Model-based test for analog integrated circuits

The basic idea of ´model-based test´ is to compute multiple test metrics from a core set of stimulus/response experiments used to fit behavioral models for the test metrics of interest. In other words, a few measurements are done to collect data used to create a behavioral model from which a number of test metrics are computed for example by simulation. One objective of model-based test is the reduction of complexity and cost of measurement systems used in test of analog integrated circuits. The method described here seeks to replace a conventional measurement system with a single broad band source and receiver which would be adequate for multiple tests, when used in conjunction with appropriate models. Application of the method to a digital receiver, including comparison of predicted and measured results, is described.