<title>Inspection of ball grid array (BGA) solder joints using x-ray cross-sectional images</title>

Abstract

The ball grid array (BGA) chip is widely used in high density printed circuit board (PCB). However, inspection of defects in the solder joints is difficult by visual or a normal x-ray imaging method, because unlike conventional packages, solder joints of the BGA are located underneath its own package and ball type leads. Therefore, x-ray digital tomosynthesis (DT), which form a cross-sectional image of 3D objects, is needed to image and inspect the solder joints of BGA. In this paper, we propose a series of algorithms for inspecting the solder joints of BGA by using x-ray cross-sectional images that are acquired from the developed DT system. BGA solder joints are examined to check the alignment between the chip and pad on a PCB, bridge, adequate solder volume. The volume of the solder joint is represented by a gray level in the x-ray images: thus solder joints can be examined by use of the gray-level profiles of each joint. To inspect and classify various defects, pattern classification method using a learning vector quantization neural network and a look up table is proposed. The clusters into which a gray-level profile is classified are generated by the learning process of the network by using a number of sampled gray-level profiles. A series of these developed algorithms for inspecting and classifying defects were tested on a number of BGA solder joints. The experimental results show that the proposed method yields satisfactory solutions for inspection based on x-ray cross-sectional images.