1022 THE BELL SYSTEM TECHNICAL JOURNAL, SEPTEMBER, 1953 



quate from the standpoint of echo, crosstalk, etc. The general procedure 

 is as follows : 



1. When a toll circuit is switched to another toll circuit at both ends 

 work it at a loss which is called 'Via net loss" (VNL). 



2. When the circuit is switched to another at one end only (the other 

 end being at the point of origin or destination of the call) work it at a 

 loss higher than VNL by an amount which we shall call ">S" C'>S" being 

 a generic term derived from the fact that it may be associated with 

 switching pads — usually called "S'' pads). 



3. When the circuit is used by itself (i.e., the origin and destination of 

 the calls are at the ends of the circuit) increase its loss by "S'' again — 

 that is, work it at VNL plus 2S. This is known as "terminal net loss" 

 (TNL). 



Via net loss is, of course, to be the "lowest loss practicable" referred 

 to above, and the next step is to establish methods of deriving VNL 

 and of selecting the best value for "S'\ 



Since it would be a very complicated process to work simultaneously 

 with all four of the limiting factors mentioned above, (echo, crosstalk, 

 singing, and noise), the practical approach has been to select one of 

 them as the basis of design and then check the results against the other 

 three, modifying the final solution as necessary so that all four are kept 

 under control. Since long experience indicates that echo is likely to be 

 the most difficult and complex factor to control, it has been used as the 

 starting point in the solution of the problem. As will be evident later, 

 there are a large number of solutions possible from the echo standpoint 

 and the one which has been selected has been affected to a considerable 

 extent by the other factors. 



"The next part of the material in this paper is, therefore, devoted to 

 an analysis of circuit design from the echo standpoint. 



DETERMINING LOWEST PRACTICABLE CIRCUIT LOSS FROM ECHO STAND- 

 POINT 



The over-all objective is to have practically no cases in which objec- 

 tionable echo will be observed by customers. 



If circuits could be precisely adjusted to the requirements in each 

 different connection the probability of echo would be the same on all 

 connections and the computations would have been carried out on the 

 basis of a very small probability — say, 1 in 10,000. However, losses 

 can be changed only in discrete steps (S) so that in a very large propor- 

 tion of cases the losses will be higher than are theoretically necessary. 



