FLEXIBLE REPEATER DESIGN 81 



connected with careful control of the parasitic capacitances and cou- 

 plings to insure reproducibility from unit to unit in manufacture. 



COMPONENTS 



The development of passi^•e components for use in the flexible repeater 

 presented a number of unusual problems, the most important being: (1) 

 the extreme reliability^ (2) the high degree of stability, (3) the limita- 

 tions on size and shape and, (4) an environment of constant low tempera- 

 ture. 



The repeaters for the transatlantic system contain a total of approxi- 

 mately 6,000 resistors, capacitors, inductors and transformers. If we are 

 to be 90 per cent certain of attaining the objective of 20 years service 

 without failure of any of these components, the effective average annual 

 failure rate for the components must be not more than 1 in a million. 

 To assure this degree of reliability by actual tests would require more 

 than 400 years testing on 6,000 components. Obviously some other ap- 

 proach to insure reliability is required. The most obvious avenue, that 

 of providing a large factor of safety, was not open because of space 

 limitations. 



Fortunately, with only one exception, the passive components do not 

 wear out. Thus the approach to reUability could be made by one or 

 more of the following: 



1 . The use of constructions and materials which have been proved by 

 long use, particularly in the Bell System. 



2. The use of only mechanically and chemically stable materials. 



3. The use of extreme precautions to avoid contamination by materials 

 which might promote deterioration. 



4. Special care in manufacture to insure freedom from potentially 

 hazardous defects. 



The philosophy of using only tried and proved types of components 

 dictated the use of wire wound resistors, impregnated paper and silvered 

 mica capacitors and permalloy cores for inductors and transformers. 

 While newer and, in some ways, superior materials are known, none of 

 these possessed the necessary long record of trouble-free performance. 

 In some cases, particularly in resistors, this approach resulted in more 

 difficult design problems and also in physically larger components. While 

 the ambient conditions in the repeater, i.e., low temperatures and ex- 

 treme dryness, are ideal from the standpoint of minimizing corrosion or 

 other harmful effects of a chemical nature, the materials used in the fab- 

 rication of components were nevertheless limited to those which are in- 



