Dan Beeker’s AltiumLive Keynote: It's All about the Space
Dan Beeker of NXP Semiconductors was a keynote speaker at the recent AltiumLive 2017 event in San Diego. His presentation focused primarily on the design of three-dimensional structures and how it’s all about space. I caught up with Dan, and we discussed his presentation, his background and career, and what kind of advice he would give to young PCB designers.
BARRY MATTIES: In your keynote presentation here at AltiumLive 2017’s inaugural event, the main message was that it's all about the space. How would you summarize that, for our readers?
DAN BEEKER: It's about plumbing. We have to design three-dimensional structures that allow us to take an electromagnetic field, not electrons, from where it's stored, either off-board or in a capacitor, to where we want to do work with it. Whether that work be logic function, or to something like driving a motor in a fuel pump, or something like that. You take the energy, you move it from where it is to where you need it to do something, and all you can do with that electromagnetic field is convert it into kinetic energy, which makes something move. It even vibrates molecules as it heats up things, like in light bulbs, or it causes other things to move, like a motor or a speaker.
MATTIES: Now, just for context, please tell us a little bit about yourself.
BEEKER: I'm an application engineer in the automotive team at NXP, so I'm responsible for supporting microcontroller embedded design for our customers, basically around the world. I have also become the designer of the missing links in the tool chain. I develop what I call widgets, which are adapters and socket interfaces that will allow customers to more easily debug their software and hardware with things that you can't buy from a generic third-party tool vendor.
From that, I had to become a good system designer for real products, because all of my parts have to be CE certified, like most of my customers. So when I design a product, I design it to be the best possible behavior that I can without any concern about CE levels. When I get the test results back for my boards, they're typically 20 or 30 dB below what the spec requires. And that way, I can show customers these designs as an example of the right way to do it.
Every time I do a new board, I'm doing it with the idea that I want help teach my customers the right way to make their products; from that, also, I became the EMC specialist because nobody else seemed to care about fields.
MATTIES: In your talk, you mentioned having a kind of “A-ha!” moment with Rick Hartley, where your thinking was shifted 180 degrees.
BEEKER: I had been at Freescale, working on an advanced pre-silicone emulator, gate array-based design for a big customer program, and I had come from the Motorola tool group that did in-circuit emulators. We had this standard called the Motorola Active Probe Interface, and it defined how you connected boards, where the signals lived. I was working with a service bureau in Dayton and I said to the designer there, "You've got to follow this spec." And he says, "Well, I don't think you want me to do that. It doesn't do what you think it does." I said, “What do you mean? This is the Motorola spec. We've been using this for years, you work for me, I am the senior engineer, and I know what the hell I'm doing, so you’ve got to use it.”
He was stubborn, too. We both have the same birthday so it's really bad; we're both Scorpios. He said, "No, you're wrong, what you're telling me to do isn't going to work, and you need to get some training." I told him that I didn’t need training, that I’d been doing this longer than he’s been alive. And he told me that I needed to go get training in California. I went to my boss and said, "I really need to get some signal integrity training because something's wrong here. I need to know this. My customers need to know this." And he sent me to Rick Hartley’s class.
MATTIES: Off you go.
BEEKER: I went to Rick's class, and five minutes into it he's starting to talk about managing the transmission lines for directing signals. Basically, I went from being haughty and arrogant and knowing what I was doing to being absolutely shattered. I knew, at that point, that everything I did worked by accident, not by design. And, as a senior engineer, that's not the kind of thing you want to have happen to you. I hoped my boss doesn't figure this out, you know?
I started hammering Rick with questions, and about half an hour into the class, he stopped, looked at me, and he said, "Who the hell are you and why are you in my class?" And I said, "Rick, I'm Dan Beeker and I'm from Freescale, and I really need to know this stuff." So I went through the two-day class, and it was just really eye-opening.
MATTIES: That sounds like it was a big moment for you.
BEEKER: Oh absolutely; it changed my career. I went back to my office and we were preparing for our first big technology forum at Freescale. My manager was involved in the creation of the content, so he had to be on another meeting, and he made the mistake of having me sit in for him on this call. I was really big on signal integrity at that point in time, and I knew we were moving to 90 nanometers and beyond, and there was a pause on the phone, so I said, "We need signal integrity training. Our customers, we've got 90 nanometers coming at us and we already messed them over with CMOS." He said, "OK, wise guy, you go create a panel of experts and we'll start that."
So, what do I do for a panel of experts? I started pulling the books off my shelf, the ones that I bought to make me look smart, and I called the authors. I called Henry Ott, Howard Johnson, and Lee Ritchey. But the same week that we had our conference, there was a big EMC consortium in the UK. So, of course all the big guns were going over there.
One of the authors who actually called me back was Ralph Morrison, and he said, "Who are you and why are you bothering me?" I told him my name and that I was looking for people to speak on my panel, and Ralph finally agreed to do that. I dragged him out of retirement. I didn't know how old he was, but at the time, he was in his 80s already. He's 92 now. I got Ralph, and then I talked to several of the instructors from PCB West, so I got Rick Hartley and Bob Hanson.
They always talked about UMR, (University of Missouri-Rolla, now Missouri University of Science and Technology) so I called them, because Freescale was a member of their consortium. We paid them bucks so I could call and say, "I'm from Freescale, one of your customers." I got Dr. Todd Hubing and the director of the program, Tom Van Doren, so that was two UMR professors. Then I got the Chief Technology Officer from X2Y Capacitors, which was doing some stuff, and then Lee Hill, a consultant who was a UMR graduate, and also a well-known EMC guy.
They were my panel. I had Ralph go first because he was the physicist, then I worked my way through Rick Hartley and finally up to X2Y, which was a solution for EMC. By the time they were done, Ralph was jumping up and down. He was so mad he could not stand still. And, I mean this is a little old man; I was worried that he was going to have a heart attack. I said, "OK, Ralph, calm down, what's going on?" And he said, "They've got it wrong. That's not how it works. They've got it wrong." I said, "Ralph, let's go have lunch and you can tell me about it."
So he started his mission to teach me about field physics, and from our discussions there, I said, "You've got something that I think is missing. I believe you." Plus he had like eight or nine bestselling books already, and so I knew who he was because I had three of his books already. I hired him to come teach his class for my customers, and I repeatedly did that and hired other people. Finally, I could pick the pieces from the other presentations, they started to make the physics that Ralph was teaching me make really good sense. I started applying that and the success was instantaneous.
When I did that first slide that I talked about for my first presentation based on antenna lengths, everything made sense. At that point, everything that I'd observed in my career made sense. Why, when we went from HMOS to HCMOS, and all the automotive modules were failing EMC, it was because the switching speeds went up by a factor of 10 and the antenna lengths went down by a factor of 10. So we had faster switching, shorter antennas, and we were gladly hooking up radio stations to antennas. It was like, “Oh, now it starts to make sense.” And the rest has been just a progression of replacing the circuit theory experience with a knowledge based on field management.
MATTIES: Was this conventional thinking at the time or was this radical thinking?
BEEKER: Ralph said nobody's ever paid attention to it. They look at his books and they shake their heads, but nobody does it.
MATTIES: What year was this?
BEEKER: This was 2008 or 2009.
MATTIES: So eight or nine years ago you did this. I see you giving speeches and presentations now; have you made an impact with this knowledge?
BEEKER: I have a lot of customers who are following these practices. I have one customer in particular, who did a training class before they started to design, and we did regular reviews as we went through the process. They went to production with their first article and they couldn't believe it. They said it was really simple to do. I said, "You followed the rules, and your board passed EMC."
MATTIES: Isn't this a common-sense approach really, though?
BEEKER: It really is.
MATTIES: And why didn't everyone figure it out if it's common sense?
BEEKER: Because for some reason, it's been clouded in the formal education process. They have made this island out of electromagnetic fields with scary math, and people don't want to take it and they're forced to take it and they forget it. Then their circuits, which are based on algebra and very easy to understand concepts, were never mapped together, and that was the missing piece. People come out of the universities, go into the job market where people are still practicing the circuit-based design philosophy that seemed to work, like I said, by accident for many years, and that reinforced perspective continues to be taught. My mentors taught me that, and I did it repeatedly in my job on a daily basis and expected my customers to do that until I finally figured out that it was three-dimensional geometry. Electrical engineers don't like to think about things in terms of geometry.
MATTIES: Why is that?
BEEKER: Because it doesn't map to the math. Even though it is ratio metric, and the switching speed of a transistor is based on its geometry. The three-dimensional structures that we put the energy into, their behavior or reaction to that energy is determined by their geometry. If that geometry doesn't match, you're going to have problems with signal integrity, radio emissions, and immunity. All that happens.
MATTIES: So, how does somebody go about hearing this and then applying it to their practice?
BEEKER: There's a book that Ralph did that I think if you were going to buy one book, it would be a book he released a couple of years ago, "Digital Circuit Boards: Mach 1 GHz." That is probably the best combination of science and practice that's out there right now. He does a really good job of, chapter by chapter, introducing pieces of the puzzle, field behavior, and mapping it into real words. Then, at the end of each chapter, a review gives you a chance to decide whether you've learned this particular idea or not. Following those techniques will help you to understand the physics better, and it's not burdened with a lot of calculus. It's there so that the people who want to see that, have the science, the deep science, but that normal mortals like me can understand.
MATTIES: What quantifiable data or empirical data do you have that says, “I did it this way and I moved to this, and this is the impact?”
BEEKER: Well, actually, I did a test board, and everybody talks about adding ground planes and how important that is from a signal integrity perspective, but nobody could ever quantify that. I designed a board with two layers, with a quad flat pack micro in older technology, and what we did was basically connect through six-inch serpentine traces, all the IO pins to a resistor to ground, and then we toggled those pins as fast as we could at the 40-megahertz clock speed. Didn't do any consideration for how we connected grounds because we didn't care before. Just wired everything up, it looked good, it functioned, and then we did some testing on it. We showed that this board had a hotspot of radiated emissions that was consistent with the switching speed of the core.
At that core geometry, 200 megahertz was the sweet spot, and I had seen thousands of failures in my customers’ designs with this technology right in the middle of the FM band, so we had conducted emissions of 35 or 40 dB at 200 megahertz. This was clock coherent core switching noise, and it was the depletion waves that I talked about. So I took the same board, put a core inside of two solid ground plains, and then didn't do any other changes to the layout other than to nail the resistors to the ground plain. No moving of components, no changing of traces, measured the same board and the worst was again the conducted emissions at 200 megahertz went down to about six DB.
MATTIES: What was your reaction to that, initially?
BEEKER: Oh, it was fantastic. It says Ralph's right, or Maxwell. We got 30 dB improvement by going from a randomly routed board to a transmission line board, and it wasn't even optimized. It could have been made a whole lot better, but it just showed the impact of going to transmission lines.
MATTIES: I really appreciate your insight. Is there anything else that you would like to share with young designers?
BEEKER: They really want to focus on the idea of fields, and try to understand that circuit theory provides some good platform for some of the behaviors, but what they need to focus on is the idea of fields that do the work. Moving fields result in current flow. Moving fields do the work in all the products that they're going to work on. If they can get a better understanding of electromagnetic fields and what that world looks like, they're going to be very strong contributors when they move out into their real jobs. And I would also remind them to be strong. Don't let the old farts tell you that you're wrong, because it's all about the space, and fields move in space.
MATTIES: Dan, thank you so much. It was a pleasure talking with you.
BEEKER: You're welcome. Have a great afternoon.
Carl Shattke: I Started Designing Boards When I was 12