Effective applications of computer-vision techniques in packaging design concept evaluation

Abstract

In packaging design concept verification, interpretations of data obtained from test vehicles are preferably supported by physics based analysis. This is vital when correlation between the thermal/mechanical test data and results from accelerated life testing is sought in order to predict the service life of a prototype. A proposed approach that integrates experimental and analytical procedures has been devised for such analyses. A feature of the approach is the wide variety of the measured data. Other than the test parameters, the approach collects temperature and time dependent deformation data at board, component and interconnect levels. Measurements in this work include surface warpage, in-plane displacements, rigid-body rotation and strains, assembling or residual stress/strain, failure loads and strength limits, etc. From the raw data, further quantities were deduced by applying the theories of beams/plates and material constitutive relations to further enrich the database. Inferred data include surface curvature, cross-sectional bending moment and shearing force, stress, strain rate and strain energy density, etc. The data variety allows better confirmation of the consistency among test results obtained with different techniques. It also facilitates applications of different theories for life prediction and failure mode and root-cause diagnosis. The comparison of different model predictions for same problem is aimed to provide a guide to the reliability model selection and application. This paper presents the experimental and analytical procedures together with some application examples to illustrate the approach.