Digital Signal Processing in a Novel Flip Chip Solder Joint Defects Inspection System

Abstract

Digital signal processing methods in a novel flip chip solder joint quality inspection system are presented. Laser ultrasound and interferometric system is a new approach for solder joint inspection. It has many advantages such as being noncontact, nondestructive, fast, accurate and low cost. Furthermore, it can be used on-line in assembly line or off-line during process development. In this system, signals recorded are ultrasound waveforms. Because noise is present, signal processing methods are developed to increase signal-to-noise ratio, and to extract solder joint quality information from those waveforms. Signals are analyzed both in frequency domain and time domain. In the frequency domain, digital filtering and Bartlett power spectrum estimation method are used, and defects can be detected from the frequency shifting. A series of experimental results are presented, showing that power spectrum estimation can greatly increase the signal-to-noise ratio than when only time domain averaging is used. This speeds up the data acquisition and analysis process. In the time domain, “error ratio” is used to measure the difference between a good chip and a chip with defect. Results indicate that error ratio method not only can detect whether a chip has solder joint defect or not, but can also locate that defect. Overall, signal processing plays a very important role in this flip chip quality inspection system.