Issue link: https://iconnect007.uberflip.com/i/1530610
12 PCB007 MAGAZINE I DECEMBER 2024 were experiencing misregistration. I would look at the punching machines as well as their punch and die blocks, lamination plates, press plates, and pins—the whole tooling from front to back registration, from the artwork to X-ray. At that time, there weren't any LDI machines yet, and Multiline's tooling was a new product; everyone used artwork, and they would have to pin the artwork. I learned how to check for accuracy and good registration. While I was doing all that deep work on factory floors, I realized that, over time, the layers were increas- ingly thinner. In 1998, a thin core was a 0.008" (0.2 mm) core, now considered a thick core. However, you may have had all the best tooling, with punching machines that were very accurate, and the lamination pins would be right spot on, but the cumulation of all the mechanical tolerances alone from the lamination plates, the lamination pins, the punched slot, and bushing were about 0.007–0.008" (0.17-0.2 mm). ese tolerances obviously do not all go in one direction, so the realistic best possible layer to layer alignment is about 0.003" (80 µm) on average. Even with the punching machine working in fantastic order and all the tooling done correctly, the best that you could possi- bly hope for is 0.003" (80 µm) layer-to-layer registration, which still holds today. Someone can purchase a brand-new punching machine today and at best they may shave off 0.0002" (5 µm); all these tolerances are still in a punch- ing and pin lamination process. That's really interesting. What else did you notice? e next big problem was the act of laying up the layers on pins. At the time, thin core was 0.008". When the operator would lay this up on pins using typical four-slot, you would place the layer on two pins, then on the other two pins, and push down. If there was a little shi, you relied on the pin to pull the layer back, but that would shi layers a bit. Of course, today, when you're dealing with 0.002"or 0.004" cores, it is so easy to tear the layers and shi them. I also noticed that every operator had their technique in doing this lay-up procedure, a lit- tle black magic tooling to get things just right. Some operators were really good at laying up and being very careful, but it only takes one slip of putting it on those pins to create a slight shi. en you're easily off by 0.005"-0.006", and there goes the whole panel. e operators are also dealing with placing 30- to 50-pound plates on top of the pins simultaneously, and there is no way to determine whether it's done correctly until aer lamination. What did you learn from all that? Looking back, I've realized that those punch- ing machines, where the punch and dies were made in the '80s, are still as accurate as some of these new punching machines. DIS refurbishes some old Multiline machines for our custom- ers who don't really want to change. at pro- cess works fine for the products they produce. Maybe the new punching machines are tighter by a tenth of a mil, but that means nothing in the scope of things based on those cumulative tolerances and operator effect, which always introduces shis and rotation. " While I was doing all that deep work on factory floors, I realized that, over time, the layers were increasingly thinner. "