Detecting stuck-open faults with stuck-at test sets

Abstract

Simulations of CMOS combinational circuits have been conducted to determine the relationship between stuck-at and stuck-open fault coverage. The results suggest that node activity is more important to stuck-open fault coverage than test length by itself. Reordering test sets so that node activity is increased resulted in increased stuck-open fault coverage. It is important to note that the reordering of the test sets requires an analysis of fault-free simulations; no fault simulations need to be done. It has been shown that all but some minimum-length test sets can easily achieve the 75% stuck-open fault coverage required by the DoD (US Department of Defense), and pseudorandom tests, which have high measures of node activity, can be expected to have over 90% stuck-open fault coverage