Judy Warner (00:01)
Hi, Eric, always good to see you, my friend. It's been a long time and I'm excited to hear about all you're doing at DesignCon this year.
Eric Bogatin (00:07)
Yeah, it has been a while. lot going on. Very excited. Looking forward to coming out to DesignCon.
Judy Warner (00:13)
Well, I know you are next deep in all your work at CU Boulder and, you know, bringing engineers into the workforce. So completely understandable why you're so busy. So I know that you always do a lot of work as we go into design con, but why don't you just take a quick second.
for anyone that's living under a rock and doesn't know who you are. And then also maybe some tech trends that you personally are tracking as we go into DesignCon 2025.
Eric Bogatin (00:45)
You know, it's interesting because I've been in this field forever. And there are a number of us who we call them gray beards out there that have been in the field for a long time. And it has constantly evolved and seems like it's accelerating. And I think DesignCon is one of those places where you get to see a snapshot of where the industry is at each year. And you can see by the examples, the demonstrations that are shown on the straight show floor,
the evolution of the different technologies from the ICs and the graphic processors and the AI network chips to the networks themselves and the interconnect technologies to do a board level interconnect as well as board to board interconnect. And one of the highlights for me is I always love going to the Samtech booth because they are always showing the latest highest bandwidth system.
with someone's chips, someone's receivers, their connectors and boards and physical systems that are working away. So you see all the pieces, your phrase of the ecosystem, you see the whole ecosystem at work in what a system is. So over the years, I remember in the old days when they were showing 10 gigabits per second, oh wow, how cool. And then went to 28 and 56 and then PAM4 was coming out and then
Judy Warner (02:06)
Wow. Yeah.
Eric Bogatin (02:14)
It was 128 or 112. And now even last year, there were some that were at 224 gigabit channels that were functioning. There's a company, we always debate in the industry this trend of at what point is copper going to crap out? When is copper going to lose steam? And you got to switch to optical. And it's one of those trade-offs between you want
Judy Warner (02:33)
Right.
Eric Bogatin (02:41)
to, you know, it's the length bandwidth product. The longer you go, the more limited the bandwidth is. And so at some point, you know, that's why we use optical fiber for long haul. And even within a building, we use optical fiber because of the distance. And when you get to the 224, how far can you go in copper and you have to switch to optical? And that's where Samtech, Mollux, and of these other connector companies have switched to the cabled.
Judy Warner (02:51)
Right. Right.
Eric Bogatin (03:10)
interconnects or cabled back planes because you can get a little farther in a cable than a circuit board because the cross sectional circumference is a little larger. But even cables have a limit. And so where do you make that transition? And I've seen a couple of companies coming out with optical transceivers that mount onto a circuit board with a BGA with they take electrical signals going in and there's a fiber bundle coming off.
Judy Warner (03:15)
You bet.
Eric Bogatin (03:39)
And that goes wherever you want it to go. Single mode fiber, really high bandwidth, know, 64, 32, 64 channels. Nubus, I think is one of the companies I saw last year. old buddy Pete Poupoulakis moved over to Nubus a couple of years ago. I've just been so impressed with what they're doing now. And it makes it transparent. So you put one of their chips on your board, you send all of your electrical signals in, it does the translation into optical.
whatever protocol you're using goes off wherever you want, another receiver turns in electrical there and then comes back in. So I really think that this idea of the system design for large high data rate is gonna be one of the fun things to look at and to see who else has that electrical to optical transition because it's absolutely inevitable that we have to go in that direction eventually. So that's one of the big things.
Judy Warner (04:34)
Yeah, I was,
well, I already talked to Matt Burns. So I'm smiling at you because I already drank from his fire hose of like, you're doing what? How fast? Because last year when I talked to him, he was like, we're showing 224. And he's like, yeah, it's not exactly mainstream, like 224, we're talking about 448. And I've just like, it's...
Eric Bogatin (04:44)
Yeah.
Yeah.
Judy Warner (04:57)
It's amazing that it never seems to stop and I don't think it's ever going to. It's just going to morph like you said and I happen to ask him about photonics because I see it popping up all over and I wonder about you know what role that's going to play in optics and you know it'll be interesting but
Eric Bogatin (05:03)
Yeah.
Yeah,
they're all tools in the toolbox. And it always comes down to what are the physical limits and what are the cost limits? At what point is it cost effective to switch from copper to photons or optical fiber? The other big challenge, think, is thermal. when you've got part of that is power distribution, you get some of these AI chips that are 1,000 amps.
Judy Warner (05:30)
Right. Yeah.
Eric Bogatin (05:45)
Right now, SI Journal, came out with this issue in January on thermal issues, thermal management. And I looked at an Nvidia last generation AI board. It uses almost a kilowatt. And 750 watts is a horsepower. And so on this one board is more than a horsepower of power.
being consumed, which you have to get rid of. And one of their chips now is more than a horsepower. It's more than a kilowatt of power. How do you get that heat out of the chip, out of the system? And I think there are going to be innovations we're going to see about distributing the DC current and the voltage, and about getting the heat out. And think combinations of
Judy Warner (06:27)
Right.
Eric Bogatin (06:42)
liquid cooling, heat pipes, and full immersion cooling. I think you're going to have opportunities going in the future.
Judy Warner (06:52)
I saw something, I don't know if it was on Instagram, TikTok or something, and it was in Japan and they had a huge NVIDIA. It was a data center and they had these like eight processors on one board, completely immersed and cooled, had a whole plumbing system and it was completely quiet. And it just looked like science fiction. I'm like, this isn't a data center, where's all the racks? And it was completely different.
Eric Bogatin (07:02)
Yeah.
Yeah.
Yeah.
Yeah.
Judy Warner (07:21)
It's like, to be able to, know, for me anyways, we talk about it all the time, but I've never had a chance to see it. And I haven't been able to track that down again. Maybe I'll go look for one of those because it's one thing to talk about. And then when you see it and see it operating, it's wild. And so I did talk to also this week, Steve Sandler, Heidi Barnes and Ben who are presenting that paper. think at least part that was out of that article you mentioned.
And what people have been saying to me is power in thermal out. Like that's a theme. Right? How do we get this much power running through these systems, but at the same time keep the thermal out and make sure that we have a good PDN. It'll be interesting. We're never bored.
Eric Bogatin (07:57)
Yeah, big challenges.
You know, one other,
you know, we spent a lot of time, you know, as engineers we're enamored with the big challenges. And so we pay attention to the high end, fastest data rate, the highest power, the highest current. But there's a lot of challenge at the other ends and distributed throughout systems. I continually, you know, cause I spent so much of the time training engineers at school and I get calls from industry. I continually see the new batch of engineers that are
entering the workforce, getting involved in this, they're still, you know, we have embedded knowledge in the industry, but it's not being absorbed or transmitted to a lot of the new engineers out there. I think we're not very efficient at helping new engineers get up to speed so that they don't have to reinvent the wheel and so that they're learning foundational engineering principles rather than special cases here and there and not being able to integrate. So I think, you know, what the other value of DesignCon is the educational
Judy Warner (09:04)
Mm-hmm.
Eric Bogatin (09:09)
value for engineers that are low on the learning curve to use this as an opportunity to accelerate up the learning curve more quickly.
Judy Warner (09:16)
Mm-hmm.
I couldn't agree more. And those principles apply to every design you do. And like you said, whether you're at 32, if you can master 32 or 56, man, you're killing it because there's a lot, lot right there, right? And that's hard to manage. So I think that's well stated. So tell us a little bit, Eric, about the things that you'll be teaching.
Eric Bogatin (09:24)
Right? Every design.
Yeah.
So, wow, I've got a busy schedule. I've got two technical talks that I've participated in. One is with one of my graduate students that just graduated. So he's off the market. He's got a great job in the company in Santa Clara. But it was an idea that we had. We were actually working with Rodian Schwartz. And they graciously lent us one of their network analyzers. And I love doing these two-port low impedance measurements. And one of the principles in PDN design is
It's nice to know what the power distribution network looks like on the board, but what you really care about is what does it look like on the die, on the silicon itself? And it's difficult to probe the die when it's on the board. And so we have developed this technique for other applications of using IOs as sense lines to look back into the die. And we've done a bunch of webinars, number of papers, we've shown a design cut in the past.
Judy Warner (10:35)
Hmm.
Eric Bogatin (10:39)
of looking at switching noise on the die by using the noise on an IOPIN. And so we said, could we go backwards? Could we use the IOPIN that when it's an output high, for example, there's a P channel MOSFET that connects the output to the power rail. And so we said, that's an external connection through the P channel onto the die. Could we use that to probe the power distribution on the die looking outward into the board?
Judy Warner (10:45)
interesting.
Eric Bogatin (11:09)
And so with the support from Brady and Schwartz, graduate student and I, we put together a project and design, and I was very surprised how well this method works. So we're going to talk about how can we do in situ power distribution network measurements from the die while it's on the board. And we can compare what it looks like when you look through the die versus when you're looking on the board. And so we did a lot of measurement and modeling of those. And I think we have a...
a pretty cool technique that we want to try in other applications as well. And then one other technical talk with my buddies at Cadence and Al Neves over at Wild River. This was a great opportunity to work with some world industry experts. And it's really a different perspective on how do you use a digital twin? How do you use the simulation environment to help you predict
Judy Warner (11:42)
That's great.
Mm-hmm.
Eric Bogatin (12:07)
the behavior of the real thing and how do you engineer the best practices in order to get better agreement in that digital twin so that you can reduce the margin in predicting the behavior of a system, adding less margin from uncertainty. And through this study, they did most of the work. was there as kind of just on the side here throwing little comments at them.
but the Cadence and folks in Wild River did most of the work. Basically what has evolved is how we have to be very careful when we make inferences about one channel, a specific channel that we measure and apply that to a bunch of channels on the board because of that manufacturing variation. And using the digital twin, we can explore what the impact of that uncertainty is from manufacturing variation and sensitivities in the design.
Judy Warner (12:57)
Mm-hmm.
Eric Bogatin (13:06)
So that we originally were looking at anisotropic material properties and we realized, my gosh, a lot of the manufacturing variations on the board hide what might be anisotropic material properties. And so it's hard to kind of distinguish those.
Judy Warner (13:09)
Mm-hmm.
Mm-hmm.
Did Bert Simonovitch did a paper on that? I feel like I interviewed him last year talking about that very issue. Okay.
Eric Bogatin (13:24)
Yeah, yeah, right.
You know, this whole thing of how do you, how do you kind of analyze the impact of anisotropic material properties that kind of got us started on this and realize, how do you go about measuring them? How do you distinguish what is an effect due to anisotropic material versus something else? And we realized in this study that it's really hard to separate out the uniqueness of
Anisotropic properties versus manufacturing variation. And that provided a lot of insight about be careful how you interpret the results that you see sometimes. So I got two papers and then I've got a bunch of little things doing a chip head presentation. I always enjoy doing those about, I love free tools and we are all lifelong learners in the industry.
Judy Warner (13:57)
Mm.
Right.
Eric Bogatin (14:20)
Even though your company may have access to all the high end tools from Keysight and Ansys and Siemens, you may have all those great tools. But when you're just playing around, when you're at home, when you're on the weekend, when you just want to try a little experiment, you learn something yourself, you want something that you can use yourself. And so I'm going to mention five of my favorite free tools that anybody can download and use. You're not going to design your 224 gigabit channel with them.
Judy Warner (14:39)
Mm-hmm.
Eric Bogatin (14:49)
but you are going to learn something and have fun with them if you're a lifelong learner. So that's a Chiphead Theater. And then I'm doing a presentation with my buddies at Rody and Schwartz. They're doing a workshop and I'm doing one on one of the popular topics of how to read S-parameters like a book. So if you're confused about S-parameters and you see these squiggly lines, you want to make sense of them, I'm going to show you how to kind of decode those patterns and what patterns to look for that tell you about
Judy Warner (14:54)
I love that.
Eric Bogatin (15:18)
features in the interconnects. So, yep.
Judy Warner (15:22)
That is a lot as usual, but
you do it with a big grin on your face. By the way, wanted to tell, give our, in addition to all that Eric is showing here, Eric and I've been working together to create an S-parameters class that will be online. So if you're not able to see his,
particular presentation. will be making announcement next month. Eric will come back and we'll tell you about that. But we will have an online class for you. So just wanted to give you a sneak peek of that. tell us a little bit about where you personally are paying attention to things. I know you work with students all the time. And so what are some things that
you know, either you're sugaring up in your classroom, you're working with SIJ, we know you have a podcast that maybe some of our audience would like to tap into again, if they're not able to attend Design Con Air.
Eric Bogatin (16:23)
Sure. Yeah, I mean, actually there are a couple of different sources. I'm always constantly creating more content. I do a webinar series with Teledyne LeCroy and full disclosure, I'm a fellow with Teledyne. I've been working with them for 12 years after they acquired my company. We've got a whole series of webinars we've done over last couple of years about kind of taking from one of the classes that I teach about circuit board design and kind of best practices and
circuit board design and measurement. So the whole series of webinars there, got a slate of them coming out the rest of this year. Of course, Signal Integrity Journal and Tactical Editor there. Got a lot of great content, doing webinars on that. And I have a new book with Artek that is just going to the publisher now. It's going to be a handbook on using oscilloscopes. And so it's really going to be, you know, mentioned before these young engineers just getting started, you know,
Everybody uses an oscilloscope, but sometimes we don't use them the best way. And I see that in my students all the time, that I will see them pushing buttons. It's like a video game. They keep pushing things until they get something on the screen they like instead of doing it with intent. And so this book is really how to do it with intent and how to understand all the features of what you're doing along the way. And then we've got some projects with you.
And so I hope some of our content will appear on your site as well.
Judy Warner (17:52)
Yep, very excited. Always happy to share you with our audience that each up and everything you share. So thank you for that. When is your R Tech book going to be published, Eric?
Eric Bogatin (18:02)
Well, they would have liked it last year. But, but, know, I wanted to make sure it was, you know, it had all the stuff I wanted. And so it's just in copyright right now. And we're looking for June to have it released on the street. And it'll also have a series of videos associated like the other two books. So one last plug. So, you know, I've got two other books with our tech and they have videos. So they're kind of multimedia and the videos are all available for anybody on the signal integrity journal site.
Judy Warner (18:04)
Hehehehehe.
I love that.
Eric Bogatin (18:31)
If you go to signalintegratedjournal.com, you pull up under education, we have a new feature of SIJ University. And we have a collection of videos there. And the two books I did with Artek, all the videos in that book are posted there. And we're growing additional videos as well. And so that's a great free resource to take a look at.
Judy Warner (18:50)
Fantastic.
Okay, well anyways, I'm looking forward to seeing you. I will track you down if nowhere else. I will see you. We don't have a booth this year, Eric, so I get I'm a little more footloose. So I will come find you. Yes, I'll be a rover this year. Okay, well, I'll definitely come see your your at least some of your sessions. So thank you for coming on and giving us a sneak peek of what you're doing for our listeners. Everything we talked about, I'm to put that in the show notes for you and something I haven't talked about it.
Eric Bogatin (19:05)
You'll be a rover. Yeah, I know I will find you.
Judy Warner (19:24)
for a long time is when I first started the ecosystem, Eric and Teledyne LeCroy gave us 90 days to like 200 hours of Eric's content. So I'll make sure that I'll share that with you. And I also have the 100 plus PCB design tips, also a little pocket guide. So if you go join the ecosystem community, you can have that for free if you're not able to.
Eric Bogatin (19:37)
Yeah.
Judy Warner (19:50)
Go to DesignCon and we'll point to SIJ University. you won't get FOMO because you can't go to DesignCon. Eric, thanks again. I look forward to seeing you in, gosh, about two weeks, I guess, or eight days. Yeah. So I look forward to seeing and thank you for coming on the show. Well, for our audience, bye Eric. For our audience, thanks so much for joining us today. I trust you enjoyed this conversation with Eric Bogaten. If you're at DesignCon, look me up. I'll be wandering the floor.
Eric Bogatin (20:03)
Yeah, less than that. great. Great chatting. OK, bye bye.
Judy Warner (20:20)
Until then, we'll see you next week and always remember to stay connected to the ecosystem. Done.