Sizes of test sets for path delay faults using strong and weak non-robust tests

The trade-off between strong and weak non-robust tests for path delay faults considering test quality, test set size and fault coverage is discussed. It is noted that strong non-robust tests detect more combinations of delay defects. However, weak non-robust tests result in smaller test sets, and more path delay faults are detectable by weak non-robust tests. A test generation strategy that mixes strong and weak non-robust tests based on this trade-off is described. Under the proposed strategy, test generation starts by generating strong non-robust tests. When the number of faults detected by each additional test drops below a certain level, indicating that low levels of compaction are achieved, test generation switches to weak non-robust tests. The authors study appropriate switching points experimentally by using a test selection procedure to construct mixed test sets. They also consider the effects of adding power constraints on the mix of strong and weak non-robust tests.