254 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1954 



as seen in the curve marked c, that should decide where best to work. 

 In planning a whole new central office which may use from 40,000 to 

 80,000 relays, and a new design of relay whose manufacturing costs are 

 not yet well established, the data to give numerical values for Fig. 1 

 can only be quite tentative, and the results can only serve as guides. 

 Yet such Avork has been most useful in recent effort on a redesigned No. 

 5 crossbar system to use the wire spring rela3^ Calculations could be 

 made with enough assurance to determine the general shape of the total 

 penalty curve (Curve c), and showed an optimum value for kinds of re- 

 lays of about 200. Even more important was the indication that the 

 curve w^as extremely flat in its optimum region, giving but a small 

 change in the total penalty if an error, even as great as 2 to 1, were 

 made in the number of codes to be used. 



6.2 THE DETAILED PROCEDURES OF CODE SELECTION 



Background information as gained above helps in guiding the early 

 steps in picking codes of relays for the job, but there is a better method 

 of picking codes when the circuit designer is actually down to cases. 

 By following the steps below, the optimum number of codes will auto- 

 matically result. 



The first step in the coding program is to pick a list of basic codes. 

 This can be done quite arbitrarily, initially, and should be based on 

 general knowledge of types of circuit functions, together with specific 

 knowledge of certain uses with very large demands. For example, certain 

 coils will be needed to cover the range from very fast operation with no 

 concern for current drain, to slow release and emphasis on econom}^ of 

 power; there will be single and multiple windings as dictated b}^ circuit 

 functions. Also, certain arrangements of contacts may be assumed to 

 span the range of needs from a single make up to the capacity of the 

 design — twelve transfers or twenty-four makes in the case of the new 

 wdre spring relay. Further, certain combinations of springs and coils will 

 be evident at once as ideally suited to particular large-demand uses, and 

 these should be on the original list. Soon the likel}^ minimum demand 

 for each will be quite e\ddent. Such a list of tentative codes and demands 

 is the first step in the coding scheme. 



Now, as each circuit application arises, this basic list maj^ be consulted. 

 For some cases, an available code will be ideal, but for many others a new 

 arrangement may be desired. In each such case a simple calculation will 

 show the economical choice between a new code or the old code with 

 more features than are really needed. Steps in the calculations are given 



