THERMAL ANALYSIS OF COMPACT SEALED COMPUTERS

Abstract

<p>Thermal performance of a cooling system for a small sea led computer is investigated. The cooling system consisted of a heat pipe unit which transfers heat from the CPU and northbridge chip to the case wall which had vertical fin channels mounted on the outside. Natural convection and radi at ion moved the heat from the fin plate to the surroundings. An initial benchmarking experiment determined the temperature drop across each component in the system and identi fied th e areas of focus for further research.</p> <p>The heat pipe unit was se lected for initial analysis based on the benchmarking experiment. Heat pipes with sintered wick and grooved wick structures were tested at different orientations with respect to gravity. Gravity was found to cause failure of the groove wick heat pipe in certain orientations while the sintered wick was able to function adequately in any orientation.</p> <p>Carbon foam is a new material with the potential for very high thermal conductivity, 2-3 times that of copper. Based on the variable performance of heat pipes, carbon foam was investigated as a possible solution. Investigations determined carbon foam had thermal conductivity in the range 30-130 W/mK, but it had high specific thermal conductivity, over fo ur times that of copper which means it could be useful in applications that are weight sensitive.</p> <p>Literature research was conducted on the topic of rectangular channel fin perfo rmance and an analytical model was found to estimate performance for different fin geometries. A radiation shape factor equation for rectangular channel fins was also found. Together the models were used to estimate the best fin geometry to maximize natural convection and radiation heat transfer. Heat spreading in the fin plate was investigated using numerical simulation validated by the analytical models and previous studies of similar geometries.</p> <p>A system model is developed to combine the temperature drops of all components of the system to quickly determine the affect of changing one parameter on the temperature of the CPU. Notable results include the highest power CPU that can be used in the current system, the highest surrounding temperature that a 35 W CPU can run in and the size of fin plate needed to run a 65 W CPU in a 40°C environment.</p>