Issue link: https://iconnect007.uberflip.com/i/1424540
84 SMT007 MAGAZINE I NOVEMBER 2021 ing at test and inspection of a more high-mix, low-volume configuration? Cain: First, I'd say design for manufacturabil- ity and design for scale are super important. I don't think you can under-invest in those, to be honest. Now, you still want to be paid back. You don't want to waste your time and mon- ey because you don't have any to waste, but there are the benefits. e leverage you get by designing for manufacturability, designing for yield, and designing for the outcomes you want to achieve is huge. Johnson: With that answer, though, you just pointed the finger at the OEM design team to set up manufacturing to succeed, right? Cain: Yes, which in a lot of cases has become a challenge because as the industry has ma- tured, electronics (and in most cases) OEMs, are not the same company manufacturing. It's contract manufacturing or outsourced, and there's nothing wrong with that; you just must work hard to span that bridge. It's not to say that the OEM must become an expert on a task, but they must know the questions to ask. I've seen customers get extremely high yield. ere is a very deep partnership, not a vendor relationship, but a partnership between the OEM and the CM. e other thing is to be sure you're measuring the right things at the right point in time. ere's not a simple formula, unfortunately, for that; there are many ways to do it. You must experiment and try things. I encourage people to run experiments, learn from them, and then in an agile fashion, make it better. I'm increasingly seeing other workflows in a lot of industries become very tuned to this agile Lean process. To me, analytics is the embodiment of W. Edwards Deming and Lean manufacturing processes which use feedback loops in processes such as PDCA. I love continuously improving feed- back loops. Johnson: What advice would you have around test and inspection to make it generally better than it is? Cain: Here's an example. We have an automa- tion system that we designed over a decade ago. It wasn't publicly available. It was designed purposely to create test automation, to move sub-assemblies, which in our case were elec- tronic test and measurement instruments like vector network analyzers. As they were being assembled, it would take that unit and move it from one test stage to the next. Previously, the reason we did that is we had racks full of in- struments and these racks would be developed to test the product. With the next variant, we have another rack of instruments testing the product. One day, about 15 years ago, we mea- sured the utilization of that equipment, and we were shocked how low most of it was. We were running out of space, which is a common thing for a lot of manufacturers, and we knew we had to do something. So, we re- invented how we performed tests by creat- ing a set of test stages so that any one of them doesn't completely test a unit. It tests only part of it, but we designed it to maximize the utilization of the equipment. en, to make that work efficiently instead of having opera- tors running around like crazy, we put a ma- chine and automation to move the equipment around automatically. Now, the approach is to take a step back when you're in a high-mix en- vironment. Utilization in space usually ends up being a bigger driver, and so you must rethink the test problem: instead of optimizing to test a unit you optimize to test the flow. Johnson: Fantastic, great. ank you for taking the time. Cain: Sure, I enjoyed this. SMT007 Christopher Cain is VP/GM for Electronic Industrial Products at Keysight Technologies.