Detection of intermittent failures in combinational circuits

An optimal fault detection experiment is reported for detecting intermittent faults in combinational circuits. Two optimality criteria were used in designing the experiment: maximizing the probability of fault detection, and minimizing the mean testing time for a given test quality. Detection of both single and multiple intermittent faults was considered. The optimal intermittent fault detection experiment was achieved by applying the input vectors with an optimal probability distribution. It is shown that a repeated application of test inputs primarily designed to detect permanent faults might be very inefficient in detecting intermittent faults and, therefore, the use of the optimal experiment is highly important.