关键词: 先进封装, 焊锡球, 粘连图像分割, 尺寸测量

Abstract: The accuracy of the roundness and particle size of solder balls determines the quality of advanced chip packaging products such as ball grid array (BGA) package and chip scale package (CSP). When inspecting solder balls, the solder balls will stick in the inspection field of view, which makes it difficult to measure their roundness and particle size quickly and accurately in batches. In order to realize the accurate detection of adherent solder balls below 760 μm, an inspection algorithm based on the combination of concave point detection and ellipse fitting is proposed. The algorithm uses Harris corner detection operator to detect the concave points of the adherent solder balls in the image, and combines the shortest path matching criterion of the concave points to realize the separation of solder balls. The edge contour of the solder balls is detected by the Canny edge detection operator, and an elliptic curve is fitted based on the coordinates of the contour points, thus realizing the detection of the roundness and particle size of the solder balls. The test results show that the roundness error of solder balls detected by the algorithm can be controlled below 3% and the relative error of particle size can be controlled below 1% compared with the scanning electron microscope measurements. This technique provides a fast and effective method for batch inspection of solder balls.

Key words: advanced packaging, solder ball, adhesion image segmentation, dimensional measurement