Fault diagnosis for mixed-signal electronic systems
S. Chakrabarti, Sasikumar Cherubal, Abhijit Chatterjee · 1999
In this paper, we propose an efficient methodology to diagnose faults in complex mixed-signal systems. The proposed technique combines the advantages of simulation-before-test and simulation-after-test approaches. Given a faulty circuit, a hierarchical fault dictionary (a list of all faulty behaviors of the circuit, used for fault location) is used to locate the faulty component. The computational requirements during fault dictionary construction are minimized by hierarchical fault clustering and behavioral fault modeling. The resulting fault dictionary is most compact in size and contains complete diagnostic accuracy. Once the faulty component is identified, a novel parameter identification algorithm determines the values of the parameters corresponding to the faulty component. The proposed approach minimizes the off-line computation required by a fault dictionary technique as well as the online computation that is typical of a simulation-after-test technique. It is also robust to component tolerances and measurement inaccuracies, as the precise values of the parameters are determined. Finally, the proposed fault diagnosis technique can diagnose a faulty component at any level of design hierarchy, or identify the exact value of the faulty parameters, according to the intended resolution of diagnosis.