philips techn, Rev. 1) Bell Lab. Record 27,234,1949 (Coaxial cable's 20th anni- 13, 223-235, 1952. (No. 8). versary). See also L. Espenschied and M.E. Strieby, H. N. Hansen A new supergroup scheme for Systems for wide-band transmission over coaxial lines, coaxial cable telephone systems, Comm, News. 12; 1-9, Bell. Syst. Techn. J.13, 654-679, 1934. 1951 (No. I). . " 142 PHILIPS TECHNICAL REVIEW VOL. 14., No. 5

General description of a carrier telephone system is associated with an equalising network, so design- ed that the net overall gain frequency curve of A carrier telephone system ,may he classified III each repeater (by which we J?ean a line ampli}ier three categories, as shown infig. 1, viz.: combined with its equaliser) will match the loss- the carrier terminal equipment, frequency curve of a repeater section of cable rather the cable, accurately. In this ~vay, cumulative lihear' distor- the line equipment. ti~n is avoided, in all channels . .: The carrier terminal equipment for each physical circuit, comprises, at the sending end, a modulation . " The number of channels per pair , system which enables each channel.from the audio- frequency range to be translated upwards in fre- The above-mentioned increase in the number of quency to some specified location in the, frequency' , channels per pair entails a corresponding increase in spectrum. At the receiving end, analogous demodu- the frequency which must be-transmitted lation , equipment translates the incoming high-. over each physical circuit. On cables of the types frequency channels downwards to their original under discussion, the attenuation increases with audio-frequency range. The .nominal width of each frequency. An increase in bandwidth thus leads to channel is usuallyd kc/so We shall not deal exten- an increase in the total amount \of gain required sively with the actual modulation process. Usually throughout the circuit .

.;

E 69982 Fig. 1. Block schematic of a carrier telephone system. E= carrier terminal equipment to which the suhscriber's lines A', A", ... B', B", ... are connected through 4-wire terrninating sets, V. The outgoing are passed to the modulating apparatus 1Uand the incoming signals to the almost identical demodulating apparatus D. Kl' K2 are the "go" and "return" cables. L = line equipment installed in repeater stations. In each station an equaliser Eg and amplifier are provided for each cable pair in each direction.

modulation is done in several steps: By a first trans- This will necessitate a shorter repeater spacing, position to higher frequencies, 12 audio-frequency as can be readily understood. The arnvmg channels are assembled to- form a group; 5 such at a repeater must not be attenuated by the preeed- groups (or.less) by a second transposition are com- ing cable section to a degree that it will suffer from hined rto form a supergroup. If there are more the thermal of the cable and the first amplify- than 60 channels, each of the supergroups again ing valve of the repeater. Hence, there is a'minimum must he shifted to its final frequency range (the permissible input level. On the other hand, the out- supergroups together form a "hypergroup"). put level is Iimited by the power available from the In quad cable systems, up to the present time it power valves. Thus, the gain to be provided by has been normal practice to provide separate "go" one repeater is fixed, and a higher total gain can only and "return" cables for the two directions of trans- be obtained by increasing the number of repeaters. mission. Nevertheless, an increase in the number of Line amplifiers must he installed at repeater channels will economically be justified by the saving stations spaced at intervals (repeater sections) along in cable cost per channel. Moreover, the number of the route; the amplifier gain compensates for the repeaters p er channel actually will also be rJeduced attènuation to "which the signals are subjected by virtue of the fact that the inèrease in cable during their propagation along the cable. In each attenuation (i.e. the required gain) is less than station, the line amplifier for each physical circuit proportional to the frequency (it will usually be NOVEJ\1HEH 1')52 COAXIAL CABLE FOR CAHRlER TELEP.HONY JA.;)

increase I' found to as thc square oot of the fre- copper and lead. In practice it IS found that the quency). optimum conductor diameter lies between 0.9 anp There are, however, several limitations which 1.3 mm. The attenuation at 200 kc/s is about prevent an indefinite increase in hand width. Since thermal noise is introduced at the input of every repeater, the minimum permissible level of the signals at the end is dependent on the number of repeaters in circuit. Similarly, the total output power available from a repeater diminishes with increasing frequency; moreover, this power has to be shared between a greater !number of channels, so that the maximum attainable level of the signals at the out- put end is reduced when the bandwidth is increased. The maximum gain obtainable is seriously affected by these effects: for example in a 48-channel system for use on quad cable, the maximum amplifier gain is about 65 dB, but in a 960-channel coaxial system it is not more than 50 dB. Conscqucntly, the repeater spacing must he proportionately shortened to allow for these level limitations, and it is fairly clear that there must he some theorctical optimum beyond which it is not economical to increase the band width. In systems in use today, opcrating on either quad

UI' coaxial cables, this economic limit has by no means been reached. The practical limit to the num- ber of channels per pair is imposed by characteristics of thc components at present available, in coaxial systems particularly the valves. This limit cannot be raised by any reasonable expedient; for example, it is useless to increase the number of valves per 72775 amplifier. These limitations will he explained presently. }'jg. 2. Typical 12-lIuaü cub!c, Each star quad con tains two diagonal pairs. Air-spaced paper irisula tion is used. The cable is sheathed with lead. Outer servings and steel tape arrnouring Quadded telephone cable are applied over thc sheath. Consider first the common quadilcd cable, con- taining pairs of conductors, every two pairs being 2,4 dB/km for 1.3 nun conductors. Assuming a combined into stars (star quads). Fig. 2 shows a repeater gain of 60 dB, the repeater spacing for such 12-quad cable of the type widely employed by the a cable is 25 km, which corresponds with normal Netherlands P.T.T. Air-spaced paper insulation is practice applicable to the 4·8-channel system used adopted, and the cable is covered hy a lead sheath as a proreetion and screen, and to prevent the dB;[,km ~ o-~ ingress of moisture. Further protective layers arc o- 2j5 -o provided to prevent physical damage to the cable a if it is laid directly in the ground. t 2,0 The cable attenuation is shown as a function of frequency in fig. 3. The absolute value of the at- 7,5 tenuation, other things being equal of course, de- pends on the diameter of the conductors. The choice of conductor diameter, within limits imposed by electrical and mechanical requirements, is governed hy economic considerations again; large diameters °o~----~~----~~----~~----~~~ m ~ m ~~~ result in costly cables, hut reduced attenuation and, _I therefore, fewer repeaters. The most economic choice Fig. 3. Attenuation-frequency characteristic of a 1.3 mm pair in a quad cable. The diagram shows the atte nun tion a ill will he greatly affected by the market prices of dB/km as a function-of frequency f (in kc/s). l'HILIPS TECHNICAL REVIEW VOL. H, No. 5

respectively III these cables, both being made of III the Netherlands and elsewere (frequency band 12~204 kcjs, repeater spacing about 26 km). copper, (fig. 5), bear roughly the ratio: So far, it has not been possible to increase con- D = 3.6 .. (1) siderably the number of channels per pair in quad cl cable due to limitations of . Crosstalk - the transfer of signals between different pairs of conductors - is due to the effects of un- balance and mutual impedance between the various pairs in the cable. Crosstalk in quad cables increases with the frequency of the signals. By suitable pre- cautions in the manufacture and installation of modern quad cables 5), it has been found possible to keep the crosstalk within permissible limits up to a frequency of 250-300 kcjs, which is sufficient to allow the eperation of 60-72 channels per pair. Possibly some further progress may yet be made, but the most optimistic estimate ofwhat may be achieved in the future does not exceed 120 channels per pair.

If such an increase in the number of channels per pair is ö9991 ever to he attained, a very efficient halancing technique will have to be evolved. For this purpose, the introduction of Fig. 4. Typical coaxial cable as used in the ·USA. Ei g h t coaxial tubes are laid up helical ly over ft core consisting of auxiliary variable balancing elements at in terrned ia tc points wit hin signal . Similar sheathing anel armouring as fig. 2. each repeater section is contemplated. Atten tion today is concentrated on possible wa ys of'diminish- ing the crosstalk because new methods of cable construction In one cable ~tandardised internationally, d = promise an increase of frequency bandwidth without a r-orr es- 2.6 mm and D = 9.4 mm (3/8"). Thc inner conduc- ponding reduction in the repeater spacing. The attenuation tor is supported along the axis of the tube by discs of present types of cable is partly attributable to or beads of polythene or other materials, spaced I" losses in the paper insulation, especially at frequencies in the apart, or altcrnatively by helical hand of polythenc order of 200 kc/s, the loss angle tan /) then reaching a value of 120 X 10-". Various manufacturers are developing new types or styroflex, etc. The remaining space is fillcd with of cable in which the insulating material consists ofpolycthylcnc, dry air or gas, sometimes under prcssure. styroflex or polystyrene foam, whose low dielectric constant and loss angle (tan cl = 3 X 10-4 or less) enable the HF at- o tenuation to he reduced considerably below the f'gUl'es applic- able to paper cable. In fact, the attenuation offoam-insulated cables at 550 kc/s is comparable with that of paper-insulated cables at 250 kc/s.

Coaxial telephone cable A coaxial pair or tube generally consists of a copper wire which forms the inner conductor or core, supported by some insulating structure, so that its axis coincides with that of a tubular outer con- ductor. The outer conductor is usually also of copper, but is sometimes made of aluminium, which is cheaper. Several such pairs or tubes may be laid up Fig. 5. Cross-section of a coaxial tube. The ratio of diameters together with auxiliary conductors to form a D/d is usually 3.6.

coaxial cable. The auxiliary conductors are required The motivation of the condition (1), as well as a brief sum- to provide a service telephone, extended alarm cir- mary of the transmission characteristics of coaxial cables, will cuits etc. Fig. 4 shows an 8-tube coaxial cable; a be found in the Appendix to this article. type has been standardised in the United States. The diameters D and cl of the outer and inner conductor We can now state in a few words the reason why it has been possible to employ such a very much greatcr 5) See L. J. E. Kolk, Het balanceren van draaggol£kaheJs number of channels per pair in the coaxial cable voor 48-kanalen systemen, PTT Bedrijf 3, 59-74, 1950 than in the quad cable: in the coaxial cable crosstalk (in Dutch). NOVEMBEn 1952 COAXIAL CABLE ron CAnnIER TELEPHONY 145

betweèn pairs does not increase, but decreases with the thickness ofthe outer conductor, but only at the frequency. expense of increased cable cost. Some quantitative This advantage can he- accounted for by a con- . idea of this relationship may be obtained by a con- sideration of . At sufficiently high fre- sideration of the depth of penetration i} of the signal quencies, the signal currents flow nearly entirelyon and crosstalk currents into the outer conductor, i.e. the outer surface of the inner conductor and on the the depth within the conductor at which the current inner surface of the outer conductor. The outer sur- density has diminished by a factor l/e. For a face of the outer conductor carries practically no conductor having a_resistivity (! (in ohm metres) current, and between two points on this surface only and .a relative permeability pr, we may write: ve!'y small alternating will occur. When 5 considering the influence of these voltages on an --0. = 10 1/ (! mm = ~ mm. . (2) adjacent tube, the skin effect again is of importance: .2n ,tr! l'f the density of the currents' induced in that tube is In the above expression, the frequency must be greatest at the outer surface ofthe outer conductor, expressed in kc/soFor copper it is found that C~2.1; it sharply decreases towards the interior, which for aluminium C ~ 2.75. In order to reduce the means that the interference is hardly capable of depth of penetration to a few tenths of a millimeter, penetrating to the transmission circuit proper. This i.e. to an amount small compared to the minimum "shielding action" of the outer conductor of each. thickness of tape which can be applied without ex- tube is most effective at high frequencies. Thus, at cessive mechanical difficulties, the frequency must sufficiently high frequencies, crosstalk will become exceed 100 kc/so In coaxial systems on multitube entirely negligible. cables, the lowest frequency transmitted is never less than 60 kc/s, often more 6). The crosstalk shielding effect of the outer conductor not The up per limit of frequency in coaxial sys- only permits the use of vcry high signal frequencies, but it tems so far installed is approximately 2.8 Me/s also makes it possible to place the coaxial tubes, used for (660 channels). opposite directions of transmission, within a common lead Standardisation by the. C.C.!.F. will in due course sheath. Owingto the large difference in level between incoming and outgoing pairs at repeater stations, so exacting require- extend this band up to 4 Mc/s (seefig. 6), and work is ments as to crosstalk must be mct with quad cables so that now in progress on a transmission band up to 8 Mc/s. in this case separate "go" and "return" quad cables must, Naturally, at these increased frequencies the at- in general, be used. tenuation per km of cable is higher, and it will be necessary to use shorter repeater sections than have From the above it is apparent that crosstalk on hitherto been customary. Nevertheless, as indicated coaxial cables is only serious at the lowest frequen- above, this is not necessarily an economic dis- cies transmitted, The lower frequency limit, there- fore, is chosen fairly high. If it were desired to 6) From equation (2) it will be seen that a high permeability extend downwards the' lower limit of the frequency results in a smaller penetration depth. For iron, CRlO.3. To save copper, the outer conductor therefore is sometimes band transmitted, this could be done by increasing covered with steel tape.

o 100 200 400 ---'-----_/ 52

,~hbNbh JOOO • 4CXXJ,l

advantage, since the attenuation only increases as allows a suitable margin), the necessary repeater the square root of the frequency. For present types spacing is thus about 9.5 km. , of coaxial cable the attenuation is given by the There is no economic objection to a considerably following formula: higher maximum frequency and a 'still smaller rc~ peater spacing, but further progress in this.direction 007 - . a = -iJif dB/km, . . . . (3) ,is inhibited by technical difficulties in the repeaters. Characteristics of repeaters for coaxial systems where f is again expressed in kcl_sand D is the inside diameter of the outer conductor in cm. Fig. 7 Fig. 8 shows a comparison between two repèatercd d%m dable systems of equal length, of the 12-quild 5 (i.e. 24 pairs) and coaxial types respectively, "Go" V I and "return" paths are shown in each c~se, In the .....~ V case of the quad cable it is assumed that each pair V transmits 60 channels, i.e. 144,0 in all, and that the 3 /'" , repeater spacing is about 23 km. The coaxial tube

, accommodates 960 channels, the repeater spacing I I -: 2 I / being taken as 9.5 km. I V It is interesting to observe the saving in line 1,/!/ amplifiers and physical circuit material per channel shown in the second diagram. Taking into account I "go" and "return" transmission paths, the quad 500 1500 2000 2500 3000 ~f 4000kcls cable requires 2/(60 X 23) R:j 1/700 amplifier per channel km, whereas in the second case the corres- Fig. 7. Attenuation-frequency characteristic of a coaxial tube; D, the inner diameter of the outer conductor, is 0,91t cm. ponding figure is 2/(960 X 9.5) R:j 1/(4500) ampli- The broken line shows the corresponding attenuation of a 1.3 fiers per channel km. mm pair in a quad cable, reproduced from fig. 3. lt is not our aim, however, once again to insist shows the attenuation-frequency characteristic of a on the economic advantage of one system over the cable in which D has the standardised value 0,94 cm, other. (To do this objectively we should have to mentioned above. The attenuation is 4.8 dB/km at consider the prices of cable and repeaters as, well 4 Mc/s. Assuming a repeater gain of 50 dB (which as those of the terminating equipment). We wish

Fig, B. a) Block schematic of go and return 12-quad cable with repeaters. Total number of channels is 60 X 24,= 14-40. b) Block schematic of a pair of coaxial tubes transmirting 960 channels. ·1

NOVEMBER 1952 COAXIAL CABLE FOR CARRIER TELEPHONY ltJ.7

rather to drawattention to some other features of sitic of amplifier valves, input and 'out- this diagram: put , etc., which are-responsible for the a) In the coaxial system each repeater handles decrease of stage gain at increasing bandwidth, also 960 ehannels, whereas in the quad cable case, give rise to a phase displacement which increases the eorresponding figure is 60 channels. with frequency; for this reason, theré is a limit to h) In the coaxial system each bi-directional phys- the number of stages' across which negative feed- ical circuit is equipped with 6 repeaters in a back may be applied; the limit in this case is three distance of 23 km, whereas in the quad .cable stages. It might he suggested that an amplifier. system there are only 2. could be built', for example, with two units, each c) Finally, in the coaxial cable there are many small consisting of three stages, with negative feedback stations, and in the quad case, fewer large ones. applied to each unit individually. The sacrifice of In consequence of these differences the line ap- 30-40 dB gain, however, would then apply to each paratus for coaxial cables in several respects must such three-stage unit. Obviously, the placing of satisfy more stringent requirements than that used units in tandem will be ofno use and no gain at all for quad cable. First, the r e.lig h il it y should he con- can be obtained if the available overall amplifica- sidered. The breakdown of one coaxialrepeater would t'Ïon per three-stage unit is already less than disable ·960 channels simultaneously, which in tele- 30-40 dB. Beyond this limit any increase in the phony work would he regarded as little short of a bandwidth transmitted over coaxial systems is thus catastrophe. Each repeater must therefore he pro- conditional upon improvements in the characteristic vided with the necessary spare equipment, and it of the components, particularly in respect of the might even he advisihle to provide a pair of coaxial capacitance and slope of the amplifier valves 7). tubes, fully equipped, as a standhy. These, then, are the principal factors which for Secondly, the stability is of importance. In the moment set a limit to the increase of the fre- a long-distance connection each channel passes quency bands employed on coaxial cables. through numerous repeaters in tandem. To main- The dispersal of the line apparatus over a large tain a constant level of all channels at the receiving number of small stations; as demonstrated in fig. 8, end, variati:ons of the individual gain of each re- is, unlike the subjects previously discussed, not peater are strictly limited. Since there are more primarily a circumstance which limits bandwidth; repeaters in tandem in a coaxial than in a quadded but an increase of bandwidth would certainly add cable, for circuits of equal length, the limits for a to the practical difficulties of feeding power to a coaxial system must be more stringent. This neces- large number of repeater stations, possibly con- sity is enhaneed by the fact that the coaxial system taining an increased number of valves. In quad is generally used for longer distances than quad cable systems it is customary to feed the line equip- cables ment from a l.ocal mains supply, a standby power Thirdly, non-linearity in the repeaters, es- supply heing provided by a battery or engine-gene- specially the valves, gives rise to intermodulation, rator set. It is, however, not economical to furnish i.e, a form of (un-intelligible) crosstalk hetween each of a large number of closely spaced coaxial channels. Since many channels contribute to inter- repeater stations with its own local mains supply, nor modulation into any given channel, this effect is with an emergency power source. The best solution apparent as a background noise which is dependent of this problem so far devised is to feed each re- upon the total numher of channels per pair and peater station from a power supply transmitted over upon the number of amplifiers in tandem. In both the cable itself. The necessary A.C. power is fed these respects coaxial cable again is at a disadvan- into the cable at stations spaced at intervals of up tage compared with quad cable. to 300 km along the cable route; these stations are The linearity. of the amplifiers and their con- provided with alternative sources ofpower supply, stancy of gain ca:q theoretically hoth be improved as a precaution against mains failure. By centralising hy the use of negative feedback. In our case this the standb,?' plant in this way, the problem of sup- requires a sacrifice of gain of 30-40 dB in cach repeater. This, however, is particularly inconvenient 7) For practical feedback and amplification' values for a given bandwidth, see H. W. Bose, Network analysis a~d hecause the stage gain of a broadhand amplifier feedback amplifier design, Van Nostrand, NewYork, 194.5. capable of handling 960 channels is already relative- See also J. te Winkel, A note on the maximum feedback obtainable in an amplifier of the cathodc-feedback type, ly small. This technical difficulty even sets a funda- Philips Res. Rep. 5, 1-5,1950. In this connection itmay be mental limit to the number of channels which can said that the use of Ferroxcube for the cores of input and output transformers in the amplifiers'already represents _ be transmitted over a single tuhe. In fact, the para- an important step forward. ' VOL. i4, No. 5 148 PHILIPS TECHNICAL REVIEW

plying. power to a large numb~r of small stations has same technique cannot be applied so easily to the been solved in a very simple way and at a low cost coaxial system. Terminating one pair of coaxial perrepeater. This system is all the more attractive tubes (go and return) at an intermediate town would since the solid copper core of the coaxial tube, which, mean providing for this. town: a hypergroup con- with the exception of the ex~reme outer layer, does taining many hundreds of channels (potentially 960), -not serve the transmission of the signals, is. then i.e. much too large a unit for convenient use. This is effectively utilized. Another - advantage of the the price which has to he paid to' secure the • method is the easy solution it offers for feeding economic advantages of transmitting a very large repeater stations in sparsely inhabited country number of channels on a single physical circuit. where there may be no electric mains for hundreds _ Some months ago this was reviewed in detail in of miles. this journal 4) and reference w:as made to the pos- sibility of interconnecting two coaxial systems on . Another important problem in the line equipment, especially a "through group" or "through-supergroup" basis, , ..•.. of coaxial systems, is the varintion of (I, the attenuation of the so that one hypergroup is no longer to be regarded cable, as a function of temperature T. Evidently, the factor as an indivisible unit, but some channels can be preceding l'Jin equation (3) and in which among other things terminated at an intermediate point and the others the resistivity of the copper is involved, will vary with tem- perature. Now, da/dT will he obtained by multiplying the can pass through in transit. The through super- temperature coefficient of that factor - whatever function group procedure requires that the incoming hyper- that may be - by JfJ. This means that Iluetuations in the group be demodulated to its component basic super- cable temperature will affect the level of the high-frequency groups, whose frequency range is 312-552 kc/so The channels considerably more than the low-frequency ones. supergroups which terminate are demodulated com- The variations in level in coaxial systems have been reduced to reasonable proportions by providing an efficient form of pletely to audio frequency, whereas those in transit automatic level regulation controlled by the line pilot signals. pass via "through-supergroup filters" to the super- The extent of such variations for the reason explained group modulators, to take their place in the outgoing being considerably less on quad cable systems, less elaborate hypergroup. regulation in that case is satisfaetory. Although in principle this arrangement imparts to the coaxial system the flexibility of 12~or 48-chan- The problem of nel quad cahle systems, the additional equipment at To complete our consideration of the relative the transit point inevitahly must detract from the merits of carrier systems on coaxial and quad economic advantage of the coaxial system. An alter- cables, we shall now examine their respective native and simpler method-of providing transit facil- adaptability to meet the demands of practical hand- ities exists; this is known as echelon working and ling of carrier systems, in other words their "flexi- it consists in allocating to the traffic between each bility". Carrier telephone systems usu ally form part pair of towns in a coaxial cable network the ex- of a national or international telecommunication clusive use of one or more of the supergroups system. At various intermediate towns or junction provided by a pair of tuhes. The intermediate points along the route a sufficient number of chan- carrier terminals then only need be connected in nels must be terminated to accommodate local traffic parallel with the main transmission paths by any requirements. This number will vary over a wide convenient method, without any ofthe special filters range, and may not be constant. Temporary mentioned above. Admittedly the available hand- changes in the distribution of channels may be width in general is not fully utilised when this necessary on account of important events; or per- scheme is adopted, and consequently part of the haps there may be steady growth of traffic due to essential feature of the coaxial cahle system is an increase in the number of subscribers or the in- sacrificed. troduction of automatic operating procedure on A completely different solution of the prohlem trunk circuits. The structure of a trunk system is that is being studied involves a rearrangement of thus always changing and demands a certain degree the supergroups ,~ith a wider mutual separation. of adaptability of the . This proposal will usually result in a less efficient Systems incorporating 12-60channels per pair such utilisation ofthe frequency bandwidth, hut will per- as those operated over quad cables, fulfil this con- mit of a new technique, the use of "band-stop filters" dition admirably; by terminating a suitable number to suppress one or more of a range of supergroups of pairs at each intermediate town or junction point, for purposes of extraction at intermediate terminals. the distribution of channels can be readily adjusted This would greatly simplify the distribution to meet traffic requirements. On the other hand, the problem 4). NOVEMBER 1952 COAXIAL CABLE FOR CARRIER TELlj:PHONY 149

The above discussion indicates that coaxial and ing end the smaller values 12and E2 are found. The attenuation quad cables each have their own characteristics of the cable is defined by the following equations:

which adapt them for specific applications, neither a = loge 1EljE2 1= loge 1111121. type having a decisive advantage over the other in From theory, the following simple formula for a is obtained: all fields. Future developments may eventually lead to one system 'gaining ground at the expense of the a=,21RIGY+Tl-c'"GIL ..... (4) .other. Ón the basis of present technique, the coaxial system undoubtedly is to be preferred for projects It contains two terms, corresponding with the two distinct in which a large number of channels must he provided causes of loss: the first term represents resistive attenuation due to R, and the second term, always considerably smaller, over long distance, and in which the characteristic the leakance attenuation d~e to G. economy in the cost of cable and repeaters is So far the theory applies to all types of line. In order to consequently obtained. In sparsely populated areas apply formula (4) to our case, we need to relate thc primary where power supplies are mostly non-existent, the constants to the dimensions and material constants of the coaxial system has the important advantage of coaxial cable. Consider first the resistance R. The high-frequency current power feed over the cable; in some areas, e.g. in the chiefly flows in' a thin cylindricallayer at the outer periphery U.S.A., this advantage has been' decisive. For corn- of the inner conductor (diameter d) and at the inner periphery paratively short distances in thickly populated of the outer conductor (diameter D). Now R can be taken country, particularly where the cable network forms as equal to the D.C. resistance of two tubular conductors a closely triangulated pattern, quad cable systems with the diameters d and D respectively and a wall with relatively few channels (e.g. 48) have generally thiekness equal to the penetration depth {}.The cross-sectional areas of these conductors are {}'nd and 1J'nD, and according been preferred on account oftheir greater flexibility. to eq. (2) we have j'(lIj! and j'(!21! respectively, (!l and (!2 There are many marginal cases between these two being the resistivities of the moer and outer conductor material extremes; the ultimate choice will here depend on respectively. The resistance R then is proportional to the progress of future development, for example the increase in frequency bandwidth of quad cable j/7e~1 + j'~2) . mentioned above, and the proposal to feed power over quad cable to repeater stations. These expected The capacitance C as well as the L of tw 0 coaxial cylindersdependsontheratio of the diameters only, not on improvements will then have to be compared with their absolute velucs.Tt is found that jlLjC, the characteristic the corresponding evolution of coaxial technique. impedance, is proportional to log Did. The resistance part Finally, it should he mentioned that the feasi- of the cable attenuation given in equation (4.) then becomes bility of relaying programmes is being proportional to advanced as an important argument in favour of coaxial cable. For this purpose a bandwidth of at . . . . • . . . • (5), least 4, Mc/s is required. A combination of telephone and television transmission over the same medium The second fraction in this expression depends on the ratio would undoubtedly he attractive. Djd and for a certain value of this ratio it passes through a (fairly flat) minimum. This optimum ratio is Djd = 3.6 for Appendix: Transmission characteristics of coaxial cable (!2!(!1= 1, the usual case (inner and outer conductors made of' The transmission characteristics of any line, for a signalof sarne material). Thus, if we regard D as fixed, the cable at- frequency! = wj2n, are determined by its resistanee R, in- tenuation will attain a minimum if d is made equal to Dj3.6, ductance L, leakance G, and capacitance C, which properties in accordance with the condition (1). When the outer conduc- are continuously distributed along the line and, which are tor is of aluminium and the inner of copper, (!2j(!1= 1.64 and expressed per unit length. The values of R, L, G and C, the the condition of minimum attenuation occurs whenDjd = 3.8. "primary constants" are usually functions of frequency, in Substituting for the second fraction in (5) the minimum spite of their name, but only to a slight extent in the case of value 4.6jloge 3.6 and, taking into account the proportionality Land C. factors, not embodied for sirnplicity's sake in the above for- The theory of transmission becomes fairly simple when the mulae, we finally obtain equation (3) for the attenuation. (In line is uniform and when wL'}:>R and wC'}:>G, which is always this case it is a~sumed that both conductors are of copper, the case in practice at high frequency, i.e. well beyond the for which (! = 0.01745 X 10-0 ohm metres, and that the audio range, According to theory, in that case the ratio of dielectric is air (er = 1).) By choosing Djd, likewise the sig nal E to signal current 1, both assumed sinus- inductance L and, when the relative diel~ctric constant er of oidal and transmitted înto an infinitely long line, is the same the insulation in the cable is known, the capacitance C and in all points and equal to ]fLjC; this ratio is independent of the l'LjC of the coaxial tube are frequeuev and is called the characteristic impedance. fixed. The characteristic impedance for a cable of optimum The losses in transmission are caused by the resistance R design with air dielectricis 77 ohms: With a typical construction and the leakance G. If the current and voltage transmitted using spaeer beads or the equivalent, this figure is reduced into an infinite line are 11and EH at unit length from the send- to 75 ohms and with solid polythene insulation to 52 ohms. 150 ;t?HILIPS TECHNICAL REVIEW VOL. VI, No. 5

The lcakance G, which is caused by dielectric losses in the . physical circuit up to the economic limit in order to reduce the Insulating medium, may be expressed in the form of a loss line cost per channel. In "quad" cables, 4·8channels per pair is now normal practice (maximum frequency transmitted angle ij for the dielectric, by the equation: . 204, kcJs). It is hoped to extend this figure in the future to 60, 72 or even 120 channels per pair. Further progress in ...... (6) G = wC tan d , quad cable development is limited by crosstalk which becomes more serious at the higher frequencies. On this score, ''\'ith most insulating materials used for coaxial cables, tan é things are fundamentally better with coaxial cable; owing to does not depend on frequency within the frequency 'range of the screening effect -of the outer conductor, crosstalk in this interest. case decreases with increasing frequency. Crosstalk at fre- Since C, except for being proportional to er, only depends quencies below 60 keJs is serious on coaxial cables and con- sequently this is normally the lowest frequency transmitted, on the ratio DJd, which has already been chosen for optimum but there is in this respect no upper frequency limit. Coaxial resistance attenuation, it follows that, with this optimum systems at present in opcration in Great Britain and the U.S.A. design, Gwill beproportional to I er' tan ij. For polythene er is ' transmit as many as 600-660 channels per pair, and standar- not large (about 2-3) anddielectric.losses are extremely small: disation by the C.C.I.F. provides for a future increase to 960 (maximum frequency transmitted 4 McJs). An economic limit a value tan ij R:i 3 X 10-4 is usually assumed. This leads to a is imposed by level considerations which lead to a reduction leakat{ce attenuation component which may he disregarded in in the repeater spacing not commensurate with the numberof ordinary cable construction, notwithstanding the fact that this extra channels obtained. This economic limit has not yet been component increases'with frequency at a greater rate than the reached, the number of channels being actually restricted due to tile aggravation with increasing frequency of requirements . resistance attenuation. in respect of reliability, stability and linearity of the repeaters. Finally, the problem of interconnection is discussed, rhis being Summary. Development of carrier telephone systems has been inherent in all carrier systems having few physical circuits directed towards inëreasing the number of channels per each carrying a large number of channels.

ABSTRACTS OF RECENT SCIENTIFIC PUBLICATIONS OF THE N.V. PHILIPS' GLOEILAMPENFABRIEKEN

Reprints of these papers not marked with an asterisk * can be obtained free of charge upon application to the Administration of the Research Laboratory, Kastanjelaan, Eindhoven, Netherlands.

1995: J. J. Went: The value of the spontaneous essentially different from that of the alloys. A magnetization of binary nickel alloys as a maximum in thc magnetoresistance is observed at function of temperature (Physica 17, 596-602, low temperatures for alloys having ahout one Bohr 1951, No. 6). magneton per atom. The positive difference be- , The Is versus T curve for pure nickel is more tween the longitudinal and the transversal resis- concave towards the T-axis than that for any tance can be explained by means of the spin-orbit binary nickel alloy, with only one exception, viz. interaction. completely ordered alloys, such as slowly-cooled At low temperatures the pure metals show an NiaFe. For pure nickel the form of the Is versus increase in resistance with increasing field just as T curve can be explained by the occurrence of an the non-ferromagnetic metals. From this the value order-disorder phenomenon of the magnetic mo- of the internal field, acting on the conduction . ments, where the latter can be placed only parallel electrons, could be determined, and was found to or anti-parallel. It is suggested that all the other be approximately equal to the flux density B. Is versus T curves must be explained as being a 1997: J. A. Haringx: De instabiliteit van inwen- result of this order-disorder phenomenon and of the clig op druk belaste dunwandige cylinders statistical fluctuations of the concentration of (De Ingenieur 63, 039-041, 1951, No. ~9). dissolved atoms. In the case of NiaFe two' different (The instability of thin-walled, cylinders order-disorder phenomena (a crystallographic and subjected to internal pressure; in Dutch.) a magnetic one) act simultaneously. It is shown that uncler certain conditions a thin- 1996: J. Sm it: Magnetoresistance of ferromagnetic walled cylinder may buckle when subjected to metals and alloys at low temperatures internal pressure. The critical value of this pressure (Physica 17, 612-627, 1951, No. 6). can easily be calculated on account of Euler's The magnetoresistauce of pure Ni and Fe, of well-known formula. Most of the formulae for Ni-Fe-, Ni-Co-, and Ni-Cu-alloys and of Heusler's pressure-loaded cylinders given in current textbooks, alloy has been measured at room temperature and however, fail to predict this behaviour and should at temperatures of liquid and liquid therefore be applied with caution for great lengths hydrogen. The behaviour of the pure metals is and/or for high pressures. NOVEMBER 1952 ABSTRACTS 151

1998:, J. A. Haringx: .De instabiliteit van in- polycrystalline semi-conductor prepared according wendig op druk belaste ronde balgen (De to the principle of controlled valency (replacing Ingenieur 63, 042-044, 1951, No. 29). (The the trivalent ion X, e.g. La, by a bivalent ion, such instability of cylindrical bellows subjected as Sr). There exists a close analogy between the to internal pressures; in Dutch.) resistivity and the ferromagnetic properties. Dis- Like thin-walled cylinders, dealt with in a pre- cussion of anomalies at the Curie temperature, vious paper (see No. 1997), also bell~ws may become of the influence of frequency, and unstable when loaded by internal pressure. The external on the resistivity, of the critical value of this pressure, which is governed, thermo-electric properties and the HaU effect. through Euler's well-known formula, by the rigid!ty of the bellows with respect to bending, is 2002: H. Bremmer: On the diffraction theory computed only for rectangularly shaped corruga- of Gaussian optics (Comm. pure and appl. tions. The result found has been checked experimen- Math. 4, 61-74, 1951, No. I). tally. The diffraction theory of optical imaging is devel- 1999: C. G. Koops: On the ofresistivity oped for objects, with arbitrary structure. The and dielectric constant of some semicon- theory is based on rigorous solutions of the wave ductors at audiofrequencies (Phys. Rev. 83, equation instead of the conventioI_l~1approximation 121-124, 1951, No. I). of Kirchoff's formula. The similarity of the wave Semi-conducting 'Nio,4Zno,6Fe204, prepared in functions in the object plane and in the corres- different ways, has been investigated. It appeared ponding paraxial image plane (Gaussian systems that the A.C. resistivity and the apparent dielectric with unlimited aperture) proves to be connected constant of the material shows a dispersion, which with Neumann's integral theorem for Bessel can be explained satisfactorily with the help of a functions (instead of the Fourier identity as in simple model of the solid: there should be well- Kirchhoff's approximations). Another solution conducting grains, separated by layers of lower accounts for the effects of optical aberrations and conductivity. Dispersion formulas are given. There of limited apertures. is good agreement between experiment and theory. 2003: H. Bremmer: The W. K. B. approximation 2000: E. J. W. Verwey: Oxidic semi-conductors as the first term of a geometric-optical series (from: Semi-conducting materials, Butter- (Comm. pure and appl. Math. 4, 104-115, worth's Scientific Publicatio~s Ltd., London 1951, No. I). 1951, pp. 151-161). The W.K.B. approximation of the solution of Non-conducting oxidic materials may be rendered y" + k2(X) Y = 0' is derived from a discontinuous conductive either by preparing solid solutions with model of an inhomogeneous medium. Higher ap- oxides of high conductivity (e.g. a poorly conduc- proximations are found by considering multiple ting spinel with Fea04) or by introducing ions of reflections. The solutions of different order form deviating valency into the lattice (method of con- à series, the convergence of which is discussed. trolled valency). A comparison is made between The well-known insufficiencyofthe W.K.B. approxi- semi-conductors obtained is this way and elemental mation in the neighbourhood of zero's of k(x) semi-conductors such as silicon and germanium. can be interpreted as a very slow convergence of The conductive properties of polycrystalline semi- the series mentioned. conductors as a function of frequency are discussed, especially the influence of non-conducting thin 2004: W. Ch. van Geel: On rectifiers (Physica 17, layers át the grain boundaries (see these abstracts, 761-776, 1951, No. 8).• Nos. 1738, 1739, 1845, 1914, and Philips techno Rev. 9, 36-54.,239-248, 1947; 13, 90-95, 1951, No. 2). , Experiments are described in which rectifiers were obtained from combinations of metals, semi- 2001: J. Volger: Some properties of, mixed conductors and intermediate layers. The following lanthanum and strontium manganites combinations have been investigated: (a) the con- (from: Semiconducting materials, Butter- tact between an excess and a worth's Scientific Publications Ltd., London deficit semiconductor; (b) the combination alumi- 1951, pp. 162-171). nium/aluminium ,oxide/semiconductor; and (c) the Discussion of electrical properties of manganites combination metal/resin layer/semiconductor. In XMnOa. This material is an example of a all these cases rectification occurs. The suggestion /

152 PHILIPS TECHNICAL REVIEW VOL. 14, No. 5 is put forward that, in áll three çases, the contact R175: J. L. H. J on kere-The angular distribution. between two Iáyers with charge carriers of opposite of the secondary electrons of nickel (Philips sign is the cause of rectification. Res. Rep',6, 372-387, 1951, No. 5).

R 173: J. C. Er an cke n and R. Dorrestein: The common equipment for measuring secondary Paraxial image formation in the "magnetic" emission (disc-shaped or spherical electrode within image iconosçope (Philips 'Res. Rep. 6, 323- a sphere) is not suitable tq obtain. data about the 346, 1951, No. 5). angular distribution of the secondary electrons. To this aim an. electrode system with two concentric In this paper a method is described for computing spheres was constructed in order to obtain a really paraxial rays in a cathode lens placed in a magnetic radial retarding electrostatic field, with which the field. In order 'tó judge the apprcximations made, behaviour 'of the secon.dary electrons with differ~nt the well-known derivation of the ray equàtion is velöcities could be studied. The distribution of the given. The' solutions of this equation are discussed secondary electrons (slow genuine -secondary elec- and a physical interpretation is given. A special trons, secondary electrons with moderate vel~èities, èase, ,approximating the electren-optical system and rapid reflected electrons) was measured as a in the image iconoscope, is computed numerically. function of the angle of incidence an.d of the bom- The electron trajectories thus found lead to a bardment voltage of the primary electrons. The discussion. of the imaging mechanism. in these construction of the measuring tube, the method of electron-optical systems. In. some respects the measuring and the results obtained are discussed. mechanism appears to differ considerably from that in ordinary magnetic lenses. R 176: N. Warmoltz: The time lag of an ignitron R 174: G. J. Fortuin: VisU:al powcr and visibility, (Philips Res. Rep. 6, 388-4.00, 1951, No. 5). n, (Philips Hes. Rcp. 6, 347-371, 1951, No. To ignite periodically a discharge in a rectifier 5). with a mercury-pool cathode, a current is sent See R 170. This part deals with the distribution' through a semiconductive rod partly j~mersed in of the visual power and of the visual types in age the mercury. Thc time lag in starting the disc~arge groups. Further the physiological meaning of the is measured on an oscillograph when- a charged constant factors in the quantitative definition. of is switched by a thyratron onto the visual power is discussed. One of these factors igniter. This is done for igniters of widely varying represents the lowest field brightness that allows resistance in a liquid and a solid mercury cathode perception of a dark object, another represents the and in a liquid and a solid tin cathode. Also the brightness at which rod vision changes into cone influen.ce of the gas pressure in the tube is investig- vision and vice versa. ated. The results are compared with those of Visibility is the subject of the last chapter. Slepian and Ludwig and those of Dow and Three possible definitions were tested by some Power s. Finally the field emission theory of Sic p i an standards, but only the ratio between the actual and Ludwig a'ud the thermal theory of Mierdel are size of the object and the size of the smallest object discussed, whereby it turns out that the measure- perceptible under equal conditions (the so-called ments of the time lags with liquid and solid cathodes size-reduction factor) complied with our demands. fit best in the thermal theory.

ERRATUM

The colour reproduction of a painting by Giovanni Bellini on page 77 in the preceding issue of this Review is by courtesy of the Trustees of the National Gallery, London. The editors regret that owing to a mistake this statement was omitted in the subscript.