Fault collapsing of multi-conditional faults

Numerous new multi-conditional fault models have been proposed in the last years. In combination with the increasing complexity of today´s designs these new fault models cause a tremendous increases of the ATPG-runtime. In this paper we present a novel fault collapsing scheme for multi-conditional faults. The objective is to significantly reduce the fault set and hence reduce runtime for ATPG and fault diagnosis. The collapsing technique consists of three individual collapsing stages, which are individually discussed and evaluated. Additionally we provide an extensive set of experimental results including runtimes of a state-of-the-art ATPG-tool applied on a set of large industrial designs. We will demonstrate that the achieved reduction of up to 48% of the number of faults also reduces the ATPG runtime significantly.