This paper summarizes an exploration of design strategies for a microproces
sor-based adaptive thermal control system for heat-dissipating computer CPU
modules. The thermal control system contains a cold plate with an air-cool
ed base mode and a thermo-electric heat pump element that can boost cooling
performance when needed. The microprocessor programming includes models of
the heat transfer in the cold plate and the thermal performance of the the
rmo-electric heat pump and the fan coolers. The microprocessor is programme
d to compare the thermal control system model predictions to the existing o
perating conditions. The results of this comparison are used to facilitate
control of the cooling system and to provide fault detection during its ope
ration. This paper reports the progress of an ongoing study that will explo
re different ways that the computational capability of the processor can be
used to minimize power consumption while maintaining adequate processor te
mperature control over broadly varying operating conditions. The investigat
ion also explores the merits of different strategies for incorporating faul
t detection features into the microprocessor programming to enhance the rel
iability and robustness of the system.