This paper presents a new parametric fault detection (PFD) approach for testing of linear analog circuits. It combines classified frequency-bands with amplitude weighting for fault controllability and test-points selection for fault observability. The test waveform sweeps on an applicable frequency-band instead of the whole band to stimulate parametric faults. The number of required samples is reduced, and the summation of samples from undesired bands is avoided. The test response is compacted for each band generating digital signature. The digital signature curve (DSC) is plotted for each component. A hybrid between MATLAB and PSPICE simulation is used to develop accurate worst case analysis (WCA). The relation between the classified DSC and the accurate WCA results in the enhanced PFD. It is found that the weighted sweeping-sinusoidal waveform is the best selection. The presented approach is applied to different linear analog benchmark circuits and shows the significant PFD improvement compared to previously published works.

本文提出了一种用于测试线性模拟电路的新参数故障检测 (PFD) 方法。它结合了分类频带与用于故障可控性的幅度加权和用于故障可观察性的测试点选择。测试波形扫描适用的频带而不是整个频带以激发参数故障。所需样本的数量减少了，并且避免了来自不需要的频带的样本的总和。对生成数字签名的每个频带压缩测试响应。为每个组件绘制了数字签名曲线 (DSC)。MATLAB 和 PSPICE 仿真之间的混合用于开发准确的最坏情况分析 (WCA)。分类的 DSC 和准确的 WCA 之间的关系导致增强的 PFD。发现加权扫描正弦波形是最佳选择。所提出的方法应用于不同的线性模拟基准电路，与之前发表的作品相比，它显示了显着的 PFD 改进。