基于动态电源电流测试的模拟电路故障诊断

动态电源电流测试对模拟集成电路故障诊断十分有效。采用斜坡电压源代替传统的恒定直流电压,从电源电流波形采样训练神经网络的数据并建立故障字典。利用小波变换具有同时在时-频域分析信号、大量压缩数据的属性,对采样信号进行小波包分解,提取故障特征来训练神经网络,简化了网络结构、提高了训练速度。实验结果表明,该方法能够实现快速故障检测及定位,具有准确率高的特点。

Fault diagnosis in analog circuits based on dynamic supply current testing

Dynamic supply current (I_(dd)) testing is very useful for analog IC's fault diagnosis. A ramp signal at the supply is used instead of a conventional constant DC signal. For training neural networks (NN), sampled data from supply current waveform is used to build the fault dictionary. Considering good properties of wavelet transform, such as analyzing signals in both time and frequency domains simultaneously, and compressing data massively. Fault features of the sampled signal are extracted by wavelet packet decomposition, which simplifies NN architecture and shorten the training time. Simulation results illustrate that this method is feasible and has many powerful features, such as diagnosing and locating faults quickly and exactly.