#392 – An Interview with Matt Duff

Chris Gammell: This is The Amp Hour Podcast. Released May 13th, 2018. Episode 392. An interview with Matt Duff. Welcome to the Amp Hour. I'm Chris Gammell of Contextual Electronics. Hi, and I'm Matt Duff of Analog Devices. Welcome, Matt. How are you doing? Doing well. Doing well. I think some of our listeners would probably know you from your... Well, I think we've recommended your videos a couple times. I know I have, personally. I always recommend your videos about instrumentation amps and everything on YouTube. But you are an analog fan, so we're connected in that way.

Matt Duff: Great. I can say that it is a little... So I'll go look at some of my videos, and they'll have like 27K views. And then right next on the right-hand side will be one of Dave's videos with like 950K views. So, you know, there is... Yeah. Right. I'm pretty impressed.

Chris Gammell: Well, you know, you should just try a new accent and see if that helps. Yeah, exactly. So you've been doing analog a long time. How did you kind of get into the whole world of analog design and working in analog devices?

Matt Duff: Well, I first... When I first got out of college, I worked for National Instruments. Okay. You know, the folks that make LabVIEW and data acquisition cards. And I did some digital stuff and some analog stuff and certainly enjoyed the analog part more. And so then I went back and after I did that for a while, I went back and got my master's and then joined analog devices. So that was a useful... Like if you can, if you're not, you know, married or have other commitments, if you can, you know, go get your bachelor's and work a few years and then go get your master's. I found that really helpful.

Chris Gammell: You mean actually to get the work experience and then kind of bounce back into the academic world?

Matt Duff: Yeah, because then, you know, like I know when I was doing my undergrad, I was like, oh, that kind of looks interesting, you know, maybe that... But I didn't really like really didn't know what to take. And then when I went back to grad school, I knew exactly like, yeah, this is exactly what I wanted to do. Right.

Chris Gammell: The hands-on stuff that you actually like doing. Yeah. Yeah. Well, and I mean, the times when you, I mean, you've been in at analog devices for a while now. I mean, you've seen not the rise of digital, I wouldn't say, but I'd say the consolidation of analog, at least. I mean, not just on a company basis, but also more and more complex components as well. Have you seen that trend as well?

Matt Duff: Yeah. Yeah, so I would say when I, so I started at analog devices in 2005. And I would say, you know, when I joined, there was still very much a performance, like a performance emphasis. So, you know, we're going to, we're going to win by having the best performing parts. And, and ADI still has that in their, you know, bloodstream or, or, you know, that's certainly baked in into analog devices. But there has been a, a, our, we've been trying to shift toward how do we make it easier for the customer? How do we combine more stuff? Because there isn't, you know, eking out that little extra bit of performance typically isn't a, a game changer for the customer anymore. Now, now we need to concentrate more on, on how to make, you know, improve time to market, make things, you know, get designed faster and that sort of thing. What was the shift there, you think? Well, I mean, I think for a whole lot of our customers, you hit a point where the performance is good enough. And there's, you know, what's the, there's some Harvard professor with the, with the, oh, I can't remember his name. Is this Dan Ariely? No, the Christensen, Christensen, the guy that. Oh, Clayton Christensen, right. Yeah, exactly. Right, right, the innovator's on that. And so he talks about, yeah, exactly. And so he talks about how, you know, companies keep innovating, innovating, and they hit a point where they have performed, you know, the performance that they've got for their products are good enough. Right. And you, you keep innovating, but, you know, you can't keep getting more value out of that. Right. And so I think for a lot of customers, the analog performance is, is, is good enough. And so then the question is, okay, what, what other access can you innovate on? And, and certainly trying to do more system type chips and, and, you know, save the customer from having to do all this debugging and so forth. I think that's what we are, are, you know, there has been a intentional shift to try to work on that some more analog devices. Now we still have, you know, we still have the components based and we still are doing, you know, high performance components because those are what, those pieces are what these more integrated pieces are, are built on. And so if you don't keep, you don't, you know, keep up with that, then you might get in trouble. But, but there has been a shift.

Chris Gammell: Okay. Yeah, that makes sense. I, I've always, I have noticed over the past couple of years as well, the kind of the rising tide lifts all ships kind of thing in Silicon, right? Everything has at least a 12 bit ADC these days, whereas that would have been something not at least, but you know, like it's more common to have higher bit ADCs on standard, standard microcontrollers. And the noise performance is better than it used to be. And it seems like a lot of that is the process improvements. Like you're, you were kind of talking or alluding to at least. And, and now that it's in everything, once you have everything, then, then what do you actually, how do you differentiate yourself? Right.

Matt Duff: Yeah, exactly.

Chris Gammell: Especially, I mean, well, I, I guess something I've talked about, about analog devices in the past is, has been the, you know, splitting, splitting time with TSMC or other foundry models. Have you noticed that impacting the availability of, of, of technologies across, across your company versus competitors? Like, is there a little bit of seeding of, of process technologies?

Matt Duff: Well, I mean, we do, I mean, I, I'm not an expert in this, but I just, and, and, you know, I should, I should throw out this disclaimer and I should have thrown this disclaimer out, you know, at the beginning of our talk here is that I am speaking for Matt Duff. These are my opinions. These are, these are not the opinions of analog devices. So, and this is just my view from, you know, one, one seat in the company.

Chris Gammell: Yeah, right.

Matt Duff: So from, from my view, you know, I guess, and this is just my guess is that, you know, kind of half of, of what we do is from TSMC and half is our internal, internal fabs.

Chris Gammell: Right. And that was the announcement I mentioned too, is that it was, it was not a complete, it was, it was not that you were giving it up all of it. It was, that it was like co-pro, co-developing new technology with TSMC, I believe.

Matt Duff: Yeah. Yeah. Yeah. Yeah. And so, you know, certainly they, you know, they're a huge fab that, that has a lot, you know, a huge amount of fab expertise and there's some things that it makes sense to partner with them on.

Chris Gammell: Yep. Definitely.

Matt Duff: Yeah. Cool. But then there's other, you know, there's other analog specific processes that, that we have in house that, you know, nobody else can do and, and we'll keep having those in house to, to be able to, to make products that nobody else can do. Yeah. Right.

Chris Gammell: That's that secret sauce stuff that you kind of mentioned as well. Yeah. That's great. Okay. And I mean, I guess, I guess I'd be remiss if I didn't also mention the, you know, just the general consolidation of the industry, right? Obviously ADI bought linear tech. Exactly. I'd love to hear how that, hopefully that's all very positive news. You know, one of my long-term favorite brands. ADI is another one, of course, but I was, it was a little bittersweet for me to be honest.

Matt Duff: Right. Yeah. Yeah. We've heard that, we've heard that a decent amount. And I would say that it's very exciting to have, you know, I think we had a huge amount of respect for linear technology. And I think one of the, you know, interesting things. So my, once we merged, my boss, actually a few months after we merged was a guy from linear tech. Oh, cool. And so I got to, we got to, you know, kind of compare notes on a bunch of different things. And it's really interesting, you know, you have these two companies and they're both kind of going after a lot of the same customers and attempting to do kind of the same things. And so it's very interesting to compare notes and say, you know, we tried to do it this way and you guys tried to do it this way. And what did you learn? And what did we learn? And, and, you know, there's this, this, you know, slogan inside the company of trying to be the, you know, the best of both or whatever. Right, right. And so, you know, hopefully, I mean, I can say at least in my little area, I think we did a pretty decent job. Hopefully in the rest of the company, it's the same is that, you know, what we're, what we have been trying to do is learn from each other and hopefully, you know, try to take on the best characteristics of each. We did a, we did a website revamp. So the linear tech parts are on our website as of about a month ago or so. And several of the things from the linear tech, I mean, I'm sure there are linear tech folks out there. They're like, ah, ADI website is still not what linear tech was, you know.

Chris Gammell: Oh, well, I was never particularly fond of the website of linear tech. But, yeah. But one thing that we will get into is that the circuit design stuff that we'll be talking about, like some of the tools that you work on. And I was like, I went to the website and I was like, oh, LTSpice, that's supposed to be here. I was like, oh, yeah, of course, right. LTSpice and analog. Yeah. So that's a very positive thing. So, yeah, there's a lot of good things that came out of it, I'm sure.

Matt Duff: Yeah. I think we've been, you know, we were always quite jealous of linear tech that they had that LTSpice and it's a great tool. And I think that we are now super excited to have it. And one of the things that we've been doing on some of our web tools is to export to LTSpice, which we couldn't do before. So now it's a nice, you know, you kind of design it in the web tool and then you can dump it into LTSpice and tweak it or whatever. And so that's nice.

Chris Gammell: I will make a callback as well. We've had Mike Englehart on the show before. Ah, great. One of my favorite guests. And, you know, Mike is, well, he's. He's Mike, right? Yeah, he's Mike. Yeah. Someone wrote to me on Twitter the other day about being at a seminar too. And apparently there's like, oh, Bodie Plotts and switching regulators. I was like, oh, man, that sounds like Mike's still being Mike. So that's great. Yeah. Yeah. So that's very exciting. So you've been there for, you said, 15 years? Is that right? Maybe 12 or 13. Yeah, almost 15. Okay. Okay. So prior to the web tool stuff that we're going to be talking about here, what were you doing there?

Matt Duff: Yeah. So I was in, you know, ADI has a big site in Wilmington. And I was there in Wilmington for six years and worked as an apps engineer covering precision amps there, mainly instrumentation amplifiers, difference amplifiers, that sort of thing. Yeah. So I covered those. And then my wife's family is from Nebraska and my family is from Texas. And I wanted to move to be closer so that we could just drive to those places. And so I got, I was able to change jobs to a job that didn't require touching hardware as much. And so because it was more of a software thing, you can be based in lots of different places. Oh, yeah. And so I changed jobs and moved to Colorado here to do that here. Yeah, that's great.

Chris Gammell: Yeah. And then actually, we, you, you had written to me and we were talking a little bit and I was actually out in Colorado pretty, probably, I probably drove by you at one point. Yeah, probably so. As we were chatting. So that's, yeah, that was good timing. That's great. That's great. I think also, I think we did, I think you actually stopped into Keithley when I was working there. And I think we chatted at that point. But the main exposure I had to you was through your videos. And so you put a bunch, I mean, early YouTube videos, not early, early, but like, you know, you were on YouTube doing a lot of whiteboard stuff. And those were really great videos. I really appreciated those. And, you know, what was that experience like? Did you get a lot of feedback on doing those?

Matt Duff: Well, the nice thing about videos is that you can, you can see like in real time, like in YouTube, you can see the views. And so you can see like which videos are getting uptake and people can comment. So, you know, I guess the traditional way that, that we get information out there, design information is we write application notes or we write magazine articles. And so I've done those as well. And it doesn't feel quite as, as personal because you go write that and then it kind of goes out in the void. And then, you know, maybe somebody mentions it, but, but it feels like the videos are a little bit more personal. And then also like I will now meet people and they'll be like, oh yeah, I've seen you in your videos where it's very rare for somebody to be like, oh yeah, I read this article or whatever.

Chris Gammell: Right. Yeah, I guess it's like a dead, well, a down sampling of information if you're writing a, writing an article, whereas a lot of your personality comes through to the video. Right. Exactly. Yeah, exactly. That's great. Yeah. And I mean, like tons of great, great, like very niche stuff too, like, like instrumentation amp noise. And I remember because I think it was when I was working at Keithley, I was just like desperate for info. Like even, you know, I worked with a lot of great engineers, but sometimes you just need to look up more and just Googling around and found spectral noise. And I think I was, I forget what kind of calculations I was doing, probably noise, noise floor calculations or something. So really good, really good stuff. Yeah. Still making videos or no?

Matt Duff: I haven't made. So, you know, in my, my old job as an apps engineer, you know, large portion of my job was answering questions that would come in from customers on various things. And so, you know, you're, you're answering questions and you see trends on, on what people are interested in. And so then you say, oh, you know, I, I ought to make, I ought to somehow generate some sort of collateral on this. Uh, cause obviously lots of people are interested in it. Um, now I'm doing, uh, web tools and I have done some videos based off of like, we did this filter wizard web tool. And so I did a bunch of filtering videos based on that. But then I've also been working on parametric search tables for, for the website and improving our, our infrastructure there and so forth. And, and there's nothing like I can't, you know, uh, I don't think a video about like, okay, when you hit 300 parts on your parametric search table, it slows down your table. And so you should take these JavaScript, you know, changes to like, that's probably not very exciting for a lot of people.

Chris Gammell: It's a different crossover audience, I suppose. Exactly. Yeah. Not, not too much, uh, JavaScript and analog, deep analog topic crossover there. I guess you're one of them now. Yeah. It's all about very niche audiences. Right. Yeah. Okay. Yeah. Well, um, I'm sure that we'll, I'm sure some of the stuff here, I think that's the other thing too, though. Like it feels like, and maybe you have input on this as well. At least I, after I left the analog kind of design realm, it, it felt like there was fewer people talking about analog stuff. I don't know if it was because there was just more readily available end user parts that didn't require as much thinking or, or what it was, but I kind of just saw less of it. Is that, is that what you saw as well?

Matt Duff: Well, you know, I'm biased because I feel like I am, you know, if you had to, if you want to work on analog and you had to pick a company in the world to work on it, like it's hard to, you know, you know, I would have picked like analog devices or linear tech and now they're the same thing or whatever. Right. Um, and so, so much of what we, you know, that's our bread and butter. And so we're just kind of steeped in it. And so I feel like I'm not the right person to, you know, have that view because everything that we do is, is analog stuff. I had, you know, when I was at national instruments, I felt like, you know, I had a broader view because you could see all this other stuff too. Right. So, right.

Chris Gammell: Everyone's using, uh, everyone's using LabVIEW for everything.

Matt Duff: That's true. Right. Yeah. Yeah.

Chris Gammell: Yeah. Okay. Well, uh, so let's move into the web tool stuff then. So, I mean, so you are, obviously you're, you've switched, you've switched into that role. Uh, what, what is kind of the idea these days? Is it, is it like you're supposed to be able to go to a website and just kind of punch in numbers and out pops a designer or what, what is, what is the idea there?

Matt Duff: Yeah. So you want, you know, we have all of these standard parts and a lot of parts have been selling. They're still, you know, there's parts that we've done 20 or 30 years ago that, that people still love and, and think are fantastic. And what we want to do is make it easier for, for people to use all of our parts. So, so as I mentioned earlier, we, we are attempting to do kind of some more integrated type parts and, and, and make it easier to do the full signal chain that way. But you need, you've got to have a big enough market to spend money to do a chip. And once you've done that chip, you're, you're locked in into that particular solution. Um, and so there's kind of this in between area where there are applications that, you know, have a decent amount of, of, of volume, but aren't enough to, to make a specific chip for, but you could make some sort of tool or some sort of collateral that allows people to very easily take a few of your chips and put them together and create that thing.

Chris Gammell: And so, you know, could you give us a tangible example of that?

Matt Duff: Yeah. So like, you know, this analog filter design. So we have this massive portfolio of op amps and there's lots of different analog filters that you could design. So you, you can design them for, for, you know, low pass or high pass or band pass. You can design them to optimize for noise or low power. You can design filters that are filter at 10 Hertz. So you can sign ones that filter at 500 kilohertz. And so there's lots of variations, but the, but the theory, I guess is, is similar across all of those. And so if you can have a software tool that has the theory and does that theory for the customer, then it makes it easy to design the analog filter. And, and I know, you know, from past experience, when I had to design an analog filter, you have these textbooks and you have these recipes and you can follow books and recipes. Well, assuming that you have ideal components, but when they, they break down when you actually try to use real components. And so the goal of our tool was to show the performance, like show what trade-offs you had with real components and, and that helps people trying to do filters. And that also obviously helps ADI because, you know, we, we put in ADI parts. And so you can see, you know, Oh great. This ADI part works really well for me. And, you know, hopefully I'll go now use that in my design rather than, you know, maybe there's an equivalent part from our competitors that would work just as well, but that would take an extra day of, of calculations for me to figure out. So, you know, I'll just go use the ADI one. So that, that's kind of the mentality there.

Chris Gammell: Yeah. Marketing through education. That's, that's very common, especially with engineering crowd, I think. Yeah. Okay. And so when you say the, the, there's not the wherewithal to, to build a, the market's not big enough rather to, to make a custom part. Is that like, is that example, like making a a hundred kilohertz custom Chevy Chevy filter or something like that? Is that like the idea of that? That's the custom thing you're talking about?

Matt Duff: Yeah, exactly. So if you knew, like if there was a market for, you know, this hundred kilohertz Chevy Chevy low pass or something that was huge, then we would make a chip for it. And, and you would just, and, and in fact, linear tech actually, you know, has a few chips and, you know, we've inherited some, uh, filter, uh, chips to do. Okay. Yeah. So linear tech has some chips just like that. So there, you know, there is a little bit of that.

Chris Gammell: What are those targeted at? I guess I've never, I've never seen those, but I've, I've also never looked for them. You know what I mean? It's like one of those, you don't know what you're looking for until you find it.

Matt Duff: Yeah. So, you know, we've got, um, we've got these analog filter products. So if you go to our, um, website, I think they're under the analog functions, I think. Yeah. So we got analog functions and we got these, uh, filter building blocks and low pass filters and filtering is used. It's kind of all over the place with lots of different industrial applications or instrumentation applications. And, and in general, you know, a lot of the applications aren't that big, but there, there's a bunch of them.

Chris Gammell: Um, and so would it be something like an anti-aliasing filter or is that kind of the idea to have a, like a, like a hard, a hard version of an anti-aliasing filter so you don't have to build it each time? Yeah. Yeah.

Matt Duff: So, I mean, okay. So filtering is part of, you can, you can filter either in the analog domain or the digital domain. And, and the reason you filter is you want to clean up a signal. So you've got some sort of the signal that you care about out there. And then there's other, there's noise or there's maybe other interferers. And so you, you want to somehow just concentrate on that, that signal. And so you want to do some sort of filtering. And so then the next question is, okay, do I do filtering in the digital domain or do I do it in the analog domain? And the digital domain has a lot of advantages in that when you filter, it's going to be the same every time. So, you know, analog filters have some variation because they have, uh, you know, component variation. Right. So like temperature you're saying or over time it drifts or that kind of thing. Exactly. Exactly. Uh, so that's one advantage of, of digital filters. And another advantage of digital filters is that there, you can kind of program it. So, you know, because it's not fixed in your PCB, you can, you can program the digital filter so that on different scenario, you know, so you want a hundred kilohertz today and 10 kilohertz tomorrow, you know, you can do that. Um, and so, you know, a lot of applications, it makes sense to do a digital filter and then you have some sort of fairly cheap. And, you know, so a lot of times just an RC filter in the front end to do the anti-aliasing. So you, you, you know, no matter what you have to do some sort of anti-aliasing before your ADC so that you're not, you know, sampling, uh, you know, this higher frequency stuff. If it's beyond your Nyquist, then it's going to come in and then you can't get rid of it. So you, you, you've got to have some sort of anti-aliasing, but normally that can be pretty, you know, inexpensive and not a big deal.

Chris Gammell: So maybe we should take a quick break here. If someone wanted to find out what Nyquist is and anti-aliasing, and do you have any like, uh, base level resources that usually point people at?

Matt Duff: I don't. And maybe, you know, that's a good point. Maybe there's a good, I mean, that's a good video opportunity.

Chris Gammell: Uh, yeah. I mean, that's the thing too, is, is like, so like you were saying with, with being at analog, right? You, you've been in it so long and you've been doing, you've seen these problems over and over. Or I, I, at least some of the stuff I've done in the past where I'm like, man, I don't even, I don't even know some of the words you're saying. Right. And it's like, I mean, I, I know these words, but I'm just saying that in other industries and until you have the vocabulary, it gets so tough. So if there's people listening right now and you don't know these things, uh, Chevy Chevy is spelled. I'm just kidding. Uh, uh, there, I guess, I mean, I guess Dave's, Dave's made some videos about filtering, I guess. And, um, and maybe Matt will be here too. So, um, yeah, there's, there, there are some resources out there, but I guess,

Matt Duff: filtering in general, maybe like, uh, I mean, are you a art of electronics fan or, or, oh yeah. So art of electronics is fantastic book. Um, and I, I did make a filtering, uh, series of videos. There's probably about 10 in the series. And one of them is Chevy Chevy versus Butterworth versus Bessel. Okay. So you can, you can go, uh, look at that. Um, I don't have anything on, on Nyquist or anything, but, um, um, yeah.

Chris Gammell: Right. And I think, I think that a lot of the, I think when you get to the, especially analog filtering, I guess both, but you know, like when I started studying analog filtering, at least it was, it got real math heavy real quick.

Matt Duff: Uh, yeah. And that's what we tried to like, you know, that's a big reason we try to do this tool is so you don't have to do that math anymore because it's painful.

Chris Gammell: It is. Yeah. Especially because you have to like go back through and then recalculate. If you change one thing, it kind of propagates through and you have to recalculate a bunch of stuff. So that, that does make it very difficult. Uh, but I, I wonder like how, how much is the, so someone starts using the tool today and they're like, okay, I know I need to filter out anything above a hundred kilohertz. Um, do they, do they need to understand like the, the, the base level functionality of it or, or, maybe what, what is the starting point of someone who is using this tool? Is it still, you know, base level analog knowledge?

Matt Duff: Well, it's interesting that you mentioned it because we attempted in the tool. So this was, this is, and I think there's probably a case in a lot of software is you're attempting to serve both the folks that don't know very much about the subject and make it easy to use. But at the same time for the people that are the experts, you're trying to put in the functionality that they want. And so I think there's always this, this tension. Uh, between those, those two poles. Yep. And so what we tried to do, you know, I think the Chevy, Chevy, Chevy, Butterworth thing is a good example is we put in a little slider and on one side of the slider, it says you can, uh, you can settle fast. And on the others, maybe I should just pull it up here. Um, but there's a trade-off in the slider between, what is it? Fast settling and, um, the cutoff. So, yeah, so you can, you can use, uh, a few stages and kind of have long, uh, long step response, or you can use a lot of stages and settle really quickly. So we just have a little slider in the tool and you can move that back and forth. And it, as you move that slider, it shows you what your filter response is going to look like in all these different views. So the nice thing about a tool is you just go play with things and you, you see what, you know.

Chris Gammell: Yeah. Right. It's like a zero investment way. I mean, that's always a big thing about spice as well. I mean, I know, I know Bob Peace used to always rail against spice, but you know, I sure as hell didn't ever create any magic smoke when I was messing around with spice. Exactly. Exactly. So that, uh, yeah, rest in peace. Uh, but yeah. Uh, okay. So, uh, so that's the idea. So where, where can people find this tool?

Matt Duff: So, uh, we've got on our website, we have a, if you go to analog.com slash design tools, we have a whole bunch of design tools there or for this filter wizard tool itself. If you go to analog.com slash filter wizard, then it will, you can pull up that tool. Um, yeah.

Chris Gammell: Okay. Yeah. And so not, not to bring up competition here, but, uh, I think I used to rail against the web bench tool and I remember that was a national instruments thing. Now TI, uh, that was like for designing switch mode power supplies. I remember it was super, super, super simple. And I used to hate it because it was just like, it felt very kid-like and it just, it kind of felt like, like, like Lego, not Lego, but like, like little tykes. It felt like, and there was only like, I mean, I think some of it was the web technology limitations as well. And I remember like, like giving this feedback to national and they're like, okay, thanks. Thanks for that. And then I talked to someone else. They're like, oh yeah, I love that. And it was, and it was, it was because for, for me, I was looking to like dig in and really, you know, get into the data around, you know, the different types of inductors and all of the, you know, the buck versus boost and like actually see waveforms and stuff I was used to in spice where someone that was coming in and they were just clueless. They're like, oh, well, like here's the answer. And it's just like a couple options. And I just want the answer. I just want this function at the end of the day. And I think that is kind of what you're talking about as well of a beginner versus the more advanced thing. Exactly. So we, yeah. Which, which does it air on for you?

Matt Duff: Well, what we try to do is we try to make it, we, we start off with this idea of, okay, we're going to make this tool really simple for the beginner. And so that we can with very, you know, in a couple of minutes, you can get to the answer and, and you can see stuff, but you don't have to know much about the theory. But then what we also try to do in the tool is add, you know, we have these little question marks with, with kind of explanation of what we're doing.

Chris Gammell: Yes. I know those. Yeah.

Matt Duff: And we have, we have a whole bunch of different views and we, that you can see, you know, you're, you know, like probably your beginner is probably not going to care about things like group delay and phase delay, for example. Uh, but, and, and so you can go through the whole tool and never look at those. Uh, but we have those in the, you know, there's a little pull down and in the pull down, you can see those. And the same thing, like we later in later tabs in the tool, you can see the noise response. And so beginner probably doesn't care about that. Doesn't have to look at it. Right. But for more advanced engineers, they really, that noise where we show the noise performance and where it comes from in some, I think they get really excited by that. Um, and so that, that's how we've, we've, we've tried to make kind of the default mode really simple and straightforward. But then if you go click on some, you know, ring control or some control, you can dig down and get more expert advice. So that's, that's been our way we've dealt with it.

Chris Gammell: Okay. Yeah, that's great. Yeah. I'm clicking through it right now. And like you said, it's the filters analog.com slash filter wizard, right? Exactly. Yep. Okay. And I mean, at the end of the day, it's basically, it is popping out of circuit. Um, what do you expect people are going to do with all this? I mean, like just go and build it. Is that kind of the idea?

Matt Duff: Yeah. So we've got, uh, at the end, we have a little eval board system. So it's not like it magically, you know, populates the eval board for you. So it is just a blank eval board, but we, based on what you've chosen, we say, okay, you should have these little PCBs. Um, and then we give you a bomb at the end and say, okay, you know, here's your bomb, here's your eval board. So you should be able to just take these things and, and solder it up. Um, and then we also export to spice. So if you want to go, you know, if you don't trust our tool and want to go try it out and spice to make sure that, you know, it works the way that you expect, you can do that. Or a lot of times what people do is they'll, they'll want to tweak it a little bit. So they'll want to add, let's say a passive filter in front or behind the tool. Um, and so, you know, when you put it in spice, then you can do whatever you want. And so, uh, uh, that that's kind of the next step is you go simulate in spice or you can go, uh, filter, you can go build with an eval board, or you can just, you know, take the schematic. You've now you have a schematic and you can go take that schematic and put it in your design if you want.

Chris Gammell: Right. I mean, Matt, it seems like, it seems like your career has been like trying to get fewer people to contact you about asking you questions about how to do things. First it was videos. You're like, well, I could just do this once. And then people don't have to explain it to people anymore. Now it's like, all right, just make a tool. And people just do that instead of calling me.

Matt Duff: Well, you know, the name of the game, like, you know, I know, I don't know about you, Chris, but like when I was working, there were some times where I would have to like contact the semiconductor manufacturer. And I'd be like, ah, you know, I want to do this because it's going to be this, you know, I'm gonna have to wait a day or two. Three days. Yep. Yeah. And then probably they're gonna come back with some more questions. And I got to ask them questions. And man, I wish I could just find this on the, you know, let me spend an extra 10 minutes and maybe I can find this somewhere on the web. Yeah.

Chris Gammell: Yeah.

Matt Duff: And so I hope that, that I'm kind of meeting the needs of those folks because at the end of the day, you know, what we're trying to do is make it so that people can, can ship their product faster. Yep. And whatever we can do. And, and I, I guess I want to give a shout out to like, you know, people that are ADI customers, like let your sales guy know or, or, or use the feedback mechanisms that we have on the web. If there's, if there's areas where we can make your life easier and, and make it so that, you know, you can get your product out the door faster. Like, like we want to do that. So, you know, tell it like, we're, we're always kind of trying to, okay, what can we do? What can we do? You know? So, you know, if you know, just let us know. That'd be, that'd be, that'd be great.

Chris Gammell: Right. Well, you know, around here where I talk about chip printers, but that's still not a real thing. So that's just a joke thing. Okay. So how much does it matter? I guess. So like, okay. So people are going through this tool, right? And I mean, they're, I like it at the beginning. It's, it's low pass, high pass, band pass. That's just kind of high level filter types. And then it's actually picking like how many poles it's going to have and, and the, the topology. So if it's Shebby Sheb or, or whatever, Butterworth, how much does that end up mattering to the customer? Is it, I mean, it really is just a, is it just a reordering of components or, I mean, how much, how much should people care about learning the underlying technology there?

Matt Duff: Well, I mean, I would hope, I mean, the hope for the tool is that most people don't have to learn the underlying technology. So the hope is that, you know, you put in kind of what you want and then the tool through these different views shows you what you can get. And you look and you say, Hey, that's good enough for me. Great. Let's, you know, let's use that. Um, you know, certainly if you want to understand how the tool works, you can dive more into kind of how that works. But, you know, we've got this, like, like we have this little optimization control where you can optimize for power or, uh, noise. And, and based on that, we pick your components for you. We pick your op amps for you. But again, kind of going back to this novice versus expert thing, there were a lot of people that said, Hey, I don't want ADI to pick the op amp for me. I have particular op amps that I really like, or that I have in stock. And I want to go, want to go use those. And so we have a mode where you say it's the, I want to choose mode. And in that mode, you go and you pick your own op amps and then the tools shows you like based on the op amp you picked, here's the performance you're going to get. And, and maybe it's good or maybe it's not. Or, or, and the other thing is, I think like, I just want to show you the performance rather than trying to pick something for you because there's some people like, let's say maybe you have a filter that you only need 20 dB of rejection. Like normally a filter, like when you're doing some sort of active filter, like you'd want like 60 dB or, you know, you, something, something relatively substantial, but you know, maybe your application only calls for 20. And so if you have a, you know, a design that would look pretty crummy to me, but if, if that's all that you need, then, then, and, and it allows you to avoid having to stock a new amplifier, then fantastic. Right.

Chris Gammell: That is, that is the engineering trade-off. Oh, yeah. And yeah, I see you have different resistor series as well because that resistor series, capacitor series, because that basically determines how hard the component will be to get, right? Exactly. Yep. Okay. This is great. Yeah. This is really nice, actually. Is there, is there a control for cost?

Matt Duff: So we haven't, and I guess we, you know, that's, that's always a debate as to internally. So, you know, part of this is you're like, you're saying, okay, I'm targeting at these, at the customers that don't want to spend a lot of time. So they, they, they're probably doing kind of lower volume. So you, you have these super high volume customers and they can spend, you know, you know, six months of some engineer's time just to optimize a filter or whatever, you know? And so those guys, they're going to care about costs to the nth degree because they're, they're shipping millions of parts or whatever.

Chris Gammell: I mean, that's really what their job is too, is to, is to, to do those optimizations on the, on the, the small, small level. Right. Exactly.

Matt Duff: Exactly. But then you have a whole lot of customers that, you know, they're shipping, you know, hundreds or thousands or whatever. And, and the, their main thing is time. Like, like the development cost is the biggest part of, of how much this thing is going to cost going out the door rather than, you know, a little extra on components. And so that's kind of what this tool is targeted at. And so kind of internally, the question is we have thought about like, Hey, should I put a lower cost thing in? But then you put the lower cost thing in and yeah, then you, you, you, you give a, it, you give a design that's maybe not quite as good. And then the customer has problems with it. And then, yeah. So I don't know.

Chris Gammell: Yeah. I mean, that's the thing, like, like a small trade-off at the beginning, if you use like a, you know, a 1980s era op amp that you guys still make, then yeah. That might work and it might be super cheap or it might even be available elsewhere, but there also might be some weird corner case thing, like, like a phase inversion or whatever. Right. Yeah, exactly. And then, and then you don't realize that until you're six months down the line and it's like, Oh, well, thanks a lot, Matt. Right. Have you gotten that yet? Have you gotten a thanks a lot, Matt in a negative voice or mostly just positive voice?

Matt Duff: I don't, I mean, I think we've done, you know, fortunately phase inversion doesn't really affect filters. That's true. And that, you know, as long as, you know, as long as you're using the voltage in the kind of operating range of the part and, and part of, you know, what the tool does is it shows you, you know, here is the voltage range of, you know, there's a view that shows you here's the voltage range that you're going to get. So we, we have, and, and we've attempted to build in stuff that we expect people to run into problems. So like a real common problem with the Salon key filter configuration is that when you're doing a low pass filter, at some point, the low pass filter stops being a low pass filter and starts going up again. So you, you'll, you'll see this low pass characteristic. And then at some point, all of a sudden the line turns around and just starts heading up again. And, and so we wanted it all. Exactly. Exactly. And so we, we know that that's a potential problem. And so we built in, into the tool and, and that's, that's based on this interaction of, so, so what happening is that the signal is kind of going around your, when it starts going up again, your signal is going around your amplifier. So it's no longer going through your amplifier and, and it's caused by this combination of what the output impedance of your amplifier is and what your components are. So, and so we've built that into the tool so that like, let's say you picked your favorite op amp and your favorite op amp actually has a high output impedance. Then you'll see that, that upward going slope and you're in the view. And hopefully at that point, rather than when you've actually built it, you'll see, oh, this, I don't want to do that, you know.

Chris Gammell: Okay. Well, let's bring this back to kind of the, the target audience for this thing. I mean, where, where are you seeing people implement? I mean, this guy, I guess it's a very broad question, but where are you seeing people implement filter designs these days in like modern applications? Yeah.

Matt Duff: Yeah. So kind of going back to our digital versus analog thing. Yeah. So the, the, I was, the digital filters have a lot of advantages, but the big disadvantage is that you have to digitize everything.

Chris Gammell: Right.

Matt Duff: Before you can use that digital filter. And so you need a relatively expensive ADC. So, you know, ADI is happy. Like you want to do a digital filter and use a relatively expensive ADC. ADI sells a lot of really nice ADC. So go to it. That's fantastic. Right.

Chris Gammell: Um, but you're also going to be doing that in order to then throw away a bunch of those samples or, or, uh, otherwise you wouldn't necessarily need that. You're saying.

Matt Duff: Exactly. So, so filtering is reducing information. And so if you can reduce the amount of information that your ADC needs to sample, that means you can either sample slower or you don't have to have such a wide dynamic range for your ADC. Um, and so if you can do the filtering before your ADC in the analog domain, then, then a lot of times you can get away with like a 12 bit ADC or some, you know, much more, less expensive ADC. Um, so I think that's, I think in the modern era, that's a big reason that you would use an analog filter. Another, if you're, if you're doing kind of, you know, let's say 500 kilohertz type filter, you may also be putting a, a, some load on your, your digital. Back end and, and, and you'd have to up your FPGA or you'd have to up your microcontroller. Um, and so that's, that's potentially another reason is to save you work in the backend there. And then some people just don't want to mess with the, you know, the way companies are structured. Maybe you've got an analog engineer and then you've got the digital engineer and the digital engineer is, is fully booked and the analog engineer has a little bit extra time. And so, you know, you can use, you can use the, right. You can use the analog engineer and they design a filter and then that saves the digital engineer from having to do this extra work. So that, you know, that's, that's actually a valid use case too.

Chris Gammell: That's an, I've, I've never heard of that one actually, but I, I could see company culture for sure. I mean, like, yeah, some people just, it's all the way we've always done it. Right. That's, that's a very common refrain. Uh, yeah. Okay. So a lot of, I guess I, I think about like audio, uh, industrial, um, you know, I guess video, but that's pretty high bandwidth to start with. Right.

Matt Duff: Yeah.

Chris Gammell: Yeah.

Matt Duff: So above kind of, I don't know, five megahertz. Say, um, you don't want to use active, like the active filters kind of peter out. So a lot of times you'll still use some sort of, you won't use a digital filter. You'll use a filter that has like inductors and, um, you know, instead of using amplifiers to do the, the thing that inductors do, you just use inductors. So that's typically what you do at the higher frequencies.

Chris Gammell: Okay. Yeah. So this is a lot of sensor, uh, sensor processing and, and such to, to really, to get that signal in the range you want it to in the first place. Right.

Matt Duff: Yeah. A lot of industrial instrumentation type applications.

Chris Gammell: Yep. That's, that feels like home to me. Okay. That's great. Um, so what, uh, so what are some of these other tools? So you've worked on this one, obviously this is kind of a focus of yours, but, uh, there's, there's other stuff in there. I mean, uh,

Matt Duff: Yeah. So like an example of, uh, of a tool that we're doing right now. So we did this tool a few years ago, the filter tool. Oh, so this has been around for a while. Yeah. So this has been around for a while. It's been around for a few years. For a while.

Chris Gammell: Okay. Yeah.

Matt Duff: Um, so a filter, a tool that we're actually working on right now is this ADC driver tool.

Chris Gammell: Uh-huh.

Matt Duff: And, um, so if you, if, if you want to go to it, it's actually still in development and we have a little area for tools that are still in development that we call the playground. Uh-huh. So if you go to playground.analog.com, it kind of lists tools that, that are in development there. And one of the tools that's in development is this ADC driver tool. And, um, when you're doing not so much, you know, kind of, well, these ADCs that will be in microcontrollers, but if you're doing a higher end ADC, like a 16 bit ADC, driving that ADC is kind of tricky because you, you have to have a driver that's powerful enough to drive it. And then you have to set this RC filter in between, uh, and then it goes in the ADC. And if you, if you, if you do things like set the resistance of the RC filter too high, then it increases your distortion for your ADC. If you set it too low, then your amplifier can't drive it. And so there's these various trade-offs. And so what we, what we're doing now with this tool is try to create a tool where you can go, you can go do these trade-offs. And I, I think a lot of it is based on, you know, you, you get this 80, this nice ADC and maybe it doesn't consume very much power. And then now you have to drive the thing and you have to have a driver that consumes just as much or more power than the ADC does itself. Right. Right. Um, and so depending on what you're doing, maybe you don't actually, maybe you can get away with having an amplifier that's maybe a quarter of the, of the, you know, so in the data sheet, we put in the best amplifier to drive that ADC. So if you, if you use what's in the data sheet, then you'll get, you know, the performance of the data sheet says, but the best amplifier maybe consumes a ton of power. Right. So in your application, maybe there's a few specs that you can, you know, you don't need as much. And so we created this tool to allow you to, to play around with different things.

Chris Gammell: Hmm. Okay. That's great. And so, I mean, maybe could you, could you talk a little bit about the, the theory of why, why, why you need a driving amplifier? I guess you kind of touched on that, but like, why, why do people need amplifiers there in the first place for these higher end ADCs?

Matt Duff: So an AC is based around, uh, the switch capacitor design. And so what you have is you have a little capacitor, maybe 20 or 30 picofarads and, uh, it samples on that capacitor. And then internal of the ADC, it does all this charge redistribution to determine what the, what the value is. And so what's happening is every sample period, you're, you're plopping this 30 picofarad capacitance on the input of whatever you're sampling. And that has to settle within the tiny sample period that, that you have allowed. And so you needs, you know, you can't just like put a sensor there that has, I don't know, 50 K, 10 K, uh, ohms of impedance there, because there's no way that it's going to be able to, to drive that capacitance to 16 bit settling in that, that short sample period.

Chris Gammell: Got it. So it's like if the, uh, 30 picofarad capacitor is like a very, very tiny bucket, you're trying to see how much water you could fill up into that bucket. Is that kind of the idea?

Matt Duff: Yeah. So you got splashing, right? So you can imagine that you have this, this, this bucket and you've got to fill up the bucket and your water level is the top of the bucket. And you have to get to whatever one over 65,000th of the, of the, you know, close to the top of the bucket in this really short time period. And, and that's very difficult to do if you don't have something that's actively like trying to push the water into the bucket.

Chris Gammell: Right. So, yeah. Oh, okay. So it's not actually, so it's only doing one bit at a time though. That's the idea.

Matt Duff: Well, you have to, you know, if you're buying a 16 bit ADC, yet you're only, uh, settling to 14 bits by the time that you actually sample, then you're wasting a lot of money, you know? So, so like you have to get settled to the accuracy of the ADC that you're purchasing. Um, and so that's, that's the, and that's why.

Chris Gammell: Yeah. Yeah. I was just asking because you were saying the filling it all the way up to the top of the 30 picofarad bucket. I thought it was, you're filling it up to a, a percentage of whatever the percentage is of the, the overall input range.

Matt Duff: So if the. Yeah. Yeah. Uh, you're, you're filling up to a percentage of the input range, but you know, you want it at this exact point. Oh yeah. Right. And so you have to fill it up to just that exact point.

Chris Gammell: Yes. I see what you're saying. So yeah. So if the, the 30 picofarad bucket represents one volt total and you have a half volt, you need to hit exactly. Oh, you're saying, oh, that's the one over 65,000. Exactly. You need to hit it with to the point 5, 0, 0, 0, 0, whatever the precision number is. Yeah. You got to hit that within nanoseconds, milliseconds, microseconds. I don't know. Yeah. A lot of times. Depends on sampling, right?

Matt Duff: Yeah. So, you know, a lot of times these are, you know, a hundred kilohertz ADCs or, or, um, you know, 500 kilohertz ADCs. So that's, that's kind of the, um, so you've got, you've got a mic, you know, you've got a few microseconds there, but then a lot of times the, the, you don't get the full, like the ADC architecture itself says, you know, out of that, let, let, let, you know, let's say you're sampling it at 500 kilohertz or something. So, you know, in theory you would have two microseconds, but in reality, the ADC architecture says, sorry, buddy, I'm taking 1.5 of it for my stuff and you get 500 nanoseconds to, to do the rest, you know? So.

Chris Gammell: Yeah. Those ADC designers, man. Right. Right.

Matt Duff: Yeah. So that's, that's, you know, that's always an internal.

Chris Gammell: Right. Because I guess the water's got to flow back out of the bucket on the other side, right?

Matt Duff: Exactly. Exactly. Yeah.

Chris Gammell: Yeah. That's, uh, sorry about all the water analogies. I'm sure there's some old analog engineer out there listening and be like, it's not like water. It's just, just the easiest way we could talk about it. Right. So. Yeah. Yeah. It's this stuff. Okay. Yeah. No, this is great. This is great. And I mean, what has been traditionally difficult about, about, you said it was because of the, oh, you, you were saying because the, the specific amplifier that's suggested in the data sheet might be overdoing what might be over spec for what people really need. So this kind of helps to figure that out.

Matt Duff: Yeah, exactly. So if, if, if you can, if the amplifier that's in the data sheet works for you, like if the, you know, package size and the price and the power consumption and, you know, the precision of the amplifier, if those are all fine, then, then use that. Cause that's, you know, that's the part that's recommended to, to make this ADC shine. But there are a lot of people that come to us and say, you know, for whatever reason, I don't want to use the amplifier that you guys are recommending. What else can you tell me? And then, you know, we're like, well, it's, it's, it's real complicated, you know, and it'll take you three days to figure out.

Chris Gammell: Yeah. First do 30 years of analog design and then, you know, have a cohort of others who have also done that. And then you can, you know, talk about this cogent. Exactly. Awesome. Okay. Well, you should sell that. You should sell 30 years worth of design. I would buy that. I would totally buy that. Yeah. Wow. These are some fancy ADCs too. I'm looking at the, so you can actually switch out which ADC you're actually targeting and boy, they, they don't get any cheaper, do they? Yeah. They, yeah. I do remember some of the stuff too. One of the things that was difficult for me was when it was a differential amplifier versus a single-ended. That always kind of threw me off too. I know that there's parts that drive differential versus single-ended amplifiers. And you could do that conversion from single-ended to differential. But I, I would just like, I would just follow the data sheet too. I would just be like, okay, well, I just got to buy another part here and that makes it what I need. Yeah, exactly. So, so this kind of does help to allow more options you're saying.

Matt Duff: Exactly. Exactly.

Chris Gammell: Cool. Okay. Are there other things you can introduce? Like I can see the drivers are inverting, non-inverting followers. So you could, you could just, you could actually invert the signal if you needed to as well.

Matt Duff: Yeah. And so I guess, you know, we've got in the upper right-hand corner, there's this little give feedback thing. And, and as I mentioned, this is in development. So we're, we're still adding features. Got it. Got it. And one thing that we're, we're, you know, debating is like, okay, what should we add? When do we say this is finished? And so forth. So we're really looking for feedback from people using the tool to say, you know, I really want this or, or why are you doing an AC driver tool at all? Actually, I want this instead, you know, stop wasting your time. Like, like, you know, whatever, whatever, you know, if anybody's listening to this and wants to go try this out and give us feedback, we would love to have it because we are, you know, we have the ability. So that's the thing about software is that, you know, unlike hardware where you're constrained, you know, you, you, you try to tweak this thing and then it makes, you know, this other spec worse with, with software, you can just keep working and working and, and, and make it wonderful. But it, you know, then the development cost for this free tool, uh, goes out the roof. And so, uh, what, what we also noticed that software developers aren't cheap. Okay. Exactly. Exactly. So, you know, the whole, the whole, goal behind all this stuff is to sell more ADI parts. That's how, that's how I get paid. Yep. Yep. And so at some point we have to say, okay, you know, I don't think any more work is going to sell any more ADI parts. So let's move on to the next thing. Yeah. But that's always a tricky thing to determine where that, where that cutoff is.

Chris Gammell: Right. What are some of these other tools in the playground here? So the playground.analog.com, I see mixer spur graphical representation. What is, I don't actually, I also see RF frequency gigahertz. So I'm already, I'm already like this, this isn't for me.

Matt Duff: Uh, yeah. So that, that mixer spur tool actually came from, uh, Hittite. So we, you know, so we talked about how, uh, ADI, you know, LT and ADI merged a year or two ago. And before that we acquired Hittite and, uh, this was actually a tool of theirs. And so. What is Hittite? It's a, they're, they were an RF, uh, company. So, uh, they specialized in a lot of high frequency stuff. And so they had this tool on their website, but it was flash based. And so we, yeah. Yeah. So we, we converted it to JavaScript, but we never, uh, you know, the part of the point of the playground is let's, let's put some stuff out there and sometimes stuff doesn't graduate. And I think this is maybe one of those that's not going to graduate, uh, that that's out there for people to use. But I think we said, okay, from based on what we can tell of, of how much interest there is in this tool, we're not going to move forward to. So, you know.

Chris Gammell: I actually, I was actually just asking, I don't know what a mixer spur is. Do you know what that is?

Matt Duff: Uh, so it's for, we hit, we sell a bunch of these mixer parts, these RF mixers.

Chris Gammell: Like a RF mixer. Okay.

Matt Duff: Exactly. And, you know, as you use these, these mixers, do you create all these various spurs and frequencies maybe you don't want? And so the point of this tool is to actually, uh, show you where those are. Now, what I don't understand is, you know, it gives you, you know, you hit calculate and it gives you this, this very, uh, uh, you know, colorful plot. And, um, you know, I am no RF guy. So we basically copied a lot of the, the information that it was already in the existing tool. So our mandate was like, okay, make it like the existing tool. So we're like, okay, I'm not quite sure what we're doing here, but it works like the existing tool. The lines line up. Yay. Exactly. Exactly. And so, so that's kind of where, where this, this tool came from.

Chris Gammell: This sounds like exactly what I would do. Cause I'd be like, I don't know what that is. I need to get more RF guests onto the amp. All right. Apparently I, uh, I'm really confused. Uh, what about other tools on the website? I mean, like there are a bunch, uh, I mean, are you doing other, so like power management stuff and yeah.

Matt Duff: Yeah. So we have, uh, um, the, there's the LT power CAD, which I can't say that I, I, there's another, there's a, another group that's from linear tech that's doing the LT power CAD, but that's kind of, you know, it's a tool to allow you to design, uh, the switching regulator circuits. And, and again, kind of adjust various characteristics of that. So that's a downloadable when you, when you go to our site, we have these little icons, you see the little cloud and that's a web thing. And you see the little arrow and that's a windows installable tool. So the LT power CAD is a, um, windows installable tool. You mentioned earlier, we've got, I've got a video out there about instrumentation amplifier, common mode range. And, and, uh, you know, the, the fact that the inputs and the outputs, what you can do can vary with the gain and, and with the, you know, the supplies and so forth. And sometimes it can be a little difficult to, to figure out. And so we've got this instrumentation amplifier diamond plot tool where you, instead of having to go read those cryptic graphs that are in our typical performance curves, you can just put in your specific characteristics for your instrumentation amplifier. And then the tool says, Hey, this doesn't work or this will work, or, uh, this doesn't work. But by the way, here are these other four amplifiers that would work for you. Um, and so again, the goal is to, to make it instead of having, people have to sit and think you just put in your stuff and bam, there's your answer. Um, and you know, again, hopefully, you know, that's less time that the, the customer, you know, I keep saying customer that, I mean, like, you know, I'm sure there's lots of people listening that like, well, that's me or whatever, but, um, you know, our goal is to make your life easier.

Chris Gammell: Okay. Well, I have a request for you then. I, I did not think of this beforehand, Matt, I want you to fix data sheets. Why in the hell are data sheets, PDFs? This is something that I used to talk to people about at supply frame. I, you know, my old job. Uh, and like, what the hell? So like, you're talking about like, Oh, you don't have to go look at the chart. Right. But currently, and maybe this is changing, but like, so there's no way to trace on top of characteristic curves. Right. So like, if there's like a, if there's something where you have a 3.3 volt, you know, power curve and you have a five volt power curve, but for some reason you're operating at four volts, there's no way to like, to interpolate between them other than like, you know, pretty much guess and check, or even, even just like to, to move my mouse along a characteristic curve. No one publishes their numbers. Right. There's no like interactive charts. I'm sure this has been done before, but I want, here's what I want, Matt. And I want you to do it because analog is probably going to buy all the other companies anyways. Uh, uh, uh, I want, I want a standard data sheet format. I want, I want like a, like, why isn't there a chart charting tool for all characteristic curves? You know?

Matt Duff: All right. So you've got that playground.analog.com up, right? Yeah. Yeah. All right. So go to, well, I don't know. I mean, this is why we have this stuff up so we can get feedback on it. Yeah. So go to the sixth, sixth tool down that says performance gallery.

Chris Gammell: Okay. Yeah. Okay. Performance gallery. And we will link many of these things. I mean, we'll link everything in the show notes, but yeah. Oh, okay.

Matt Duff: Okay. So then this is, this is kind of the goal is to have interactive typical performance curves. And so you can go, you know, mouse over stuff, see stuff. You can zoom in. We have some filters where you can go, you know, see various, various things. So I don't know. You might look at it. Yeah.

Chris Gammell: Um, okay. But I need this for every data sheet that you guys make. Yeah. Yeah. For the past 40 years.

Matt Duff: Right. Yes, exactly. So that's the, that's the tricky thing. And, and, um, you know, I think there is a little bit of hesitation, you know, so you said, Hey, why don't you interpolate between, you know, so when we characterize a part, we characterize a part under certain conditions. So, of course.

Chris Gammell: Um, and this is, you're, you're about to explain exactly why this doesn't exist yet. And I get that, but go ahead.

Matt Duff: Yeah. That we do have a little bit concern, you know, so let's say we, we take temperature measures and said, I don't know, 25 degrees C and 85 degrees C. And that's what we've got. Well, you know, theoretically you can kind of linearly interpolate between the two, but in reality, we don't know how the part exactly performs at 55 degrees.

Chris Gammell: It's 57 degrees. Yeah. There's some weird anomalous polite behavior and yeah, you wouldn't have known about that.

Matt Duff: Um, so that's, that's one concern. Um, and then, you know, the other concern, like, I think this, this is quite feasible going forward, but then, you know, kind of to your point about all the parts going back 40 years, you know, the, the, the guy that did that part 30 years ago is no longer with ADI. And, you know, the little engineering paper or whatever that he did, you know, like that's no longer in some file drawer or any, you know, so, so the. Oh, I get it.

Chris Gammell: I'm just saying this is what they make interns for. I'm just saying, you know, there's a lot of, there's a lot of kids up the street and, uh, well, I guess you're in Colorado, but the, what, ADI is out in Massachusetts. MIT keeps graduating new people. I'm just saying.

Matt Duff: So again, if, if you're listening to the show and you go try this performance gallery and you like it or you don't like it, or you have other things, we've got that little feedback thing. Feedback. So just, you know, like we're driven by what people are saying we ought to do. Okay.

Chris Gammell: So how much feedback are you getting these days?

Matt Duff: A little bit, a lot of bit. Uh, so we get, you know, like, like on the tools that are more popular, we get, I don't know, I don't know, a fair amount of, you know, so maybe some feedback. Every week or something. Um, okay.

Chris Gammell: So that's not that much though. That's, uh, it's not, not. So what I'm saying though, is if someone is using, this is good. And, uh, this is good, good for our listenership. If you are listening and you have feedback, you'll be a squeaky wheel and Matt will probably do your bidding. I'm just saying. Exactly. Or his minions will. Uh, right. So, well, not much for minions. So yeah, yeah, exactly. Well, you know, maybe you'll get more minions.

Matt Duff: You need to get more minions. That's right. And especially if somehow you can say, you know, I, if, if. Right. I had this. I'm a customer. Right. Then I would buy these parts. But you know, since you don't have it, I might have to go to this other guy.

Chris Gammell: Yeah, no, I think, and I think this is kind of pointing out too, right? So I'm guessing that when you started at ADI 15 years ago, you probably wouldn't have guessed that you'd be working on web tools now. Right? No, no. Yeah. And I'm guessing the same, like, I never would have thought that that actually would help to sell more parts. But at the end of the day, like you started saying before, is like, this does help people get their job done faster. There's tangible value here. Yeah. That makes a lot of sense. Cool. Well, this, no, I'm actually, I'm now on the ADA 4522-2, which is an op amp. And I'm looking at zoomable charts. This is very Web 3.0, I've got to say. I don't know what web version we're on. I don't know either. Yeah.

Matt Duff: But it's got all the, you know, it's got the fancy JavaScript. You know, so one of the things that we do a lot in our tools is we try to do, have the tool run a decent amount in your browser. Yeah, I like that. Because for anything that runs in your browser, you don't have to have this round trip to the server and then back again. Yes. And so hopefully it's nice and snappy.

Chris Gammell: Yep. Yep. And I like that too, because it's, I mean, you know, during the days I'm usually on a Mac and at night I'm on Windows or Linux and like web-based stuff is just a little bit more cross-platforming than. Yeah. Yeah. You know, like sometimes I'll even, with all apologies to Mike Englehart, you know, sometimes I'll use the false dad simulator because, you know, LTSpice doesn't, it works on Mac. I know that, but like not that great, you know. Yeah. So, yeah, sometimes it's just easier to have a web tool. So, yeah. It was interesting before the show too, you were mentioning, what was the thing you were saying about books between the web?

Matt Duff: Oh, right. Yeah. Yeah. So the pace at which the web technologies are evolving. So it's amazing. So we, you know, as I was mentioning to Chris, I did this filter tool and all the electrical engineering theory, like I had all these books that were from the 1970s, you know. And the. That stuff has not changed much, you're saying. Yeah. The stuff hasn't changed much. And so, you know, I could use a lot of these and a lot of these books from then. And then I ordered some books on the JavaScript and the web technology so I could get up to speed on how that stuff worked. And I realized really quickly that that was just a dumb idea because by the time the books were printed, the stuff was out of date already. Yep. And so you really just have to go to the various websites and learn directly that way. And so I think that it is interesting what you can do now versus what you could do just five years ago in terms of web technology.

Chris Gammell: Yeah. Yeah, it's great. It's great. What do you, I mean, on the electronic side of things, are you seeing any trends in the analog world that people should keep an eye on? I mean, changing ADC specs or process tech that's making op amps change characteristics wise? I mean, I guess you're seeing more on the website these days.

Matt Duff: Yeah. I mean, I would say that, like, you know, in my little wheelhouse, I see a lot of kind of the web technology. But I see, unfortunately, you know, because I'm not unlike what I was doing seven or eight years ago where I was interfacing directly, you know, answering customer questions all the time. Um, now I'm, I'm more on this web technology, so I'm probably not the, uh, the best guy to ask. You know, the only thing I know is that, that we are driving, you know, how can we figure out how to make, you know, folks lives easier? And that's like, we're having to force our, so what ADI wants to do, what ADI is, you know, DNA is, is let's make better products. I want to make the next fan, you know, that's what everybody really wants to do is like, just let's do this new exciting thing and get this, you know, extra few DB of SNR or whatever. Yeah, right. And, and I think we are, you know, uh, having to be disciplined and saying, you know, look, if we, if we release this, this product, it may not, it may not sell as well as we want versus some other product where instead of doing that, we, we, you know, so, you know, maybe the next product we do and, and we're actually working on, you know, we've released some of these things. These products that, that the ADCs are easier to drive. So the, the, the vector that, that we improve was not the ADC performance itself, but that it was now you can use these very low power op amps and you can see some of these in that ADC driver tool. You can use these low power op amps to drive these new ADCs that we have that you couldn't do before. Yeah. Um, and we have another line of, of parts where we've integrated in the, the amplifier with the ADC. So now you can just connect an amplifier straight to it and, and you don't have to worry about that.

Chris Gammell: So it's almost like a, it's like a, the feature is that the data sheet is getting shorter almost.

Matt Duff: Yeah.

Chris Gammell: Uh, I mean, it's just less complicated, not necessarily shorter, but like just that there's less caveats, that there's less need for measurement because more is taken care of for you.

Matt Duff: Yeah. So the hope is, you know, before you needed three parts and three different data sheets and now you just need one. Yeah. So, yeah.

Chris Gammell: Are you seeing that on the, so I remember there was an ADC part that was like a all in one sensor chip type of thing with like a ARM processor on it. And one of the versions even had like a little, I think Bluetooth radio or something. I mean, does that end up impacting the tools that you're working on as well or no?

Matt Duff: Uh, so normally what we try to do when our, so the tools that we focus on is we look for some sort of problem that a bunch of customers, you know, a lot of times a bunch of smaller customers have. So like if there's a problem that a couple of vertical customers have, then we don't bother with that because we have sales guy and we're just going to direct support them directly. And so, you know, anything, the vertical, the massive customer says, we just say, yes, sir. You know, whatever you would like, sir.

Chris Gammell: Please just keep paying your bills. Right, right. Exactly. Um, yes, I too have worked with the oil and gas industry.

Matt Duff: So when we, when we do a tool, we try to find something that, that hopefully spans a bunch of parts, a bunch of problems that, you know, maybe these, you know, customers that aren't these massive customers have that we can make pretty graphs out of it. That, that, uh, has hopefully a lot of math because computers are really good at math. Um, and so those are kind of the characteristics. So if, if you have a, you know, if, if you've released a part that is, you know, this one kind of complicated part, it, it may not, it may not be worth trying to develop some sort of web tool around that one, one particular part.

Chris Gammell: Got it. Yeah. Yeah. These are like, these are all like self-service tools here, right? This is trying to help the customer help themselves versus, yeah, the expensive part. Then the application engineer helps that person because that's what they're there for. Yeah.

Matt Duff: And there, there's a total another. So the kind of tools that I work on are kind of more the, uh, when you're first looking at, at whether you want to buy a part and, and you're trying to evaluate, you know, this part versus another, you're trying to kind of build up, you know, what, what, what's this little circuit block going to look like. And so we do these web tools because there's very low barrier to entry and so forth. But then there's another part of the company that works on drivers for, for particular parts so that, you know, when you, when you buy your ADC, you have some sort of driver so that you can get it up and running. And, and, you know, so the low level, you know, so I'm doing these web tools, but there's other folks that are doing the, some, some example microcontroller code, uh, to make it easy that way. And so I would imagine a part like the one you mentioned, somebody there is making microcontroller codes so that you can easily interface to it. Yep. Yeah. That makes sense.

Chris Gammell: And yeah, and that's a, that's a common thing in the, in the micro industry specifically. Yeah. I think, uh, but it, it wouldn't necessarily be a visual tool like you're saying it's, uh, that's just a, a quick start resource type thing. Yep. Awesome. Well, uh, Matt, anything else we should know about it in analog devices or, you know, around the, the analog world that you see these days?

Matt Duff: Oh, I'm sure there's, you know, if I, if I had like looked at my, you know, I'm sure there's a lot of stuff that the corporate guys would say that I ought to be pitching, but I don't know. So I'd say that's, that's all I know.

Chris Gammell: So that's, that's good. Yeah. We'll, we'll hold you to that on that one here. That's good. All right. Well, uh, where can people find you online?

Matt Duff: Uh, so I know that, you know, lots of folks are out there on Twitter. I actually am just lurk on Twitter. I don't, uh, don't actually have an account there. So the, probably the, the, the, the two best.

Chris Gammell: He only follows the amp out. He follows the amp out of Twitter account and nothing else.

Matt Duff: That's right. Exactly.

Chris Gammell: Exactly.

Matt Duff: Um, so I'd say that the two, you know, the, the two things that you could do is one on our tools themselves. We have this feedback thing. Please use that feedback. And on, on most any page on the analog website in the lower right, there's a little triangle. Sometimes you, you miss it, but you can click on that little triangle and you can give feedback that way. And it may seem like it's going into a black hole, but there, we, we do actually read that stuff.

Chris Gammell: It's going to Matt's inbox. Yeah. Yeah, exactly. Go, go to analog.com and say, click the button and then say, hi, Matt. Yeah.

Matt Duff: So, so that's, that's definitely one way. And the other way is you can just contact me directly at Matthew.duff at analog.com. Um, that works great too.

Chris Gammell: All right. Well, thanks, Matt. I mean, thanks. I, I really did appreciate all the videos you've made over the years and it sounds like you're still making lots of, uh, useful things. So I, I appreciate that and I appreciate you coming on here to talk about it.

Matt Duff: Oh, thanks for, thanks for having me. This has been fantastic.

Chris Gammell: Okay. Well, we'll talk to you soon and we'll look forward to trying out your web tools. Okay. Thanks. Thanks.