Ultrasonic Inspection of Thin Multilayers

This paper presents a new approach for the detection of multiple ultrasonic echoes in the thin multilayer structures and the estimation of their locations. The basic idea is to make use of the two‐dimensional property of the time scale (frequency) representation and develop a knowledge‐based signal processing technique. A dynamic wavelet fingerprint is generated at each sample point to characterize the local time‐frequency structure. A distinctive DWFP pattern with a well‐developed circle inside was found and used to automatically detect multiple echoes under different simulation conditions. The performance of the algorithm was evaluated with simulated ultrasonic signals. It shows that the proposed algorithm has advantages of high precision, low ambiguity, robustness and that it can be used to detect partially overlapped echoes. Further evaluation was carried out by applying the algorithm to ultrasonic waveforms obtained from a plastic encapsulated IC package. Similar DWFP patterns were found for these pr...