Yield estimation for BGA assembly

Abstract

Assembly process yield is an important measure of the manufacturability of the design of an electronic product. It is also a measure of the capability of the process. Several factors affect the yield of a ball grid array (BGA) assembly process, including the materials involved, the methods adopted, equipment and tooling, and the personnel. In this research, the focus was on factors that relate to the selection of pad design parameters of a printed circuit board (PCB), component characteristics and solder paste deposit attributes. The primary indicators that were used to suggest the loss of yield were solder bridging between adjacent solder joints ('shorts') and electrically discontinuous solder joints following assembly ('opens'). This research exclusively studied the variations in component solder ball diameter/height, attachment pad dimension and stencil aperture opening, and their effect on the resulting assembly yield. This paper discusses the use of Monte Carlo simulation and regression to estimate the loss in process yield due to the occurrence of opens and shorts. Monte Carlo simulation was used to generate the attributes of the solder balls for a million BGAs, their component pads, and the corresponding PCB attachment pads and stencil aperture openings. Regression equations were used to calculate the height and radius of the simulated solder joint. The regression equations were generated by simulating a sample of BGAs using Surface Evolver, an interactive program for modeling of liquid surfaces. The height of the simulated BGA was used to determine the occurrence of an open and the radius was used to determine the occurrence of a short. If a single solder joint resulted in an open or a short, it would result in a loss of yield. Yield loss was calculated in parts per million (ppm).