ECONOMICS OF TKLKPHOXK HKT.W A PPIJCATIOXS 253 



most economical, over-all. The problem will first be discussed in a gen- 

 (n-alizod form, and the moans \'ov <);ottinji- the economy will then be re- 

 \'ic\ved. 



G.l GENERAL DISCUSSION OF CODING 



If enough information were available to the circuit designer, he should 

 be able to choose a relay for any particular application by considering 



(a) The penalties in performance (expressed in terms of first cost) re- 

 sulting from having only a certain limited number of available combina- 

 tions of relays, as against complete flexibility, i.e., the penalties due to 

 standardizing, and 



(b) The penalties in first cost resulting from many variations of a 

 basic type, as compared with one single standard type, i.e., the penalties 

 due to not standardizing. 



By weighing both the penalties and the advantages of standardization 

 in each case, they may be maintained in approximate balance, and give 

 the economical number of codes. 



In the development of an entire new switching system using relays, 

 there will be approximate ideas on the number of relays needed in the 

 system, and on the number of kinds of relays required to fairly well 

 satisfy the circuit functions. Based on past experience, the statistical 

 distribution of kinds of relays for the various uses in relation to annual 

 demand may also be approximated. The number of codes provided will 

 determine the number of cells into which the statistical distribution is 

 divided, and also the relative annual demand for each. In the extreme, 

 only one code might be provided. It would have the advantage of very 

 large lot-size, but also it would have to do the hardest as well as the 

 easiest function. Such a requirement could lead to such absurdities as 

 requiring that all relays have three windings, twelve transfers, best grade 

 of contact metal, and the like; or to a greatly increased number of simpler 

 relays. In addition, with but little flexibility in the design, there would 

 be the added performance disadvantages due to imperfectly matching 

 certain needs for power economy, speed, and the like. Hence, there is a 

 large performance disadvantage, if but one kind of relay were provided. 



At the other extreme would be cell divisions of kinds of relays to give 

 the perfect match for every circuit use. Then, with no spare contacts, 

 the best possible power economy, and every other condition at its ideal 

 value, no performance penalty at all would be entailed. But every code 

 increase would further subdivide the manufacture, till the lot-size penal- 

 ties grew excessive. 



These variations are shown by the curves in Fig. 1, and it is their sum. 



