Probabilistic fault detection and the selection of measurements for analog integrated circuits

New methods for analog fault detection and for the selection of measurements for analog testing (wafer probe or final testing) are presented. Using Bayes´ rule, the information contained in the measurement data and the information of the a priori probabilities of a circuit being fault free or faulty are converted into a posteriori probabilities and used for fault detection in analog integrated circuits, with a decision criterion that considers the statistical tolerances and mismatches of the circuit parameters. An adaptive formulation of the a priori probabilities is given that updates their values according to the results of the testing and fault detection. In addition, a systematic method is proposed for the optimal selection of the measurement components so as to minimize the probability of an erroneous test decision. Examples of DC wafer-probe testing as well as production testing using the power-supply current spectrum are given that demonstrate the effectiveness of the algorithms