In the presence of multiple delay faults, automated diagnostic procedures that make a single fault assumption may give an incorrect diagnosis. In this paper, a systematic approach is proposed for delay fault diagnosis under a multiple fault assumption. Information from the failing test vectors are used to construct a list of single and multiple fault suspects that may have caused all of the observed faulty response. The list of suspects is then pruned and ranked in a novel way by using information from the passing test vectors combined with static timing information.
Jayabrata Ghosh-Dastidar, Nur A. Touba