Issue link: https://iconnect007.uberflip.com/i/1327102
JANUARY 2021 I DESIGN007 MAGAZINE 61 Happy Holden: Stackup has been a fundamen- tal characteristic of PC boards ever since we moved off double-sided. That's always impor- tant because, even a long time ago, certain signals were transmission lines, which means they needed a reference plane for the return. What has always been in contention was what thicknesses people used because, with mate- rial that you buy, it depends on the process what the final thickness is going to be, which then determines the impedance. Electrical engineers, not knowing enough, looked at the material specs of the raw material to deter- mine thicknesses. Those in the know realized they had to go to the PC shop to say, "For this particular material in your process, what will be the finish thickness?" Gaudion: We still see that, Happy. Somebody says, "But I've seen the spec and this is what it is. It's 75 microns plus or minus four microns." And you say, "Yes, but that's pressed. You're already going to find the finished thickness, and it's going to depend on the copper density, the layer, and vary from fabricator to fabrica- tor." Certainly, within our tool, we can sim- ulate the pressing, but the simulation is still only a simulation. Holden: I remember the big change when we finally had to go to field solvers from equations, because the errors in the equations were begin- ning to be significant. Suddenly, field solvers were required to really get the right numbers. Gaudion: That was a big change for us. Cer- tainly, back 25, 30 years ago, we had an equa- tion-based tool and then moved on to field solv- ers. Then, as the speed pushed up past 2, 3, 4 gigahertz, then bringing insertion loss in again, that became an ever increasing and important part of the equation. It's a thing we're seeing now where the focus is more on how smooth the copper is, and actually, we've done some tests for some customers where they want to look at the pretreatment of the copper to see when they're using adhesion promoters; more and more, newer adhesion promoters are being made smoother and smoother, so that they're not actually roughing the copper up as much as they used to. Holden: That brings up an area of stackup that we never talk about, but we ran into a lot of problems with—the stackup has to include what's going to be on the printed circuit board, through assembly and afterwards. It took a lot of education for designers to understand that these magnetic fields go out into space. Num- ber one, we assume that it's air, it's dielectric constant. But if you put on a thermal grease, and then put on a heat sink, and it's cover- ing over traces, then it is more than just the solder mask and air. Now you had better add in your thermal grease or adhesives, and your heat sink, because they're going to affect the transmission line. It's not just the material, it's a three-dimensional issue. Gaudion: That's right. Sometimes, we recom- mend that you maybe put critical traces like that on inner layers so that they're away from the surface. If they're running as striplines— rather than microstrips—you have better con- trol over that kind of thing because you've got at least a plane between you and what's going on the surface to keep the surface stuff out of harm's way. But it's not always possible, and if you are doing stuff on the surface, you need to route around some of those problems. That's an area we're doing some work on— not with conformal coatings, but with shields. Certainly, we see a significant number of PCBs where there's an EMC shielding material on the outer layer, so looking at how that impacts things is an area, and actually documenting just the sheer fact that whether it does have a shield on the outer layer is something that we're finding increasingly important. It's something we're looking at adding into our tools next year, because certainly in cellphones and tablets and things like that, you see a lot of shielding. You have to think about designing the whole of the Z-axis, whereas the CAD designs the X and the Y. We're looking at the Z-axis all the way through to the enclosure, the air, or what- ever is at the boundary layer.