Valérié Eveloy and Peter Rodgers
University of Maryland
College Park, MD 20742
School of Mechanical and Manufacturing Engineering,
Dublin City University, Ireland
Numerical predictive accuracy is assessed for board-mounted electronic component heat transfer in free convection, using a computational fluid dynamics code dedicated to the thermal analysis of electronic equipment. This is achieved by comparing numerical predictions with experimental measurements of component junction temperature and component-PCB surface temperature, measured using thermal test chips and infrared thermography, respectively. The printed circuit board (PCB) test vehicle considered is populated with fifteen 160-lead PQFP components generating a high degree of component thermal interaction. Component numerical modeling is based on vendor-specified, nominal package dimensions and material thermophysical properties. To permit both the modeling methodology applied and solver capability to be carefully evaluated, test case complexity is incremented in controlled steps, from individually to simultaneously powered component configurations. Component junction temperature is predicted overall to within ±5°C (7%) of measurement, independently of component location on the board. However, component thermal interaction is found not to be fully captured.
Complete article is available to CALCE Consortium Members.