Don’t be afraid of Manufacturing Process Variation (2/2)


And what about Central Midori?

We have been using statistical tools in the following areas:

  • Process validation
  • New product qualification
  • Process monitoring

Process validation

At the end of 2018 CMI relocated to a new building in Singapore and the challenge was to speed up the process validation while minimizing the downtime. Intel was the first company in the 80s to develop a method to duplicate production lines all over the world, so called “copy exactly”. Our approach was strongly inspired by Intel experience.

In CMI’s case we had to change the tunnel ovens and we knew that the process would be greatly affected. After weeks of head-scratching we finally decided to purchase one new piece of equipment and install it in the old plant. We could then produce parts on the old and new equipment in parallel and fine tune the new equipment to match the capability of the old one. When it was time to move to the new building we had collected enough data to ensure that the new plant had the same or even better process variability than the old one.

New product qualification

As described in a previous article CMI is a typical customized solutions manufacturer which means that we are making the products of our customer and that each design is unique and comes with its own sets of requirements. During a new product development phase we conduct capability studies on all critical product characteristics, typically mechanical dimensions and electrical properties. The data are submitted to the customer in our final qualification report.

Process monitoring

Part of the solvent contained in the ink evaporates during the printing process which leads to viscosity increases and change in the printing structure. For the majority of inks it does not affect the product coming out of the curing oven with one exception, the force sensing ink. In that case solvent evaporation affects the stability of the process and requires 100% monitoring with closed loop feedback to the printing technician. For all other inks a statistical sampling suffices.

Don’t be afraid of process variation

We should not be afraid of process variation and on the contrary we should use it to control our processes. As Dr. W. Edwards Deming declared in one of his executive seminars in Chicago in 1989, “Variation there will always be – between people, between products, between machines, between vendors. The key question is, What is that variation trying to tell us about the process, and about the people who work therein?”