Deterministic pattern generation for weighted random pattern testing

Weighted random pattern testing is now widely accepted as a very economic way for external testing as well as for implementing a built-in self-test (BIST) scheme. The weights may be computed either by structural analysis or by extracting the required information from a precomputed deterministic test set. In this paper, we present a method for generating deterministic test patterns which can easily be transformed into weight sets. These test patterns contain only minimal redundant information such that the weight generation process is not biased, and the patterns are grouped such that the conflicts within a group are minimized. The quality of the weight sets obtained this way is superior to the approaches published so far with respect to a small number of weights and weighted patterns, and a complete fault coverage for all the ISCAS-85 and ISCAS-89 benchmark circuits