Accurate thermal characterization and optimization of high lumen LED array fixtures

High lumen light emitting diode (LED) arrays have been preferred in automotive, building, and street lighting applications due to their low energy consumption. In order to maintain the high performance of LEDs, the heat generated by the LED chips should be removed effectively and the junction temperature should be reduced. Junction temperature of the LED chips in LED arrays can be estimated by the thermal models utilizing constant thermal resistances provided in LED datasheets. However, the constant thermal resistance used in thermal simulations often neglect the tempera- ture and thickness dependent thermal conductivity of chip layers, which becomes more important in high lumen applications of LED arrays. To include this effect, a two-step multiscale thermal model will be built via finite element solver. First step will include important package features such as LED array and the heat sink and the second step will include the detailed analysis of an individual LED chip by including temperature and thickness dependent thermal conductivity of chip layers. The model will be used to analyze large and complex high lumen LED array fixtures' junction temperature sensitivity to thermal optimization steps. Today, heat sinks that provide passive ther- mal management with fin structures are commonly used for effective heat removal from LEDs. Optimization of the heat sink and fin geometry based on thermal needs is cru- cial for better design of LED arrays. High lumen LED array structure has different priorities such as mass, efficiency, and the base temperature of the LED arrays. The optimization model will be used to optimize the high lumen LED structure by using only three different inputs such as width and length of the fin, and the amount of the heat that is desired to spread. And the results of these optimization are applied to the two-step multiscale thermal model to see the real effect of these optimizations on the junction temperature. The results of this study show that the junction temperature estimation is different at high solder temperatures than using constant thermal resistance values from the product datasheet. By using multiscale thermal model, junction temperature estimation is more accurate. In addition, to design a high lumen LED array, different priorities can be used according to user demand and different initial LED array designs are obtained from the optimization result. Finally, to see the real effect of these optimization designs, optimization results are applied to the multiscale thermal model and better junction temperature estimation is obtained than the optimization model.