Abstract
In a previous contribution we highlighted the potential use of spectral analysis of acoustic signals reflected from solder bumps as a means to identify anomalies in the bump. With numerous interfaces in contemporary solder bump geometries, time domain interpretation of acoustic signals is not straightforward. Frequency domain analysis is hence another route to utilizing information contained about defects in acoustic signals emanating from suspect bumps. In this contribution we highlight the use of Wavelet transforms and power spectra in analyzing acoustic signals and demonstrate with a few examples how the transform may be used to obtain unique finger prints of anomalies. We also discuss how cross correlation technqiues may be used to classify a fingerprint once a comprehensive library of finger prints is empirically constructed.
