[00:00:12] Hello everyone and welcome back to the Cisco Optics Podcast where we talk about pluggable optics for networks. In the pluggable optics industry, every next generation is led by countless presentations and proposals at standards bodies meetings along with lots of lab development and testing behind the scenes. Right now, the industry has its sights set on 3.2 terabit transceivers with 400 gigabit per electrical lane. All those meetings and all that testing are happening as we speak.
[00:00:39] Today, I am joined by John Calvin, Strategic Planner and Datacom Technology Lead at Keysight. John has been bridging the measurement science gaps of emerging telecom and datacom development efforts for 20 years. He has chaired multiple physical layer interoperability working groups and is currently serving as contributing member to IEEE 802.3, OIF-CEI, InfiniBand, and PCIE development efforts. John holds a BSEE from Washington State University.
[00:01:07] And his graduate-level studies are in signal processing from Stanford University. In Episode 69, we continue our conversation with John. We get into the measurement science and precision instruments needed to characterize signals at femtosecond timescales. And now join me as I talk with John Calvin.
[00:01:35] John Calvin is one of the fundamental tools that's used to characterize the core components of a network. So you said the customer is involved in the process. Does that mean that customers said, like, hey, we need a 250 gigahertz VNA? Yeah, absolutely. And then you just had to go figure it out? Or were they, was there any collaboration? Well, you know, so our last VNA was 120 gigahertz VNA, right?
[00:02:01] So it's not uncommon for people to say, all right, we need to have twice this performance here in four to five years. And we have large customers that are on the cutting edge of these design cycles. They watch high-speed surities design and channel architectures the same way we do. So they have their own planners there and they communicate regularly to key sites saying, all right, we need to make sure we have a timetable here where we get to a 250 gigahertz VNA.
[00:02:29] And to be honest with you, people are asking for a 300 gigahertz VNA. We just don't know how to build a 300 gig one, but we can build a 250 gig one. And we did as a proposition of build it and they will come and they do come in this business because these are the tools that allow you to get ahead.
[00:02:47] Remember when I was talking about that $1.6 and $1.8 trillion market, these tools are what enables you to get in and address those design requirements. So if you can get a two-year head start over your competitor and a $2 trillion market, you're going to do it, right? It's absolutely essential to your survival to be there at the cutting edge. And these tools are what gets you there.
[00:03:13] So, but when they, I mean, you're making these projections, it's really on just your team to just figure it out? Or is there, do you benefit from going to these IEEE meetings and having a community to just banter with? The community is important here.
[00:03:36] And in fact, these are, they give our primary contacts for the people we value the most here and trying to track high-speed design. You know, if you take a look at this IEEE session that we have going right here, we have 120 of the best engineers on the planet focused on high-speed networking in a room together. And we stand in the hallway, we talk, we do joint projects and papers together.
[00:04:02] And it's that kind of collaboration that drives the industry and drives the test and measurement industry and lockstep moves us forward. And that kind of interaction with the customers is essential for this to work. On the other hand, you know, this industry is, and IEEE is particularly bad at this, is IEEE is really good at focusing on the problem that's right in front of them.
[00:04:29] So they're just trying to get 200 gig out of the way, then they got their sights set on 400 gig. They got to make 400 gig work. And then from our standpoint, the timelines are being compressed to the point where it's hard for us to focus on the next speed grade. We have to focus on the next speed grade after that. So that's why we're thinking about, you know, what does it take to make 800 gig work?
[00:04:55] And there are people out there that are, who are pushing this forward saying, you got to get 800 gig scope and you have to have 800 gig capable of VNA ready by 2028 because it's coming. The technology is advancing that fast right now. The, uh, you don't, you're not going to see a hundred gig transmission system. That would be a 6.4 terabit link, um, anytime soon, soon, but the R and D process is starting now.
[00:05:25] And that's, uh, what's driven by the availability of tools. So you can inspect and design and, and root cause, uh, the physics challenges that you're going to see as you go to these speeds. I had written down, uh, some questions. Um, I'm not sure if you have a single answer to these, but if you don't mind, let me, can I shoot, shoot some of them? Oh, absolutely. Go ahead.
[00:05:53] Uh, what has been the most difficult problem to solve? Uh, you can take this in any context you want. Oh, you know, um, you know, there's a, so Keysight in my current role at Keysight, we make multiple instruments, right? So we'll make a sampling scope. Uh, traditionally sampling scopes have been, uh, the performance leader.
[00:06:19] And that's, that dynamic is slightly changing here now, but sampling scopes are highly sought after in the communication space because of their extraordinary fidelity. They're cost effective. And, uh, the communications industry just loves their sampling scopes. Um, moving forward, uh, we have been challenged by the ability to build physical clock recoveries that will work at these speeds.
[00:06:47] Um, and, uh, as a result of that, we've had to transition and we brought our real-time scope performance, a real-time scope portfolio to bear on the same market. So if you look at Keysight's measurement capability here for 100 gig or 200 gig, or even going back to 50 gigabits per second, uh, you'll see that you can buy a real-time scope or you can buy a sampling scope. Uh, they're two very different instruments, very different architectures.
[00:07:15] And yet people still expect you to give the same answer regardless of which instrument they're using. And people, and rightfully so, they pay a lot of money for these instruments. They expect them to be traceable to NIST standards. And, and unfortunately there are no NIST standards for things like jitter decomposition. So it's up to us to make sure the answers come across consistent.
[00:07:41] So one of the challenges that we've had is, you know, if you measure jitter on a sampling scope and you get a hundred milli UI of jitter, and then you go measure it on a real-time scope and you get 120 milli UI of jitter. And people want to know why. And, uh, uh, we're, we're inclined to say, wow, you got jitter numbers within 20 milli UI of each other? Have a party. I mean, that's, that's great. And people go, no, no, it's, it's a different number though. You don't get it. It's, it's not a hundred. They should both be measuring a hundred.
[00:08:12] So, you know, working through these sorts of dynamics and having two instruments that solve the same problem, um, has been a blessing. And has it been a curse? And the curse is, you know, trying to make sure the two give you the same answer. But one of the questions that we get, people ask us all the time, which is how do you know this jitter is correct? And cause again, there's no NIST traceable standard. You can't go by a standard cell from NIST.
[00:08:42] That's the 50 milli UI jitter source. You can't do that. And for us, the, uh, the, the fact that you can come out the problem from one architecture and you get an answer here and you can come at the same problem from an entirely different architecture and get an answer. That's almost exactly the same to within a certain error tells us we have converged on the same, the correct answer.
[00:09:11] It's, um, you want to have accuracy or do you want to have consistency in this business? And you really want both, right? And when you have one instrument in the game, you get to play the consistency game, right? As long as instrument to instrument performance gives you the same answer, you're fine. But if you have two different architectures, you have to play the accuracy game. You have to make sure that they both give you the same answer.
[00:09:38] And, and that's actually the, one of the more challenging undertakings and, and Keysight to their credit has overcome this. Um, we have SQE teams and design teams. They're testing high speed certies solutions and they have a quality metric that they go through where they, they have a common. A common transmitter, a transceiver that's putting on a test pattern. They'll hook it up to the DCA with the new firmware. They'll hook it up to a real time scope with the latest firmware.
[00:10:08] You got to get the same answer. Or if it's different to within a certain margin, it sets off flags. We go look at why. So it doesn't take much to cause these numbers to be offset. Things like PLL behavior or vertical dynamic range or noise issues all contribute to the variation that we get in these instruments. So on both systems, completely different architectures and both systems, everything has to be working just right for you to get the same answer.
[00:10:38] Because you think about it, you do the math on these. And so we're, we're joking about 20 milli UIs, right? How, how big a number is 20 milli UI? Let's just do the calculation. So at, at 200 giga... What is that unit, milli UI? What is, what is UI? Well, a unit interval is a time interval over which a, a series of bits might get sent, right? So a bit time in effect. Okay. So the periodicity?
[00:11:07] In the old days, the NRZ, it was typically a, you know, a clock pulse width, right? Mm-hmm. That would be one unit interval. One bit time would be one unit interval. Nowadays, where we have higher order modulation in effect, we're now sending two bits for PAM-4. We're sending two and a half bits for PAM-6. We're sending three bits per unit interval for PAM-8.
[00:11:28] So we have, these are, this is the window of time where you have a series of different levels where the, the bit can transfer through and that communicates a message of the signal bit, whether it's going to be a gray coded symbol of a one zero or a one one zero zero or something like that. Mm-hmm. But a, a 212 gigabit per second signal has a unit interval that's 4.7 picoseconds wide.
[00:11:59] That's a very, very narrow window of time. And if you do the math, that's basically the time it takes light traveled that far. Okay. That's a really, really short time. Yeah. And if you do 20 milli UI of that, that's, that's roughly, that's a hundred femtoseconds. Wow.
[00:12:19] So to, to say that the two instruments are giving you the same answer to within a hundred femtoseconds, if you break it down to absolute terms and take the milli UI out of it, take 20 milli UI out of this thing, 20 milli UI is really a hundred femtoseconds. That's actually an extraordinary capability.
[00:12:39] If two instruments, different cal constants looked at the timing properties of a signal and triangulated within a hundred femtoseconds of the same answer, that's definitely party material right there. And that's an extraordinary accomplishment. Yeah. That, that's crazy. We, we, you know, I went to grad school and ultra fast optics are at that time. Our, our fastest optical pulses with our laser was around 80 femtoseconds.
[00:13:08] That's the width of the pulse, not, not the jitter, the width of the pulse itself. Yeah. Well, that's the other thing. The jitter was, was much more, I think. Yeah. Yeah. And the ability to resolve the edges there and actually extract a clock and get timing information has been a challenge there as well.
[00:13:26] So I think, uh, back to your question though, Pat, I think in terms of the greatest challenge is if you have, when you have two ways of solving the same problem, you're forced to be brutally honest with yourself and your customers. And, uh, it's a luxury that people who make a single instrument have that we don't. And, uh, from that process, we're, there are folks that will escalate, have a red phone next to their scope.
[00:13:55] If that scope even hiccups, they're calling someone at Keysight and giving us the riot act. Uh, so, um, we've learned the hard way and, uh, we keep these two instrument portfolios side by side and they, they track exceptionally well right now. But on the flip side, if you, if you do have confidence because they both give you the same answer and there's no NIST standard, couldn't this be the de facto reference? Okay.
[00:14:25] At this point in time, it is a de facto reference. This is kind of a unique spot that we worked ourselves into here. Um, Keysight has got, you know, hundreds of man years invested in some of the measurement science behind these standards measurements. And, um, the, uh, it's, it's, we've gotten to the point where, um, you can go out there and buy reference kits from analog devices that give you, you know, 80 gigahertz scope performance.
[00:14:55] Uh, you can buy low end scopes with A to Ds with the proliferation of the communications market being A to D based. There are now tons of A to Ds out there with performance. It's pretty, pretty impressive. Uh, so building a system that can acquire a signal is actually fairly straightforward. Being able to do any meaningful measurements with that data is where the real challenges are. And that's what differentiates us.
[00:15:21] We spend a extraordinary amount of time really perfecting the measurement science behind these measurements and capabilities that go into these standards. Whether it's measuring TDECQ or SNDR or JITR decomposition. These are things where Keysight has made a lot of really substantive advancements into the space and we're, we're happy to do it. It's all part of advancing the science of measurements. And that's what Keysight really stands for today. Okay.
[00:15:50] Uh, next question. And again, take it in any context you want. It's pretty open ended whether, whether you want to talk about test and measurement or go back to IEEE. Or just, you know, modulation schemes, whatever. What is something that would surprise most people in your opinion? I, uh, I think something that would surprise most people.
[00:16:26] That was the sixth part of my conversation with John Calvin. Next time we'll get into the growing need for engineers who understand analog behavior, signal integrity, and the physical layer. Please follow the Cisco Podcast Network on Spotify or Apple Podcasts. Better yet, leave a review on Apple Podcasts. Remember, you're listening to the Cisco Optics Podcast, which is part of the Cisco Podcast Network, where you can find other great Cisco Podcasts too.
[00:16:54] If you want to listen to us on a website, go to optics.podcastpage.io. We also have educational videos on YouTube. Just go to youtube.com and search on Cisco Optics. Thank you for listening. This is Pat Chow, Director of Technical Marketing Engineering at Cisco Optics. The next episode is part seven of my conversation with John Calvin. See you there. But in your shot at명 panel you can call yourself and see you next time with the software. See you there and see you on the menu.