1 Introduction
Extensive use of design for testability (DFT) techniques, including scan and test points, and non-nominal test methods such as low-voltage test and test [1], [2] lead to overtesting, i.e., the IC is demonstrated to fail, but under conditions which cannot occur in its normal operation mode. One reason for overtesting is the presence of latent defects, which are too small to cause a failure under nominal conditions or logically redundant. A further reason is the elevated level of IR drop and crosstalk effects which is caused by atypical power consumption during test that does not correspond to the power consumption profile in normal operation [3]. Last but not least, behavior which does not contradict the specification could be classified as “faulty behavior” by the test process if design tricks such as cycle stealing are employed.