Thermal uniformity of packaging multiple light-emitting diodes embedded in aluminum-core printed circuit boards

The gallium nitride (GaN) based white light emitting diode (LED) luminaire system presents a pathway to enhance efficacy, functionality, and the possibility to cut down pollution from traditional lamps. Much effort has been done to improve its performance/price ratio to penetrate the market, including designing and optimizing the LED package structure. Multichips embedded on the aluminum core printed circuit boards (Al-PCB) architecture is one of the features of the LED package method, along with low thermal resistance, low price, and easier development compared with luminaire systems. However, the thermal resistance of the packaged LED varies with the stochastic parameters of the LED packing process, thus posing challenges in packing cost and reliability. In the present paper, the thermal uniformity of packaging the LED array embedded on Al-PCB technology is explored using the Bayesian probability theory. The stochastic thermal characteristics of Al-PCB are investigated by an active infrared imager, and its thermal time constant matrix-based probability is suggested to estimate uniformity. A package design procedure, involving multi-physical finite element simulation, Al-PCB infrared checking, and probability estimation, is proposed. Finally, a 12 LED array for MR-16 lamp is demonstrated. The results show that the thermal uniformity of Al-PCB plays a significant role in packaging multiple LED embedded on Al-PCB, which can be physically indicated by the thermal time constant matrix-based probability. The suggested package procedure sheds light on solving the cost-related thermal uniformity of the LED array. In addition, a proposal to further reduce cost by replacing Au wire with Ag bonding wire is not deemed to be possible based on scanning electron microscopy structure analysis. (C) 2012 Elsevier Ltd. All rights reserved.