38 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1957 



From a later estimate of L = 1,955 nautical miles, S calculates to be 

 36.9. The repeater design was based on this spacing in the deep sea tem- 

 perature and pressure environment. Subsequently, after better knowl- 

 edge had been obtained of the cable attenuation in deep water, the actual 

 repeater spacing for the main part of the crossing was changed to about 

 37.4 nautical miles for the eastbound (No. 1) cable and 37.6 for the west- 

 bound. Only 51 repeaters were required in each cable. 



Number of Channels 



The number of channels which could be transmitted was determined 

 by the upper and lower boundary frequencies. In this system, the bot- 

 tom frequency was established at 20 kc, primarily because of the loss 

 characteristics of the power separation filters.^ 



Preliminary studies were made to estimate the usable top frequency. 

 For a system having a fixed number of repeaters, this frequency falls 

 where the maximum permissible repeater gain equals the loss of a re- 

 peater section of the cable — which varies approximately as the square 

 root of frequency. The repeater gain is the difference between the re- 

 peater input and output levels. The minimum permissible transmission 

 level at the repeater input depends on the random noise (fluctuation 

 noise) contributed by cable and repeaters, and on the specified require- 

 ment for random noise. The maximum permissible transmission level at 

 the repeater output may depend on the modulation noise contributed 

 by the repeaters below overload, or on overload from the peaks of the 

 multi-channel signal complex. In this system, overload was found to be 

 the controlling factor. 



An important consideration was to provide enough feedback in the 

 repeater so that at the end of 20 years the accumulated gain change 

 (Mu-Beta effect) in all the repeaters would not cause the signal-to-noise 

 performance to fall outside limits. The usable feedback voltage was 

 scaled from the Havana-Key West design according to the relation that 

 this voltage varies inversely as the f power of the top frequency. Electron 

 tube aging was estimated from laboratory life tests on Key West-Havana 

 type tubes. 



Concurrently, detailed theoretical and experimental studies were be- 

 ing conducted on the transmission design of a repeater suited to trans- 

 atlantic use with the chosen type of cable, as discussed in a companion 

 paper.' Intimate acquaintance with the repeater limitations led to a de- 

 cision in 1953 to develop a system with a working spectrum of 144 kc 

 (36 channels) and a top frequency of 164 kc. Use of compandors would 

 not increase this top frequency appreciably. 



