Masters Abstracts (2000)

MUNIKRISHNAN, MANIKANDAN
(August 2000), Supply chain - Senior Consultant, Cap gemini Ernst& Young Consultants LLp, New York City, NY

Email: Manikandan.Munikrishnan@us.cgeyc.com

A Monte Carlo Simulation to Predict the Placement Yield of Area Array Devices

The performance of any manufacturing process needs to be predicted so as to effectively design the process, control the system parameters to achieve desired outcome(s) and to make other business decisions. In fine pitch Surface Mount Technology (SMT) arena, assembly yield is a critical process concern because of the emphasis on reducing manufacturing costs. A variety of factors can contribute to assembly errors. Incorrect placement is one of these factors. The hope is that contributions from placement related defects may be kept negligible, to obtain good assembly yields. Hence, the ability to predict the placement yield of an assembly process is very important for the development of the process and to make strategic decisions regarding the materials and equipment required. The placement yield of an area array device depends on a multitude of material and machine parameters. An enormous amount of data would have to be collected directly from the placement process to calculate the yield on a part per million level, which is not practical. Because of the number of parameters that influence the yield, developing a complete and exact mathematical model is also not feasible. Hence, the need for a computer simulation of the process. This research effort concentrated on the development of a simulation model to predict placement yield of area array components and to implement it in user-friendly software to conduct "what if" kind of analysis on various placement situations to 'optimize' the design and the process for achieving 'good' placement yield. "Area Array Placement Yield Prediction" is a Monte Carlo based simulation program, developed in Visual C++, to predict the defects that could occur during the placement of an area array device due to in-plane substrate variations and placement machine accuracy. The machine, component and board details are taken as input from the user and a Monte-Carlo simulation is performed, to calculate the yield. In this simulation, the variations are generated by the computer and need not be collected from the ¡®real world¡¯. The model was validated and found to be theoretically sound by re-examining the formulation of the problem and uncovering possible flaws. All mathematical expressions are dimensionally consistent and the model was executed for various test cases that resulted in plausible defect rates.