Abstract
Feature vectors for health monitoring of electronic assemblies under repetitive mechanical shock have been developed for assemblies subject to 3,000g acceleration levels. The resistance and strain measurements of the PCB are acquired during each drop to analyze the changes in the values during the experiment. Analysis on the progression of failure was carried out using frequency-based techniques on the strain signals from different locations of the board and failure of the package was identified from the increase in the resistance values of the package during the drop. Feature vectors selected were based on the time-frequency data as well as the logarithmic decrement of the strain signals during the different drops. Different statistical approaches on identifying the changes in the damping characteristics of the package during drop were also carried out. Statistical analysis on the changes in the resistance values were quantified in accordance with the changes in the strain and correlation of the both were attempted. The dependence on position of the strain gauge on the PCB were also studied by comparing the variation of the feature vectors of the corresponding strain signals. The before and after failure strain signals were compared on the frequency components and as well as the changes in the damping characteristics of the strain signals.
