Reliability Prediction of PBGA Solder Joint under High Temperature Torsional Composite Loading Based on BP Neural Network

Le résumé fourni par la source

The electronic products are inevitably subjected to high temperature and torsional load at the same time in practical work, which brings more rigorous test to the reliability of electronic products in service. In this paper, the common PBGA chip is chosen as the research object, and its geometric model is built up. The chip size is 25mm × 25mm my × 1.16 mm, and it has 18 × 18 solder joints array. There are 324 solder joints in total. When conducting high-temperature torsional composite loading analysis on PBGA solder joints, the direct coupling method is used. Apply a displacement load of 0.5mm in the same size but opposite direction to the four corner positions of the PCB substrate in a high-temperature environment of 150 °C. After solving and analyzing, obtain the maximum van Mises stress and strain of the solder joint. Next, MATLAB software establishes a BP neural network prediction model for driving quantity items, and predicts the stress values of PBGA solder joints under composite loading. 40 combinations of solder joint morphology parameters are designed. After 1000 iterations of training; The correlation index R of the input and output is 0.96757, with a maximum error of 4.04% and a minimum error of 0.131%. The use of neural networks can effectively predict the stress value of PBGA solder joints under high-temperature torsional composite loading.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.