Skip to Main Content
In our approach, we use ATE (automatic test equipment) to teach a neural network (NN) to correctly classify a set of worst case input patterns with respect to the maximum instantaneous current. This can be thought of as learning a behavior of chip power consumption change due to different input patterns applied. We then further optimize this set of (NN) worst case patterns using a genetic algorithm (GA). The final set of worst case patterns can efficiently identify a defective or weakness due to power supply noise as well as locate the defect or weakness within the design. To the best of our knowledge, this is the first NN learning and GA self-optimization ATE-based approach for practical application in silicon analysis automation. Our practical experimental results demonstrate the enlarged fault coverage obtained with this approach.