Abstract
A computer-aided design technique to assign tolerances to system components is introduced. The essence of the method is to ac curately represent the allowable margins of the system's perfor mance functions in the percent deviation parameter space (PDPS) and to determine the point in the PDPS corresponding to the most efficient tolerance assignment. The criteria used in volve minimum cost-tolerance assignments.
The technique presented solves the tolerance assignment prob lem for a significant class of systems for which one hundred percent yield is sought and whose components are statistically independent. The most salient feature of this technique is the simple and straightforward manner in which the "curse of dimen sionality" associated with tolerance assignment problems is overcome.
A computer program has been developed; it accepts a system's topology, nominal component values, a figure-of-merit for system performance, and information relating a component cost to its tolerance. The output of the program includes the set of com ponent tolerances that will produce the least expensive design with a hundred percent yield. As an example, a rather complex low pass filter has been simulated to solve the tolerance assign ment problem. Computation time is about seven seconds on the Univac-1100 computer.
Get full access to this article
View all access options for this article.
