Development, validation, and application of thermal modeling for a MCM power package

Abstract

This paper addresses the differences in thermal characterization and modeling for a multi-chip package with respect to a single-chip package. Since no thermal measurement system is commercially available to power more than one device during the test, the thermal model can be validated only with a single-heat source. Wires, which are used to connect the package under the test to the measurement system, should be included in the thermal model to account for the heat sinking effect. The modeling results from the validated thermal model showed that the thermal behavior of a multi-chip package with a single heat-source does not truly represent the behavior in the operating mode. The thermal resistance difference between the measured one from heating a single device and that with three powered simultaneously might be as large as 100%. It is pointed out that the commonly adopted linear superposition method, which can be applied to thermal resistance characterization and measurement of components with multiple, independent heat sources within a package, might result in significant errors in a natural convection condition. A modified superposition method was developed in this paper, and has proved to be accurate in estimating the junction temperatures for a multi-chip module (MCM).