in the airbus,� the power converter for the air-conditioning controls is cooled by the very system it controls. the module's� base plate is mounted to the die cast housing of the turbine that circulates conditioned air throughout the cabin. heat is conducted through the housing to stator blades that are immersed in the air drawn through the turbine.�

because the� module is built on a thin ceramic substrate instead of a thick copper base plate, heat dissipated by the igbt chips and diodes must be spread into the turbine's die casting. the� die casting must be thick enough to spread the heat to the blades, and the number and pitch of the blades sized to dissipate the heat at 70 deg. c ambient and inlet temperatures. the air-conditioner manufacturer had developed a proposed configuration, but contracted with thermal consulting firm, ams technologies, ag of munich, germany to analyze its design and make recommendations.

the approach velocities in the turbine are above 50 m/s and the velocity between blades approaches the speed of sound. the heat transfer coefficient to air is close to maximum and can not be improved. therefore, the thermal resistance junction-air is governed by the heat conductance from the module to the blade surface and by the effective cooling surface area.

because a detailed simulation of the turbine's geometry would have exceeded available computer capacity, a simplified coolit model was built consisting of a segment of the unrolled annulus. all the details of the segment were modeled, including the curved stator blades.� the inaccuracy of this simplification, regardless of the die casting thickness, was estimated to be within +/- 5%.��

based on the sensor reading, the temperature increase between case and junction was calculated to be 10k.� however, coolit analysis proved that heat dissipation beneath the module was not uniform. varying the number of blades, blade pitch, etc. produced a case to junction temperature difference of up to 17 k.� if engineering had relied on a single sensor, it would have missed the hot spot. the analysis also showed that optimum heat transfer occurred when the components were positioned precisely above the blades and that even a small axial offset proved� detrimental over the entire design temperature range.

ams recommendations, which were incorporated in the final design, included moving the igbt module 15 mm axially and positioning it directly over the blades and thickening the base plate 5 mm� beneath the module.

ams also investigated the heat dissipated by the electrical windings in the core of the annulus.� initially, there was concern was that this heat might raise the igbt temperature significantly. however, the coolit analysis showed that even maximum temperature conditions only slightly increased the junction temperature of the igbt.

for more information, visit : www.daat.com or contact: [email protected].

maximum cooling occurs when chips are mounted directly over stator blades.

coolit-based power converter design flies high with airbus.