Analog Circuit Fault Diagnosis Based On Boundary Scan

Analog circuit fault diagnosis plays an important role in circuit testing field,has been the object of researchers attention. With the development of science andtechnology, circuit board has entered the large-scale and highly integrated circuitera. The circuit fault diagnosis become more and more difficult because fewer ofnodes, it is especially important that solve the problem of less nodes and obtainfault information of large integrated analog circuit efficiently and accurately.This paper proposes a method of analog circuit fault diagnosis based on boundaryscan, use physical probe for "virtual probe" instead of the traditional probe,improving the testability of circuits greatly. Using Wiener series to describe thenonlinear analog circuit, and by training the neural network to realize the circuitstate online identification, completing the circuit fault diagnosis.This paper analyzed the IEEE1149.4standard for a mixed signal test of theboundary scan technique, gave instruction in the diagnosis of analog circuitsusing boundary scan. In order to be better to apply in the simulation of boundaryscan circuit that input forward three kinds of design for testability of boundaryscan circuit on this basis based on IEEE1149.4standard, which laid thefoundation for the use of boundary scan for fault diagnosis of analog circuit.In order to solve the nonlinear problem of analog circuits, analog circuitdescribed by the Wiener functional series, study method to obtain the Wienerkernel, to extract the Wiener core of the boundary scan circuit. MATLABsimulation circuit of each state of the extraction of specific Wiener nuclear andanalyzed, after the analysis of the Wiener kernel can reflect the characteristics ofdifferent circuit under different conditions. Make the Wiener core and thecorresponding state before the three order encoding as the training sample ofneural network, the circuit state through the neural network on-line identificationtraining well, achieve the purpose of intelligent fault diagnosis. In the theoretical research and simulation at the same time, this paperintroduces the design of fault diagnosis system based on ATmega128single chip,the hardware includes JTAG controller, signal generator module, data samplingmodule, data storage module. The software design includes the system of faultdiagnosis, PC software and the generation of neural network library. Wienernuclear extract can be obtained, the nuclear characteristics and training of neuralnetwork learning and other functions. Finally, make the actual circuit include ofSTA400chips with boundary scan characteristics, to diagnose kinds of the faultstate of the circuit, the PC interface display the fault types and the number offault components, experimental results verify the feasibility, which is based onthe circuit fault diagnosis method of boundary scan and Wiener core.