#340 – An Interview with Jason Cerundolo

Chris Gammell: This is The Amp Hour Podcast. Reported previously, but released March 19th, 2017. Episode 340. Interview with Jason Sarundalo.

Dave Jones: Welcome to the Amp Hour. I'm Dave Jones from the EEV blog.

Chris Gammell: And I'm Chris Gammell of Contextual Electronics. And I'm Jason Sarundalo from CastAR.

Dave Jones: Hey, Jason.

Chris Gammell: Welcome, Jason. Hey, how's it going?

Dave Jones: A CastAR is a CastAR alumni.

Chris Gammell: CastAR, yeah. What is something-something CastAR? Who we always talk to about that.

Jason Cerundolo: Yeah, so we're making an augmented reality gaming system, so you can play 3D holographic games in your living room. Awesome.

Dave Jones: Can we have an update here? Because we've had Jerry on before, of course, who founded CastAR. Or was it? No, it wasn't called CastAR back then. It was Technical Illusions. And then you sort of pivoted, changed names, and went in a different direction. Are you able to talk about that direction change and how you refunded backers and stuff like that? Because I don't think we've-

Chris Gammell: Oh, we covered all that stuff in the past, though. Did we? Have we? Yeah, we've covered that before.

Dave Jones: Did we have Jerry back on after that fact?

Chris Gammell: I think so, yep.

Dave Jones: Oh, okay. Well, how long have you been there, Jason? Give us the background.

Jason Cerundolo: Well, I've been here just over two years. Yeah. And I started in 2015, right after the company moved down from the Seattle area. So I was here when we raised our Series A.

Dave Jones: Moved to Silicon Valley, right?

Jason Cerundolo: Yeah, right now we're in Palo Alto.

Dave Jones: Which is as Silicon Valley as you can get, really, isn't it?

Jason Cerundolo: Yeah, right. Yeah, it's pretty much the- It's actually the I-caric. HP garage. A ride.

Dave Jones: We're practically in the garage, yeah.

Jason Cerundolo: Yeah. And so a quick update is we're still planning on launching the product this year. The schedule is going to be a little tight, but we think we can hit it. It'll probably be a bit of a soft launch. It won't be a huge deal like hitting all the retail stores and everything right up front. But yeah, I know you guys covered the Kickstarter situation. I feel like that was a pretty good summary there. And so right now we're trying to raise our Series B. We're in the process of closing a round right now. And soon we'll have the money that we need to go through launch and aggressively hire a bunch more engineers and other people.

Dave Jones: And everyone gets a smaller cut as a result. Yeah. Well, you've got to build a bunch of hardware too, so that ain't easy. So how much did you get in Series A? I forget. Was it $20 million?

Jason Cerundolo: Yeah, it was $15 million. $15 million.

Dave Jones: Right. Okay. So that wasn't enough. You need more. It's a money thing. Now you need more money to expand and take it to production.

Jason Cerundolo: So we need money just for the bill of materials is one thing. Yeah, right. But in addition to that, a big cost obviously is the engineering talent and salaries in general. Just take a fake cut, Jason. Come on, man. Just go for it.

Dave Jones: Yeah, work for pizza and the good of humanity.

Chris Gammell: I hear it's really cheap to live in Palo Alto or nearby.

Jason Cerundolo: Well, we're over 100 people right now. Wow. Holy crap.

Dave Jones: Wow, that's big. Yeah.

Jason Cerundolo: Yeah, we've been growing quite a bit recently.

Dave Jones: That's a lot of cash. That's a lot of cash flow.

Jason Cerundolo: That's a lot of burn. Burn, baby burn.

Dave Jones: Burn rate, yep.

Jason Cerundolo: Yeah, it adds up pretty quick. So we have our bulk of our engineering team here in Palo Alto and we also have a first party game development studio out in Salt Lake City. Right.

Chris Gammell: That's right. I remember that was like former Disney people or something. I remember hearing about that, right?

Jason Cerundolo: Yeah, I started with them and then Eat Sleep Play and a couple other, not acquisitions, but we've gotten a lot of developers from other studios and they're going to be developing content exclusively for Cast AR.

Dave Jones: Got it. Cool.

Jason Cerundolo: That's great.

Dave Jones: So they're out there because that's where the talent was that you wanted. So you just left them out there.

Jason Cerundolo: Yeah, it was kind of a good opportunity when Disney Infinity was shutting down.

Dave Jones: Oh, we haven't heard. Have we discussed that before, Chris? I think we mentioned it. For those who don't know.

Jason Cerundolo: Well, they were doing, I believe, the Skylanders type stuff. What, Skylanders? Yeah, little physical toys that interact with the virtual games.

Dave Jones: Oh, okay. Right.

Jason Cerundolo: So, yeah. So that sort of thing is right up our alley. You know, we want to merge the digital realm and the physical realm. Cool.

Dave Jones: Yes, because you interact, the whole idea of your thing is that you don't just sit there with the goggles on and just watch. You can interact with the wand and everything else. Is it still like that? Is that still one of your key differentiators, I guess you could call it? Is that, you know, you can interact with the wand and all that sort of jazz?

Jason Cerundolo: Yeah. So the major components that we have is the headset, which has the projectors and tracking system. We have a content hub that has our main CPU, our battery, Wi-Fi, all the hot and heavy stuff. So we keep that away from your face. And then we also have the game board, which has the magical retro-reflective material that makes it all possible. And the wand is like our game controller, but it also has tracking markers on it. So we can track it in 3D space. And that's one of the major ways that you interact with the system.

Dave Jones: Right.

Jason Cerundolo: And so all the features and examples that were shown in the Kickstarter videos, we're still intending to hit most of that. The technical details of how we actually accomplish that will be changing. So I know we talked about the RFID grid.

Dave Jones: Yep.

Jason Cerundolo: We're not going to be using the RFID in the future. We have other methods that are more integrated with our existing tracking systems.

Chris Gammell: Got it. So what is your role there, Jason? Tell us about that.

Jason Cerundolo: Yeah, well, I think my business card says hardware engineer right now. It better, man, if you're on the amp hour. Yeah, definitely.

Dave Jones: Not the software rubbish.

Jason Cerundolo: You do a job firmware sometimes.

Chris Gammell: Sometimes.

Jason Cerundolo: Yeah, I do firmware when I have to. Right. And most of the time I'm doing electrical engineering. And we actually just hired another electrical engineer. So it's been great to have a little bit of relief there. It's given us a great opportunity to spend a lot more effort figuring out the details of the design and make sure we actually have a product and not a prototype.

Chris Gammell: Gotcha. Yeah. Like taking it to actual production and stuff like that.

Jason Cerundolo: Yeah. It's like the last 20% of the design effort, that's like 80% of the work. Yeah. Right. Yeah, we're in that stage right now.

Chris Gammell: Great. Are you guys working with manufacturers at this point? I mean, again, whatever you could talk about, but are you doing the larger scale type stuff?

Jason Cerundolo: Yeah. Yeah. We've done, I think we've built somewhere on the order of hundreds of prototypes so far. Uh-huh. We're partnering with someone in China. Yeah. And for the glasses, we're taking on most of the responsibility for the design. The hub and the wand are pretty much like a cell phone without the screen and a game controller with a couple extra things. So those were pushing more of the design effort onto our partner because they're relatively simple. And then that allows us to focus our limited resources on the glasses, which is really the Yeah, the magic part, right? The key part. Yeah. Yeah.

Chris Gammell: Yeah, that's good. I mean, and then especially like if you're trying to target lower costs. I mean, you guys are trying to target low cost, bomb costs, right?

Jason Cerundolo: Yeah. I think the key phrase I'm allowed to say is we're trying to hit a price point that's on par with a modern console.

Chris Gammell: Oh, oh, that's higher than I thought. So what, like three, four and a bucks kind of thing? I don't actually buy consoles anymore.

Dave Jones: That's typically the psychological impulse buy point is like sub $400.

Jason Cerundolo: Yeah. We're trying to create something that almost anybody can get access to and have fun with our system. Yeah. And of course it plays into the business models and so forth.

Chris Gammell: Yeah. Yeah. If you're working with an industry that people understand and games and stuff like that. Yeah. It looks like I just, I actually have not seen, I've not been paying attention to the Switch thing at all, but 300 bucks ish looks like. So that's cool. Yeah. That's great, man. So, well, so I mean, technology has been changing over time. I remember when we've heard about it, it was a lot of FPJs and like not DLP drivers, but there were like those projector drivers and stuff, right?

Jason Cerundolo: Yeah. We're using a L cost panels and a, and a Pico projector that we've custom designed. Cool. Nice. That's great.

Chris Gammell: Is that like working from reference designs and stuff? Like just kind of customizing from there?

Jason Cerundolo: Yeah. I'm not too familiar with the optical design. We have a great optical engineer here that's been managing that for the last over a year now. Yeah. And we have some advantages over most projector designs because our light output is somewhere around a lumen, like that order of magnitude. Yeah. And that's much, much less than most projectors are. Yeah. And because we have the retroreflective surface, so all that light or like 99.9% of that light comes right back at the user's eyes. So we don't actually need to be too terribly bright. And that gives us a huge advantage for, you know, battery life, the design of the projectors, so forth. Yeah. Right.

Chris Gammell: No, no huge fans in the face.

Dave Jones: No. Big fans to cool your head down. Yeah. The thing I've been noticed about the Cast AR is that there's so much little technical detail that could almost become a showstopper, you know, like, and you've got to overcome all these little hurdles. Otherwise it's game over, man. Game over, man. And have there been any other, any stuff like that that you've worked on or was that all sorted out like right at the beginning by Jerry and the early cohorts?

Jason Cerundolo: I think a lot of that was sorted out pretty early. I'd actually disagree that the system's pretty brittle. And the way that I've seen it, we have all these, we have multiple ways of doing a lot of the things that we want to do. Right. Okay. Yep. And I've seen a lot of like corrections, the techniques that we have to correct for head movement and so forth. And we have multiple different paths. And every demo that I've seen where we enabled just that one thing makes it a magical experience. And so I'm really excited when we turn on all of these things all at once. Right. And it'll be that much better.

Chris Gammell: So you're talking about like sensor fusion type stuff is almost? Or what do you mean by that? Yeah.

Jason Cerundolo: So sensor fusion is going to be part of it. We can do some prediction as well because your head can't accelerate that far. That's fast. There's limits. Challenge accepted. Yeah. Yeah. And because we've constrained the design space, it really relaxes a lot of the other design constraints. You know, we're not trying to do room scale tracking just yet with this version. Right. Yeah. So we just focus on the game board right now. And then that simplifies the problem quite a bit. Like we always know we're going to have those tracking markers because if you can't see the tracking markers in the game board, you can't see the game board and you can't see the retro reflective surface. So it doesn't matter that much if we lose tracking in that scenario. Got it. So it's little things like that that are going to help make the system very feasible.

Dave Jones: So all the elements really haven't come together yet in a complete package. You're sort of like maybe a few of them have, but it sounds like you haven't actually got the complete package together yet as a whole.

Jason Cerundolo: We have the whole system running end to end right now. Yeah. With the latest prototypes for the hub and the glasses. What hasn't come together is all of the enhancements that we've been making.

Dave Jones: Got it. Yeah.

Jason Cerundolo: Yeah. So when those come online, the holograms will appear sluggish right now, but when the enhancements come online, they really appear to be locked to the table. And there's this certain threshold that's a little hard to describe, but once it gets good enough, it just, your brain locks in on it and just treats it like a physical object. It might as well be a physical object.

Dave Jones: It's kind of like, what's that magic response rate? Like 0.3 of a second or something. And anything under that appears instantaneous, but anything slightly over that. Yeah. 30 milliseconds. Yeah. It's like human perception. 30 milliseconds. The human perception. Anything slightly over that. And it appears, oh geez, that's a bit sluggish. You know, your brain just sort of switches modes kind of thing.

Jason Cerundolo: Right.

Chris Gammell: Yeah. Well, there's also like the averaging of like LEDs and blinking and stuff like that too, right?

Jason Cerundolo: I don't know exactly what you mean by that.

Chris Gammell: So I think that same number also works with like, if you're pulsing an LED, you can't actually see the individual pulse if it's below 30 milliseconds. Like stuff, that kind of thing.

Jason Cerundolo: Oh yeah. Like persistence of vision displays and that sort of thing. That's the right word. Yeah.

Dave Jones: That's what it was often. That's what it was fishing for.

Chris Gammell: Got it. Got it. All right. Yeah.

Jason Cerundolo: So the human eye is actually not that quick, at least compared to the audio, like your ears. Your audio channel into your brain's pretty low latency and your eyes are pretty sluggish in comparison.

Dave Jones: Right. Well, because there has to be a lot of, like it flips the image and it sub-processes it before it goes in or something. There's lots of stuff that happens there visually.

Jason Cerundolo: There's like the V1 and V2, I think they're called. It's been a while since I took that class on the human vision system.

Dave Jones: There's a class on human vision systems? Sorry. I just have to know this. What did I miss at uni?

Jason Cerundolo: Well, there was one when I went to college. Yeah.

Dave Jones: Really? As part of what course?

Jason Cerundolo: It was actually an elective and I was taught by Professor Christophe Koch. Where did you go to school, Jason? I went to Caltech for undergrad.

Chris Gammell: Oh, cool. Okay.

Jason Cerundolo: Yeah.

Dave Jones: In engineering, I assume?

Jason Cerundolo: Yeah. Actually, mechanical engineering. Right. So, yeah. And then was getting interested in control systems and that sort of thing, which involved writing computer algorithms and firmware. And then eventually soldering up robots and that sort of thing. And that's how I got involved with electrical engineering.

Chris Gammell: That was the beginning of the end. Yep.

Jason Cerundolo: Yeah. That's cool, man.

Chris Gammell: So, you guys hang out at Playground all day, right?

Jason Cerundolo: Yeah. Playground Global. It's Andy Rubin's new VC firm. Well, not that new.

Chris Gammell: Is it?

Jason Cerundolo: Not that new anymore, I guess. So, that's where the company's based? Yeah. We're based right here. So, part of the deal with Playground is we get the money, but we also get a space to set up shop. Right. We got all of our engineering staff here. And then probably the coolest thing about it is there's also engineers hired by Playground that work for Playground and consult with the portfolio companies.

Dave Jones: Ah, they can consult for any of the portfolio companies. Right. That's interesting. And you guys have a sweet shop there, too.

Chris Gammell: I mean, that's nice.

Jason Cerundolo: Yeah. We have a really nice machine shop. And we just got like a five-axis, I think it is, mill that can make injection molding tools. We have a 3D printer that can print directly in titanium. Yeah. Obviously, a laser cutter. Totally necessary. Absolutely.

Chris Gammell: The titanium thing.

Jason Cerundolo: Yeah.

Chris Gammell: We haven't quite found a use for that yet. Trinkets. All of our field pieces are titanium printed. Each one costs $400,000 each. That's cool. So, what else? I mean, I got to visit once. That was very nice of you to invite me there. And I appreciated that. It seemed like kind of like a clubhouse. But there are other hardware companies there, too, which is cool.

Jason Cerundolo: Yeah. There's a bunch of portfolio companies here involved in all sorts of different markets. But the one unifying thing is they're all hardware companies. And that's probably the distinguishing feature of the Playground Investment Firm. Yeah.

Dave Jones: So, is the idea that getting all these hardware companies together, do you all hang out in the same canteen and mingle at lunchtime? Or do you get into your own tribal groups and you have food fights with the other hardware companies? Or is that the idea, is to get mixed hardware companies to mingle together? Is that part of the concept?

Jason Cerundolo: Yeah. I'd say that's part of it. We do hang out together, like Chris was saying, at Happy Hour.

Dave Jones: Right.

Jason Cerundolo: Friday afternoons. That's probably the best opportunity to mingle across companies. Right. During the week, though, we're pretty focused on our own work. But we have had a few times where, because we're consulting with the Playground engineers, they see all the engineering that's happening across all the portfolio companies.

Dave Jones: So, they're the ones you want to hang out with at Happy Hour, so you can talk to them to find out all the secrets the other companies do.

Chris Gammell: I was really surprised. Not really surprised. I think that, so like, I really loved visiting there and it was really nice. I got to show up at Happy Hour. That was great. But I'm always surprised when people don't talk about the projects. Like, I talked to, I met someone there and they're like, oh yeah, well, we're in stealth mode. I'm like, who the hell cares? You know, like, I will never understand that.

Dave Jones: Most engineers don't care about that. Most engineers just want to talk about their shit. They don't.

Chris Gammell: Right.

Dave Jones: I'm surprised. Were you hanging out with real engineers? I mean, usually they're, you know, as loose lippy as you get.

Chris Gammell: Right. I mean, Jason, did you hear about that there? I mean, are there people that are like pretty tight-lipped about stuff or what?

Jason Cerundolo: There's definitely some of that around here. I can kind of see some of the benefits of keeping things close to the vest. But in general, I haven't really found a good reason for having or being in stealth mode, as it were.

Chris Gammell: Right. Well, I mean, you're working on these glasses and you understand all the supply chain stuff and all the other difficulties. And it's like, not to mention that, like, when you're doing hardware, you're going to have to get your bomb quoted at some point. Right. It's like, it's getting out of the world somewhere. You know, it's like, yes, there are secrets. And yes, there are things that you should, you know, definitely not talk about. And I'm sure Jason will not tell us things if we asked him. But like, I just, but like even the idea, like, oh, you're working on a drone thing? Ooh. I don't know. I get, I just get, I start getting antsy when people do that.

Jason Cerundolo: Yeah. It's kind of a conversation ender. Right. Right. When he asks like, hey, what are you working on? It's, oh, I'm at an aerospace stealth startup. It's like, okay.

Chris Gammell: Big bumps, huh? All right. Well, see you later.

Dave Jones: I think they're more likely to fail than not because they think they're onto something so disruptive that their odds of success are probably lower. I'm probably generalizing here.

Chris Gammell: Yeah. But I'm, you know. No, I think I agree with you. And like, yeah. Well, like, so like, yes, they are. I like what you guys are doing because I understand it. Obviously, you know, we've seen details.

Dave Jones: But they're pretty open with what they're doing. Right, right.

Chris Gammell: They're pretty open with what they're doing. But then compared, what's the other one? Magic Leap? Is that the other one?

Jason Cerundolo: Yeah, that's, yeah, the big AR company in this space. Right.

Chris Gammell: And that's been a hype machine for five years. And yes, investors have seen it, but they're also hype machines. And it's like, no one's seen a real thing. No one knows if it actually exists. And it's like, I don't, you know, people want to talk about it. It's like, just fine, but show me a real thing. You know, like hardware talks, you know? That's, I don't know. Sounds like you, Beam. Anyway, no, yeah. Touch points.

Dave Jones: So you would agree that these companies are probably, just from a sheer statistics point of view, more likely to fail?

Jason Cerundolo: Well, I'm talking to Chrissy. A lot of startups at that stage are pretty likely to fail.

Dave Jones: Yeah. But yeah, no, I just, that's the, that's the sense I get is that if you're in stealth mode, it's like, you know, it's, you're probably more full of crap than someone who's not in stealth mode.

Chris Gammell: That's quite the assertion, Dave. All right. Anyways. So, Jason, anything else we should know about Playground?

Jason Cerundolo: No, I think we pretty much covered everything. Cool.

Dave Jones: Although stick around for the end of the show, because he's going to tell us how you can work there too.

Jason Cerundolo: Oh, yeah.

Chris Gammell: Well, let's talk a little bit about Reclaimer, because this is your other thing. This is the real thing that I, so obviously I get to see you every month. You are very diligent in driving up to, or taking the train up to the meetup that I do in San Francisco all the time. But you've been, you've been working on Reclaimer as like your personal brand. What is Reclaimer?

Jason Cerundolo: Yeah, I think personal brand is probably the best way to put it. It's basically all the hobby projects that I work on if they're going up on Tindy or something like that. I just kind of put it under that umbrella. And I have a bunch of pretty interesting and diverse projects right now that I'm working on.

Chris Gammell: Well, let's hear about some of them. Well, wait, leave the newest thing for last, because we'll definitely talk about that for a while.

Jason Cerundolo: Okay, yeah. So one thing that I was getting into was LEDs. And I actually gave a talk at one of the San Francisco meetups on LED driver circuitry. And I have a product right now that I'm putting together that involves over 25,000 LEDs. What?

Jason Cerundolo: Mike Harrison, eat your heart out.

Jason Cerundolo: Yeah. So it's basically a small segment of like a video billboard that you'd see up on the highway or Times Square. And so I took a small section of that and hooked it up to a Wi-Fi connected microcontroller. And you can program it over the internet and it can connect to websites and pull information and display whatever you want. So I've set up mine so that it's counting down to the next rocket launch, whenever that is. Nice.

Chris Gammell: Wait, but you said 25,000 LEDs?

Jason Cerundolo: Yeah. So it's 128 pixels wide by 64 pixels tall.

Dave Jones: You don't need many pixels in a square grid for them to add up. Yeah. It multiplies up pretty quick. It's multiplication, Chris. Right. Yes. That thing you learned about in primary school.

Jason Cerundolo: Yeah. So, you know, 128 by 64 doesn't sound that great, but that's why I usually lead with the 25,000 LEDs. Yeah, yeah. It sounds cool.

Chris Gammell: What's the size of that panel?

Jason Cerundolo: It's about the size of two sheets of paper. So maybe about two feet wide and about a foot tall.

Chris Gammell: Oh, okay. So still pretty spaced out though.

Jason Cerundolo: Yeah. It's a three millimeter pitch. Okay. One of the smaller pitches. And then these are little panels that you can buy on AliExpress or Alibaba. Adafruit sells them as well in various pitches. And so what I did to address all these LEDs, the panels themselves come with a shift register interface. And so I programmed up a CPLD to convert a SPY packet from the microcontroller into this parallel shift register interface that the panels use. Mm-hmm. And then that way I can just do a DMA transfer from the microcontroller out over SPY and then leave most of the CPU time available for doing whatever else you want to do. Hmm.

Chris Gammell: Wait. So are you building the panels itself? Are you sourcing the panels and then building the drivers? Or how does this all work?

Jason Cerundolo: Yeah. I'm sourcing the panels. Uh-huh. And then I just design a little backpack board that manages the power and the data.

Chris Gammell: Hmm.

Jason Cerundolo: And then the core of it right now is a protocol photon.

Chris Gammell: Oh, cool.

Jason Cerundolo: Which I think you guys have probably talked about before. Yeah, we've talked about the potting. Yep. Yeah. And so just using the protocol IDE for now and using some of the libraries that other people have written and contributed.

Chris Gammell: What is it? I see another micro on board or is that a driver on the upper right of this thing where the power comes in? Is that just a switcher? Or is it like a PSOC or something? I can't quite see the details.

Jason Cerundolo: Yeah. So the only other ICs on the design right now are going to be the CPLD.

Chris Gammell: Ah, that's probably it. Yeah.

Jason Cerundolo: Yeah. And then I've also had to add in a buck converter. This thing requires quite a bit of power and it runs at five volts. So that winds up being about 16 amps if you turn on all the LEDs.

Chris Gammell: Okay. So yeah. So you got like a laptop power supply or something like that?

Jason Cerundolo: Yeah. I found a 24 volt power supply and then I have an onboard buck that'll take that down to five volt and then that way the connector doesn't melt on me.

Chris Gammell: Oh, right. Right.

Dave Jones: That's often easier to do. Like a higher voltage input, like a 24 volt input and then local sub, like then local regulation at the points required is often a good way to do these large distributed projects.

Chris Gammell: Are you doing 16, you're not actually doing, you're not delivering 16 amps though. You're like, you're like multiplexing or something. How do you deal with that?

Jason Cerundolo: Yeah. So the, the 16 amps is with multiplexing. Oh. Yeah. Yeah. Yeah.

Dave Jones: It's with, yeah. It's, but that's total. So you wouldn't like, you know, it's not like you feed in like 3.3 volts to the entire thing. Cause then you need the, you know, to run the high current. That's why I was saying, you know, have local regulation across the board. So you don't have to run.

Jason Cerundolo: Wait, but how many buck, how many buck?

Dave Jones: The current right across to the other side of the panel or something like that.

Jason Cerundolo: Yeah. No, there's just the one buck converter on my board. And then I think the panels themselves have the shift register. They've got their own local. ICs have a, they're an open drain, you know, constant current. Yep. Set up. So they just burn whatever the extra voltage is off as heat. So like a linear regulator. Got it.

Chris Gammell: Wait, but I'm sorry. I'm still confused here. So, so what you're, you're switching, you're switching supply, your buck converter goes 24 to five at 16 amps.

Jason Cerundolo: Yeah. Five volts, 16 amps out.

Chris Gammell: What's the efficiency on that thing?

Jason Cerundolo: Uh, according to the calculation, somewhere around 95%. I'm getting the prototypes in soon, so I'll be able to really calculate that.

Chris Gammell: So like 75 Watts, I guess. Okay. So like five, one 20th of 75 Watts.

Jason Cerundolo: Is that right? Uh, I think it's a little bit closer to like a hundred. And 40 Watts.

Chris Gammell: Wait, 16 amps is on the 24 side or it's on the five volt side? On the five volt side.

Dave Jones: Chris is getting bogged down in the daytime.

Chris Gammell: My math is just, sorry. Yeah. 15 times, isn't 15 times five. Am I doing that wrong? 15 times five is 75, right?

Jason Cerundolo: So that's where I got that. Yeah. I guess. Yeah. Somewhere around there. Okay. Um, yeah. When I had a five volt 10 amp supply that turned out not to be enough. Yeah. If you, if you actually turn on all the LEDs at the same time. Yeah. Uh, so when, when I say that, when I say turning all the LEDs on at the same time, uh, like you said, it is multiplexing and it turns on a one 16th of the LEDs at any given time.

Chris Gammell: Yeah.

Jason Cerundolo: Uh, but still, you know, it's cycling through that. Uh, there's enough LEDs in a row to draw a lot of power.

Chris Gammell: Yeah. Hmm. That's crazy. Okay. So yeah. But the calculation is not bad though. That's only burning four Watts then. So if it's 75 Watts out at 95%, it's only four Watts. So that's not bad.

Jason Cerundolo: Yeah. It's still pretty significant for a buck converter. Yeah. That's more along the lines of what you'd expect to dissipate with a linear regulator and most other projects.

Dave Jones: You actually need a reasonable heating on that, you know, or, and or airflow.

Jason Cerundolo: Yeah. Um, uh, I've got some heat sinks that I found that are, uh, SMT compatible and they just have, they're kind of shaped like, uh, like two T's butted together. So they, they only contact the board at two points and you solder them down and it gives

Dave Jones: me a lot of flexibility. They're very cool. They can go over the package. They can go over like, um, you know, TO220 type SMD packages and stuff.

Jason Cerundolo: Right. Yeah.

Dave Jones: Sorry. TO220 is the through hole. I'm talking about D squared packs. Right.

Jason Cerundolo: Yeah. Uh, so I think, uh, so obviously I'm using a, a buck controller. I think it's called not a regulator, uh, technically. And then, uh, Right.

Chris Gammell: Cause you mean external MOSFETs. That's what you mean.

Jason Cerundolo: Yeah. And for the two fats that you need for a buck converter, I found this part from TI that has both of them in one package. And so they're connected. The source of one's connected to the drain of the other and you just get a, the gate for the high and the low and, uh, helps with the layout and, um, also helps with the efficiency cause you, you get less losses, uh, between the two fats.

Dave Jones: Yeah. Cause they matched. So.

Jason Cerundolo: Right. Yeah.

Chris Gammell: That's cool. So you're going to start selling this thing soon though?

Jason Cerundolo: Uh, yeah. So we'll see, uh, what the testing, how the testing goes for this, uh, prototype. Uh, I got, uh, ordered two boards from macro fab and they're, uh, they just shipped the other day. So I should get them this week. And, uh, if the testing looks good, I'll order a few more. And then I have a custom design frame, uh, and the whole thing screws together pretty easily. So I should be able to get a bunch of them up on Tindy and get to do some documentation and that sort of thing as well. Uh, but yeah, next couple of weeks or so it's, uh, is the plan. Cool, man.

Chris Gammell: Wait, that's great. Sounds like you, sounds like Cast Air must be easy, huh?

Jason Cerundolo: Well, I've been working on this for, uh, what? Probably over, over a year. Yeah. Definitely. Okay. Yeah. Nights and weekends when I can. Um.

Chris Gammell: Nerd. Yeah. Best kinds. So what else you got? Uh, you have another one, you have a, the, the thermocouple thing. You sent me, or you showed me one of these or you sent one with your, your friend.

Jason Cerundolo: Yeah. Uh, yeah. So this is, uh, kind of a, a weekend project that I did for a friend of mine. Uh, he's, uh, finishing up his PhD in material science in Chicago.

Chris Gammell: Woo, Chicago.

Jason Cerundolo: Yeah. And, uh, he's, uh, was telling me about some of the work that he's doing in the lab. And, uh, when I visited with him, he was showing me around and, uh, we were both talking about how manual this process is. And it's, you know, for the most part, the material science research hasn't really advanced much beyond the 1890s. Uh, and, uh, so they, one of the things that they do is they have these furnaces and they will anneal samples for a while, uh, like three months at a time. Yep. And they need to maintain a constant temperature for that entire time. Uh, the problem is with the, uh, lab infrastructure that they have, they lose power on average about once a month. Uh, and even if it's only for a couple of minutes, it's enough to reset the furnaces. And so somebody has got to go in there and like push a button, turn the furnace back on. Right. Right. The thing is so massive, you know, you don't lose that much heat.

Dave Jones: Yeah. The thermal mass of it.

Jason Cerundolo: Yeah. So if you get it quickly, it's fine. So anytime they get an email that powers down in that region of campus, they have to go down and physically check on it, uh, and look at the digital readout on the furnace. And so with this device, they can just hook up a thermal couple and it might not be good enough to collect scientific data, at least not yet, but, uh, they can at least get an idea of if the furnace is still on and if it isn't, they, they don't need to go in. Um, like on Thanksgiving where my buddy had to walk 30 minutes in the cold into the lab just to find out the furnaces were fine and then turn right around and go home.

Chris Gammell: So, uh, sounds like, uh, we, we found it, Dave. I think we found it.

Dave Jones: Sounds like a job for the internet of things.

Chris Gammell: I was going to say this is the first, the first practical version. I think we may have said that in the past, but this is the somewhat practical thing.

Jason Cerundolo: It seems like. Yeah. Um, I also call it the internet of thermal couples. Okay. Uh, but yeah, this was a project, you know, I thought about it and I was like, you know, this is like a weekend project, you know, basically one day to, design the board and then spent about another day, uh, writing up the firmware libraries and it's all open source. And, uh, just got it, the product approved on Tindy as well. So I made a couple extra and they're available for sale. If, uh, if you happen to need one.

Chris Gammell: Cool. We'll put a Tindy link in there, of course. Yeah. Great.

Dave Jones: Yeah. Speaking of stuff getting hot, why did you make a USB C easy bake oven? Because you can.

Jason Cerundolo: Yeah. More or less. Uh, I was, uh, thinking about what you could do with a hundred watts of power, which is the maximum power that you can transmit over USB C. Yep. And so I was kind of thinking, you know, what else is around a hundred watts and. Light bulb. Yeah. You know, a light bulb.

Chris Gammell: Tiny will check. You can't even get those anymore, right?

Dave Jones: Well, I would not have guessed an easy bake oven. I would have said that's a thousand watts. Like aren't ovens like thousands, like order of magnitude more than a hundred watts? I mean.

Jason Cerundolo: No, Dave.

Dave Jones: I don't know. I don't, I don't use easy bake ovens, I guess.

Jason Cerundolo: Yeah. Well, the, the original easy bake oven just had a hundred watt incandescent bulb. Oh, did it? Okay. The heat source. Yeah.

Dave Jones: I didn't know.

Chris Gammell: That's why it was so cheap and it was a toy. I mean, like, yeah.

Dave Jones: Ah, right.

Jason Cerundolo: Yeah. Yeah. And it was considered safe for kids, even though it got up to 350 degrees Fahrenheit. This is an American thing. I don't know. Yeah. Yeah. I mean, it can, it can bake cookies. I mean, just, it's a really small chamber. So that's how it gets the power down.

Dave Jones: Got it. Yep.

Jason Cerundolo: Yeah. When I did the final test on my modified oven, I measured it with a thermocouple and I was reading somewhere around 300 Fahrenheit, which is a little low for an oven, but it was hot enough. You know, I leave the cookies in there a little bit longer than usual.

Chris Gammell: Great for a sous vide, right? I mean, you could do a sous vide in there. No problem.

Jason Cerundolo: Yeah. I could actually probably do a little computer control. Yeah. And target a specific temperature. Make some delicious steak.

Dave Jones: You get into PID control.

Jason Cerundolo: Yeah.

Dave Jones: Oh, fun, fun, fun.

Jason Cerundolo: Yeah. I already did that for a smoker that I built. Nice.

Dave Jones: Right. So has it, I have not worked on USB-C. I don't have anything that uses USB-C. It's pretty new, right? I mean... Yeah, it's pretty new. Yeah. Yeah. I'm just working on all of my computer shit's like five years old. You know? Yeah. Um, is there an issue with, because there has been with like USB-C and stuff like that, was with that, you know, like if you buy the cheap shit USB cables, they're just rubbish. You know, they've got no copper in them. I can't imagine, you know, like the USB-C cables that are supposed to deliver 100 watts and then you buy the cheap ones for two bucks on eBay and they're going to deliver. Have you had any issues with dodgy USB-C cables yet?

Jason Cerundolo: Well, like you said, the standard is pretty new. So the market hasn't really yet been flooded with all the low-cost and not-cost. I'm sure that's coming. Yeah. One thing that's in the spec actually for this high power is if you're going to draw over three amps, the cable itself has to be what's called electronically marked, which means there's a microcontroller in the plug of the cable that responds to a certain packet.

Dave Jones: Oh, but they'll fake that too. They'll fake that as well. And then to get the cost down, they'll put, you know, one-tenth the amount of copper in the cable and they'll fake the ID. You know, it's going to happen. Yeah.

Jason Cerundolo: I mean, yeah. Yeah. Obviously it'll happen. And hopefully, you know, there's, I forget his name, but that Google engineer on Amazon that goes around reviewing all the USB-C cables. Oh, yeah. So hopefully the online reviews and so forth.

Dave Jones: I haven't heard about this. Does he, like, is he so, you know, passionate about USB cables that he goes around and tests all the cheapies?

Jason Cerundolo: Well, he's been working on USB-C at Google, I'm pretty sure. Right. And so just kind of as a hobby, he buys one of every cable on Amazon and plugs it into his little USB-C debugger and finds out if they're connected. Right. There's certain pull-ups and so forth that can get connected. Like if you pull it up to VBUS, which you assume is five volts all the time, when you set up VBUS to be 20 volts, you can start frying inputs, input pins on your microcontrollers and that sort of thing.

Chris Gammell: Oh, goodness. So, Jason, you wrote a series of posts. We've been posting your posts to the subreddit and mentioning them here once in a while, but you wrote a series of posts. Can you kind of just give us a rundown of how you got started in USB-C, but also kind of what it involves and the challenges you've seen with it?

Jason Cerundolo: Yeah. I got started with USB-C, actually, with CastAR. This was pre-Series A days. We were trying to figure out how we're going to design a product that can carry the video and so forth with a removable cable so you can unplug the two devices. Until then, we had just sort of hardwired them and connected them internally with a connector that's good for, say, 10 or 30 cycles.

Dave Jones: Right.

Jason Cerundolo: And we wanted something that the user could unplug and plug in every day.

Dave Jones: Kids. Oh, my God. Slam those cables together. Wham! Yeah.

Jason Cerundolo: So, we were looking around some off-the-shelf connectors like HDMI and DisplayPort, and it was interesting to figure out that a lot of them are only rated for about 300 cycles.

Dave Jones: Yeah, which is not many if the consumer's whacking them together.

Jason Cerundolo: Right. So, you know, once a day, 300 cycles is less than a year, and we definitely want our product to last more than that.

Chris Gammell: Is that just because of the gold contacts, like the gold plating and stuff, or is it something else? Yeah, I'm not exactly sure what wears out first.

Dave Jones: There's a whole bunch of mechanical inter... Like, you know, the wear. If you insert them at a not optimal angle and stuff like that, it can wear out part of the connector, then it just gets worse and worse. And there's a whole bunch of standards for testing connectors in this sort of way.

Jason Cerundolo: Yeah. Yeah. Yeah. And especially when you want to run DisplayPort over the connector like we have, you can wind up with the signal integrity degrading over time. Oh, okay. Yeah. So, one of the connectors I was looking at that had a really high cycle life was the Micro B USB connector. Mm-hmm. And that's rated for something like 10,000 cycles. Yeah, it's nuts. And so...

Dave Jones: A lot of people claim that's bullshit, and it doesn't take into account angled connections when you plug it in at an angle, slight angle constantly, and all that sort of stuff.

Jason Cerundolo: Yeah, but... Anyway. Even if it actually was 1,000, that's still... Yeah, it'd still be pretty good. About three years. That's probably good enough. And then I came across USB-C, which was really new at the time. This was like two years ago that I was looking at this. And it seemed to have everything that we wanted. You could run video over the super speed pins by going into an alternate mode. We could provide enough power because your power rating, even staying at 5 volt, you can get up to 15 watts.

Dave Jones: Wow.

Jason Cerundolo: And yeah, and on top of that, we could also do the USB 2 high-speed communication. We can do all these things over one cable, one connector, and the cycle ratings, at least on the data sheets, were 10,000 cycles, like Micro B.

Chris Gammell: Yep. And it's reversible. One of my favorite things about it.

Jason Cerundolo: Yeah, that's been a little bit of a pain to deal with. Yeah, right. Why is that? Yeah.

Dave Jones: Well, because your power connections are backward, Chris.

Jason Cerundolo: Actually, it's not the power. No, it's not. So the power pins, there's just four power pins and four ground pins. So they're radially symmetric. It doesn't matter which way you plug it in. Okay. It's the super speed pairs. So the USB 2 is slow enough. You can just short the two options together. So either way, you wind up with the positive and negative connected. But with the super speed pairs, if you're at gigabits per second, you can't just short them together because you'll have a huge stub. Yeah. Yeah. That'll ruin your signal integrity.

Chris Gammell: And what is the speed of that? I mean, is it like five gig or something?

Jason Cerundolo: The connector, I believe, is rated to 10 gigabit per second, depending on which specific model connector you're getting. But the maximum in the spec is 10 gigabit per second per lane. And there's four lanes.

Jason Cerundolo: Oh, wow. That's great.

Jason Cerundolo: Yeah. Yeah. So you can run, theoretically, 40 gigabits per second of data and 100 watts all at the same time.

Chris Gammell: That's just a lot of stuff going on, you know?

Jason Cerundolo: Yeah. So unlike traditional USB, which is four, maybe five pins, this one has 24 pins. Yeah. And some of those are redundant, like I said, for the power. And then there's actually a dedicated configuration channel that's used to determine the orientation of the plug and a connector power source. And yeah, there's a lot going on.

Chris Gammell: Right. Right. So it enables flipping and all this stuff at the expense of like brokering who's the master, who's the slave type stuff? Or how does it actually do that handshaking?

Jason Cerundolo: Yeah. All that handshaking is done over the communication channel. And the very basic way to do it is just with some pull-ups and pull-downs and you just do some analog reads or set some analog voltages. And then if you want to do more complicated things like swapping power roles or swapping data roles on the fly, you need to do what's called the power delivery communication. And it's called power delivery, but it does a lot more than just power. And that's right now the only way to do it is with biphase mark coding, I think is what the acronym is for. It's a BMC communication. It's a one-wire protocol. It's all time-based. And until recently, you kind of had to do some tricks with some spy hardware to get it to work. But now there's a bunch of fives on the market that just convert that to something like I squared C that's a lot easier to plug into your Arduino.

Chris Gammell: So it just abstracts that out kind of thing? Yeah. That's good. That's great. Yeah. I was really confused. So I've been looking at those new MacBooks too, where they have like four ports on board. And I was thinking like, well, what if you plug two chargers in? Or like all of that stuff where it's like you got to kind of, like there's four USB-C ports or whatever they call it. I don't know. But it just seems like there's a lot of who's in charge here? You know, who's sending what? Who's receiving? You know, like all that stuff. It seems like there's a lot of handshaking that needs to happen in order to not blow things up.

Jason Cerundolo: Yeah. And I think that's one of the biggest risks with the USB-C because you can put everything on one connector. You could have a connector do anything. So your laptop might have only one of the four ports you can actually charge the battery with. And you better have the charger plugged into that port. Or you might have multiple ports, but only one of them can do video out. And so you need to make sure you plug your monitor into that port and not any of the other ones. And to get around this, there's this series of markings that you can put on the case, which the user will probably ignore. And then there's also this new standard that they've created called a USB billboard class. And what that's supposed to be able to do is deliver over USB a human readable string that can let the user know why it's not working. You're breaking it.

Chris Gammell: You're breaking it.

Jason Cerundolo: Yeah. So, you know, theoretically, you could plug your monitor into the ROM port and then the computer could send a message to the monitor and the monitor could display a message that says, like, hey, plug this into the ROM port. Try port number one or port number two.

Chris Gammell: Hey, dummy.

Jason Cerundolo: Yeah. Yeah. So it'll be interesting to see how that problem is resolved. Obviously, the best solution for the user would be just make every port able to do everything. But now you're talking about a lot of extra hardware internally.

Chris Gammell: Right. Because then you have to, like, broker that at some point on the circuit board and still do it relatively quickly and have all the power and signal integrity stuff pass all the way through, right?

Jason Cerundolo: Right. Yeah. So you have to add switches and MUXs internally, which you already have to do right now in certain circumstances for monitoring the cable flip. But that's kind of been the general strategy behind USB-C is make things easier for the user and push all that effort back on the engineers.

Chris Gammell: What about the... So it doesn't start delivering... What does it take to get to the 100-watt standpoint? So you said there's some kind of electronically mark in the cable. But, like, so can you walk us through the steps to kind of get to 100 watts? Sure.

Jason Cerundolo: So the first thing is the source will be powered up because it's preventing the power. So it's monitoring both of the CC lines. There's CC1 and CC2. And it'll activate pull-ups on those lines. And so when it sees the voltage drop on one of them, it knows that, number one, a plug has been inserted. And also it'll figure out which position the cable's in, which of the two, in normal or flipped orientation. And then at that point it can provide 5 volts to VBUS, which is just like you'd expect from USB. Pretty standard. And then the source, because it can do 100 watts, it has to send out these source capabilities messages that are just unsolicited. So every couple of seconds, I think it is, it'll just send out a packet and it says, hey, I'm a charger. I can do 5 volts at this current. And then it'll list... After that, it can list all the other voltages and current limits that it has. The device on the other end, so in this example, it's the Arduino that I plugged into my breakup board. So it's either waiting for those packets to come through over power delivery, or it will send out a sync... No, sorry, a source request packet. So it asks the source, hey, what are your capabilities? And then once it gets that packet, it can go through the capabilities. And the code that I have right now basically looks for the highest voltage. And then in that just grabs the highest amount of power that it can. And I just program my code as like anything up to 100 watts, go for it. Nice. So it'll see the 20 volts, and I think it's 89 watts, so it's like 3.6 amps or something like that. So it'll see that and I'll say, okay, I want that index, index 2 in this case. So it'll send back a request message with that. And then it's up to the source then to grant that request. And what can make things a little bit complicated, in this case, it's really simple with the Easy Bake Oven because I'm just using a MacBook Pro charger. So it only has the one USB-C port. But you could have a hub that could be supplying power to a whole bunch of different devices. So it has to keep track of what's its total input power, and then how much is it promised to each of the ports.

Dave Jones: You know, if it's got 10 ports, you'd need, in theory, a kilowatt if you wanted to supply full power to each one.

Jason Cerundolo: Oh, wow. Yeah. That'd be insane. It's insane.

Chris Gammell: Not to mention efficiencies of each, you know, is it just doing pass-through at that point? I mean, there's probably switching on board then too, right? So you have efficiency losses and whatever.

Jason Cerundolo: Yeah, I haven't really seen a teardown for the new MacBook Pro chargers, but yeah, I assume there's some sort of a buck or similar switching converter in there. So anyway, the source gets the request, and then it will do all these calculations and figure out if it can actually fulfill the request. And in this case, I know that it can because I'm the only device plugged into it. So it'll send back an accept packet. And it basically says like, yep, that's good. Let's do it. And then I'm still doing a little bit more research into exactly how it works with the electronically marked cable. But it's up to the source basically to interrogate the cable as well using a different header. And it asks the cable like, hey, what are your capabilities? And before it actually provides the power, it needs to verify that the cable has the capability to transmit up to five amps or whatever it is.

Dave Jones: Have you had to like experiment to find out this stuff? Or is it all documented? Or is it like a closed standard that you have to pay to, you know, get the standards document to figure all this out or what?

Jason Cerundolo: So all the basic USB stuff is available on the USB website. You just go to the developers page and you download the zip file.

Dave Jones: Got it.

Jason Cerundolo: Yeah. Some of the more specific stuff, so the alternate modes, how to use those, that can be proprietary.

Dave Jones: Oh, okay. Right.

Jason Cerundolo: So for example, getting into an alternate mode, that's in the USB spec. But the alternate mode for DisplayPort, for example, that's run by VESA. I think that's how it's pronounced.

Dave Jones: VESA is how I pronounce it.

Jason Cerundolo: VESA, VESA. Yeah. V-E-S-A.

Dave Jones: Yeah.

Jason Cerundolo: And they publish a spec. You know, they publish the DisplayPort spec, but they also publish a DisplayPort over USB-C spec. And that you have to be a member of their club to get access to the dissertation.

Chris Gammell: They're probably cheap to be a member too, right?

Jason Cerundolo: Yeah. I'm sure. I think it's like $5,000. Yeah.

Chris Gammell: That's actually cheaper than I would have guessed. Okay.

Jason Cerundolo: All right. Yeah. And, you know, sometimes you can find bootleg copies on Baidu or something like that. No. To get the real spec, you have to pay the fee and become a member. Yeah.

Chris Gammell: Well, no. If you were making tables or whatever, if you wanted to mark it as compliant too, I'm sure you'd have to pay your licensing for that, right?

Jason Cerundolo: Yeah. If you want to put the logo on your product and tell people that you're a DisplayPort device, yeah, you know, they own all the trademarks. So that's basically what you're paying for is a license to use their trademark. There's also a more of HDMI and a few other standards that are becoming pretty popular for as alternate modes for USB-C.

Chris Gammell: That's great. Yeah. I think that, I mean, I'm super, I was excited when they came out just for the power stuff, but this other stuff, I mean, like, honestly, the fact that Google and Apple kind of co-developed, they were like part of it too. And like, I don't know if people saw the iPhone 8 is rumored to be switching to USB-C and everything. Like, I just think it's, I thought it was such a great move when the European Union standardized on USB micro. And like, thinking about this kind of thing, like taking it now a step further, like it is kind of a one cable to rule them all. And obviously there's a lot of challenges like you're talking about here, but kind of the idea of like, it's definitely more of a universal cable than a lot of things that are out there. So I really like that.

Dave Jones: But micro USB is still going to survive because it's micro. It's much smaller. Like you can put it in tiny little products and stuff like that.

Jason Cerundolo: Well, actually the USB-C is not that much bigger than a micro USB connector.

Dave Jones: I've never physically seen one. It's not that much bigger?

Jason Cerundolo: Yeah. It's much less than twice the area, I would say. It's like just a little bit taller and a little bit wider. Yeah.

Chris Gammell: Right. More board space for the PHY chip and stuff like that, right? All the handling and whatever.

Jason Cerundolo: Yeah. And there's actually a lot of options on that side. So there are microcontrollers that you can buy that just have an M0 and the PHY and all the hardware that you need and probably like a two by two millimeter package. Mm-hmm. There's FPGAs I think that I've seen that have been incorporating some of this hardware. Okay. Interesting. TI I know has a lot of integrated products. So they'll integrate the BMC PHY, the MUXs, and maybe even a USB high speed to serial converter all in one chip. And so there's a lot of options out there. The stuff that I've been working with, I prefer to do more of a piecemeal solution, which is great for prototyping. So I've been using a Fairchild part.

Chris Gammell: Is that that FUSB302 that you have a board for?

Jason Cerundolo: Yeah. So another Tindy link to put in the show notes. I have a little breakout board that basically breaks the pins of this PHY out to headers. And that's what I used for the USB-C Easy Bake Oven. Just jump it over to an Arduino. And then, yeah, if you want to talk about complexity, the code to get all this running, I was porting a library that Google Chrome has for their embedded computing. And I sort of lost count of the number of lines somewhere around 4,000. Good Lord.

Chris Gammell: 4,000 on an Arduino? Is that right?

Jason Cerundolo: Yeah. Well, of course, it only compiles the stuff that it actually needs. But I think it uses like 60% or 70% of the program space right now. Wow. Okay. Just to do the USB-C. So it is possible to run it off an Arduino. Yeah.

Chris Gammell: Yeah. So in terms of complexity of the code, is it just handling lots and lots of scenarios? Is that kind of the thing or what?

Jason Cerundolo: Yeah. So basically, there's a big state machine is kind of the core of it. And it has something like 29 possible states. And the state machine itself is the majority of the code. I think that's like 1,500 to 2,000 lines of code.

Chris Gammell: Sounds like a job for CPLD.

Jason Cerundolo: Yeah, maybe. That is interesting.

Chris Gammell: Huh. So you got up to 300F, which is what? That's like 190C? Is that right? My mathelizer is broken today. So that's not bad, though. I mean, temperature-wise. I mean, kilowatt showed 80 watts of power being drawn, it says. Yeah. What were you sourcing? You were just sourcing into a heating element? Is that right? Yeah.

Jason Cerundolo: So the first thing I had to do was take apart the oven, because I knew they had moved away from the incandescent bulb. So I was kind of curious what they had replaced it with. And it turns out it's just a nichrome wire. So it's just a fancy toaster, basically.

Chris Gammell: Wait, wait. This is an actual Easy Bake Oven that you're using here? This picture that you have?

Jason Cerundolo: Yeah. Yeah. I bought one on Amazon.

Chris Gammell: I didn't know they still made them.

Jason Cerundolo: They do. They've had to, like I said, they've had to move away from the bulbs, but they have a new design now that uses a nichrome wire. And basically, when you flip the switch, it just connects the 120 volts across the nichrome wire. Safety. Sorry.

Dave Jones: But that's what a toaster is. I mean, it's mains across an element. I mean, geez.

Jason Cerundolo: Yeah. And this is a lot safer than a toaster because the heating element itself is so far buried. There's no possible way, even with small hands, to get in there and touch it. And you have to use this plastic tool to move the baking tray through the unit. So they've definitely thought about safety and reliability there. So the one thing I had to do, basically, the major modification I had to do was chopping up the heating element and dividing it by six. So it had six equal elements and then wired them all up in parallel. Yeah. So cut the resistance by a factor of 36. And that winds up being about the same amount of power. I wasn't quite perfect with how I divvied it up. So I only got like 75 watts out.

Chris Gammell: Mm-hmm. Right. Yeah. It's just a, that's going to help determine the load, right? It's only going to ask for so much. Does it, so, okay. So that's actually an interesting question. So the source and the, and the, oh, I've already forgotten the name of these things. But it says. The sink. Okay. So one of them says, I have this much power to offer them and says, I can accept this much power. Right. But in your case, you just have an open load. So, or not an open load, but just a dumb load pretty much. Right. Right. So how does it handle when it's not the maximum power that it's trying to deliver? Does it, does it, does it deal with that okay?

Jason Cerundolo: Yeah. Well, it seems to from all the experiments that have been running. But yeah, basically when you send the request message, you're sending along a couple of numbers. And it's, I think it's the number in tens of milliamps for two currents. And so one is your sort of normal operating current and how much you're normally going to use. And then the other is a, like a surge current or a peak current that indicates the maximum possible that you might need at a certain time. And so normally those will just be the same and it'll be just whatever the sort of average max power you're going to need. And so in this case, you, or my code sets the, those values to match the maximum that the source is offering. But theoretically, you know, the source could say I have 20 volts and I can do five amps and you could send back like 20 volts sounds great, but I only need one amp. And then, uh, like we said, for a hub situation or multi-port device, you know, it takes it away.

Dave Jones: It takes it away from the power budget.

Jason Cerundolo: So, yeah. And this gets into terms that USB calls like policy managers and policy engine, maybe. I don't know. There's a lot of fancy buzzwords.

Jason Cerundolo: We call it the power of bureaucracy. One, one must handshake with another. Sink must talk to source. Yeah.

Dave Jones: A lot of people got that wrong on USB, um, two. Is it they, sorry, or is it USB one or whatever? Like the original, like, you know, 500 milliamp USB limit, you know, there's a lot of people that claim that, oh, you know, you have to commute, you know, you have to communicate to enable the 500 milliamp load, but that was bullshit. You could have just ignored it. Um, all the devices were just capable of providing 500 milliamps anyway. You know, it was like, it's only if you wanted to play nice, um, that you would actually request, you know, from windows, you would request like the power budget, you know, say, oh, I only need a hundred milliamps, you know, so you're free to give 400 milliamps to somebody else, you know, or to four other devices, for example. But you could just ignore it. It wasn't a, an enforced thing.

Jason Cerundolo: Yeah. And that's what happened in practice. And I, I think some of that was just, it was not worth the complication of trying to figure out how to kick devices off. So it was easier to just wire everything to the five volt rail. Yeah.

Dave Jones: That's it. And, and, and in most ports that have a 500 milliamp fuse on there anyway. So they were like an RTD or something.

Jason Cerundolo: Yeah. Yeah.

Dave Jones: So it was all good.

Jason Cerundolo: Yeah. Yeah. I've found more so than with the USB two with USB C, uh, I've definitely gotten kicked off buses for sending packets and so forth. Like, uh, the circuits are complicated enough that it's just a matter of firmware to enforce these sorts of things. So they get enforced a lot more. And, uh, yeah, it's having a protocol analyzer or something like that is definitely helped because, uh, keep wondering like, why do I keep getting reset and keep getting a hard reset? It just turns off V bus on me every time.

Chris Gammell: Oh, really? Okay. What do you use for that kind of thing?

Jason Cerundolo: Uh, well, we have a really nice one at work from, uh, Teledyne LaCroix. It's like the Mercury T2C, which is just their T2 model with all USB C ports. Uh, and it'll, uh, record and trigger off all sorts of various things happening with power delivery and can keep track of the pull-ups and pull-downs and, uh, what got plugged in where. Uh, and then you can pay another fee and then also do high-speed USB. Of course you can.

Dave Jones: It's Teledyne LaCroix. Yeah.

Jason Cerundolo: Yeah. Um, I know there's one that, uh, Google released and open source the design of, and I think you could get it on Amazon last I checked. That one was only like $120. Okay. Um, yeah. So it's been kind of a dream project of mine to put together a, an open source USB-C protocol analyzer.

Dave Jones: Analyzer. Yep.

Jason Cerundolo: Yeah. It's, uh, a lot of work.

Dave Jones: So there's a lot of detail in that. Yeah. Yeah. It's, yeah.

Jason Cerundolo: It's not just hardware. You need firmware. You need software.

Dave Jones: Uh-huh.

Jason Cerundolo: Yeah.

Dave Jones: And, and you're selling into a niche market that, you know. Right. So don't really care about the open source thing. All they want is something that works, you know. So. Yeah.

Jason Cerundolo: Yeah. So I'm just starting with just the breakout board and starting with the Google's open source code that I've modified and also made open source. So it's a kind of a start and hopefully other people get involved and can help me with a lot of the firmware.

Chris Gammell: Yeah. Uh, you're on GitHub under what name? Uh, under Reclaimer Labs. Okay. Cool.

Jason Cerundolo: Yeah. So right now I have two libraries that I put together. One is just to talk to the FI and then the other uses that library and does all the power delivery.

Chris Gammell: Mm-hmm. Yep. That sounds good. Uh, so what, uh, what was the most surprising thing about USPC? I mean, like what, what, what should people know if they're getting started into it?

Jason Cerundolo: Uh, I'd say the, the first thing is that it might seem really complicated and it can be, but the most basic things that you'd want to do with USPC, uh, it's pretty simple to get started. Uh, so basically you need one extra resistor. If you're going to use, uh, replace a USB two port with a USB C port, uh, you can just tie the, the two positions for the USB data lines, uh, to each other and then wire that into your device, just like normal. Uh, basically just add a pull down resistor on the CC lines and, uh, you're good to go. Uh, and everything should work just like you expect.

Chris Gammell: So that's without a FI you're saying?

Jason Cerundolo: Yeah. No FI, no extra firmware. Uh, you just, just changing the connector. Yeah. Using the new thing.

Chris Gammell: So that will give you what you said that's USB 2.0, 2.1 or something. And then.

Jason Cerundolo: So that'll give you just like you expect the five volts, 500 milliamps, um, USB to high speed communication. So the 480 megabit communication. Uh, and then if you want to get a little bit more power, uh, all you have to do is monitor that CC pin. So that pull down that you added, um, you know, just monitor the voltage and depending, uh, which range it's in that can tell you whether or not you have the 500 milliamps, uh, 1.5 amps or three amps available to you. Uh, and again, this is without a FI, uh, you don't need to do any of the power delivery communication. Uh, you basically just need, uh, an ADC, which you probably already have on your microcontroller.

Dave Jones: Yep.

Jason Cerundolo: Uh, so that'd be a simple way to get 15 Watts, uh, and USB 2.0 communication. And then if you want to go up from there, if you want to do different voltages, uh, connect to battery straight across on V bus or, uh, use any of the alternate modes, then you need to start getting into the 4,000 lines of code and all that.

Dave Jones: But, but to have 15 Watts capability, that's a lot.

Jason Cerundolo: Yeah, that's nice.

Dave Jones: So, you know, that's, that's plenty for most, um, projects, I think.

Jason Cerundolo: Yeah. And, um, I think the seven and a half watt version, the 1.5 amp seems to be kind of the new standard, like 1.5 amp is the new 500 milliamp. Yeah, right. Cool. And, uh, yeah, the, the MacBook that I have that has a USB-C port, uh, when I plugged the Arduino into it and read out the capabilities of the port, it was advertising that it could do the one and a half amps, uh, at five volts. So that's seven and a half Watts from your laptop. So it's cool. Yeah. That's pretty handy. More than enough to run, run some pretty interesting widgets.

Chris Gammell: Definitely. Yeah. Yeah. I, I, I had thought in the past that this would be interesting, like, especially at the higher, the higher voltage stuff. Like the 20, the a hundred watt thing, but like doing some kind of like power supply type thing would be interesting at some point. I talked to some people about that. Um, obviously you need to clean up the power, but, uh, you know, like that's kind of, that's kind of a, that's, that is a serious amount of power for, especially for mobile devices and just for most things that you need these days. You know, it's, it seems like it's in the realm of possibility just to use that.

Dave Jones: I might use it on my new, uh, power supply. Okay.

Chris Gammell: No, there you go.

Dave Jones: Hmm.

Jason Cerundolo: Yeah. And, uh, if you're plugging this device, your power supply into a laptop, you have USB communication. So you could just open up a USB serial, uh, adapter and, you know, be running a Python script or something on the, on the laptop and changing the voltage and current limits, measuring, uh, things, getting the readings back.

Chris Gammell: Oh, cool.

Jason Cerundolo: Um, so there's a lot you could do with, uh, automation with even just the seven and a half or ideally the 15 Watts and a USB connection. Very cool.

Chris Gammell: Well, Jason, where can people find out more about you and what you're working on and everything?

Jason Cerundolo: Yeah. Uh, well, you can find me on Twitter. My handle's, uh, at Ascended Daniel. Um, and my website for all this hobby project, uh, stuff is reclaimer labs.com. Um, and, uh, assuming we'll have links in the show notes. Oh yeah. Yep.

Dave Jones: And, uh, you're hiring, aren't you?

Jason Cerundolo: We are hiring. So, uh, I know we're, uh, in particular, we're looking for a firmware engineer, uh, electrical engineer, and a whole bunch of other positions. Uh, so if you check out, uh, cast AR.com, uh, there'll be a link to the jobs or careers.

Dave Jones: No, no, come on. We want direct email to bypass, you know, HR. Come on.

Jason Cerundolo: Uh, yeah, well, uh, my email's, uh, no. Put it on the spot, huh, Dave? Yeah, we'll, we'll put a link in the, in the show notes.

Chris Gammell: He blinked. He blinked. That's okay. We don't blame you. We don't blame you.

Dave Jones: We, we have a very good track record here on the Amp Hour of finding people for companies. So, yep. Mm-hmm. We should be charging a fee. Shouldn't.

Chris Gammell: Maybe. Yeah.

Dave Jones: Mm-hmm.

Chris Gammell: Well, Jason, thanks for telling us about all this stuff. I'm really excited. I'm, this USB-C stuff. I think it's really great that you're helping open that stuff up. It's, it was definitely daunting when I first looked at it. Yeah, yeah. It looks complex. Yeah. And the fact, yep. The fact that you dealt with those 4,000 lines of code before I ever had to think about it, I do appreciate that. Yeah, well, you're welcome.

Dave Jones: All right. Thanks for joining us, mate. It's been good.

Jason Cerundolo: Yeah. Thanks for having me.

Dave Jones: All right. We'll talk soon. Catch you next time. Bye. Bye.

Dave Jones: Bye.

Speaker ?: Bye. Bye. Bye. Bye.