Modeling quality reduction of multichip module systems due to uneven fault-coverage and imperfect diagnosis

This paper proposes new analytical approaches for evaluating the quality reduction of repairable multichip module (MCM) systems due to uneven fault-coverage and imperfect diagnosis in the chips during assembly. They occur due to the procurement and quality levels of the different manufacturers of the chips. In the proposed approaches, we employ a novel Markov-chain based model which is solved analytically in O(N³) (where N is the the number of chips in the MCM) to effectively compute the quality by relating different measures, such as known-good-yield, test fault-coverage and diagnosis. Unlike a previous method which computes quality as a function of yield and even fault-coverage, our method computes quality more accurately by including the effect of diagnosis as well as the unevenness in fault-coverage at the assembly phase of MCMs