Diagnosis by Image Recovery: Finding Mixed Multiple Timing Faults in a Scan Chain

In this brief, we present a robust new paradigm for diagnosing a scan chain with multiple faults that could have different fault types. As compared to previous methods, the major advantage of ours is the ability to not only target mixed multiple types of timing faults in the same scan chain but also tolerate non-ideal conditions, e.g., when these faults only manifest themselves intermittently. Unlike the previous matching-based algorithms, we formulate the diagnosis problem as an image recovery process featuring a dynamic windowing technique and a running sequence handling technique. Experimental results on a number of real designs show that this paradigm can successfully deal with some situations beyond existing methods.