SYSTEM DESIGN — NEWFOUXDLAND-XOVA SCOTIA LINK 237 



in the nature of an art than a science and usually requires an intimate 

 knowledge of the behaviour and peculiarities of the particular cable 

 concerned. While the problem appears at first sight to be simple it is 

 complicated by: 



(a) The presence of ground-potential differences along the cable, some- 

 times amounting to hundreds of volts; these vary with time. 



(6) Electrolytic e.m.f. generated when the center conductor is exposed 

 to sea water. 



(c) Absorption effects in the dielectric of the cable. 



When repeaters are added the position is further complicated by: 



(d) The lumped resistance of the repeaters, which is current-dependent 

 and exceeds the cable resistance. 



(e) The lumped capacitance of the repeaters with an absorption char- 

 acteristic which differs from that of the cable. 



It is a great advantage of both-way transmission over one cable that, 

 by introducing some form of freciuency changer at each repeater, signals 

 outgoing in one direction can be looped back to the sending terminal. 

 There have been a number of developments based on this principle, and 

 in the Clarenville-Sydney Mines link two methods are available for use. 

 Of these, the so-called 'loop-gain' method uses steady tones and depends 

 on selective frequency measurements to discriminate between repeaters ; 

 the second is a pulse method in which repeaters are identified on the 

 basis of loop transmission time. 



The use of these methods under fault conditions depends on the pos- 

 sibility of keeping the repeaters energized. Work is in progress to de- 

 velop methods of fault location which are of general application and do 

 not depend on the activity of the repeaters, but these are outside the 

 scope of the present paper. 



Loop-Gain Method. 



In the loop-gain method, the frequency changer in the repeater takes 

 the form of a frequency doubler and each repeater is identified uniquely 

 by one of a group of frequencies spaced at 120 cycles and located im- 

 mediately above the lower main transmission band in the frequency 

 range 260-264 kc. Since the frequency changing is in an upward sense, 

 the measuring terminal is Sydney Mines, which transmits the lower 

 band. On the Clarenville side of the directional filters in each repeater 

 is connected, via series resistors, a crystal filter accepting the test fre- 

 quenc3^ appropriate to the repeater [see Fig. 12(a)]; this frequency is 

 doubled, filtered and returned to the repeater at the same point at which 



