Test Point Selection and Multiple Soft Faults Detection in Linear Analog Circuits Based on Multiple Frequency Measurements

KEYWORDS analog circuit – fault diagnosis – test vector –tolerance –test node – multifrequency-modified nodal analysis ABSTRACT A method to identify and select test points for analog circuit testing and to detect multiple soft faults in linear analog circuits using multiple frequency measurements is proposed in this paper. Modified nodal analysis (MNA) method is used to simulate the circuit under test (CUT) and to derive the circuit parameters. With the knowledge of circuit topology and the component values, the test vectors associated with each components of the CUT are derived. The group of potentially faulty components, suitable test nodes and diagnosis variables for testing are identified and selected based on the test vectors. To solve the component tolerance issue in analog circuit testing, fault detection is performed based on the diagnosis variables obtained from multiple frequency measurements for fault free and faulty conditions of the CUT. Effectiveness of the proposed method is validated through the results obtained from simulation of benchmark circuits. 1 INTRODUCTION Analog circuit test process involves in developing methodolo-gies to detect the variation in component value or open and short circuits called faults which leads to variation in system performance. Analog circuit faults are classified as soft faults or parametric faults and hard or catastrophic faults. Soft or parametric faults cause variation in system performance and are hard to detect whereas hard or catastrophic faults cause complete variation in system performance. The factors such as nonlinearity of circuit components, tolerance and the number of test nodes to locate the faulty elements, limit the development of standardized methods for testing. Different methods have been proposed to detect multiple soft faults in analog circuits. In [1], test vector based method is proposed to identify parametric or soft faults in linear analog circuits. Multiple parametric faults are identified based on thresholding technique. A two-dimensional fault model is proposed in [2] to simplify the algorithms for test point selection and potential fault simulations, which are primary difficulties in fault diagnosis of analog circuits. Further to reduce the process of identification of faults location, a 3D complex space is proposed to achieve better fault detection ratio against measurement error and parametric tolerance. In [3], mathematical model based on normalization algorithm to reduce the dimension of the fault samples and to improve the accuracy and efficiency of fault diagnosis is proposed. A method to locate faulty components with lesser test points and test …