Issue link: https://iconnect007.uberflip.com/i/1116895
40 PCB007 MAGAZINE I MAY 2019 is somewhat limited. Now, it's starting to get harder and harder to have an incremental im- provement in the loss due to resins. Weldon: We're seeing an increase in the num- ber of issues in high-speed digital. As they go to higher data rates, skew, signal levels, and SNR are becoming more of a problem, which goes back to what John was mentioning with low-loss materials. Glass-free materials have also started to become a hot topic as skew is no longer manageable by adjusting the glass weave or using traditional techniques. Even homogenous materials, such as polyimides, are becoming interesting to the high-speed dig- ital community. From the traditional RF, high-frequency side of things, we're seeing an increase in the fre- quencies of interest. Traditional 4G LTE and consumer electronics applications were rela- tively low in frequency between 1–6 GHz. They're still operating there today, but now we're looking at the 28–39 GHz bands for 5G, which is changing the game. It used to be that dielectric loss wasn't the primary player while constant and conductor loss would have been fairly dominant, espe- cially in thin constructions and flex. But with those higher frequencies now, dielectric loss it- self starts to rear its ugly head, and traditional materials that might have been relatively lossy from a dissipation factor standpoint may not be sufficient for all applications. There is a def- inite need for improved materials. Further, we're seeing some interesting things in terms of design space where people are starting to increase package densities. There have been a lot of changes in substrate height requirements for certain designs that may have been very thick before but are trying to get thin- ner as well as things that have been thin that people are trying to make thicker to improve performance. Overall, we're seeing changes in how people are approaching design challenges due to these higher frequencies. Feinberg: What is the most used thickness of dry-film resist right now? Back in the day, it was 0.5 mil, 1 mil, or 1.3 mils. Andresakis: Not much has changed as far as thickness; the range you mentioned is still very common. We still sell a lot of traditional Ris- ton® products because the lines and spaces for servers and routers haven't changed apprecia- bly either. I think it switches over when you go to the consumer world with modified semi- additive processes, etc. They're using the thin- ner resists and switching from subtractive to additive processing. doing a plate up versus an etching. However, more people are using laser direct imaging for regular products because of the benefits you have removing the photo tool. Almost every board shop now has some de- gree of laser direct imaging, even if it is just for quick turns and prototypes. Feinberg: We used to look at 10-mil lines and spaces being very, very fine line with 15–20- mil lines and spaces being more typical. Is that still the same for both the standard circuit boards as well as the ones where you're try- ing to shrink all of the circuitry down? Because the average person wants more space in their phone for a battery than they do for a circuit board, but yet still want all of the functions. What kind of resolutions do you see now, and what are your customers asking for? Andresakis: I haven't checked in that space in a while. But when it comes to traditional high layer count, servers and routers, they've been doing 3-mil lines and spaces for a long time with traditional photoresist, so that's not an issue. When it gets to the consumer, the big switch is that the resolutions have improved so you can do the subtractive process down to From the traditional RF, high-frequency side of things, we're seeing an increase in the frequencies of interest.