The increasing need for control of every process during SMT manufacturing can be attributed to the miniaturization of components such as 0201's, chip scale packaging (CSP), flip chip (FC) and the stride toward increasing capacity and throughput. With the emergence of the automotive industry standard in electronics manufacturing (QS9000/TS16949), the need for a common language for machine characterization (IPC-9850) and verification of machine compliance with manufacturers' specifications is apparent. It is also obvious that we need to find a system to identify root-cause of machine drifts; use statistical methods to characterize a process or a machine; and eliminate human error in machine calibration, measurement, and the interpretation of machine capability.

As an industry, we have come to realize that the best way to accomplish more with less is to focus on line optimization and to use the equipment already in place to its fullest potential. This is the reason why machine users must find solutions to characterize the accuracy and repeatability of the process. Thus, the first step must be done on SMT equipment. This document summarizes the tools and methods used in Solectron Bordeaux. A complete capability measuring system hereafter called Cm System was acquired in July 2000 to measure "machine capability". Since that date, many tests have been performed on similar machines from different manufacturers.

Process capability can be defined as the spread within which most of the part values within a distribution will fall, generally described as within plus or minus three standard deviations (���3 Sigma). This baseline definition enables us to compare process capability under real manufacturing conditions with specification tolerances. Process capability is calculated over a long period of time and is influenced by the manufacturing environment. Machine capability, on the other hand, is calculated within a short period. The impact of all materials and parts used must be eliminated. This is also called "short time capability". The examination of machine capability is used to audit the quality behavior of a single machine.

A process can be viewed as a series of actions or operations influenced by several elements or factors, all contributing to the eventual outcome. These elements or causes of variation can be generally broken into the following typical categories:

• Material
• Machine
• Method
• Manpower
• Measurement
• Environment

Each of these elements contributes some degree of variability to the process. Nevertheless, when we are talking about machine capability, we only determine spread and centering of machine parameters independent of the influence of other factors. This can be represented by the Ishikawa diagram below:

Fig. 1: Ishikawa Diagram

Thus, machine capability coefficients are calculated as follows:

C_p =  (Spread)

C_pk = min()  (Centering)

C_p, C_pk: Machine capability indices
USL: Upper Specification Limit
LSL: Lower Specification Limit
X: Mean
s : Standard deviation

Although the distributions of many processes may assume a variety of shapes, many random variables observed in nature possess a frequency distribution that is approximately a normal probability distribution. The normal distribution is the most important continuous probability distribution in the field of statistics.

Fig. 2: Normal Distribution