Testing linear and non-linear analog circuits using moment generating functions

Abstract

Circuit under test (CUT) is treated as a transformation on the probability density function of its input excitation, which is, a continuous random variable (RV) of gaussian probability distribution. Probability moments of the output, which is now the transformed RV, is used as a metric for testing catastrophic and parametric faults in circuit components that make up the CUT. Use of probability moments as circuit test metric with white noise excitation as input addresses three important problems faced in analog circuit test, namely: 1) Reduces complexity of input signal design, 2) Increases resolution of fault detection, and 3) Reduces production test cost as it has no area overhead and marginally reduces test time. We develop the theory, test procedure and report SPICE simulation results of the proposed scheme on a benchmark elliptic filter. With the proposed scheme, we are able to detect all catastrophic faults and single parametric faults that are off from their nominal value by just over 10%. Method reported in this paper paves way for future research in circuit diagnosis, leveraging moments of the output to diagnose parametric faults in analog circuits.