#614 – Reunion Impedance Matching and 2023 Predictions

Chris Gammell: This is The Amp Hour Podcast. Released January 8th, 2023. Episode 614. Reunion Impedance Matching and 2023 Predictions. Welcome to the Amp Hour. I'm Chris Gammell of Contextual Electronics.

Alvaro: And I'm Alvaro from the Unnamed Rehearsing Sharing Podcast.

Ariel Briner: And I'm Ariel Briner from Electron Labs. And I'm Chris Osterwood from Capable Robot Components.

Chris Gammell: And this is a reunion show. Welcome back. You guys have all been on the show before. I think so. Thank you.

Ariel Briner: Been a while, yeah.

Chris Gammell: Nice to be back. You saved my bacon. I had a guest cancel and you guys came to my rescue. And now we're going to talk about the state of electronics in 2023. So we'll see where this goes. I don't know. First, in the same order that you just introduced yourselves, let's get a little update. What's new and hot in your world? Where are you guys working? What are you working on? Well, we'll get about what you're excited about later. But since the last time you've been on the show, perhaps, any major life changes people should know about?

Alvaro: Okay, I guess I'll start. Yeah, I don't know when the last time I was on, but I... So many times. Yeah, people might remember from other podcasts, but many years ago, I was at Apple. Then I went to Planet Labs and then I rage quit and I took a five-month sabbatical and traveled the world and it was great. So I decided to do that again and I quit my job. I was working at Loon and Google X, well, it used to be an X company. I decided to quit my job to travel. And so my last day was March 3rd, 2020. So that didn't really work out. Yeah, that worked out well, huh? Yeah. Yeah, it worked out great. So I had a sabbatical for five months in my apartment and then I got a job where I am now doing ocean science stuff. So I write firmware for buoys that are all over the ocean measuring things.

Chris Gammell: Yeah. I think you've talked about that. I think you went back on Embedded maybe and talked about that at some point or...

Alvaro: I have no idea. Yeah. It's been two and a half years now.

Chris Gammell: I'm a frequent podcaster, so yeah. Yeah. It's very possible to find out Rose stuff. Yeah.

Alvaro: I've done a few conference talks on it. It's mostly about me troubleshooting things and how to fix things, but related to the work. And yeah, I do have a podcast that I sometimes record.

Chris Gammell: Cool. Cool. I'm feeling that.

Alvaro: All right.

Dave Jones: Me next. Yes. Chris, I was last on the show in 2015. Okay. So it's been a little while. And back then I was running my wildly successful startup, Cartesian Co. No, that was not successful. Printing circuit boards. You may remember. I was trying to get me to print a circuit board on your arm. I'm sorry that never ended up happening.

Chris Gammell: Oh yeah, it didn't. Well, I'm sure you still have some in the basement or something. You probably have a printer somewhere. We can get it to go.

Dave Jones: Yeah, yeah, sure. Totally. Yeah. I didn't get rid of all the mouse bite. But since then, a lot has happened. There's not time to catch up on all of it. But I worked at a couple of startups. One of them being Zipline doing drone delivery, emergency medical delivery via drone. I joined a consulting company called PCH International. And then I left PCH to go and do independent contracting. And that became Electron Labs, my company now. And over the last two and a half years, that has grown. We currently have a headcount of eight. So it's really snowballed on me.

Chris Gammell: Yeah, that's great. That's great. Wow. Awesome.

Dave Jones: Thank you.

Ariel Briner: And Chris, what about you? Yeah. Yeah. So I think the last time I was on, I was about to launch the USB hub that I've been working on on CrowdSupply. That launched actually at the Teardown Conference that they hosted in 2019. That went very well and been working on a bunch of different things since then. Working with Ant Micro on some embedded Jitson carrier baseboards, synchronization systems, camera systems. And now back to USB hubs and working on the new generation of those. So yeah, still doing capable robot components and still building stuff here in Exeter. Awesome.

Chris Gammell: Yeah. I mean, how is the robot market, I guess? That is like a thing that I guess pops up. Every list of the future is robotics. And I was just listening to things about like Japan losing workers. And there's all like, oh, we can't hire people in the US at least. And so they always talk about automation. But is that coming true? I mean, Alvaro, Ariel, you guys seeing that stuff too? I mean, robotics, is robotics going to be a thing?

Ariel Briner: A bigger thing? It will continue to be a bigger thing. Like it's, I don't think globally we're going to see like a hockey stick growth in robotics like we have in like the internet taking off or anything like that. But it is a very useful approach to solving many different kinds of problems. And where the safety and price and performance and flexibility of automation or lack thereof makes sense, it will be well employed. But yeah, I mean, demand for automation is going to continue.

Chris Gammell: The other two guys live in San Francisco. And I think they would have the reflex to like be like ready to pitch to an investor. Be like, oh, yes, it's going to explode next year. You know, like I think you're just living that East Coast life, Chris.

Ariel Briner: This is why I've never moved to California, I guess. Yeah. I mean, 2023 is not the year of Linux on the desktop. You know, to use an old phrase.

Alvaro: No, that was last year. That's right.

Chris Gammell: We've achieved it. Yeah.

Ariel Briner: Yeah. So, no, I mean, this is an area of growth. I wouldn't say rapid growth. But in particular niches, there definitely will be rapid growth for particular verticals. So, yeah, I'm continuing to be very excited about building robotic components, about the industry writ large, and about like the democratization of it. You know, things are easier than ever to build. There's more that you can do at a lower budget. You know, development timelines are decreasing, but still long. And I think there's still a lot of opportunity out there. There's still a lot of dirty, dull, and dangerous jobs that could benefit from automation and flexible and intelligent automation, not just repeat the same thing ad nauseum like a robot arm in a factory.

Chris Gammell: Yeah. Yeah. Huh.

Dave Jones: Okay. All right. Well, I was going to say, I was thinking, I'm just pulling up like my list of clients. It's difficult to keep track of how many clients we've had. Flex?

Chris Gammell: Are we talking about flex circuits here or are we just going to flex that much?

Dave Jones: I just wanted to count how many of them were robotics because I'm trying to think how many clients have come to me recently or clients we've had over the last like two and a half years that have been robotics companies. I don't think it's any of them. So, maybe that's just me personally. I'm not appealing.

Chris Gammell: Well, it could also be the location. I mean, I'm guessing you probably get a lot of Bay Area interest because there's a centralization.

Alvaro: Half of my clients are robotics. I have two clients. So, I am doing some robotics.

Dave Jones: This is good. I wanted to know, Alvaro, are you seeing more than me? So, I guess you are.

Alvaro: I'm seeing one.

Chris Gammell: That's the other thing too. I wonder about, you know, like in a consulting context generally as well, like it does feel like robots are often very vertically integrated. Maybe, you know, maybe they're buying components from like a capable robot, but Chris's company, you know, to be a system integrator in certain parts of it. But it does seem like, you know, it doesn't seem like they're kind of subcontracted out here and there where, you know, if it is a, you know, I think about like the Amazon factory robots. Those things like they're all one unit and they're just kind of made by that company. Maybe that's wrong though. I don't know.

Ariel Briner: I think it varies a lot by the domain, the constraints of size, weight, and power in that domain. Like if it's, if you're really constrained in those regards, it's harder to integrate off the shelf components and you need to go to a more embedded and integrated design, which makes it more difficult to integrate outside components or outside firms in your development. But it's possible. You know, the, the whole point of, you know, why I started capable robot components is I saw this gap in the market between, you know, building everything in house yourself or buying things off the shelf that were designed for other markets. So the whole point of this company is to create a whole suite of components that are easier to integrate and more purpose built for robotics, but still flexible for use in a wide variety of different kinds of robotic applications. So, you know, there's a nice middle ground there, but I don't think many companies are aiming for it. So there's, there's still an underserved part of the market there. But, you know, I think, uh, you know, from a consulting perspective, I think there is lots of opportunity to help robotics companies get to market faster. You know, the power systems in these things are not that different from other sorts of power systems and other kinds of high power electronics. Same thing with radios and computing and cameras. And, you know, if you have expertise in a particular vertical within this domain, not from an end user perspective, but a technology vertical, I think there's absolutely opportunity to work with robotics companies to help them bring to market a product sooner.

Chris Gammell: Yeah. I always see like, you know, so see, I should mention CES is going on right now and I'm kind of viewing it from afar via social media. And it's like, every time I see robotics kind of in popular media and, you know, just, it seems like really trite, but I feel like that's also because that's what gets covered, especially at a CES. You know, you think about like the things doing the hard work in the robotics realm are probably the stuff that I'm not seeing really, you know, if you broaden it to automation, then it's like, oh, that really explodes. You know, automation versus just robotics. I mean, Albro, that's like, you're basically doing buoy automation, right? Yeah.

Alvaro: Yeah. I mean, remote sensing, you know, we're, we're measuring things that no humans could reasonably do.

Chris Gammell: Send an intern in a dinghy out into the ocean, peanut butter sandwich and a couple jugs of water.

Alvaro: 15, 15 meter waves off the coast of California. Holy crap. That would not be fun. Yeah. That's like 40 some feet for the Americans.

Chris Gammell: For freedom units. Yeah.

Alvaro: Uh-huh.

Chris Gammell: And that's just the recent storm or like regularly?

Alvaro: Yeah. It's a current one.

Chris Gammell: Oh, wow. Woo. Yeah. You guys are getting slammed. I guess we should mention if people are listening to this later and you haven't looked at the title, this is being recorded in January, 2023.

Alvaro: Well, we're going to be getting just lots of water over the next couple of weeks, which is rare, but we'll take it.

Chris Gammell: Yeah. Yeah. Awesome. Grow us some food, please. I hear you guys do that.

Alvaro: It might be drowned. So we'll see. Yeah.

Chris Gammell: Yeah. I mean, that's like a, so like ag tech is a sector that I'm, you know, doing, I'm doing reference designs for, like ag tech type stuff. And like, you know, it's just, and I'm sure you probably see stuff in that space too, right? Or just generally consulting robotics. That's kind of where it all comes together. A lot of that stuff kind of comes together in an industrial way in like a very practical manner. But it's just like, it's so insular, not insular, but kind of insular. It's just like, there's so many big providers that they just go and design their own stuff. And some of them buy it off the shelf and some independent farmers need that stuff. But it's like, you know, it's like its own industry by itself.

Dave Jones: Sort of modeling itself a little bit after the like automotive industry where it's all very insular.

Chris Gammell: So yeah, it kind of feels like that.

Ariel Briner: Yeah. Just connecting that back to CES. I saw today, I've not watched the whole keynote, but apparently John Deere did a big part of the CES keynote talking about their automated tractor development and rollout. So yeah, I mean, there definitely are frivolous robotics at CES. I would say the vast majority of robotics displayed there are rather frivolous, but there is sort of the real productive robotics on display there as well, which is nice to see.

Dave Jones: I suppose, unlike many other industries that we would deal with, robotics is one where there's a practical application of it, but then there's also a marketing application of it. And they're pretty disconnected from each other, right? You get to, same thing you were saying, Chris, where you're seeing a lot of stuff in the news about robotics, but that's usually these like trite examples that are flashy. I mean.

Chris Gammell: We're not even always flashy, but I think like, I think graspable, right? That's what I would think about it. Like the, the one we talked about in the show a couple months ago now was the, the tortilla cooking and deep frying robot, basically, you know, like that's going into like fast food restaurants. And it's like, because people can understand the end application. You don't have to be like, oh, well, this is, this is doing, this is like a, you know, robotic combine. It's like, oh, what's a combine. You don't have to ask that. Like people know what tortilla chips are. And that's like, oh, I like those. You know, Chipotle.

Dave Jones: You just reminded me of the, I just remembered. I have had one robotics client. Oh yeah. Just one, one of the robotic barista companies. Uh-huh. Yeah. Yeah. In fact, that was my very first one. It's such a long time ago now. I forgot, but I'm looking at my list and I'm like, oh yeah, there we go. That's the one on the list.

Chris Gammell: And what was it? What part of the barista process was it automating?

Dave Jones: Well, if you've seen any of the robotic baristas, it's a robot arm that basically moves a cup back and forth between the customer and standard coffee machine, right? Like an espresso machine. So it was doing that. It was automating the process of putting a cup under an espresso machine nozzle and then bringing it to the customer in a very novel fashion.

Alvaro: Yeah. They have one at the airport here. Yeah.

Dave Jones: Yeah. May or may not have been the one.

Chris Gammell: I think that that's the thing. Like that's like kind of like general robotics hits specialized robotics. And that's still on the general side of things, probably because they didn't want to develop a fully custom solution or there wasn't enough of a market need for that sort of thing. You know, it's just like, what can you get funding for versus what can you actually make money at it? I'm amazed that people make money at coffee in general, personally. I mean, it's just, it's crazy to me, but it's obviously very addictive and needed, you know, growing market, I guess.

Ariel Briner: And if you were to design something purpose built for that in an integrated way, it wouldn't be interesting to watch at all. If you were the end customer, right? There would be a box and out would pop your perfect cup of coffee.

Chris Gammell: It would be the vending machine that delivers coffee, right? It just tastes better. More of that.

Ariel Briner: I did have to laugh when I saw those robotic pizza startups a couple of years ago with, you know, like a robotic kitchen to make a pizza as some like new invention that they have graced to the world. And what do you think, like, where do you think the frozen pizza in your grocery store comes from? Like that is a factory built pizza done way more efficiently than any of these Silicon Valley startups doing it, you know, on demand.

Alvaro: But you don't get to see that.

Ariel Briner: You don't get to see it.

Alvaro: Speaking of robotic food things at the last three inch meetup here in the Bay area, we did it at a food truck park and they do have one food truck. That's that's a robot, but the food truck doesn't make the food. It just plates it. So they bring in the pre-made food and then the, you kind of select what you want. It'll mix it, warm it and put it in a plate for you. But you don't know that, right?

Chris Gammell: I think the ultimate, the ultimate like San Francisco startup would be that except underneath it's just like a struggling artist with a, you know, a costume robot arm, you know, behind the curtain and they're just moving it in a robotic fashion.

Alvaro: So they did do that during Halloween at one of my old jobs. I was sharing, I was sharing space with a company that made these robots and somebody dressed up as a robot arm and somebody pushed them around in a rolling chair and walked around with a laptop behind them taking notes.

Chris Gammell: Nice. Yep.

Alvaro: It's pretty good stuff.

Ariel Briner: At Modex a couple of years ago, Modex is a big trade show for fulfillment and automation and fulfillment. And I saw a demo of a robot arm piecemeal sorting application where they were taking like a stack of stuff and putting it into bins to sort it all out. And they were, you know, they kept talking about how good their AI was and how good their vision was and machine recognition of everything. And you'll, you walk around to the back of the booth and you see a guy sitting in a chair with a laptop watching all of these different cameras and clicking on what the robot arm should pick up. It's like, it was literally pay no attention to the man behind the curtain, except it wasn't even a curtain. Like he was just there next to the machine running it.

Chris Gammell: Training the models. Curtains are so expensive at trade shows. Like they charge you a robot arm and a leg for that. So yeah, well, I've become a dad since the last time you guys have been on the show. So, you know, dad jokes are allowed now. It's, it's part of my, it's my birthright. They're required now.

Alvaro: Not yours, but yeah. Yeah, that's true.

Chris Gammell: What about other trends? I mean, so robots always make the list. What else? What else are you guys seeing in the marketplace?

Alvaro: I'm not the only one doing IOT. That is correct.

Chris Gammell: You are not the only one.

Alvaro: I'm doing industrial. IOT or, or, or the ocean of things, but, but yeah, the, I mean, my life is remote sensing satellite communication, cellular. And, and because of that, I, you know, I talked to other folks who are doing the same stuff, but, but I know y'all are, are also in that kind of area and it just keeps growing.

Dave Jones: Yes. Yeah. Lots of IOT interest. I'm quoting on another IOT project right now.

Chris Gammell: Okay. Is that, uh, how much, how much of this, so you have many clients I hear, uh, which I'm very, I'm very glad for you. Uh, how, how many of those products are connected versus not? I mean, so Alvaro and I both work on connected stuff all the time, but so you have kind of a broader swath. How many of those are talking to the, to the web?

Dave Jones: So to put some, to put some perspective on it over the last two and a half years, while I've been growing electron labs for a lot of that time, it was just me. Right. And then I've only really grown the head count in the last year. Oh, sorry.

Chris Gammell: I didn't, I wasn't making fun of you. I just, I like it. That's good. I want you to be successful. Come on, man.

Dave Jones: Thank you.

Chris Gammell: Thank you.

Dave Jones: But in that, in that time we've had just over a dozen clients, I think. I think, with like multiple projects per client, usually sometimes just one and I'm going over the list, trying to calculate in my head, like how many of these were IOT? Uh, probably like a third of them. Yeah. Or how many of them were connected? Probably half of them. I mean, some of them, one of them was like hardcore networking equipment. So yeah. Right, right, right. Do you define that as connected? Yes, I would. I would. Yeah. It does the connection. So yeah, almost all of them are connected. So where everything's going. Yeah.

Chris Gammell: Like whether or not the, I think, I think the other piece is like, cause from my experience in the consulting world too, it's like often you're, you know, you're called in to, to do this thing and build this piece that, you know, like, like Chris is doing too, right? Where you're, you're basically making like an appliance that goes into something else. Does the end customer necessarily have a connected business? Well, like, are they making money off that connection? Eh, maybe, maybe not. But you know, like, but if it's still part of the, the overall setup, then yeah, it's, I'd still listed as, you know, if you're, if you're making a display board for a device that's connected to the internet through a Linux computer, like, yeah, I'd still say it's connected device, even though you're doing a part of it that's maybe not connected, you know?

Dave Jones: Yeah.

Chris Gammell: Fair enough.

Dave Jones: I'm interested to see what you guys have found, like for number of inbounds, like we've seen a lot of layoffs recently, right? And we've seen a really cooling of the tech sector. And I'm wondering if you guys have noticed that with the amount of inbounds you've been getting with the amount of work, how many people have been beating down your doors? Because I have definitely noticed a downward trend in amount of interest, but luckily that's only gone from when I couldn't handle it at all. It was overwhelming to now it's just the right amount for our headcount. That's good.

Alvaro: I mean, for, for, for work or customers are mostly industrial or scientific. So those haven't really noticed, I guess, cause they, they, they, they think in years, but personally I, I don't like the clients I have where somebody asked or connected me to them. I'm not really looking for, for, for work.

Dave Jones: So you've got the same clients you've had for, for a while. You've got like a long time.

Alvaro: I had one client for a couple of years and then he told his friend, like, Hey, you just talked to Alvaro and now I'm working for the other one.

Dave Jones: Yeah, that makes sense.

Ariel Briner: And I think some of this is hard to separate from the normal yearly cycles of, people, you know, doing end of year wrap up instead of beginning of year planning. And, you know, there's not a bunch of people kicking off projects in Q4 generally.

Alvaro: Yeah.

Ariel Briner: Sort of a tidy everything up, finish things out, get ready for the new year and then start the new 2023 efforts in Q1.

Chris Gammell: I've been surprised by down, down cycles in the past too, where it's like, sometimes the, you know, hiring consultant is actually because they, they did lay off people internally, you know, like that's also a possibility.

Alvaro: Oh, that's happening for sure. Yeah.

Chris Gammell: That's exactly, I was wondering if you guys had that same experience. Yeah. They want flexible labor, that sort of thing. You know, that makes a lot of sense, but personally I'm not actively soliciting. So I'm not even, yeah, no, nothing, nothing for me on that, on that front. Yeah.

Dave Jones: The only down cycle I went through was at the beginning of 2020. And in that instance, I only had one client at the time. I'd only just started independent contracting and that client laid off a bunch of their full-time employees and then gave me a bunch more work because they didn't have the full-time employees anymore. So I'm expecting that to happen again, but that's why I was kind of been surprised. I haven't really seen that over the last couple of months, but I think what you said, Chris, makes a lot of sense. It's more of a yearly cycle that's impacting things.

Alvaro: And I think, I think the companies that just did layoffs, which is a lot of them, they're probably still realizing how much work there is to do. Cause you know, there was the holidays and it's probably the folks that are left are trying to, you know, get stuff done and then they're going to realize they can't do it. And then they're going to go looking.

Dave Jones: That makes sense.

Chris Gammell: What about, well, what else in the IOT sector? So Alvaro, I mean, you're, you're connecting stuff to satellites. That's, that's interesting. That's actually an emergent thing that I've seen. And I was really surprised about the like phones are starting to talk to satellites too. I've seen articles about it. I don't have fancy phones. I don't know how that works, but.

Alvaro: Well, the new iPhones can do global star.

Chris Gammell: Okay.

Alvaro: And I think T-Mobile is working with Starlink. And then I also heard that. What's the other one? Um, uh, Iridium is going to start working. They didn't say with who, but it was Scott Manley on YouTube. It was speculating. I mean, Samsung, but yeah, that's becoming a thing.

Speaker ?: Hmm.

Chris Gammell: It only took them 30, 30 years from the, uh, the, the elliptical orbits or whatever it was called that book where they talked about phones, phones regularly connecting to the satellites.

Alvaro: That was a pretty good, that was a pretty good book. Yeah. I don't expect to be doing web browsing, uh, with satellite internet on your phone. It's at best, it's going to be text messages. I think the, for example, the modems we use, we use Iridium, uh, which is the best constellation constellation right now. If you, if you just want to send messages from anywhere in the world at any time, but they have the SBD modems like short burst data. So I can send 340 bytes up and get 240 bytes down per message. And it takes a lot of power to do it.

Chris Gammell: Oh yeah. How much are you talking here?

Alvaro: Oh, like over a watt to transmit over several seconds. And that's when it actually gets through.

Chris Gammell: Got it. Yeah. Yeah. And you have to do it multiple times to like, do you get confirmations back down or no?

Alvaro: So you'll know if the message was transmitted successfully. The problem is you can't just receive messages. So you, in order to receive data, you have to send data. So you have to check in with a satellite to, to get anything back. So downloading is extremely expensive. So we don't do firmware updates over satellite just because of, of, of costs, both in power and data. But we also do cellular now. So for the coastal applications, we can actually do firmware updates.

Chris Gammell: Oh, that's nice. Yeah. And is it cat, cat M1 or is it cat one or what?

Alvaro: I think we're using the blues that I owe the, the note cards.

Chris Gammell: Yeah.

Alvaro: Yep. And it's, it's been, I have fixed a few bugs in their code, but, but they've been fantastic. Just, just weird ones where, you know, I, we get a customer in Chile, they get the unit and then it's just crashing and crashing and crashing. And occasionally we get some reports into the crash, but it's just kind of not quite a boot loop, but every hour or so it crashes. And eventually, you know, long story short, the cell tower, the name of the location had an accent in it, which was a Unicode character. Oh no. Oh no. And then their driver was ignoring. Extraño. Well, it thought it was noise because they had a very naive filter. So if, if the, the character of the byte was zero or greater than 127, they're like, oh, this must be UART noise. But no. Yeah. Unicode characters can happily be, you know, 232. And so it was, it was crashing. And so, so that was a fun one to debug remotely.

Chris Gammell: I should mention we had a Ray Ozzy on the show of Blues Wireless. That's one of my favorite shows. Yeah.

Alvaro: He wrote a lot of that stuff early days. Now he has a team of people.

Chris Gammell: Yeah. That's great.

Alvaro: But he's still, he, he, he looks at my pull requests and merges them in. Oh really? That's great.

Ariel Briner: That's awesome. Alvaro, that reminds me of my favorite weird field bug ever, which was back when I was doing sewer robots. We got a report from a field crew that they found a street where the operator control unit, the OCU would blue screen. And we're like, oh, come on.

Alvaro: And that street.

Ariel Briner: On that street. Yeah. Yeah. What a street. They're like, come on. Like, what are you, what's going on out there? Lacky pattern matching. Yeah, exactly. Exactly. So we sent them a new rugged laptop. They went back to that location later that week. They said, yep, same thing's happening. This street crashes the laptop. So we send our software engineer out and he reproduces the problem. You go to the street, the laptop blue screens.

Chris Gammell: Oh, wait, wait, wait, wait, wait, wait a second. Let's all take some guesses here. What, what is it guys? Does it have GPS?

Ariel Briner: No GPS. Well, no, no, there was GPS. We, you know, we were recording locations in the manholes that we were inspecting. It's gotta be something like somebody, somebody's wifi network.

Alvaro: Yeah. Wi-Fi network.

Ariel Briner: Yes, that is, that is the button. Yep. So there was someone on that street that had the empty character as their SSID name. Oh my God. Like null string. And whatever version of Windows and whatever version of the, you know, 802.11b driver we were using, you know, 12 years ago, couldn't handle that. It would blue screen the whole Windows laptop when you would get that, you know, null character SSID.

Chris Gammell: So it'll be fixed in Windows 12.

Ariel Briner: Yeah. It was a location specific blue screen of death.

Alvaro: I put emojis in all my SSIDs and then it keeps most of the IOT stuff out of it. Cause they can't connect.

Dave Jones: Well, do you think that that was malicious on the, on the part of the access point?

Alvaro: They probably just didn't want to be visible.

Ariel Briner: Yeah. I think they were trying to be hidden and they didn't know that there were other ways of doing that. Yeah.

Alvaro: Wow. That's a good one. Yeah. Unicode, man. It's a, it's great, but.

Chris Gammell: It's a thing.

Alvaro: Yeah.

Chris Gammell: It's a real thing. I think the international thing too, like, like you mentioned, it was Chile, right? So like, just like when your product gets to the point where it is actually expanding outside your home territory, you know, you can test so many things, but you just don't know what, you don't know everything you're going to run into until you, you do start going more. Obviously your thing is literally drifting, but like, you know, if you get into a new market or something, there's going to be a bunch of new, new things that might crop up as a result of that.

Alvaro: I mean, I can tell you a couple other weird bugs that could only have been tested by, you know, moving. And one of them was the GPS. They, everything looks fine. The units are reporting in. And then we test our buoys out, out ocean beach. So go around the golden gate and out to the ocean. And we more of them out there for testing. And the unit we took, everything was fine until we took it out there. And then the position reported was like way off. And then it restarted and everything was fine again. And then we couldn't reproduce it. And the only way to reproduce it was to move that buoy in a particular direction. And, and what would happen is we had, uh, because of the data restrictions, we do bit packing. So if, you know, if it takes, you know, 11 bits to describe a location, call it, we're only going to send 11 bits. We're not going to round it up to two, two, two bytes. And the bit packing function was expecting to have a buffer full of zeros. So it would only set the ones. It would not clear the zeros. So as it moved, the ones would get set. And then I kept moving and more ones would get set and more ones would get set. And I would just get thrown out.

Chris Gammell: It's an accumulator issue, huh?

Alvaro: Yeah. And that was the only way that we could have tested that. You know, I, I didn't, I don't know how with spoof GPS data, perhaps.

Chris Gammell: That's crazy. I would love to hear more about your, uh, so Alvaro, I know you do free RTOS. Yep. Not sure about Chris and Ariel, what you guys are doing for, I feel like my, my personal experience has been with RTOS is this year. That's been like my year of learning RTOS is. So like, I kind of see them everywhere, but that's not necessarily the case. I mean, what do you, what do you guys see in terms of, uh, you know, levels of processing? I remember Chris also said something about the CM4 in an email we had before this as well. So like, you know, where, where in the software stack of the worlds are you guys living?

Alvaro: Yeah. I, I, I can go with, uh, I've been using free RTOS for like over two years now. We did take a look at suffer, but I vetoed it because, uh, uh, for, for a different project, but one is to, you know, stay in the same thing for all of our projects. But the other one is it's great if you're doing things in kind of the domain of, of suffer that, but if you try to do like ultra low power on a platform, that's not been developed or do things different way, it's a lot more complex, but we're, yeah, we're doing everything, uh, free RTOS these days and yeah, you have to write your own drivers. It doesn't come with drivers. That's the, the nice thing is suffer comes with everything.

Chris Gammell: Not everything, but yeah.

Alvaro: Well, everything, it comes with more. Yeah. It's, I mean, it takes like half an hour to clone the repo the first time, uh, which kind of drove me crazy, but, uh, the, with yeah, free RTOS we're, we're doing, uh, Cortex M4, uh, actually now we're doing M33 for a new product and it's the STM 32 U5 series, which is kind of neat. Oh, okay.

Chris Gammell: All right. I'd love to get into parts to what you guys are using it for parts. That's that's, that could very easily be tied to our tosses. Well, it's which parts can you get?

Alvaro: And that's why we're switching to the U5 because it's newer and I think less people have designed it in. So that was a trick. Yep.

Dave Jones: Counter to Alvaro. We have been using Zephyr across two of our products, two of our projects recently, but they have, I agree with you. I would not want to do this when we're power limited or trying to do ultra low power. These are both plugged into the wall products. So that's just like, it's much easier. It's much, much more straightforward.

Alvaro: Absolutely. Yeah. Yeah. We, we, we, for, for prototyping the, the, the team was using Zephyr and they were able to get network SAP network stack up and running and all that stuff. But as soon as we started looking at power and performance, we were trying to use some high speed communicate. Well, you know, 10 megabit, but you know, high speed communications. It's, it started getting very complicated to debug, figure out where, where things are, are slowing down. And then the power management was, was going to be rough. Yeah.

Dave Jones: I want to attempt that with Zephyr. That sounds awful.

Alvaro: Yeah. But, but if you're doing like a Bluetooth sensor on a NRF 52 or 53, that's probably the best use case.

Chris Gammell: Yeah. I feel like that's how most people are coming into Zephyr these days too. It's just Nordic is kind of leading the charge. Obviously they're not the only ones in this, in the project, but like that is most people that I, I'm, I, I would take a very educated guess that Ariel is probably also doing some Nordic stuff. Most people aren't choosing Zephyr out of just, just for funsies, but maybe you don't know. Yeah.

Dave Jones: Actually we're running. I can't think right now of what it's running on, on the second project. Cause I haven't been involved technically with that one for a while, but the one I can remember are actually running on STM 32. Oh, there you go. Prove me wrong. All right. Great. But that was because it was like our internal favorite already at that point. Yeah. So yeah.

Alvaro: And they do have pretty good support for their peripherals for STM 32. Like they even had the U five, which is pretty new that they had that supported. Having said that the, the U five has a lot of really cool, weird features that I don't think are supported. And I'm going to have a hard time getting it to work just because they're so complex, but it has, you can do DMA in stop mode. So even when the processors, yeah, which is bizarre, but it has a low power DMA that it'll do magical things. Cause usually people think, oh, DMA, it's going to be low powers. I'm like, no, your clocks still have to be running. But this one has a special DMA that can power down most of your clocks and just do the DMA in your peripheral. But you know, these are so obscure of features that you can't generalize them. So you have to do it all custom and that's where it starts breaking down.

Chris Gammell: What would you guys say your hardware software split is these days as well? I mean, that's another thing I feel like that's a, you know, it depends on, you know, where you're, where you're focusing in on, but obviously this is a hardware mostly show, but you know, I'm, I'm digging more into firmware and Dave does a little bit of firmware here and there. And it's just like, you know, you can't escape it these days. So I'm always curious about what, what the, uh, what the percentage of time people are spending on different things like that.

Alvaro: Professionally, I I'd say it's 95% firmware. We have a hardware team and, and, you know, they do, I'll do board bring up and troubleshooting. That's kind of hardware-ish, but most of my hardware is on my side projects and, and I'll make tools for work by myself, like, uh, uh, be able to debug SLED over USB-C. So, so I don't have to bring a 10 pin connector or I've been making front blades for the, for the new Julescope, which, you know, kind of basic. Uh, and then the last thing I made was a USB cable tester, just cause that makes me angry when there's charge only cables.

Ariel Briner: I guess my time is mainly, it's pretty well split. Uh, I mean, it sort of depends on the week or the month, you know, this year I've done a fair bit of FPGA work, a lot of embedded development, a bunch of Linux development and a bunch of hardware design. So it's, it's sort of been all over the map and on the, uh, the embedded software side, uh, I started playing around with Rust on microcontrollers for the first time for real this year and, uh, excited to be doing more of that. You know, there's a bunch of different paths and options there, and I'm still sort of exploring that landscape, but had one effort recently that used an ESP 32 C3, the RISC-V, you know, ESP 32. Yeah.

Chris Gammell: Yeah.

Ariel Briner: And Espress has done a good job of porting the standard Rust library to that part. So you can actually do like thread dot spawn and, you know, mutexes and, you know, the full standard library, uh, basically exists for that target. You know, you're not, uh, down in, in the weeds with like an embedded, no standard Rust. It's, it was pretty feature rich. And so I was able to, you know, develop a fairly simple, you know, power ons monitoring, you know, watched a kind of system for a main processor on this little ESP 32 C3 in a couple hundred lines of, uh, of Rust and, uh, works really, really well, really stable. Awesome. And it was easy to debug, easy to write. I was impressed.

Alvaro: They also have pretty awesome logging capabilities. Um, I was talking to the folks over Oxide and actually I think it was a Memphold did a local meetup, but, but they were saying how, and there's some libraries to do this and see, but usually, you know, we put print apps in our code and then it goes to our UART or USB, but it's really slow. You're not supposed to do an interrupts and it takes up all your, your flash space. And, and there's a Rust library that lets you do all these logging prints, but they don't get compiled into the binary. It's just kind of symbols. And then you're only basically sending pointers to the strings, even though the, the point, the thing they point to is not in the micro control itself. You, you have it on the other side. So now you're only sending a few bytes back and forth for very verbose messaging or logging and, and, um, very lightweight and fast way. That's cool.

Ariel Briner: I'm, yeah, I'm actually looking at using their hubris, uh, embedded RTOS for a new project.

Alvaro: Yeah.

Ariel Briner: So I've not actually started using it yet, but I've got the hardware in design right now. I've chosen a chip that they're already using on their, uh, backplane system so that it should be fairly easy to port their whole, uh, you know, embedded, you know, RTOS, uh, repo over to this new application.

Alvaro: This is for folks who don't know that the oxide computer companies are TOS for this random Rust.

Chris Gammell: Yeah. People may remember Laura Abbott has been on the show talking about all of this stuff.

Alvaro: So, and I'm sure you've had Rick author on as well, right? I'm not sure. Yeah.

Chris Gammell: Boy, I think so.

Alvaro: MX shipped.

Chris Gammell: Yes. Yeah. But before he was here, I think before he was at, uh, oxide. Ah, well. He was talking about cars. Yes. His cars. That's right.

Ariel Briner: I do love that they're, that they're custom debuggers named humility, which is, it's just an amazing pairing of words that the RTOS called hubris and the debugger called humility.

Dave Jones: Sounds like we have the full spectrum of breakdowns between hardware and firmware then because I market electron labs as primarily hardware engineering, primarily electrical engineering. And, um, we just do a little bit of firmware engineering where necessary. I think that's probably because I personally see myself as being much weaker on the firmware side than the hardware side. I feel comfortable leading a hardware team, but I don't feel comfortable leading a firmware team. So that's why I guess I ended up marketing it that way. And that's the type of work we ended up getting. The team actually wants me to try and pick up more firmware work, but I'd say at the moment we're like 90%, you know, like EE mostly focused on circuit board design and then 10% firmware. Yeah.

Alvaro: It's hard not to do firmware though, right? Like most, a lot of things will have a microcontroller on there. And even if it's simple, you have to program something.

Dave Jones: I mean, it depends on the size of the client, right? Some clients have. Oh, their own team. Have their own team. So they don't necessarily want you to do it.

Alvaro: Yeah.

Dave Jones: But most of the time, if a client is coming to you, it's because they want the whole enchilada, right? They, they want the whole thing done for them. So I agree with you. It's impossible to it.

Chris Gammell: I think it's the best, the most fun clients are the ones that want the whole enchilada. And yeah, then in that case, then you definitely need to offer other services and yeah, be capable and all of that.

Alvaro: How do you deal with board bring up with those clients where they're doing all the firmware and, and, you know, bring up as kind of a hybrid of the hardware engineer, you know, powering it up, but you also might need firmware support to do it. How do you deal with that?

Dave Jones: Alvaro, it's like you're in my brain right before I started recording. I was starting to plan out exactly how we're going to do that with one of our clients. So all I can say is what we've done to date, not what I would prefer we do, because I haven't quite figured that out yet. But what we've done to date when they have internal firmware teams is that we will just basically, you know, we'll finish the design. We'll help manage the fab and assembly. We'll get the boards sent. Usually we'll send like either all of them to us or a sample to us and the rest of the client. And then we'll just test the voltage rails test that we can JTAG scan the chips. That's it.

Chris Gammell: Right. And then, and then you solder your 30 gauge wire to this QFN over here.

Dave Jones: Yeah, you're absolutely right. So usually what would come out of it is like, oh, okay, well here's some rework instructions. Send all of your boards back to the shop. Yeah. And, and have them reworked and then we'll track them with a new assembly number and we'll send them back to you. And then should be good to start JTAgging from that point on. Good luck. Contact us when you find something wrong.

Alvaro: Yeah. That's rough when, when your power rails are turned on by the microcontroller, right? Like a lot of my, a lot of my jobs in the past, I've been doing firmware, but with hardware. So, so I, I, the board bring up was my job and I didn't design it, but I had to get it working. And, uh, at least I, you know, had access to the firmware engineer, to the hardware engineer, but having them remote sounds, sounds rough. If anything goes wrong. I mean. For folks that don't know hardware, I guess, I guess, cause I've worked with firmware engineers like me who come from electrical engineering. So it's not as scary, but also you have the folks that came from software engineering down to firmware and electrons are terrifying. Right. So there's depends on who you're dealing with.

Dave Jones: We are, we've been lucky. We haven't had that second character, second type of person yet. You'll get there. Right. That would be a problem. The only time we've really had problems like that with the whole like remote bring up process is when we have a client that is non-technical. Yeah. But really wants to insert themselves into the product development process and wants to like power it up and test it for themselves the first time or want to involve their team in that. And then it becomes quite a problem. Then we usually end up going to the client site.

Alvaro: You know, I was like, you should charge extra for that. It's like, you want to pirate yourself.

Dave Jones: Yeah. Yeah. I've been thinking about various ways that we could like, you know, ship it with a laptop that already has zoom on it and like a cellular connection. Because inevitably they'll be like, oh gosh, we've had so many times where the client is testing the product in like effectively like a server room. That's like too loud to talk inside of. And I want to just ship them some good noise canceling microphone headsets so that we can have a zoom conversation while they're in. They might not even be in the US, right? They might be in the country.

Chris Gammell: Yeah, exactly.

Dave Jones: And we've gotten real close to shipping like full laptops with a cellular modem.

Chris Gammell: I mean, at a certain point you just ship someone, right? That's the, I think that's the option. Yeah. And then you end up just shipping someone.

Alvaro: Have any of you done embedded, like a Linux, embedded Linux systems? At what level?

Dave Jones: Of like all the way?

Alvaro: Like creative one, like make a board with an SOC and then get it up and running?

Dave Jones: There is a board we're working on at the moment, which I'd have to say is like the most complicated one. Most complicated of my career. Certainly the most complicated that I've ever come across, which we're working on an ethernet switch using a, I think it's Micro Semi is the brand, Jaguar SOC, which is just running embedded Linux. But again, we're not doing the firmware for that one. That's one of the companies that has its own internal firmware team. And I mean, the funny thing is they're not even doing the firmware for it because it's too complicated for that part.

Alvaro: Yeah. Because that's too complicated. That's my question.

Dave Jones: They brought in holes.

Chris Gammell: Yeah. And I think it's also like if you're, if you're using like a module or if you're doing like, you know, your own like build all the way to the, to the metal, that's another like defining characteristic, I feel like. Like, so I put like a CM4 on a board. Right. But that doesn't, I wouldn't really call that like a Linux system so much as a, you know, it is a embedded computer that I just need to get the pinouts. Right. And then everything's.

Alvaro: Yeah, it is. But that one, like you and I know how to do it. Oh, we just put a Boon to on it. Right.

Chris Gammell: Right. Yeah, exactly.

Alvaro: But, but that's because Ubuntu comes pre compiled for it.

Chris Gammell: Right. Exactly. Yeah. I think when you're doing Yocto or build root or whatever, you know, that's when it gets more complex.

Ariel Briner: Yeah, the CM4 project I was involved with this year, which is a embedded camera used build root for the distribution creation and, you know, did a nice power on self test and fail back to the previous image if certain things didn't work on that first. Oh, nice. Oh, wow. That's awesome. Yeah. You know, it, it took a little while to get that all working and make sure that, you know, an update if you interrupted power would, you know, not brick the device, but, you know, you test all that thoroughly and it all works well. You know, uh, the, the build root support for the CM4s is really good and, uh, you know, fairly easy to configure the right kernel and, uh, include all the software we needed. And I think in, I've not done much with Yocto, but my impression at least is that build root, you sort of have to do more yourself, but it's a lot simpler than Yocto with all of its giant overlay, you know, uh, systems.

Alvaro: And yeah, no Linux is, I I'm a Linux user, but I never have developed like things like that.

Ariel Briner: Chris, this was a project you did for a client. Yeah. So it's, I can't talk much about it yet because my involvement hasn't been published. And at some point the whole project will be open sourced, which is why, uh, I was excited to get involved with it.

Alvaro: Cool.

Ariel Briner: But yeah, it's a embedded camera using the Raspberry Pi high quality camera and the CM4 and a custom carrier that I did, uh, and an enclosure. And I sort of got it to the prototype stage, you know, hardware, uh, from both electrical and mechanical, and then software, uh, got it to prototype level and then handed it to a larger team to take it to production and do industrial design and injection molded enclosures and, you know, redesign the electronics for machine robot arm assembly and, um, you know, automated focusing and UV cured staking of the lenses and, you know, all of that, which

Dave Jones: was, that's very cool.

Ariel Briner: It was a very, yeah, very impressive, uh, you know, final product and way too big of a project for a single person to tackle, but it was neat to be on the sort of the start of that and prove out the concept and get the initial software stack running and do the initial hardware design and bring up.

Chris Gammell: Yeah. It's kind of an interesting question of like how many, you know, like when you're going with a more, you know, I just think Linux systems are generally are more complex. I think that's a pretty basic, but important statement. Uh, but then like, you know, what are, what are minimum team sizes to like properly start to do this sort of thing? You know, like, I feel like embedded small little fiddly bits that I make with the, you know, little micros. I can handle that myself, right? My code hardware production, all that stuff. But then like, there are some certain size projects where it just kind of gets, you have to buy necessity. Do you have to grow it if you want to deliver it in a, in a normal time span?

Alvaro: Yeah. And that's a part of defense and experience, right? Like I'm sure if I had worked on 10 embedded Linux projects, I would have kind of a set of tools that, that I know how to use like, oh, build root. Okay. I just need to tweak these things and, and then speed things up. And I'm sure you could do it with a, like one or two people, but, but like you said, it's all about times.

Chris Gammell: Yeah. I do feel like every time Jay Carlson talks about it, it's like, sounds like the simplest thing in the world. I'm like, I don't even know what you're saying right now, Jay.

Alvaro: Yeah, no, exactly. The people do this. And that's why I'm always like, oh, I should try this sometime. But no, I, after every time I start contemplating it, I'm like, no, it's easy. Boom. Two. I, that's fine.

Dave Jones: Yeah. Togetherness. Chris, I'm impressed that they gave you the, your clients gave you the like room to work, to put that whole build, build root process together and make it so clean. You must have been working for like very technical clients that they understood the importance and the elegance of the system that you were building.

Speaker ?: Yeah.

Ariel Briner: There were very good clients in many ways in that particular, or you didn't tell them about it. Maybe that, maybe that was, yeah. It's closer to what Chris is saying where like, we knew that that was the right thing to do. So we did it and just slow. It just happened. Right. But it wasn't like a major thing that we were putting in front of them as here's this giant scope of effort first, because it wasn't giant. And we knew that if we discussed it too much, it would have gotten pushed down on the priority list. Like everyone agreed to need to happen prior to product launch, but it helped with testing and bring up and regression testing so much. So I'm really glad that we did that really early on in the prototyping stage so that we had reproducible builds and were able to test different versions of software and different hardware in a really clean way. So you use the, you use the lies to children approach with your client at that time? No, we didn't lie. We just, we just did it upfront and, you know. Did you make a digi-wish? I mean, I definitely do this.

Alvaro: Did you call it DigiClause? I do this kind of thing at work. Well, I knew there's something that we, like you mentioned reproducible builds, right? Super important to be able to build the exact same binary on my machine, on the cloud, on someone else's machine, but it takes so long to get it right. And if I told the team like, oh, this is going to take me a month to figure out, I mean, not a month, but however long, they'd be like, no, no, no, this is not important right now. It's one of those things that I'll sneak off on my own, like a few hours a day for, for a few weeks, get it working and then tell everyone, hey, we have reproducible builds, by the way. And nobody will appreciate the amount of work that went into it, but, but it's the right thing to do.

Ariel Briner: Yeah. And it saves so much time later on. Later on. Yeah.

Chris Gammell: I don't know, guys, if you just made a Gantt chart and, you know, put it in front of your leadership, I'm sure everybody would be super happy about it.

Alvaro: Yeah. Same, same with bootloaders, you know, like, you know, we should. Implement a bootloader. It's going to make things easier.

Chris Gammell: Yep. Yep.

Alvaro: And now that we have cellular.

Chris Gammell: It's going to be important at some point. Yep. Yeah. What about, so, so this is something, so, you know, Chris and I have both been mentioning CM4 and other things like this is something I ran into recently. Actually, not even recently. I've been, you know, I, I have been not making a lot of custom hardware in, in so many words. I've been using a lot of modules, a lot of like breakout boards, things like that.

Alvaro: Like trying to get stuff done.

Chris Gammell: Yeah. Yeah. There's that. I find it kind of frustrating that I'm, I'm, I haven't found a good reason to really dig in at least, you know, some of it's just the type of work that I'm doing, but like, you know, using, using pre-configured modules, there's so many of these days, it kind of helps you to abstract some of the problem. Are you guys seeing that in your work as well? Like, like in my case, I was talking about the Pico and, you know, I'm using the Pico on a, on a board I'm working on right now, but like that same kind of thing.

Alvaro: I mean, we, we did it for the cellular modem, right? Instead of getting a cellular module and writing the driver, I'll sell, well, I mean, there was some drivers, but instead of talking AT commands to the modem, like we do for satellite, we just got the, the note card, the blues that IO module, and they deal with all the, you know, security, the updates, all that stuff on there. And, and SIM cards, right?

Chris Gammell: SIM cards on there too. So that's nice.

Alvaro: Yeah. Yeah. It's, it's built in and, and, and they have a North America model and then a rest of the world model. And, and we didn't have to deal with any of that. And yeah, we were paying a premium, but the engineering time we saved was like, we wouldn't have been able to ship in time. So, so there are a lot of benefits of just not rolling your own, even though you, you lose a little bit of the control, it's usually the right call. I think.

Ariel Briner: Yeah. It depends on the particulars of the project and the volume and, you know, what kind of module this is. I'll mention that like on this camera, there's a Laura radio in it as well. And for initial prototyping, we used a module, but we quickly went to building that ourselves for supply chain reasons. And also to get moved to. Like chipped down then versus, versus like a Hope RF style. Exactly.

Alvaro: SM32WL. There you go.

Ariel Briner: Well, no, but we actually wanted just the radio transceiver because we wanted to do all the software on the CM4. So we didn't want a microcontroller in between the radio and the CM4. Fancy. Which actually made it harder to find a module because most of the modules have an MCU built in. So for that project, you know, doing chipped down made sense. You know, the volumes were there. It was the right engineering call, but you really need to consider all these trade-offs for the particular effort you're working on.

Dave Jones: Yeah. I think by virtue of the work that we do, it's all custom. We, we always are, you know, spinning boards of our own for a few of the projects. Yes. It will do prototypes with eval boards. And then the final pro the final product is really just those eval boards smashed together onto a custom PCB, but they had to be smashed together because this is something that's going to production. And it, you know, it, there were size constraints, whatever. That's only a few of the projects. Most of them. They're just, how do I put this into words? They're just way too custom or there's some, you know, form factors usually a big deal. It's usually, it's a big, it's a big one form factor. Yeah. Well-defined.

Alvaro: How about?

Dave Jones: Yeah. Yeah.

Alvaro: Yeah. I mean, they always want it tiny, right? Like the one thing I'm working on right now, it needs to be microscopic and, and I don't know how we're going to fit it, but we're going to figure it out. And, and yeah, you have to go custom.

Dave Jones: For the first time in my career, I inherited a project from another engineer. That's not first, that happens all the time, but I inherited a project from another engineer where the brief from the client was the other engineer made it too small. Oh, really? Wow. Yeah. Like to have to troubleshoot or something? It was too small to easily handle and assemble into the, like the brief they gave the previous guy was make it extremely small. So he went off and made it extremely small and then they came back to, oh, this is too small.

Alvaro: Wow.

Chris Gammell: So now we're making it bigger. I guess it's a DFM type thing. If you were actually like hand assembling and need to test and probe and all that other stuff too, right? Yeah. That's exactly the problem. Yeah. Yeah. Yeah.

Dave Jones: We go through, I mean, that's a lot of our work as well is we go through DFM. Some of our projects in like the hundreds of thousands of units being manufactured now. And we were doing it all the way from early R and D. So lots of, lots of boring compromises you have to make when you go to production.

Chris Gammell: I feel like that's, you know, if you're calling a, you know, a hardware consulting shop that, you know, you're probably only doing it if you want to, you're not doing it just for funsies, right? You're doing it because you want to take something out in the world and you want to make it custom and make it low cost or, you know, small, whatever you guys are saying there. So that makes a lot of sense. That would be there. Yeah. I think I'm just not, I'm not in the high enough volume space guys. I'm just in the, I'm in the fives, the tens. Not yet. Yeah.

Alvaro: Yeah. We're, we're, we're medium like to low. Cause you know, there's not a ton of, of buoys out there. I mean, we're probably in the high hundreds to low thousands kind of thing more or less. So it's not enough to justify custom designs for everything.

Chris Gammell: And buoys aren't really size constrained either. I'd imagine generally.

Alvaro: Ours kind of are like the, they're about basketball size. The ones we're doing, but the electronics are kind of much smaller.

Chris Gammell: They're smaller than I thought. I think buoy, I think like the one that the, like the walruses get up on sort of thing. No, no, no.

Alvaro: These are, these are pretty small. That's the nice thing about them that we can just chuck them. The we'll give them to cargo ships and they'll throw them out the back. It's kind of a fun activity for the crew to just throw. And now we've also been throwing them out of airplanes and helicopters without a parachute. And they survive.

Chris Gammell: What is the terminal velocity of a buoy?

Alvaro: Oh man. I don't remember.

Chris Gammell: But it must be calculated at some point though, right?

Alvaro: Well, we measured it more than calculated. There was more of a. We have accelerometers on there. Let's throw it out and see what happens.

Dave Jones: That's the difference between engineering and science.

Alvaro: We did write it down. We actually, so we have barometric pressure sensors and we thought about using that to calculate it, which is possible. But the problems started spinning around and then you have all these weird noise artifacts. But the GPS worked great on the way down. And we're doing like two and a half Hertz update rate. And we were able to get pretty good.

Chris Gammell: What else you need? Yeah.

Alvaro: XYZ. But that is one of the limiting factors for the electronic size. They need to be small enough to fit into this enclosure. And you can't use two big components because they'll fly right off or you have to glue them down. It's a pretty high G situation instead of hitting the water. That makes sense.

Chris Gammell: Well, we were almost at an hour mark here, guys. What other trends are you seeing in the marketplace? Or what are you excited about? I guess we've talked about the things we've been doing, the things that we've talked about often in the space. We didn't really talk too much about the shortage, but we've talked about the shortage a lot.

Alvaro: I'm excited about it to end.

Chris Gammell: That's true. Yeah. What else are you guys excited about? New parts, new technologies, new things in the world. What are you, 2023 and beyond? What are you thinking about? Crickets. I have such a piece. Nothing. Absolutely nothing. All right. Yeah. Same here. The future is bleak. I'm such a piece of this. I do. You know, like all of us have been on, you know, I've been on the show for a while and you guys have all been on the show. We're all getting older here. You know, like.

Alvaro: Speak for yourself, Chris.

Chris Gammell: Well, you know. Yeah. Not all of us can live in the Bay Area and what blood transfusions. Is that what the new hot thing is?

Alvaro: At best, I would be able to sell it, but I'm probably too old. I'm looking forward to Chaos Camp, hopefully, if it happens this summer.

Chris Gammell: Oh, okay.

Alvaro: Yeah.

Chris Gammell: That was a lot of fun last time. Yeah. That was 2019, I guess. Right? Yeah. Chaos Camp.

Alvaro: Long time ago.

Chris Gammell: Yeah.

Alvaro: Yeah. No. I think catching up with people is, I mean, I keep saying that the past three years, but I did get, I cut off with the, with the Mighty Home himself when I was on my way to Tor Camp. And then Tor Camp was a lot of fun. And yeah, just catching up with folks is always, always nice.

Chris Gammell: I agree with that. Yeah. I was excited about catching up with you, except you didn't go to Supercon. You instead went to. The other one.

Speaker ?: The other one.

Chris Gammell: Anti-conference. Yeah.

Alvaro: Yeah.

Chris Gammell: Yeah. Okay.

Alvaro: They didn't schedule at the same time. Wouldn't be a problem.

Dave Jones: Yeah. I have such a businessy answer to the question. Go for it. Chris.

Alvaro: Yeah.

Dave Jones: Like specifically for Electron Labs, I am excited that we may in 2023 finally reach like the size and stability level where we could do a project of our own that isn't just for a client.

Chris Gammell: That's a big deal though. Right. I mean, I don't know.

Dave Jones: I don't know if it's going to happen even in 2023, to be honest, because it is really hard to do, but that would be really nice. That's kind of like a stretch goal that I'm pushing for in the back of my mind.

Chris Gammell: How are you going to source those ideas though, too? I feel like that's, I guess, you know, it's not, it is a good thing to think about, but it's also then it's not just a business achievement, also a business risk. So like, how do you actually source that idea that might be useful?

Dave Jones: That's a big question. I think I'll let the team pitch ideas, right? As we come closer and closer to this happening, because I want everyone to be involved. I already have an idea. I haven't told them about this yet. So if they're listening, this is when they're first hearing it. Yeah. I was going to say. I have already an idea in my back pocket and it's, but I want to be clear. This is not something where I think it's going to be something that we sell or money making. This would just be something that would be like a demonstrator, a technology demonstrator, fun for the team, something to-

Chris Gammell: Open source maybe sort of thing. Oh, definitely. Get the build skill and market and show off. That sort of thing. Yeah. Okay. Got it. Yeah. Yeah. It's almost like hackathon, but like a little bit more pro.

Dave Jones: Yeah. Yeah, exactly. I want to bring a DFM aspect to it because it's kind of one of the things that we specialize in. I want to make something that's not just like a fun little prototype, but that we actually take through the process of DFMing. And I hadn't thought about it much, but maybe we document that process a little bit and make that documentation available. Cool. He's excited about marketing, folks. I mean, I had a very business-y answer.

Alvaro: If that gets people making stuff, to me, in these kinds of situations, I always end up going for the tools. Like, oh, what is something I wish I had during the past five projects that would make things easier? And then I'll do this for myself, right? I'll make tools for myself, but that's usually good. Like, I would buy this for me. I'm going to go make it.

Dave Jones: Well, Chris, you call it marketing. I call it education. So the idea that is in my back pocket is you guys have all seen Mark Rober's Glitterbomb, right? Oh, yeah. Yeah. And you've seen probably, like, he's done, what, five videos on it now? And he said this last one was the final episode. And in the last two episodes, he started including the concept of car break-ins in San Francisco. And this is something that is near and dear to my heart and probably yours too, Alvaro.

Alvaro: I don't have a car anymore, so.

Dave Jones: Breaking into a lot of cars there? That's one way to get rid of car break-ins is get rid of your car. My wife had her first car break-in since moving to the city, since moving to America. And it really rattled her. And it was right about the same time as the last Mark Rober episode came out on the topic. And so it was, like, fresh in my mind. And I was thinking about how, like, he was unsuccessful in being able to... Like, he was very successful with porch pirates, with people stealing packages from front doors. But he was largely unsuccessful with pranking, which is, I guess, the outcome, what he's really looking for. He was unsuccessful in pranking car thieves. They were... It just didn't work out. I don't know. I don't think he spent enough time on it.

Alvaro: Well, they wouldn't take the really obvious boxes. They wouldn't take the bait.

Dave Jones: Yeah, part of the problem. And I was trying to think about how I could basically start where he left off and make a purpose-built product that was designed to be stolen and, for lack of a better word, terrorize the car thieves. And I have some ideas about how to do that. I have to look into the legality of some of those ideas. But I was thinking that not only just make a little proof of concept and, you know, make it into, like, a little video or whatever, which is cool. I want to do that as well. But, like, really design it with DFM in mind where we can design it down to a price, make a little payload that can be loaded into a fake Amazon box, loaded into a fake backpack. No blowing people up, okay? All very non-lethal, all very funny pranks, you know, nothing too serious. Parts great. But, like, you know, put a little bit of the fear of God into these professional thieves that might make it so that they decide they don't want to do this anymore or at least make their business model no longer profitable because it is a business for them that would maybe then, you know, win the streets back. I think that's something that people in this city can get behind.

Alvaro: If this weren't, yeah, people would buy it.

Dave Jones: I think people would buy it, right? Like, it's not something I want to make money off. I would just want to break even and have the good feels of knowing that we're actually striking back in a way that is both hilarious and effective and non-lethal. Very importantly, non-legal and legal. So that's the tricky part. But, yeah, that's what I was thinking is, like, go through the process of actually DFMing this and showing, like, setting up a production line for a new product and designing down to a cost and designing for reliability and availability of components, putting it in the field, getting feedback on where it doesn't work, revving it, et cetera, et cetera. I think that would be pretty good. I reckon if someone came to me with that offer, like, I was just a person in a city that had car theft and they said, hey, for, like, 50 bucks, you can buy this product that you put inside, like, a cheap $5 handbag or something and leave inside your car. Well, your car's going to get broken into then. So maybe you have a work truck, right? And your tools get stolen all the time. So you leave this on the back of your work truck in, like, a cheap plastic toolbox. And then five minutes later, you're inside the work site and you get a little text message saying your package has been stolen. And then 10 minutes after that, you get an audio recording of people inside a car with, like, a dye pack going off and exploding inside the car, staining the inside of their interior. And you hear the commotion that goes off. Of course, an accelerometer to make sure that the car is stationary at the time. But I think that would be well worth $50.

Chris Gammell: I think that's a buy. That's a buy right there. Exactly, right? Please, who else does? I'm interested in terrorizing car thieves in 2023. Chris, what about you in 2023? What are you excited about?

Ariel Briner: I'm excited to launch some new products. I've got two things that are done that I just need to actually put on the website and write some documentation around. And then hopefully two to four more products launching later on this year. We've done a lot in the works.

Alvaro: Getting more hubs?

Ariel Briner: Yeah, new version of the USB 2.0 hub, a new USB-C version of that. Sort of, like, souped up from a speed and monitoring perspective. The two recurring requests I've had on the USB hub are, you know, I want faster speed and I want better power resolution. So the new USB-C version will do both of those and some other fun things as well.

Alvaro: Oh, I'm going to get a bunch for work. Me too.

Chris Gammell: You should license Alvaro's cable tester in there, too. You should just add that as a little fun add-in.

Ariel Briner: It could be an add-on accessory.

Chris Gammell: There you go.

Ariel Briner: Yeah. Yeah, you know, the supply chain has thwarted a bunch of things that I started two years ago. And it seems like it's easing a bit, but definitely not out of the woods yet. And, you know, some of this will be able to be produced because I've been buying parts over the last two years in anticipation of the things that I was designing over the past year. And that I'll be designing early this year. But it's been horrible from a cash situation to be, you know, buying parts for things that haven't been designed yet. Because I know that these ICs are critical and that they will be in the design. But, you know, it's not an effective use of money except that it has to be done this way in this order right now. So it's nice that things are easing a bit, but there's still big categories of products that are difficult to source. So, yeah, sorry to end the show on that down note. Supply chain.

Chris Gammell: Supply chain.

Alvaro: No, we're ending on new products.

Ariel Briner: New products are coming. Yeah. No, we're coming out the woods.

Chris Gammell: I'm also interested. I'll take one. I'm, you know, I am increasing. I just actually bought a Pine64 board, the RISC-V, with a three core RISC-V processor on it that runs Zephyr. So, like, you know, obviously I'm on the Zephyr train, but also, like, I'm real pumped about RISC-V. You know, every time I, you know, I'll read about it here, there, anywhere. You know, there was an article about, like, Google targeting it for Android phones. It's like, okay, well, we'll see. We'll see that when we see it. But, like, but just generally, like, you know, like you guys mentioned with the CM3, or sorry, the ESP32C3, like, all of the parts that are coming out. I'm real jazzed about RISC-V stuff.

Speaker ?: So.

Alvaro: Absolutely.

Chris Gammell: See where that goes.

Alvaro: I mean, even if you're in the ARM camp, it's just going to mean ARM's going to have to make better stuff, which means yay.

Chris Gammell: Yeah. Yeah. That's great. I have, well, I haven't done anything with it yet, but I did get in the RISC-V 10-cent micro as well. The, what is it? I'm trying to grab it here. Sorry.

Alvaro: Like 10 cents or 10-cent, the company?

Chris Gammell: 10 cents. It costs 10 cents. It is the RV32, I think. RV32V003, but it's like a super tiny 48 megahertz processor with a bunch of peripherals on it.

Alvaro: Nice.

Chris Gammell: Yeah. There's not much of a tool chain for it yet, but it's real cheap.

Alvaro: You just put in ones and zeros.

Chris Gammell: That's right. Yeah. How hard could it be? I don't think there's any Rust support for it yet. How about that?

Dave Jones: That's awesome. At that price point, that's going to unlock so many more, so many more projects, so many more possible products.

Alvaro: I don't know how AdMail is going to sell their ATtinis anymore. Like, except for the fact that they can sell it. Yeah.

Chris Gammell: They'll sell the legacy stuff, but I think people still are just comfortable with them. I don't feel like, I feel like chips don't really die. They just kind of slide into the comfort zone of certain engineers. You know what I mean? Yeah. That's true. But the ones who are comfortable with it, they keep designing them in, and they should, right? That's a skill set that they have that they could just deploy over and over again.

Alvaro: Yeah.

Chris Gammell: It's just, they get hosed when supply. Yes. When it's not prioritized for supply, then it's like, oh, okay. That's when the rubber hits the road, I feel like.

Alvaro: Oh, you didn't buy years worth of parts? Yeah.

Chris Gammell: Yeah. Yeah. It's a darn shame. It's a darn shame. What about, I guess, other hot button issues? AI as well, I guess, as Dave just made a video about that. I'm sure we'll talk about it when Dave's back from vacation. Yeah.

Alvaro: I mean.

Chris Gammell: All the art for this show will be using, what's it called? Mid-Journey.

Alvaro: Someone posted someone using the ChadGPT style thing for Ghidra for disassembly, for reverse engineering. Yeah. Where it will tell you what the code, what it thinks the code is doing, which that might be interesting.

Chris Gammell: Yeah. Like kind of like a first guess sort of thing.

Alvaro: Yeah. Like it tries to explain. Because you have disassemblers, you have decompilers. Sorry?

Ariel Briner: Yeah. What if it explains wrong? Like.

Alvaro: I mean, it's. Yeah.

Chris Gammell: That's when you have to use your artificial, artificial intelligence.

Alvaro: But reverse engineering is trial and error. Like when you're, you know, decompiling or disassembling. You have to make up or make guesses as to what is this variable mean? Because you just know variable one, variable two, variable three, and you don't really know what it does. But there are tools already to figure out. Oh, this looks like a printf function, right? This signature, this assembly code looks like a printf. Or it looks like a Linux syscall this or that. So I think it's going to be more of recognizing those patterns than telling you what it does. I don't know.

Ariel Briner: There is potentially value in that. I mean, there's the danger that it's slightly wrong and there's major implications to that and you don't catch it. Yeah. You know, I've heard these systems described as. Yeah. I think it was. Trusting it too much. Stochastic bullshit. And that's not that far off from what they're doing, unfortunately. Oh, absolutely. There isn't a deep understanding, right? Like it is pattern matching and often it pattern matches wrong and it's, you know. So do I though. Right, right. But at least if you're writing something, I find your mind is like formed around the thing that you're doing. And I think it is therefore easier to debug. Like it's harder to debug someone else's code than your own code. And catching a mistake in one of these chat AI things, I think would be harder than catching a mistake in your own work or even in human generated work. Because the way that they phrase things and the confidence that it has in what it's saying puts more, you know, more gravitas or weight on it than I think should be there.

Alvaro: I think the best use that I've heard people is to kind of go back and forth with it just to get you, your brain out of a spot. Like if you need ideas, if you're stuck, you can have it spit out nonsense. And maybe some of that nonsense will be like, oh, idea. And then you can go off and do a different thing. Because apparently you can have arguments with it. You can tell it to argue against your argument. And yeah, it might be BS, but it might get you thinking like just rubber ducky 2.0. I don't know.

Ariel Briner: I guess I think there's danger in truth as a source of truth. Oh, it's absolutely not. Right. But people will treat it that way.

Alvaro: Oh, that's a problem. Yeah. Yeah.

Dave Jones: So you have to teach it humility so that it's a little bit less strong of itself. Yeah.

Chris Gammell: Yeah. Right. Maybe it's this. I kind of like what Alvaro was saying too. Like I think the, there was a group that I was on a, on a chat with and they said, you know, one of them said they, they used chat GPT as a, like a way to break writer's block. And I think that as a, as a mental construct is actually really strong because it's like, I'm not going to use it to write, you know, an article for me, but if I need to think about some of the things that kind of is in the zeitgeist around a topic, like that might give me some bullet points that I could, you know, start to research and, and dig out of there. And like that, that actually was really useful in at least one case for me. That's probably the main time time that I've used it so far.

Ariel Briner: Yeah.

Chris Gammell: That's, that's a very good point. I wanted to get to argue with me though. Like argue with me about Ohm's law. You're wrong, Chris. Well, yeah, you guys will do it. Come on. I want a computer to do it.

Alvaro: It might.

Chris Gammell: Yeah, it might. Yeah. All right. Well, I'm really glad you guys have been here. Thank you for coming back to the show to catch up with me and talk about 2023. It's been really great to catch up with y'all and hope you have a, hope you have a prosperous year. Likewise.

Ariel Briner: Thanks for having us on.

Chris Gammell: All right. We'll see you soon. Cheers. Bye.