Advanced Acoustic Microimaging Using Sparse Signal Representation for the Evaluation of Microelectronic Packages

Abstract

Acoustic microimaging (AMI) has been widely used to nondestructively evaluate microelectronic packages for the presence of internal defects. To detect defects in small devices such as BGA, flip-chip, and chip-scale packages, high acoustic frequencies are required for the conventional AMI systems. The acoustic frequency used in practice, however, is limited by its penetration through materials. In this paper, a novel acoustic microimaging technique, which utilizes nonlinear signal processing techniques to improve the resolution and robustness of conventional AMI, is proposed and investigated. The technique is based on the concept of sparse signal representations in overcomplete time-frequency dictionaries. Simulation and experimental results show its super resolution and high robustness.Index Terms-Acoustic microimaging (AMI), microelectronic packages, overcomplete dictionaries, sparse signal representations.

Resolution Improvement of Scanning Acoustic Microscopy Using Sparse Signal Representation

“…Equation (5) is the blind source separation problem and can be formulated as follows when ignoring noise: observed A-scan signal y and an overcomplete dictionary Φ in which A-scan signal has sparse representation are given. The problem is to obtain the reflectivity function r={r i } such that y=Φ r and the constrained condition is "as sparse as possible" which means the number of non-zero elements in the source vector r are so few [9]. Since we supposed Φ is an overcomplete dictionary, solution of this equation is not unique.…”

“…This technique proposed by [9] has high resolution and high robustness comparing with conventional SAM method. The SSRSAM can be summarized in three major steps:…”

“…Various kinds of timefrequency dictionaries such as wavelet packets dictionaries, cosine packet dictionaries and Gabor dictionaries have been proposed over the last few years. According to the previous research [8][9][10] Gabor function can model the ultrasonic echoes very well. Therefore the most suitable dictionary is Gabor dictionaries.…”

“…2 and 4. According to the literature [8,9] a real Gabor function can be used for producing the incident ultrasonic pulses in Eq. (4).…”

“…This method was based on the concept of sparse signal representation in overcomplete dictionaries. In 2006 [9], the results of more studies on sparse representation of SAM signals were presented and some algorithms such as best orthogonal bases (BOB), basis pursuit (BP) and matching pursuit (MP) were applied to decompose A-scan signals of microelectronic packages into different wavelet packet dictionaries. The performance of MP on Gabor dictionary was also evaluated and introduced as the best overall.…”