248 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1957 



under all working conditions.'^ Factors involved in assessing these mar- 

 gins and in planning the equalization and level diagram for the s^^stem 

 are as follows: 



(a) Temperature. — The final assumed sea-bottom temperature was 

 2.3°C, with a maximum annual variation of ±3°C. The maximum change 

 in attenuation might therefore be ±4 db at 552 kc. The land section 

 change would be ±3 db at 552 kc due to a possible ±10°C change on a 

 mean of 7.5°C. The effect of these seasonal changes would be reduced 

 by the provision of manually adjusted equalization at Clarenville, 

 Terrenceville and Sydney Mines. 



The repeaters show a small change in gain (less than 0.05 db) during 

 the warming-up period after energization, but the effect of ambient- 

 temperature change is negligible. 



(6) Repeater spacing. — The repeater-section cable lengths were to be 

 cut in the cable factory such that the expected attenuation at 552 kc 

 when laid at the presumed mean annual temperature of the location 

 should be 60.0 db. An anticipated decrease in attenuation of 1.42 per 

 cent at 552 kr was assumed when laid. The assumed mean annual tem- 

 perature of sections of the route varied between 1.7 and 4.0°C. Tempera- 

 ture corrections employed an attenuation coefficient at 552 kc of -f-O.lG 

 per cent per degree centigrade. It was expected that the total error at 

 552 kc after laying seven repeaters would not exceed 1.5 db, and this 

 could be largely corrected as explained in (c) . 



(c) Cable Characteristics. — The cable equalization built into the re- 

 peater was based on a cable attenuation characteristic which was later 

 discovered to be appreciably different from the laid characteristic. Cut- 

 ting the cable as described in (b) overcomes this difficulty at 552 kc, 

 where the signal/noise ratio is at a minimum. The new shape of the 

 characteristic, however, indicated that at about 100 kc the error would 

 reach 7 db on the complete route. To reduce this deviation it was de- 

 cided to introduce a submerged equalizer in the middle of the sea section 

 to correct for half this error and to insert in each of the four-wire paths 

 of the transmit and receive equipments equalization for one-quarter of 

 this error. There is an appreciable signal/noise margin in hand at this 

 frequency, so that the system would not be degraded below noise speci- 

 fication b}^ these equalizer networks. 



It was also decided that the splice at the equalizer which would con- 

 nect the halves of the link together should not be completed until after 

 the laying operation had commenced. An excess length of cable was pro- 

 vided on the equalizer tail, and this could be cut at a position indicated 

 by measurements taken during the laying of the first half-section so that 



