An algorithm for diagnostics with signature analyzer

An algorithm for fault diagnosis that makes use of the information from a faulty signature is presented. The idea is to search the likely error locations before the tests are performed. The method reduces the number of tests required to diagnose the errors with the probability of aliasing. Such probability is always smaller than that of error detection in signature analysis. When matching tests are difficult or impossible, the method provides an estimate of where errors that caused the incorrect signature might have occurred. Also, the case of 'don't cares' at the input sequence of signature analysis is discussed. The algorithm can be readily implemented in software with the faulty signature as the only input variable. The proposed fault diagnostic scheme has an advantage over exhaustive testing in that it leads to fault isolation in a homing-in manner.<>