Issue link: https://iconnect007.uberflip.com/i/972396
42 FLEX007 MAGAZINE I APRIL 2018 But the dielectric properties tend to be general- ized, I would say. We have a common dielec- tric constant that we use for impedance control type calculations, and of course when you get a design like that from an RF engineer or from a high-speed engineer, they're using different calculators than we are and typically they get really close. The bigger problem that we have when we're talking about dielectrics and stack- ups for that sort of design is the thickness of materials that are built. We have a standard line of thicknesses, and if the customer has designed it and put it through the calculator and maybe it says there needs to be 5 mils in between copper layers, and that's just not a common thickness for us. That's when we get into situations where literally we will go out and purchase special material just for that. Dielectric properties, though, are or have been historically basically generalized in our industry. Goldman: I've heard that if you've changed your parameters or your temperatures, for instance, then the dielectric properties change and so you almost have to pick a point and go from there. Could that be why they're so general - ized? Talbot: I would assume so. For instance, in the summer when there is high humidity, we get a lot of moisture in the materials and that changes the dielectric properties as well. We've had to live with that for a long time, but the truth is if we stick to the normal typical stan- dards and design a circuit around them, and adjust as necessary after measuring impedance coupons, we see very little failures after that. And these are the general pub- lished numbers. Yeah, you're sort of just jumping in the middle and using those specifications, but they work well in the field. Shaughnessy: You mentioned the very long flex. Do you guys do any of that? Talbot: We do. Probably about 10% of the circuits that we build are out- side of the standard length. We can go up to a meter pretty comfortably and have gone much longer than that for several others. The last one that we quoted for somebody, they wanted a 12-foot or 14-foot long circuit, and they wanted 5 mil traces and 5 mil spaces. And that's just not possible, for us, along that length. Shaughnessy: How do you even know how to bid a job like that? Talbot: We don't bid it. We can't. If we thought that we could build it with a yield better than 1 out of 10 we would. But if not, then the cost is enormous, and we can either bid it that way and take up a whole lot of time and effort, or just say no. This design just doesn't fit our capabilities well. We've had some custom- ers who kept coming back to us because they couldn't find anybody else to quote it. Shaughnessy: What do you think about the new technology at CES, like the drones and 3D viewers and virtual and altered reality that have flex inside. Talbot: That's great. But there's one thing I haven't seen, Andy. Every year, the overall circuit industry publishes the amount of cir- cuit boards that are built every year versus the amount of flex, and I haven't seen the amount of flex really increase much. It kind of hovers between 10 and 15% of the overall industry. When that gets up to 25% or 35% or some- thing like that, we'll know that it's into a lot of products.