880 the bell system technical journal, july 1953 



Introduction 



In a transmission system such as the L3 coaxial, the degree to which 

 system objectives are achieved is largely dependent on the quality of the 

 amplifiers which compensate for the cable loss. To a considerable extent 

 the same statement applies to the similar flat-gain amplifiers used to 

 make up for the loss of the equalizers at various points along the route. 

 The development of an amplifier which would meet the exacting re- 

 quirements of the L3 system was in turn dependent on new develop- 

 ments in the fields of vacuum tube design and circuitry, network design 

 techniques, element and network fabrication, and statistical quality 

 control. To these new tools were added the lessons learned in years of 

 manufacture and operation of the preceding LI system. 



The importance of some of these factors can best be illustrated by 

 examining the implications of the amplifier requirements which follow 

 from the material in the companion papers on system design^ and equali- 

 zation.^ Obviously the figure of merit, modulation coefficients and life 

 of the vacuum tubes will be determining factors in setting the amount of 

 feedback that can be obtained and the signal-to-noise ratio of the system. 

 It is not so inomediately apparent that system requirements could not be 

 met with the present 4-mile repeater spacing if it were not for the use of 

 quality control at every stage of manufacture from elements, and even 

 the raw materials entering into components, to complete amplifiers. As 

 the companion papers show, however, the equalization plan of the sys- 

 tem is predicted on a degree of reproducibility of amplifier gain and delay 

 characteristics obtainable only by quality control applied at every stage 

 of the manufacturing process. 



The present equalization plan is based on the assumptions that the 

 gain of the average line amplifier will match the loss of the preceding 

 line section to within 0.15 db, and that the average amplifier gain will 

 not vary from one batch of new amplifiers to another by more than 

 about 0.06 db. Under these assumptions a system equalization plan can 

 be worked out which results in reasonable spacing between equalizers, a 

 tolerable signal-to-noise penalty due to misalignment and equalizer loss, 

 and a practicable procedure for adjusting long systems. Any gross de- 

 parture from the basic assumptions as to reproducibility of amplifiers 

 would seriously compromise these objectives, which even now are 

 achieved only by using e(iualization which requires a flat gain amplifier 

 for every four or five line amplifiers. Now it turns out that with the most 

 precise elements that can be made, the gains of individual amplifiers 

 will vary by al)out dbO.O db. We need, therefore, a tool which will permit 

 us to control the gain of the average amplifier to an order of magnitude 



