The Application of Designed Experiments 

 to the Card Translator 



By C. B. BROWN and M. E. TERRY 



(Manuscript received October 27,^1953) 



In the course of developmeyit of the card translator for use in the No. 

 JfA toll crossbar system it was necessary to evaluate in detail the perform- 

 ance characteristics of all phases of the translator operation. One of the 

 most important phases was that of the action of the translator cards. Since 

 the action of these cards is controlled by the simultaneous influence of many 

 independent variables a study of the card action was made using statistically 

 designed experiments. This study made use of Graeco-Latin Square and 

 Factorial Designs; herewith is presented a detailed description of their 

 application to the problem. Also the method of analysis of the data and 

 resulting conclusions are described in detail. 



I. INTRODUCTION 



111 order to permit circuit designers to use the full capabilities of card 

 translators' it has been necessary to conduct tests to determine (1) the 

 time intervals required to drop the cards into reading positions and (2) 

 the maximum number of cards which a translator can operate reliably. 

 It was also desired to establish the maximum and minimimi number of 

 cards that could be operated efficiently in a card translator. Early esti- 

 mates indicated a machine capacity of 1 ,000 cards but if a higher capacity 

 could be demonstrated, a considerable cost saving in the 4A system 

 would result. Since some differences in card drop time known to exist 

 are due to card position and machine loading this study also was to 

 consider whether or not any special loading instructions were necessary 

 or desirable for field use of card translators. As the study commenced 

 the Western Electric Company Installation Department requested in- 

 formation as to the need for leveling card translators on installation and 

 it was felt that such infoiTnation could readily l)e obtained in the course 

 of this study. In investigating the effect of the many variables related to 

 card drop time, a consideration was given to the possibility of the need 

 for any design changes. 



369 



