Fault detection and identification techniques for embedded analog circuits

The advent of multichip module (MCM) technology has enabled the integration of complex digital and analog functions into a single electronic package. Such mixed-signal systems are hard to test and debug due to the complexity of testing embedded digital circuits and also due to the presence of analog circuits and embedded MCM RF-passive components requiring analog and high-frequency test stimuli. Detection and identification of faults in these analog subsystems is necessary to assure high-quality and cost-effective MCMs.;This dissertation considers the test and diagnosis problem for classes of active and passive analog circuits. For embedded active linear circuits, it is proposed to use a built-in self-test/on-line error detection scheme to detect faults in the analog circuitry. Fault diagnosis is automatic as each embedded circuit is equipped with its own dedicated error detection system. For embedded passive components, two test techniques are proposed for detecting and identifying faults: sensitivity analysis using S-parameter measurement scheme and pole/zero analysis using Y-parameter measurement scheme. These schemes are supported by a hierarchical statistical fault modeling technique to predict behavioral-level fault manifestations from process-level statistics. Test synthesis based on this statistical modeling work is used for diagnosing of catastrophic and parametric faults in active and passive high-frequency filters.