Abstract
For years, there has been a ongoing debate in the industry regarding the effectiveness and costs associated with functional versus DFT-oriented testing and ATE. To answer important questions that arise from the functional vs. DFT debate, we consider actual production data analysis to evaluate the value of functional and DFT tests. Production test data from 10,000 parts randomly sampled from over 1 million total datalogs of the Motorola MPC7410 microprocessor, a high-volume design, is used to evaluate the value of ATE features in terms of test escape rates across different tests. We also examine the value of features that cannot necessarily be quantified in terms of test escape rates. This work will help to provide some insight into how to lower the cost of test by making the right tradeoffs in terms of ATE features without compromising overall test coverage and quality.