Testability Analysis And Fault Diagnosis Of Analog Circuits Based On Multi_Signal Features

With increasing integration of circuits and reduction of accessible test-nodes, it is more and more difficult to accomplish testability analysis and fault diagnosis. Furthermore, as for analog circuit, due to the nonlinearity, component tolerance influence, the development of testability research also processes at a slow pace. Traditional methods extract single feature on a test-node for analysis, and ambiguity sets determination is dependent on the corresponding transfer function. Because the existing method introduces incomplete information, the method based on ambiguous set owns the shortcoming of low efficiency and high error rate in the process of testability analysis. Thus, in this paper, a new method based on multiple features on one test-node is proposed for analog testability and fault diagnosis. The research content focus on 4 aspects: testability index calculation, ambiguity set determination, fault diagnosis and design of testability analysis software.1. Testability index calculation. In order to utilize the information on a test-node adequately, the research focus on the testability index calculation through collecting multiple features on a test-node. At first, the combination of fault dictionary technology and multiple features collection is presented for testability index calculation. Then, the testability index is calculated when the multi-signal model and multi-features technology are put together. Experiments indicate that these new methods get better circuit testability analysis result and can be effective in the tolerance-influencing circuit.2. Ambiguity sets determination. Firstly, features optimization is discussed based on theoretical analysis of features in both time domain and frequency domain. According to theoretical analysis, in the time domain, the feature vector composed of the peak value, peak time and steady time are near-optimal features combination. While the magnification factor and transitional zone are near-optimal features combination in frequency domain. Secondly, the relationship between the undetectable components as well as components belonging to same ambiguity set and different components’ feature functions are obtained. Experiments and other methods comparation indicate the proposed method could get the accurate result with a simple calculating procedure. At the same time, a “threshold” is given in this paper to deal with the tolerance problem, which helps us find ambiguity region of two feature functions. This region is significant for fault diagnosis of analog circuits with tolerance consideration.3. Fault diagnosis. The normal feature vector is obtained based on the information of ambiguity sets determination and stored previously. Comparing feature vector from analog circuit test with the normal feature vector, we could decide whether the circuit is good or not. If the circuit is in fault state, the fault region is determined based on the difference between the feature vector obtained presently and stored feature vectors.4. Testability analysis software. The software is built on VC++ platform. The module of calculating testability index, ambiguity sets determination and fault diagnosis has been designed. Besides, the functions including data acquisition, data processing and data storage are also established in this testability analysis software.