Tolerance Of The Analog Circuit Testing And Diagnosis Of Complex Faults

Since the 1980's, fault diagnosis of analog circuits, other than network analysis and network synthesis, has become the third main branch in the network theory. The chief task in the fault diagnosis is to locate the fault components and determine their parameters in accordance with the excitation signal and the response of a part of the network.There have been quite a lot of achievements in the research of the fault diagnosis of analog circuits. The existing diagnosis methods are basically in two categories, namely, the SBT (simulation before test) and the SAT (simulation after test). The fault dictionary method is the representative of the former, while the component parameter identification method and the fault verification method are the representatives of the latter. Since the fault dictionary method and the fault verification method are not feasible to circuits with tolerance, they cannot be applied to the fault diagnosis of practical circuits that are actually with tolerance always. Though the component parameter identification method is applicable to circuits with tolerance, it is difficult to establish and solve its diagnosis equation. Therefore, the feasibility of this method is much affected.This thesis mainly deals with tests and diagnosis of complex faults in analog circuits with tolerance. It keeps the merits of the existing component parameter identification method that it is applicable to circuits with tolerance, while improves the drawback of this method that it is difficult to establish its fault diagnosis equation with a high nonlinearity. A fault diagnosis equation based on node voltage equations, which can be established with much ease, is proposed. With the inaccessible node voltages taken as the assistant unknowns, the nonlinearity of the equation is much reduced. The improved Newton-Raphson iteration algorithm speeds up the solution procedure. Fault diagnosis of linear analog circuits as the basic of other work is mainly researched, while large-scale linear circuits and nonlinear circuits are also analyzed and discussed. Decomposition diagnosis method is proposed to large-scale linear circuits, which transforms a large-scale diagnosis problem into some small-scale diagnosis problems, and the forenamed method to linear circuits is feasible to solve the problem. Nonlinear circuits include nonlinear static circuits and nonlinear dynamic circuits. DC excitation method is proposed to nonlinear static circuits, which transforms the nonlinear resistances into DC equivalent components, and the circuits degenerate to linear static circuits. Two steps method is proposed to nonlinear dynamic circuits. By add different test excitation signals, the nonlinear dynamic circuit is decomposed to two simple problems, namely, fault diagnosis of nonlinear static circuits and fault diagnosis of linear dynamic circuits. Finally, practical examples validate the feasibility and effectiveness of the proposed approach.The research on fault diagnosis of analog circuits is a frontier of science. There are still some deficiencies of the methods proposed in the thesis, for example the theory about fault diagnosis of nonlinear circuits is still to be deeply researched and improved.