Ge Quality

Donald Splaun, Six Sigma Black Belt, lead a team at the GE Fanuc manufacturing plant to examine Black Belt Project #P52320. The objective was to determine that whether the company needed high-priced Ni-Au finished boards which were being used in the past as mounting platforms for fine pitch surface-mounted devices (SMDs). His team consisted of ten members – seven people, mostly engineers, primarily responsible for assembling, purchasing, testing and evaluating SMD boards, one financial representative responsible for cost calculations, one Master Black Belt reviewer responsible for preventing gaps in the analysis and one Champion responsible for resource allocation and ensures project visibility. A 12-step DMAIC process was used to solve and analyze the problem. Team members used a cause-and-effect matrix to identify three CTQs – two business factors and one project factor, and their respective weighted rankings. The team then developed a team charter in order to clearly define the problem. Then, they identified the tools and the databases which were going to be used in this process. Two statistical tools – Minitab and Excel were used along with a plantwide integrated database (SAP) containing all the necessary information about the boards, their specifications and costs. A complex experimental design was implemented based on a 29-step process flowchart in order to determine the effects of supplier differences and finishes. There were three suppliers – Vendor G - Singapore/China, Vendor P – Taiwan/China and Vendor D – USA. This experiment tested 96 hot air solder leveled (HASL), nickel-gold (Ni-Au) and, silver (Ag) boards, where 32 boards of each of the three finish types came from each vendor. The team had proposed two hypotheses: Firstly, there are no significant differences in number of defects, regardless of finish. Secondly, there is no significant interaction between the vendors and finish. An 8-step process was conducted over six days to collect and carefully analyze the data. After data collection, 15 ANOVA computer runs were performed to test the above mentioned hypotheses. Some of the key findings were that the company can save money by not using Ni-Au boards as they are not significantly better than HASL or Silver boards, Ni-Au is worse for wave soldering, and the company should not used vendor D because of its higher mean number of defects. The team concluded that GE Fanuc did not necessarily require Ni-Au boards for fine pitch SMDs, resulting in a total estimated savings of $190,000 per year.

1. The experimental design had to be so complex to ensure that everyone