Issue link: https://iconnect007.uberflip.com/i/978458
MAY 2018 I PCB007 MAGAZINE 23 ing to more complex, thicker PCBs. We're talk- ing about boards that can be 10, 14, 16, 18, or 20 layers, which we wouldn't have seen in the past on previous generations of mobile com- munications or what have you. So yes, you do see that complexity, and you do see much more challenge on the PCB manufacturing side, and that's what I was referring to when I was talk- ing about the stacks starting to look like more complex high-speed digital boards than they have in the past. Shaughnessy: As far as materials go, is it go- ing to require a whole different set of new ma- terials? Hendricks: I think that depends on whether you are talking about the sub-six gigs or the millimeter wave. The sub-six gigs can broad- ly work with the materials that are available today. What happens at 28 gigs and higher is that you start to require, for example, ex- tremely smooth copper. What happens in the millimeter wave range is that as the materi- als become thinner, which is simply a func- tion of the smaller wavelength, then it's more than just having to have just a low loss di- electric. The components of insertion loss on a microstrip circuit are both conductor-based and dielectric-based, and as you get thinner the copper component of that loss becomes more important. And that's driven primarily by the smoothness of the copper, because at very high frequencies you have the skin ef - fect and the current travels along the bottom of the copper, so things like the copper foil roughness become more important. When you start having smooth copper, that means What happens at 28 gigs and higher is that you start to require, for example, extremely smooth copper. ducing power requirements can greatly simpli- fy things. But it does increase other challenges for passing electromagnetic compatibility stan- dards and certifications. That's going to be a huge challenge for a lot of the designers out there. And they'll have to learn more anyway, even if they're doing everything with a simple chip and a basic carrier printed circuit board, all the way to maybe a more specialized hub style device, like a tower or a node for aggre- gating connections. Those boards are going to be the complex ones with exotic materials. But then hopefully the actual end-user devices will be simpler, be- cause a lot more of the magic secret sauce will be on the chip, or in the package, or in the die. And in my mind, the problems are equally on the semiconductor companies as they are on actual board-level design engineers and PCB designers. Shaughnessy: Right, they're supposed to start releasing commercial chips, this year and next, designed for 5G. Jordan: I have a friend who is an RF engineer at a company that begins with Q. I was speak- ing with him just this morning, knowing I was going to join this discussion. And he said he's working on 5G right now, and it's very, very difficult for those guys doing the actual chip design and the analog front end. We'll see how well that goes. Shaughnessy: One of the things that I've seen is that they're saying some of these are go- ing to require much thicker boards. You know, they're going to demand thicker PCBs and they're going to have higher aspect ratios and it is going to be a lot harder for everybody, but harder for the fabricator to put these boards to- gether. John, have you seen that in your findings? Hendricks: Yes, in truth we see a lot of people trying a lot of different solutions for the same type of problem. But there are some com- mon threads that run through them, includ- ing a greater degree of integration that is lead-