Machine Learning Techniques for System Level Test and Diagnosis

The gap between working silicon and a working board/system is becoming more significant and problematic as technology scales and complexity grows. The result of this increasing gap is failures at board and system level that cannot be duplicated at the component level. These failures are most often referred to as “No Trouble Found” (NTF). The result of these NTFs can range from higher manufacturing cost, and failure to get the product out of the door. The problem will only get worse as technology scales and will be compounded as new packaging techniques such 2.5D/3D extend and expand Moore’s law. The speaker will describe the nature of this problem and present recent advances in using machine-learning techniques to facilitate accurate and rapid board repair. In particular, the speaker will describe how techniques such as artificial neural networks, support-vector machines, decision trees, and information-theoretic analysis can be used for addressing root cause identification in boards and systems. He will present Imputation methods that allow us to carry out reasoning using incomplete data. Methods for knowledge discovery and knowledge transfer for early-stage diagnosis will also be presented. The presentation will include a number of case studies using telecom boards from high-volume manufacturing.