Errors in testing
Richard Hays Williams, Charles F. Hawkins · 2002
Previous testing error models with VLSI chips are extended to include the effects in the testing process that cause good products to fail. The probability of this error (an error conventionally designated as Type I) is combined with the yield of the manufacturing process and the test coverage of the testing process to find a theoretical expression for the fraction of VLSI chips which have been failed erroneously by a testing process. Practical examples are given to illustrate the magnitude of the testing error. These theoretical results are consistent with reports that 30% to 70% of VLSI chips failed by a test process are, in fact, good.>