Skip to Main Content
Heat management and reliability is essential for high power LED packages, e.g. for high temperature application like automotive lightning or application with very long lifetimes like street lightning. To reduce the LED junction temperature a low thermal resistance is realized by mounting the LED packages on heat spreaders or boards with good heat conduction. The joint between package and heat spreader, very often a solder joint due to the good thermal conductivity of the solder material, need to be void and gap free to achieve a good heat conduction and high reliability. The quality of solder joints of LED packages is usually controlled in production by X-ray and acoustic microscopy (CSAM). From a good solder joint, i.e. detection of no bad soldered area, a good thermal performance is concluded. The Thermal Transient Testing provides a method to measure the thermal resistance by measuring the forward voltage Vf(t) time dependent after a thermal power step, i.e. switching the drive current from high drive to low drive current. However, the k-factor, the linear dependence between Vf(t) of the LED and the real thermal power step needs to be measured to obtain the correct thermal resistance. We have developed an algorithm to enable inline thermal transient testing of LED modules without the need to measure the k-factor and the thermal power step. Instead of calculating the structural response function from the transient forward voltage, we evaluate the forward voltage in the time domain. For the development of the method we have set up a finite element (FE) model for our LED packages and performed transient thermal simulation. The FE model was fitted to the experimental data. We simulate the influence of void sizes and positions, gaps and joint thickness on the transient temperature curves. By comparing the measured sample with a known good sample we can evaluate the quality of the solder joint and calculate the thermal resistance. We apply the mea- - surement method for quality control of the solder joints of our high power LED packages. The measurement method targets to replace X-ray or CSAM inspection within production. We compared CSAM inspection with the thermal resistance measurements. Thermal resistance and non soldered area are correlated for larger bad soldered areas. We achieved a detection limit of roughly 30% of bad soldered area.
Date of Conference: May 31 2011-June 3 2011