THERMAL PERFORMANCE PREDICTION OF PLASTICS BALL GRID ARRAY (PBGA) USING ARTIFICIAL NEURAL NETWORK

Abstract

Artificial Neural Network (ANN) based on feed-forward backpropagation model is used to predict junction temperature in PBG A package. The limited results obtained from FEM (using IDEAS software) are used to train the neural network. The effect of source power, substrate and mold compound thermal conductivity, die size, substrate thickness and air velocity on junction temperature and thermal resistance has been investigated using ANN. The predicted junction temperature using ANN agrees closely with the prediction from FEM. ANN method takes a small fraction of the time and effort compared to that required by HEM for prediction.