New Aspects of Fault Diagnosis of Nonlinear Analog Circuits

Abstract

The paper is focused on nonlinear analog circuits, with the special attention paid to circuits comprising bipolar and MOS transistors manufactured in micrometer and submicrometer technology. The problem of fault diagnosis of this class of circuits is discussed, including locating faulty elements and evaluating their parameters. The paper deals with multiple parametric fault diagnosis using the simulation after test approach as well as detection and location of single catastrophic faults, using the simulation before test approach. The discussed methods are based on diagnostic test, leading to a system of nonlinear algebraic type equations, which are not given in explicit analytical form. An important and new aspect of the fault diagnosis is finding multiple solutions of the test equation, i.e. several sets of the parameters values that meet the test. Another new problems in this area are global fault diagnosis of technological parameters in CMOS circuits fabricated in submicrometer technology and testing the circuits having multiple DC operating points. To solve these problems several methods have been recently developed, which employ different concepts and mathematical tools of nonlinear analysis. In this paper they are sketched and illustrated. All the discussed methods are based on the homotopy (continuation) idea. It is shown that various versions of homotopy and combinations of the homotopy with some other mathematical algorithms lead to very powerful tools for fault diagnosis of nonlinear analog circuits. To trace the homotopy path which allows finding multiple solutions, the simplicial method, the restart method, the theory of linear complementarity problem and Lemke’s algorithm are employed. For illustration four numerical examples are given

A method for fault diagnosis of nonlinear circuits

“…which is termed a restart homotopy equation (Zangwill and Garcia, 1981;Nazareth, 2003;Tadeusiewicz et al, 2015aTadeusiewicz et al, , 2015b. Observe that at t = t l , h l (p (l) , t l ) = 0, whereas at t = 1, h l (p, 1) = h (p, 1) = 0 has the solution that meets equation (1).…”

“…Thus, the problem of fault diagnosis is of great significance to improving reliability and safety of analog electronic circuits. Although numerous publications including books, (Gizopoulos, 2006;Kabisatpathy, et al, 2005;Sun, 2008), and journal publications, (Binu and Kariyappa, 2017;Deng and Liu, 2017;Ji and Hu, 2018;Papakostas and Hatzopoulos, 2010;Sindia et al, 2012;Tadeusiewicz et al, 2015aTadeusiewicz et al, , 2015bXie et al, 2015;Yuan et al, 2010) refer to this problem the diagnosis still heavily relies on the engineer's experience and is a difficult task, especially for the soft fault diagnosis of several parameters. The complete fault diagnosis includes three aspects as follows: detection of any fault in the circuit under test (CUT), identification of faulty elements, and determination of the parameter values relating to the faulty elements.…”

A Novel Analog Circuit Fault Diagnosis Method Based on Multi-channel 1D-Resnet and Wavelet Packet Transform

“…Electronic systems typically consist of digital circuits and analog circuits. Unlike digital circuits, analog circuits are difficult to diagnose and slow to develop due to their own continuity [6], non-linearity [7], tolerance [8], and high sensitivity to the environment. According to relevant experimental investigation, analog circuits, which account for only 20% of the electronic system, cause 80% of the electronic system faults [9,10].…”

Archives of Electrical Engineering

“…However, due to the complexity of analog circuits, it is very difficult [1]. At present, the main problems are difficulty in establishing fault models caused by continuous parameters [2], complex calculation caused by many non-linear problems [3], high difficulty in fault distinguishing caused by component tolerance [4], difficult location caused by limited test points, high sensitivity to the environment such as temperature and noise, and existence of feedback loop. Restricted by these problems, the accuracy of fault diagnosis is greatly affected.…”