Issue link: https://iconnect007.uberflip.com/i/1116895
66 PCB007 MAGAZINE I MAY 2019 tions, we have to look not just at our dielectric material but also at coppers that are starting to play a very big role in the overall loss of the product. Johnson: What are some of the market dynam- ics that you're currently seeing? Cochrane: Almost six years ago, it used to be low Dk to try to get thinner. Everything is shrinking and there's more miniaturization. Packages are getting smaller and smaller. To maintain that proverbial 50-ohm characteristic impedance or 100-ohm differential applications, we wanted thinner dielectrics with lower Dk so that we could maintain 50–100-ohm applications. How - ever, we started getting into too low of a line width with resultant resistance to the overall losses, which became a major problem. Our fo- cus for the last five years has been to look at lower loss for our high-speed digital customers. Loss is the big driver followed by skew. The skew from trace to trace with differential pairs is determined a great deal by the glass weave that is utilized. What used to be a single-ply construction with a flat glass is evolving into multiple ply. The magic-ply count seems to be about a three-ply construction and to get in- to 5-mil dielectrics that we're looking at for a 5-mil trace, that means we have to use a thin- ner glass. Now, we're crossing over to HDI. The 1027 glass—one and a half mils in overall thick- ness—used to be for laser applications for thin, stacked microvias. Today, we're using a lot more of that in high-speed digital. Within the last two years, we've gone from the stan- dard glass being a 1078, for example, to a 1035 and now down to a 1027. We're getting thin- ner and thinner glasses to support that skew requirement as well as the loss requirements now even moving to 1017 and 1010 glass styles. Johnson: Where do you see TUC's current sweet spot in your product portfolio? Cochrane: We have two main areas right now. We've gotten very close to the magic number of the 0.70 dB loss per inch at 25 GHz, which is for the next generation of chips that are out there for the 112-Gbps or 400-Gbps switch. That's the lowest loss material that we have that is not in the PTFE range. We're getting a lot of traction in that area because of that need. We're also seeing a resurgence of halogen- free. With the offering that we have in the hal- ogen-free world, we still have a very low-loss material. We're seeing a lot more utilization of the super low-loss, halogen-free materials. There may not be a corporate mandate to use halogen-free, but designers are finding that they can get the kind of performance out of a halogen-free that they couldn't a number of years ago. Johnson: Does the drive for the halogen-free seem to be customer demand? Cochrane: Yes, in addition to their ability to have very low-loss and super low-loss materi- als. Then, we can easily get good performance in the 56-Gbps applications and the 14.5-GHz area. They're being good corporate citizens. We see demand coming out of Europe, Cana- da, and North America as well. Johnson: That's an interesting point. Give design teams glob- ally the opportunity to get high performance out of something that is greener, and they will choose it. Cochrane: Definitely. That seems to be the trend. Another area where we're making in- Prepreg manufacturing at TUC.