Issue link: https://iconnect007.uberflip.com/i/1116895
MAY 2019 I PCB007 MAGAZINE 13 work. I tested some samples, and the material was fine. But I did notice that the boards had been rotated 90 degrees on the design. Instea d of running against the warp direction of the glass fabric, they were running against the weft direction instead. The warp and the weft, by the way, are the X and the Y direc- tions of the glass fabric, and they have a dif- ferent density of the weave. That density of weave change was enough to cause the board to stop working. It sounds crazy, but that was enough. They were probably on the edge any- way, but that switch of warp and weft stopped the board from working. Matties: If you're on the edge, what's the next step to break the barrier? Morgan: Take out the glass. There are materials without glass—unreinforced materials—but the problem you have is that they're not rein- forced so you could have stability problems. Polyimide films—flexible films—are done that way. Flexible materials also work that way; they don't have a glass weave in them. You can also attempt to mitigate this by using different kinds of glass. Patty Goldman: Like flat glass? Morgan: Yes, that's just a tweak of the edge. I mentioned before that the warp and the weft are very different densities of glass. You can do what's called a square weave and make them the same density. You can play around with that a bit and make it more uniform. You can also spread the fibers out more. The fibers are built in bundles like ropes. And the ropes have hundreds and hundreds of individual glass fil- You might think of the dielectric constant of the resin, and it could be a value in the order of three. The dielectric constant of the glass is at least double that, so around six, and it has a capacitive effect. How much charge can the material carry and hold? Glass holds a lot more than the resin does. If you try to send signals down there with a pair of conductors—like a differential pair—and one of those conductors runs along a bit of laminate with a lot of glass underneath it and another part of the pair runs along a bit that's got a lot of resin underneath it, they see a very different dielectric property of the material. One of those conductors has a significantly slower signal speed compared to the other one, and you wind up with mis- matches in the signal integrity. Johnson: That means designers could have variations in the performance from board to board as they go to higher and higher frequen- cies simply based on what part of the weave, for example, is in the material underneath. Morgan: Yes, and the reason for that is the bulk properties are uniform. If you take the average, everything is fine. But using today's very fine conductors, the microstructure underneath is very significant. We're talking about the weave of glass fibers—the actual scale of that weave. You may end up with a trace running in a good or bad area along the board. There are ways to mitigate that, but it is a fundamental prob- lem. As long as there is reinforcement with dif- ferent electrical properties, you will have this kind of issue. When you design the board, you don't know where the glass is going to lie un- derneath. When you build the board, you can make the boards one way, and they will be fine. Another day, you'll make them and they won't be fine. I'll give you a very practical example of that. Some years ago, we had a customer in Ireland making transponders for automotive alarms. He suddenly said, "My boards stopped work - ing. I can't get them to communicate with the car anymore." They thought that somebody had used the wrong material, so they specified the material but used a cheaper material that didn't Instead of running against the warp direction of the glass fabric, they were running against the weft direction instead.