The effectiveness of different test sets for PLAs

Abstract

It has been theoretically demonstrated that the single stuck-at fault model for a PLA does not cover as many faults as the single crosspoint model. What has not been demonstrated is the real relative effectiveness of test sets generated using these models. This paper presents the results of a study involving presenting a number of test sets to fabricated PLAs to determine their effectiveness. The test sets included weighted random patterns, of particular interest owing to PLAs being random resistant. Details are given of a method to generate weights, taking into account a PLA's structure.