Mixed-Signal Circuit Testing Based On Signature Distance And Performance Analysis

With the development of technology, especially micro-electronics, the density of LSI and VLSI is increasing. This makes the test of ICs more difficult, and the application of digital and analog mixed-signal circuits also makes the test much more difficult. In the same way, the research of the fault dectection of mixed-signal circuits is always a difficulty, and it is also one of the important research aspects. It is urgent for us to use novel methods and novel theories to solve this problem.The pseudorandom testing technique is a new and efficient testing method for Linear Time-Invariant(LTI) analog compoments, which analyze circuit in signature space for classifying faulty-free and faulty circuit instead of directly measuring performance parameters. The signature space is get from analyzing the statistical character of the input and output stochastic processes of the CUT (Circuit Under Testing).This main works are as follows:This paper uses a distance concept of every circuit to circuits set in signature space through analyzing the covariance matrix of circuits set. So we can calculate the distances of a CUT to fault-free and faulty circuits set, and can distinguish the fault-free and faulty circuit through comparing the two distances, and directly analyze the CUT in m-dimension signature space instead of selecting the signature-pair of m-dimension signature space to analyze the CUT.This paper also discusses a circuit parameter estimate method based on mixed-signal circuit performance analysis. This method can set the different performance range according to the different demand on circuit performance parameter. So this method can calculate the component value range, and it has strong practicability and flexibility. We set up a mathsmatical model based on optimization theory, and solve this problem by the mathsmatical software Matlab.