Too much delay fault coverage is a bad thing

Abstract

Delay fault test application via enhanced scan and skewed load techniques is shown to allow scan-based delay tests to be applied that are unrealizable in normal operation. Rather than higher coverage being a positive feature, it is shown to have negative impact on yield and designer productivity. The use of functionally justified tests is defended by both a motivating example and data from benchmark circuits. Implications on overhead, yield, timing optimization, and test debug are discussed.