Health Monitoring and Feature Vector Identification of Failure for SAC305 Solder PCB’s under Shock Loads up to 10,000g

This paper focusses on the prognostics of damage for a PCB under varying conditions of shock loads. The test board is a multilayer FR4 of JEDEC standard JESD22-B111 specified dimension with 12 packages that are arranged in a rectangular pattern. Health monitoring of the bare printed circuit boards under different drop and shock loads are studied in this analysis from the strain signals acquired from different locations of the board. The test boards are subjected to different shock loads of 3,000, 5000, 7000, and 10,000g acceleration to understand the failure of the packages during these various shock levels. Further, the analysis of varying shocks loads on the same test boards was also carried out to understand the effectiveness of the feature vector in predicting failure during varying load conditions. The strain signals that are acquired from four different locations of the test board during each drop are used for the identification of feature vectors that can predict the failure. The resistance measurements of the packages are used to identify the failure of packages during the drop. The health of the PCB is studied from the strain signal acquired from the strain gauges fixed at four different locations of the test board. The strain signals are processed, and various feature vectors are identified to predict the failure of the package during drop based on the time-domain and frequency-domain characteristics of the strain signal. The time-domain characteristics quantified the variation of the shape and profile of each peak of the damped strain signal, and the frequency-domain studied the characteristic change in the frequency components during each drop of the test board. Different data processing algorithms are used in dimension reduction and feature extraction from the strain signal and frequency components of the strain signal. A comparative study on the difference in the four strain signals under varying conditions of the load and sustainability of the feature vectors at different conditions of load and differences in the feature vectors with the position of the strain gauge is also studied.