Issue link: https://iconnect007.uberflip.com/i/1419905
22 PCB007 MAGAZINE I OCTOBER 2021 smaller scale, less control of the machines, and their price point per via becomes more. at's challenging. Happy Holden: e one thing that VeCS has going for it is that you can get HDI density without having a laser drill, which ought to be a big advantage to smaller North American shops when all they have to do is change their drill bit in their processes, rather than buy new machinery. Dickson: Right. Two years ago, we were look- ing at doing it with conventional routers, con- ventional machines, just like Happy men- tioned. Now we have very high-speed rout- ers, approaching high drill machine capabil- ity, CCD alignment routing machines, and depth-control routing machines. ose have all been innovated in two years. No one in the West, from a board shop standpoint, really understands how that works. I see on Linke- dIn all the time these celebrations that these via fill machines can do 3.0 millimeter thick with 0.2-millimeter vias and can via-fill them. We've been doing that for five years with the processes we have in Asia, but the equipment is high-volume, dual-step processes. High vac- uum screen print, second fill, and automated squeegee, and we've been able to do it in a rel- atively high volume for years. It's because the machines are large and expensive, so it's diffi- cult for a small-volume shop to rationalize that. Lamination, drilling, everything is going to large highly automated equipment and it's get- ting more challenging for a small shop to jus- tify those. I don't envy an engineering man- ager or somebody who is trying to do that any- where. It's very tough. I talk to my friends that are still in the industry in the West, and they have to get very creative to find these capabil- ities. Happy, I'm sure you've seen that on the HDI side. It's very challenging. I think the days of the 3D stacked system-on-chip, sys- tem-on-interposer, system-on- PCB have just begun. We are just barely getting into it. e idea that a PCB is a commod- ity is an antiquated concept that most customers are just starting to understand. And I think this shortage in the die market and the initial supply has allowed us to look at the ecosystem a little differently. Johnson: Joe, you just said something very insightful: a printed circuit board is moving from being a commodity to being a specialized product as a part of the design. Dickson: If you look at where we were 25 years ago, we had data coming into a chassis and it had to be managed into a single CPU. We created a carrier, a server, or a motherboard and this entire network went out to what's called a midplane or a backplane. We were all happy and thought, "Wow, this is the com- puter age. is is what it's going to be for- ever." As the signal speeds got faster, then we started running signals on optical cables and it got faster. But that single logic management was still there. Now we're at another fork in the road where the main PCB is becoming more like the back- plane again. e transference layers may not have much of the active capability directly on it, but it houses everything else, and you're stacking 3D chips, memory, and accelerator systems. Everything is becoming more closely tied together but also interdependent. at's where you have what you'll call a substrate interposer. Now you have a system-on-chip where you have multiple chips, CPU, GPU, memory, accelerators, whatever it is, and you're putting them closer and closer, stacking them, and manipulating the data. Yet, as soon as you get it to where it you've got that system Joe Dickson