Stuck-Fault Tests vs. Actual Defects
We've known for a long time that stuck-at faults are not a particularly good model of real defects. However they have been effective for ATPG tools and fault simulation, so except for adding the delay fault model, we've been stuck on them. How can we improve the chance that a test created using the stuck-at model detects unmodelled defects? This paper, based on extensive experiments with real, if small, chips, found that a test that detects faults multiple times catches defective parts that a test with 100% stuck at coverage misses. It has changed the way we think about faults.
| Go to Paper | Back to Anniversary Page | ITC Home |