Abstract

An extensive study was performed with the aim of developing compact thermal models of a variety of electronic packages used in conduction cooled scenarios. A nonredundant set of boundary conditions suitable for generation of compact thermal models for packages cooled by conduction to the board was developed by formal mathematical principles. A design of experiments method was used to reduce this set to four conditions allowing the creation of CTMs that were independent of board and underfill characteristics. The accuracy of CTM generation by applying external resistances representative of underfill and board resistances was critically examined. The technique was found to be convenient for optimizing the model parameters on both junction temperatures and heat flows through the prime lumped areas. Detailed thermal models of about thirty components, representing thirteen different package types, were created from physical component data extracted from X-ray, SEM and high-power microscopy images. Using optimization techniques allowing constrained nonlinear global optimization, compact models of different network topologies were generated for all the packages. To optimize the thermal networks, a genetic algorithm-based commercial code was employed in a standard spreadsheet environment. It was found that for most of the packages only network topologies that included a floating node provided satisfactory accuracy for both the junction temperatures and heat flows through the prime lumped areas.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call