Research On Analog Board-level Circuit Test Techniques Using Signature Analysis

With the rapid development of modern electronic technology, the density of circuit isincreased, the package of circuit is more diverse, and the number of PCB board layers is alsoincreased. With the wide application of mixed analog digital circuit, the electronic equipmentdetection and diagnosis is more difficult, and has high maintenance costs. Under the conditionsof new technologies, researching on circuit fault detection technology is of great significance onelectronic equipment maintenance and security.This paper is under the background of a certain Department Project "XXXX electronicdevices automatic test and diagnostic system". For analog circuits which are prone to failure anddifficult to test, we have launched a research on board level circuit test technology, the mainwork and contributions are as follows:In this paper,we introduce the research of system fault detection and diagnosis, board levelcircuit test technique. We summarized analog circuit fault detection method based on signatureanalysis. In accordance with the method of fault detection process,this paper focuses on the testsignal generation, feature extraction methods and fault detection algorithm, we propose signalgeneration algorithm based on the Gauss Newton iteration and faults detection method based ontime series comparison. At last, we design a board level testing system based analog signatureanalysis, the main work of this paper includes:1. The test excitation signal generation technology based on Gauss Newton iteration. Byintroducing circuit fault sensitivity model, a test signal selection strategy is introduced, and wealso improve the Gauss Newton iterative algorithm. Based on the algorithm the test stimuligenerated have high efficiency and high fault coverage.2. For the time domain and frequency domain features can not effectively reflect the truesignal problem, a signal model based on time series is proposed. We analyze the pros and cons ofthe time series similarity measuring method, finally Euclidean distance is selected as the basis ofthe signal similarity measure. Because of Euclidean distance similarity has weakanti interference ability, dynamic distance metric algorithm is proposed, which solve the problemof time series matching effectively under interference.3. For the Tolerance characteristics of analog circuits, circuit fault detection method basedon similarity matching is proposed. Monte Carlo computer simulation is conducted under thecircuit failure, which we get the response characteristics in circuit tolerance state. Then we injectfault to get differnces between the fault state and normal state. Based on this, analog circuit faultis detected. Experiments show that this method can effectively detect common faults in analogcircuits. 4Lack of circuit schematic, the circuit under test can’t be tested in the normal way. Theanalog board level circuit test system is implemented using analog signature analysis technique.And the test system is successfully integrated into XXXX electronic equipment automatical testand diagnostic system. The actual results show that the use of this system can detect the circuitfault on the analog board level, and have a higher fault detection rate.