A system for automated BGA inspection

For the advantages of high I/O count and small size, ball grid array (BGA) package techniques have been developing very fast in recent years. In this paper, a vision-based algorithm was proposed. The BGA grayscale images captured by CCD camera, and some image processing methods were applied to segment and enhance the image of solder ball. Three kinds of feature size were employed here: area, centroid and roundness of solder ball. The defects were detected by matching these features to the template data. The inspection items included missing ball, extra ball, ball bridging, oversize or undersize, offset and deformation. In this research, the speed execution time of inspecting one BGA image is also taken into consideration