A comparison of methods for supply current analysis

It is demonstrated that AC supply current analysis can be an effective method for testing complex digital devices in a technology other than CMOS. Experiments using a microprocessor demonstrate the effectiveness of this approach and its potential for reliability analysis and mixed-signal testing. Results from two different methods of waveform analysis are presented. The application of statistical signal detection methods is shown to provide superior fault detection, compared to AR (autoregressive) modeling. Because the signature is analyzed as a continuous-time signal and no assumptions are made regarding waveform shape or fault effects, it is believed that this method may be applied to the testing of both digital and analog devices, at various levels of integration. An additional benefit may be the capability of detecting minor current aberrations, resulting from activated failure mechanisms prior to functional failure