Automated peak detection for analysis of error propagations and validation of structure functions in reliability tests of LED systems

In this paper, a workflow for an automated peak detection for analysis of error propagation using structure functions is shown. These functions’ peaks of several LEDs were studied as they were affected by different conditions, such as different operators or thermal interface materials between LED and heatsink. Over 1000 curves were collected and analyzed in repeatability and reproducibility tests. In order to keep the internal error caused by manual analysis as low as possible, automated data analysis was implemented. This analysis includes similarity analysis of the repeatability data, automatic detection of peaks for both tests with subsequent clustering and determination of the characteristic intersection between curves measured with different thermal interface materials. A statistical analysis enabled to evaluate the reproducibility and repeatability in these cases.