Data analytics approach for optimal qualification testing of electronic components

In electronics manufacturing, required quality of electronic components and parts is ensured through qualification testing using standards and user-defined requirements. The challenge for the industry is that product qualification testing is time-consuming and comes at a substantial cost. The work reported with this paper focus on the development and demonstration of a novel approach that can support “smart qualification testing” by using data analytics and data-driven prognostics modelling. Data analytics approach is developed and applied to historical qualification test datasets for an electronic module (Device under Test, DUT). The qualification spec involves a series of sequentially performed electrical and functional parameter tests on the DUTs. Data analytics is used to identify the tests that are sensitive to pending failure as well as to cross-evaluate the similarity in measurements between all tests, thus generating also knowledge on potentially redundant tests. The capability of data-driven prognostics modelling, using machine learning techniques and available historical qualification datasets, is also investigated. The results obtained from the study showed that predictive models developed from the identified so-called “sensitive to pending failure” tests feature superior performance compared with conventional, as measured, use of the test data. This work is both novel and original because at present, to the best knowledge of the authors, no similar predictive analytics methodology for qualification test time reduction (respectively cost reduction) is used in the electronics industry.