论文部分内容阅读
为了降低模拟电路参数型故障的测试难度,提出了一种基于奥克塔夫(Octave)-Haar小波结构的模拟VLSI电路故障诊断方法。将测试响应经小波滤波器组完成子带滤波,随后对各子带滤波序列计算故障子序列与正常子序列的互相关系数,对每一故障,可确定出互相关系数最小的子带,并将此数值作为该故障的特征,对应子带的正常响应序列的自相关系数作为无故障特征,用故障特征与正常特征的对比可诊断故障。对国际标准电路的实验表明,该方法对参数型故障的诊断已具有高分辨率。
In order to reduce the difficulty of testing parametric faults in analog circuits, a fault diagnosis method based on Octave-Haar wavelet for analog VLSI circuits is proposed. The test responses are sub-band filtered by the wavelet filter bank, and then the cross-correlation coefficients of the fault sub-sequence and the normal sub-sequence are calculated for each sub-band filtering sequence. For each fault, the sub-band with the smallest cross-correlation coefficient can be determined Taking this value as the feature of the fault and the autocorrelation coefficient of the normal response sequence corresponding to the subband as the faultless feature, the fault can be diagnosed by comparing the fault feature with the normal feature. Experiments on international standard circuits show that this method already has a high resolution for the diagnosis of parametric faults.