Reading time ( words)
Editors Andy Shaughnessy and Nolan Johnson recently spoke with Lee Ritchey of Speeding Edge about the direction of PCB design. Lee also discusses some of the changes that he has seen in this industry over the past 40 years and some of the technological drivers that are causing designers to think more like electrical engineers than ever before.
Andy Shaughnessy: Lee, can you start with a 30,000-foot view of the PCB design? You said you have an example of how drastically speeds have increased in the last 25 years.
Lee Ritchey: When I started working at 3Com on internet products, it was hard work to make things run at 10 megabits per second. Today, I have clients who ship products at 400 gigabits per second every day; that’s 40,000 to 1 in about 24 years.
Shaughnessy: That’s crazy. When I first started covering this in the ‘90s, designers had started to focus on signal integrity, but nobody talked about power integrity until about 10 years ago.
Ritchey: I’d say that happened longer than 10 years ago because I started with terabit routers in 1999. But you’re right that most of the world got into power about 10 years ago. And there were no books, etc., of course; we had to make it all up. To give you an idea of what a two-billion transistor IC might look like, the last one John Zasio and I did was 0.9 volts, and the current was 160 amps. It takes 80 amps to start your car, so how do you get that kind of current to an IC? First, how do you create it, and second, how do you get it where it has to go? However, there is a silver lining. Fifteen years ago, everybody had problems with EMI, but it has virtually vanished as a problem because the villain that created the noise on PCBs was parallel buses, which are gone as well as electromagnetic interference (EMI). I used to work on an EMI problem every two weeks, year-round, but haven’t had one in two years.
Shaughnessy: Designers seem to have a love-hate relationship with their EDA tools, but they’ve come a long way in the last decade or so. Do you think the tool companies have done a decent job keeping up with technology and leading the way?
Ritchey: I would say they’re good enough, but that isn’t really the problem. If you give a chainsaw to a two-year-old, you’ll wind up with a ruined house, and that’s the problem I face every day. I have a client in Boston with somebody who does simulations and has no idea what hardware looks like; they don’t even get the models right, and then they make bad design choices, which is the problem. We need people who understand how to run things.
Nolan Johnson: We’re talking about all of the changes that have happened in the last 40 years, and the rate of change that you’ve outlined so far. Do you see that same pace continuing?
Ritchey: I used to think I knew where the limit was, and so did lots of other people. But the only thing I can say is if there’s enough money in it, someone’s going to figure it out. When we reached gigabits per second, we thought we were at the end of the line, and when we reached a rise time of 20 picoseconds, we thought we were at the end of the line. But we have gone right past both. For example, I heard about a new way to make transistors so that they switch in less than a picosecond. I’m not sure on the data rate exactly, but it’s going to allow about 100 gigabits per second on a single pair of wires. And that 400 Gbps that I talked about earlier? That’s actually eight 50 Gb/S lanes. Next, you’re probably going to see four 100 Gb/S lanes. I don’t know where the end is.
To read this entire interview, which appeared in the October 2019 issue of Design007 Magazine, click here.