Issue link: https://iconnect007.uberflip.com/i/1428512
NOVEMBER 2021 I PCB007 MAGAZINE 65 In my experience, there are a few things that should be used with the process; 3D microscopy is a big one. When I mentioned earlier about the phi- losophy and everything being a data lake upfront, that's nothing too special. A lot of industries are doing it right now. Cloud com- puting costs have come way down. It's not a big deal to buy yourself some cloud space, start one process at a time, and start building up your data lake. Start with the highest net present value (NPV) investments, and grad- ually work your way down. Next thing you know, you're a digital company, and you can forecast. Because I've recently gone to school for this, I'm extremely indoctrinated, plus I built a fac- tory recently that was quite advanced, so my perspective may be a little bit different than brownfield sites and people who aren't going to these types of courses right now. Nolan Johnson: People have been boxed into their thinking. Isn't that part of the challenge, philosophically, with process control, and test and inspection? Stepinski: ere are a lot of forces in the mar- ket right now. One of the main forces affecting PCB fab is digital transformation. In the next 10 years, we're going to see a lot more auto- mation, but the soware side is really the big thing that's going to change. We see it other markets right now too. I would encourage all PCB shops to get a free Coursera account; pay $100 and take some courses in digital transformation. Learn Python and how to apply it to your business. Do some data engineering, data science. Start now, because the people who do this first are going to be those who stay in the long term. at's my feeling on this. Johnson: You talked a lot about non-destruc- tive sensors, measurement from inside the pro- cess, not on the board or the finished product itself, right? Stepinski: Yes. Johnson: at's a shi in thinking, I think, for a board fab. If your processes are right, then the product that's coming out is right. Stepinski: eoretically. is is, in large part, driven by legacy specifications. Fabs are going to use methods that the customers accept. If you go out of line with this, you're a fool because your management will put the ham- mer down, and the customer is not going to accept product. At the end of the day, though, this is a big fallacy. It makes a lot of sense to develop other methods. e challenge is the legacy data that says, "We measure things this way," so you must have additional labor to do those other things. In my experience, there are a few things that should be used with the process; 3D micros- copy is a big one. You can avoid cross-sections in process by doing a lot of 3D microscopy, and then just using the cross-sections for final acceptance. ere are offsetting benefits there. If you can make agreements with your cus- tomers to use the 3D microscopy, the 3D back drilling, and things like that to model your sig- nal integrity, then just sample at the end and do a regression; this also makes a lot of sense in an existing shop. But the headwinds are asso- ciated with legacy specifications. at's really what it comes down to—that and the culture of our market and of our industry. Johnson: You were doing some things differ- ently at the fab you recently built. You had