Issue link: https://iconnect007.uberflip.com/i/1116895
18 PCB007 MAGAZINE I MAY 2019 Morgan: What usually happens is a designer goes to their favorite board shop and says, "I want to build this board. What do you recom- mend?" The board shop decides what material to propose, and they typically propose things they're used to that work with other OEMs. The board shop recommends it because the designer doesn't know. Thus, the board shops carry most of that information, and they're who people go to for speci- fying materials. That's how it works. The designer knows them and is going to buy boards from them, so there's a cus- tomer-supplier relationship there. The board shop is going to help them out if they can, but they tend to recommend the materials they know best or have used most. Howev- er, they may be missing out on some other area, or they don't have a relationship with a supplier who has generated a load of great, new products. Johnson: So, there is an added responsibility now for the designers to have a better under- standing of the materials their designs will be built upon. Morgan: I think so. And all that I can do is try to help them. The recent AltiumLive 2019 PCB Design Summit was great with all of the de- signers in there, but designers need to under- stand that the material interacts massively with their design, and the properties of the material impact the final performance. Many didn't un- derstand that until now. Again, it didn't really matter as much in the past, but people are per- haps waking up to this issue now. Many of the PCB design software tools now contain material libraries from the suppliers, so you can pull down the loss characteristics, etc., and get that data in there and then design on it, which is fine. The only question is, "Is it really true?" The properties that people pub- lish and the ones that work in designs are not necessarily the same; you have to test them. Let me give you an example of how these things work. I mentioned before about glass fabric and resins. Resins typically have quite high loss, and glass fabrics have a lower loss. Of the two components, the resin has the high- est loss. So, when you specify a value for loss, you put a number down, such as 0.010. We supply a range of material, so if I'm supply- ing X material to somebody, I can supply that having a res- in content around 45%, for example. And I could go as high as 75% resin content on that material. If I'm quoting loss figures, I test it on 45% resin con- tent because that gives me the best value. And I will get a value of 0.01 at 45% res- in content. If I take a board and build that material, for high-end circuitry, it will be probably around 70% resin content. My loss of those materials might be 0.015 ,so 50% higher loss than the tested value. I still call the loss at the lower value because I'm allowed to as the specification sheets say quote the value in the resin content range between 40% and X%. People choose the material to write the data sheets based on what gives them the best re- sult. Classically, for rigid material, it's 0.5 mm. It has a fixed resin content—50%. You're nev- er going to use that in a final design—that's out of the question—but that's the data sheet values that you're going to find. It's not really cheating because the values are correct, but they're not necessarily the values of the prod- uct you're going to receive from the supplier. Also, there are different ways of measuring these things. Entire projects run on comparing different values. You can measure the loss in three or four different ways and get different numbers, for instance. You can measure differ- ent Dk values and frequencies as well; it's not like a melting point of 10. Instead, it's a broad range of numbers, and it depends on how you measure. Therefore, you can't rely on the data