Jason Cerundolo of Reclaimer Labs

Hello and welcome to the macro fed engineering podcast. I'm your guest Jason sermon below.

And we are your host,

Parker, Dolman. And Steven Greg.

This is episode 158. So a quick announcement before we jump into the podcast macragge will be at South by Southwest in Austin, Texas. We are teaming up with particle.io to put together a hardware happy hour. It will take place on Friday, March 8, from 4pm until 8pm. At the super cool jester King brewery, check the show notes for full details and RSVP. Join us for light bites and refreshments and network with fellow hardware nerds and kick off your South by Southwest weekend.

So Jason Cirilo is an engineer experience in mechanical, electrical and firmware engineering. Jason has over nine years of experience as a hardware engineer, and working on electronic designs in Silicon Valley. In his free time, he works on open source projects, such as USB C tools.

So yeah, today Jason is here to demystify the Universal Serial Bus. So Jason, what about your day job makes you qualified to talk about USB?

Well, I've been working with USB and USB C now for a couple of years, started really in earnest when I was at CAST AR. So it's a startup doing augmented reality. And I was focusing mostly on the electronics for the headset, and we were working to select a connector that would connect the content hub, which had most of the processing up with the headset, which had all the projectors and the microphone and speakers and everything. And we're looking for a connector that could carry some extra power, carry video signals, and then also have a line open for more housekeeping stuff and audio. And so right around that time, USBC was just starting to go mainstream, and seemed like a good fit. It had something like 10,000, mate D might cycles and could do everything in this small connector. So I had to go and get really familiar with it. So we could implement it in our design.

Yeah, that's actually kind of convenient. Cuz USB type C does all that.

It does. Yeah, so a lot of people don't realize you can run upwards of 100 watts of power. You can also run DisplayPort, or four channels of Thunderbolt, two channels, a PCIe, HDMI, not all at the same time, of course, but they're each available as alternate modes. And then on top of all that you still have a regular USB high speed connection. So that's good enough for doing audio sending data back and forth. control signals. So it's really a capable connector.

I don't know exactly how it closed down. But that's not what you're doing anymore. Right.

Right. Yeah. So yeah, the short version is the cassiar ran out of funding and had to shut down. So almost everybody was laid off on this one fateful Monday.

Oh, man, that's unfortunate. There. Was there any, like precursors for workers or anything? Or just

know, I mean, it was pretty sudden, what had happened. I mean, obviously, a group of us there were, we knew things weren't going well. But you know, we didn't know exactly when

the snack machine runs out of snacks. Yeah.

But yeah, so after that, took a little bit of time. And it was actually over at zipline for about a month, month and a half, helping them with some of their firmware. And their new drones flying around Rwanda, delivering blood. So that was a lot of fun. And that was always interested in space and aerospace in general, and found a space startup in the Bay Area, Capella space. And so then, I was there for about a year, building software defined radios. So the radio that they use is one radio that does radar, and high data rate communications. And so I was responsible for all the electronics pretty much aside from the power amplifier.

And so like what frequencies that run on,

it's an X band radio, with about somewhere upwards of 500 megahertz bandwidth, with some options to expand it a little bit higher. A lot of that is really dependent on the FCC licenses that you can get. Okay,

so it can do anything pretty much but it depends on how much money you I guess you get the FCC.

Yeah, it's not all about money. I mean, there's there's a band set aside for doing radar at X band from space. And so a lot of it is just making sure you file it Applications far enough in advance that you get the approvals when you need them. And

yeah, not like right before lunch. Yeah.

Yeah, so few companies still struggling with

that. You were in charge of all the electronics for that. You said mine is the power amplifier.

Yeah, so I was working on. We call it the baseband board, which had the SOC from Xilinx, the ADC and DAC all the clock chips that you need to run everything, all the power supplies, and then all of the connections from that out to the main spacecraft bus. So we were not the flight computer were shut off most of the time. And then the flight computer was responsible for turning on the radio at the correct time and giving it the instructions on what to do. So you know, that way, took a little bit of the pressure off of the radio design, it didn't need to be 100% working all the time, if you know for power and thermal is a lot easier. Yeah, so then after that, recently, I'm moved over to lift bikes and scooters team. So Lyft has the little rideable scooters, the stand up ones, and a few markets. And then all of the bikes around San Francisco here where I work, New York, like the city go bikes. Those were all operated, built and operated by motivate. And then it was recently announced that Lyft acquired motivate. And so all those bikes are now under lift technically, and they're going to be rebranding them at some point. And that will probably be coinciding with the release of the bikes that I'm working on now. And so that's, you know, a bit of a change, but going from building one satellite that absolutely has to work to building 1000s of bikes.

Yeah, mass mass production product, pretty much.

So when you say they're rebranding, or they're potentially rebranding it, do, they have to go to the field and retrieve all the bikes.

I mean, the bikes don't last forever, obviously. So as they come in for maintenance, or get swapped out, the new ones will almost certainly have to lift branding. I'm not too privy on all the details of you know what the ID is going to look like. But I do know that that has been announced.

So even someone does have to go around and collect them, like every day and recharge them usually. So I guess you can just slap a new sticker on it, then.

Yeah, that might that might be it. And I know that there's some funky rules here in Denver with them. Such that if they're left cache, if they're left in a street than the company who manufactured the bike, it's a $200. Fine. So cops can actually find them directly for that, which is kind of crazy. I guess it's whoever owns it. Yeah, yeah, exactly.

That's kind of a weird rule. Because from a if your left or I guess the other ones, bird or all these other ones? Like, do you just like tell? I guess you just have to tell your customers. Hey, don't leave them in the street.

Yeah, basically. Yeah. I think they're trying to they, they've seen them as a nuisance here. But I don't know, it's it. I think they're gonna kind of neat.

I like using the one I got to use in Austin. So

yeah, they're very convenient. The current bikes are a docked model so that you only go between two docking stations. And then the scooters are the dockless con. So there's no like set station that you need to drop them off to just so you're supposed to put them off to the side. So they're not blocking the sidewalks. But I know around here, even in Oakland, so we don't have to lift. We don't have to lift scooters here in Oakland. But the other brands, you know, inevitably you find them in the street or they fell over there blocking the path and just big

pile of them in the corner. Yeah.

Yeah. You know, we're gonna have to figure out I guess, as a society,

but they all talk to the mothership, right?

Yeah, so at least for lift and I don't know about the other ones like, Yeah, they're pretty in pretty much constant communication. So that, you know, they have GPS and cellular connections. And so they're, you know, in pretty constant communication. And, you know, there's ops teams that go around to collect them and I see them sometimes outside, you know, from lime or bird or whatever, going around and picking them up recharging them.

So can you tell us anything that you're doing over there?

Yeah, rather not get into a lot of the details of what I'm working on. Yeah, even things like you know, timelines and quantities and so forth are pretty sensitive topics. So yeah. Not not not at liberty to talk too much about what I'm working on. And obviously, you know, here with you, I'm just representing myself. So I'm not, you know, here officially as a member of Lyft or anything like that.

Sure. Sure. Sure. Sure. Yeah. Well, let's go ahead and kind of reel it back to talking about USB type C, again. Yeah. So, so you have will actually, let's go ahead and talk about Reclaimer Labs, which is sort of your personal projects, right.

Yeah, that's the name that I use for all my hobby projects. Some open source stuff I'm working on, sell a few of the boards on tindy and working on some other projects that I haven't made as public yet. And we can definitely talk about those a little bit later, for sure. Yeah, so Reclaimer Labs is just the it's actually an LLC at this point. Just a umbrella for all of that.

And I'm so I'm on I'm on reclaiming labs right now. Just your front page. And it looks like you have a pretty extensive blog, it seems like you write quite a bit and go into a good bit of detail, especially on USB type C work.

Yeah. So I realized that as one of the engineers that was, you know, really using USBC, and was sort of on the front edge of the adoption of USBC that I was answering lots of questions from other engineers, whether it was at Cafe, our go to meetups, and inevitably USPC would come up and people had asked me, How does it work? What's going on how to how to get started. And so I took a lot of those questions that I kept answering over and over again, and just turn them into blog posts. So mostly, so don't have to keep answering.

If someone asked you that question, you hand them a QR code now.

Yeah, I wish. But yeah, I can, you know, give them a brief overview, and then tell them you know, hey, wrote this blog post that kind of explains the basics. And so I felt like there was a need for more of a, like a narrative description of what's going on with USBC. Because obviously, all the specs for USB and USBC and power delivery, that's all publicly available, it's freely available from USB if but when you download that you get, you know, several 100 megabytes with 1000s of pages of detailed specifications. And it's not exactly clear where you would start or what's going on, because it just as listing voltages and currents and timing diagrams, and so forth. All the fun stuff. Yeah. Yeah. I mean, there's a lot of detail, and it's everything you need to start it. But the USBC like the type C document itself is like 600 pages or something. It might be more now because I keep adding to it.

Oh, brutal.

Yeah, they had like 3.1 came out like last year, and that, you know, added another couple 100 pages onto it.

So yeah, I think they're 3.2. Now actually,

honestly, now I'm behind.

Yeah, I think 3.1 was 2014, or I don't or 16, something like that.

Yeah, yeah, that sounds about right.

So so you've created the USB C explore board, which is kind of shown in a large image on the front of your webpage. You want to go ahead and talk about that and kind of let our listeners know what it is.

Yeah, so it's an evolution of the breakout board that I designed. So the breakout board was just a USB power delivery fi. And you need to find generally, because the power delivery communication happens over BMC at this sort of weird voltage level that you can't just direct drive off of a normal microcontroller. So it's somewhere around like point six volts, and it's all very timing dependent.

Is that a differential signal? No, it's

single ended. Okay, so the BMC is by phase Mark coding. So just, it's pretty much related to the phase of the transition. So if you have like a long pulse, and then a short pulse, that's either a one or zero forget which right now, but then if it's the other way around, then it's the the other bit. And so having a fire to like handle the timing for you on that handle the voltage levels, there's all sorts of pull ups and pull downs that are used for Type C to indicate different current capabilities without doing a communication protocol. But that means you need like all these different resistors that you can enable or disable. And so this one little chip, it's at the smallest versions like a millimeter square. That can just do all of this interface between I squared C, which is easier to talk to, and then the type C port.

Gotcha. And that's the on semi F USB 302, I think.

Yep, I think it's a 302 b right now is like the latest version. And then there's like a couple of different part numbers that give you a couple options for I squared C addresses. Because there's no like address pin on it. The the BGA version only has nine pins, so they gotta be economical.

Gotcha, gotcha. Yeah. Oh, this is used to be a fair child part.

Yeah, I remember dealing with Fairchild on this. And then shortly after I left cast AR maybe while I was there, they were transitioning over to being on semi, the

phone number changed, right?

Yeah, the best description I've heard it's, it's reverse Moore's law. It's the number of semiconductor companies has every 18 months.

That's not far from the truth.

Yeah. The worst one lately is linear and analog. But yeah, so I had this breakout board was useful, even around cast iron, because we could develop firmware on like, Discovery Board from S T, or something like that. Just jump around I squared C and have everything we needed. And then got, basically I got tired of having a jumper over to an Arduino or some other dev board. So I figured I'd make another board that just put the microcontroller on board. And then added, obviously a debug interface to program at UART. And I was told about these really small OLED screens. And so this one's about one inch on the diagonal. And gives you I think it's 128 by 64 pixels. It's just black and white. But what it lets you do is just have a really nice user experience. So right now the default firmware will interrogate the other side of the cable, so

interrogate give me that power, or else.

Yeah, in a very non threatening way, as you know, what, what are the capabilities, so something like the 87 Watt, charger from Apple might have three or four different voltages and current pairs of what it can do. So almost always five volts is one option, and some of them go up almost to 20 volts, or, yeah, the 87 Watt one is, I think it's like 20 volts and 4.3 amps or something like that. And then it can just display those numbers right on the screen, so you don't need another computer and you don't need any scripts or software on a computer, you just plug this thing into your laptop charger, and I'll just list out what capabilities you have. And that's really the beginning of what you can do with this. So some other things I've done with it, I jumper, jumper it over the debug port to a breadboard with some load switches, which you need if you're going to do be a USB power source, and so reconfigured the software to be a source and then I actually charge my MacBook through my USB C explorer board from power supply on the bench. So that was a lot of fun. And a little bit scary.

How much how much juice was the laptop bowling

I have one of the really lightweight ones. So it's I think 15 volts at better like up to 1.8 amps or something like that. And it really depends on the state of charge of the battery. So I remember that the first time I went to test it was like really confused why it wasn't pulling any current and then I realized I just had the laptop plugged in all day and the battery was fully charged and it basically didn't need any current

I've had that same experience doing USB micro back in the day and you know plugging in your you know, smartphone to it and like I hope this this charger works that built

Yeah, it's a very expensive way to test

Yeah, that's actually what kind of want to get to is USB testing evaluation as well. Because that's actually my biggest problem that I have with with these kinds of USB designs especially going to power delivery with with Type C is like the am I doing this right question? Because when you start looking at like testers and verification stuff, like sure you can read a blog posts that says put a 5.1k Pull down on the CC pins and it will work right but you build it and you're like, okay, it works in this one case that have on my bench? How do you make sure it's going to work to the standard?

Yeah. So that's obviously a big question. And a lot of the, a lot of the spec talks about, maybe not exactly how to test but what all the conditions are. So there'll be tolerances on that resistor, for example. There'll be voltage ranges. And, you know, the, the easy way to test it is, you know, get one of the compliance testing machines, that's probably like hundreds of 1000s of dollars and just plug it in. But obviously, it's not a very accessible for hobbyists working at home, or even like a small company, that significant amount of money. So fortunately, there's a bunch of other options that are on the market right now. The one that I've been using a lot lately is this debugger from Cypress. So it's essentially a pass through for USB C, and it just taps the CC lines, which are the configuration channels. And that's what's used to coordinate the power and alternate modes between the two sides. And then it just has some free software you run on your computer, and I'll print out a list of every message that goes over the bus. And that thing, I think, is about $200. Which, you know, in the grand scheme of things, not too too much money. The next level up the one that we had at CAST AR was one from Teledyne Lacroix. And I was I think, closer like $3,000. But it had two different ports, which is great, because you could tell exactly which side was sending the messages. So even if once I was sending messages that should be coming from the other side, maybe your firmware backwards or something like that, you could tell a lot more about exactly what was happening. And it would track things like the voltage change on the CC line to let you know something was inserted. It could check the all your pull ups and pull downs were working. So yeah, that's probably more of an option for like a small company. But it was very valuable in the early days getting things working. Because it's tough. When you have a new hardware design, maybe that you made you have brand new firmware you're trying to integrate. And, you know, something goes wrong, or like it doesn't all work right off the bat. Right. So how do you figure out what's working and what's not?

Exactly? Yeah. And that I did a little googling it's the Cy for 500. Easy PD. Yeah. Yeah, I'm gonna put that in my shopping cart.

Yeah, cuz it's, I guess it's just not as simple as just throwing your oscilloscope on it and just watching stuff, right? Like, that just doesn't, it's not gonna cut it.

Yeah, so the, the, well, if you have a saline A, which a lot of engineers around here do, it can definitely record the raw bitstream going through. And then I think, last I checked, it wasn't like quite ready for primetime. But there's a group out there that's are someone on their own is working on a decoder option for say, Lea. So you would just clip your input onto one of the CC lines. And then it would do the IC and right now it can do the BMC decoding. So it can get you ones and zeros based on the transitions. But then there's a five B for B overhead, so you need to convert it for from that using a table that's in the USB spec. And then from there, you need to know, you know, okay, so we have the ones and zeros that it's trying to send. But, you know, what does that mean? And how how do you decode like a header packet, for example, to get the length of the packet and, you know, so one way you can do that is just go through bit by bit and, you know, start up at zero and work your way up from the spec. And I've definitely done that, from time to time when I'm like really trying to figure out what's going on. Or maybe there's something that actually there was one case where the protocol analyzer didn't know what a certain bit position did. And it was like something new that was added to the spec that my hardware was using to like look up, okay, what does bit eight mean? Because that should be a zero, but according to this protocol analyzer, but now it's a one for some reason. And yeah, so just being able to get access to the raw, the raw packets being sent back and forth, can help quite a bit. And then if the protocol analyzer did a lot of the work for you, so either get some software in Sapa or the tuner dollar Cypress one I'll save you a lot of time.

I had to look for that that software package then.

Yeah, I don't know if it's like actually released yet but definitely saw people chatting about making one. So yeah, unfortunately, I don't have enough time to really go in and help them. So hoping some industrious firmware and software engineers can go and get that working.

Sounds like a good option is the Cypress board and your explorer board. So you kind of get the whole picture.

Yeah, and, you know, just like with any sort of debugging, you know, if you can start from known good hardware, software, electrical, and then just change one thing at a time, that can be really helpful, too. So, you know, one option would be, just get the USB C explore board, I test them all before they go out. And you can test it yourself when you get it and then put your new firmware on it. And then at least it's just the firmware is new. And if something goes wrong, you can just revert back. And maybe even, you know, going line by line if you need to, but you know, start start with something that's good. And then work your way to where you want to be.

Yeah, I actually have one of those breakout boards right over there. So I've been messing with it. Oh, great. Yeah. The, my favorite thing is actually plugging in that explorer board into power supplies and seeing what they actually say they will do versus what they're advertised. Because they're hardly ever Correct.

Yeah, there's a good tendency to round up. So it might be, you know, 2.4 amps, but they'll just say, two and a half. Yeah, yeah.

I'm actually actually on power supplies for Type C power delivery is there one that you would recommend for testing that is like, it adheres to the spec. And if you mess up, or you decide or you try to pull more than you requested, it will shut you down.

I haven't quite gotten that sophisticated with testing. One of the ideas for board and that I've had as a sort of an add on board, that would be programmable load for the USBC. Explorer. So it could just in addition to telling you what the power supply says it can do, it could just run the load and double check that it can actually pull that amount of current. Yeah, the best strategy that I've found is just having a variety of USBC sources around. So I use that for kind of, at this point, now it's just regression testing the firmware. But in the beginning, certain power supplies were tolerant of like different violations of the spec. So if you took a little bit too long to respond to a message, some of the power supplies would allow that and just keep functioning and other ones would, as soon as you were a millisecond late, they would issue a hard reset and kick you off the bus and then your device would just be in a boot loop. And it'd be kind of hard to figure out what was going on. Gotcha.

I guess you could use the Explorer to build like it. You can build a source with that correct? Yep, yeah, I've done that. And then just monitor the power out and say, hey, if that device, if your device requested an amp, let's just use that for an example. And it went over that? You can you can have a, it can notify you that you know, something's up.

Yeah, yeah. And there is the debug port on the side there. So you can do a eyes could see your UART out and have it send a notification or just print something on onto the screen. So there's a lot of options with that board. And

so, one of the things that I'm actually a bit interested in myself is some hardware layout tips. Do you happen to have any suggestions or things that you have found that work better or don't?

Yeah, so for USB until you get into high speed, and for sure, once you get to super speed, you need to lay out your traces appropriately. So that's all the classic high speed differential pair layout impedance control. Yeah, so yeah, you need impedance control, you need to make sure, try not to have right angles, keep everything length matched. Super speeds, pretty forgiving. I think it has an embedded clock. So that makes a thing routing a lot easier. You don't have to like match between lanes, but the intro pair spacing, so that's between the P and the N. still matters quite a bit for high speed have never had a design where, like I've been so far out on the spacing that I've had to like add bumps or anything because that only runs up for 80 megabits per second. Yeah, that's

USB 2.0. Yeah, yep. Yeah, it's

high speed. Because USB 2.0 So covers full speed. That's a lot more forgiving. Basically. Just be like slightly careful, you're not routing it directly underneath a switching power supply, and you're probably going to be okay. Yeah, another trick is for high speed USB, or really any high speed buses. If you can keep that connection short enough, you can, you're all in near field effects. So you don't actually need to worry about anything like impedance control.

What's short in

short, is generally, I think it's like 1/10 of a wavelength. And that you have to account for the fact too, that it's the rising edge that's really defines kind of your fundamental frequency. And so, yeah, I have to look up the numbers exactly. But generally, if you can put two chips next to each other, and there's only like a 20 mil trace, really, unless you're doing like 100 gigahertz stuff, that's going to be close enough. And so the, basically, the pencil just coupled with each other, and the board design is not as important. But obviously, I can only work in very specific situations. So you know, that's if you have like a USB 2.0. It's just a single pair. And you can put your connector and your microcontroller, like right next to each other. That can work. I've definitely seen it done with things like MIPI, which is a high speed camera interface. And if you're using an FPGA that can be a little bit flexible about which pins you can use. You could set up your camera connector and your FPGA right next to each other, then you can avoid a lot of issues with impedance control and having to test that.

Have you ever had hardware just flat out not work due to the layup?

Not that I can recall. You know, I'm usually pretty careful to follow all the guidelines. And the guidelines I've found in general have quite a lot of margin in them.

manufacturing tolerances.

Yeah. Yeah, but uh, I've seen MyPie run through, like hand cut micro coax that's like a couple of meters long, and it worked just fine. And obviously, all the specs are saying is like, you can't, can't just like hand cut it, you're never gonna get accurate enough with the link matching or, you know, just holding it in the air instead of like, properly grounding and shielding it and, and everything.

It's like, USB extenders like those are not in this. Like you can't have those according to this USB 2.0 standard. Yeah, but they work.

I've seen some long extenders before.

Yeah, you do have to be a little bit careful with some of that stuff now, because not everything's five volts anymore. So if there are a few like adapters and cables that are explicitly like not allowed in the USB C spec, but obviously you can buy them. And you just have to be really careful if you use them because sometimes you could send 20 volts from your laptop charger into like a flash drive or something that's just not ready for it.

Yeah, it's like building your own generator cable. If you don't know what, that's when you put the mail in of like a of electrical cable or an extension cord on a one cable. So you can plug the outlets in.

Yeah, yeah. When I like the picture comes up every year in the winter. And it says, if, if you need one of these cables that has the males on both sides, then you've hung up your Christmas lights wrong.

Yeah. Or I remember going to a hardware store and they had a picture of it. Like that cable on the like hanging up and the electrical aisle and they said, if you need help building one of these we are not helping you.

Yeah, yeah, there's a reason for that. I mean, the the mains is pretty dangerous. At least the USBC stuff probably won't be able to kill you but might kill your device.

Yeah. Poor MacBooks. Yeah,

I'll be entirely honest. My my my brewing rig. Currently it will change but it is a 220 version of that. That cable to 2050 volt a 50 amp version of that game. Oh, yeah. Yeah, that's so it's like so so but you know, Parker and I've talked about this a couple of times. In fact, we've we've kind of envisioned some interesting projects with it, but but USB type C it has the capability of actually delivering up to 100 Watts, right. So 20 volts, five amps, right.

So a little side tangent on that. Yeah, I can't find all these on. I've only looked on Amazon. I can't it find 100 Watt USB type C power delivery one, like 96 Watts is like why see? I wonder what that four watt difference is? That's that manufacturing

tolerances you're talking about?

Yeah, had the had the same thing. So the for a while the highest power you could get was the 87 Watts from Apple. And I think yeah, there are a few that have cracked 9090 watts. Still not exactly clear to me why nobody's done it. But yeah, I want one thing I would like to do is take the USBC explorer board and hook it up to a power supply and have it be able to actually deliver the 20 volts five amps. Yeah, I

think this one it's 90 watts. And but why hooked up your explorer board to it says 20 volts, three amps. So 60 Watts, but it will allow you to pull up to 90.

Oh, okay. I think I might know what's going on there.

Yeah, that's, that was a little sketchy right there. works. But

yeah, generally, when the when you do the negotiation for the power supplies, you're setting up nominal or maximum, like that the source has to be able to provide. And I think if you start drying over that it's maybe it's usually okay, but it's just the source isn't obligated to keep giving you more current after a certain point, you got beyond what it signed up for. Yeah. And I

did make sure to I have the I guess we can talk about this. I have the E rated cables. Well,

that was actually going to be my next question is what do you need to actually deliver a boatload of juice?

Yeah, so obviously, you need the source, like we talked about, you need a sink that is capable of asking for it. Because a lot of laptops will, you know, they'll just ask for what they need to charge the battery. And that might even change, it might ask for less current if they're near fully charge. The other thing you need as a electronically marked cable. And this is something that freaks a lot of people out is that with USBC, you have microcontrollers in the cable. So it's right there in the plug. And it will be able to tell both sides, what its capabilities are. And so that can be the higher power and the superspeed pairs. So that's the higher power anything over three amps, you need to have an electronically marked cable. And so that's probably why if you just use a regular cable off the shelf with the USBC explorer, you're not gonna see anything advertised over three amps. And then it's a little bit complicated on the source side, because it has to talk to the cable and find out what capabilities it has. And if it doesn't get a response and knows, you know, it's got to be less than three amps, and then actually change what it's advertising. So as an experiment, I took the apple charger and plugged in the that Cypress power delivery analyzer and then plugged in the cable that I knew was rated for five amps. And as soon as I plugged it in, so I didn't have a sink, attach just the charger in the cable. They had a chat about what they can do. And so then the charger knows Okay, when it sees the sink plugged in. You know it can advertise the 4.3 amps. But yeah, some of them. They'll see a sink get plugged in and they'll advertise three amps. And then they'll say like, oh, wait, is the cable out there? Can you do more? And then if it does, it'll re advertise higher current?

Yeah, cuz I've got a USB cable checker as well. So making sure that everything's legit. Because you don't know when you buy stuff on Amazon.

So wait, wait, you're telling me that they the USBC cable I bought at the gas station the other day won't actually give me the full 100 watts. I feel cheated.

Yeah. might not

use the shielding. It might.

Yeah, actually, that that got us into some trouble at one point with the EMC testing. Because the cables that we were using didn't have an internal ground wire. They were just using the shield as the ground. And so we saw like the spike at what was it like 72 megahertz. And like, what is that and then we tried it with the shorter cable and the frequency went up. And like okay, there's obviously the power cable. And then we did a more thorough teardown later and we found like too nice. I think it was like 22 gauge. I think there's like 222 gauge wires for V bus. And then we couldn't find anything that was ground. So that was obviously just The shield

on I'm that this goes a little back to the layout stuff is the shield on USB connectors. And I've looked up like how you should terminate this and the USB spec, I can't remember the exact wording, but it's like your mileage may vary is its explanation? Yes, it needs to be connected to ground. But it doesn't. Like Cypress says you should do like a low pass circuit or something like that. And other people have different ways of doing it. So what was What's your preferred method of connecting the sheath, the connector shield to your, I guess, ground on your your PCB?

Yeah, so everybody's got a different idea of what's the right way to do it? And I think you know, the correct answer is it depends. You know, the simplest thing to do sometimes is just put, maybe put a pads for resistor and a capacitor, and then you can change what the stuffing is later. So you can put a zero in there if you want or do a an RC filter. So that gives you a lot of options down the road. If you get into FCC, pre scans, and so forth, and you're having trouble. I generally just go for grounding the shield. Generally, I think the best way to do it is you ground the shield on one side, and then maybe have some loose connection on the on the other or just leave it floating.

That's the Yeah, that's if you go into the RF world that's like how you would shield the cable. But in the USB spec it you don't know if like, if your hub did that, and then you didn't do it, then your whole shields floating.

Yeah. And I think they might specify it for ESP too, when you'd have one device was very clearly like the power source and the data source. And then the other side was the Data Sync and the power sync. Now with USBC, those roles don't have to be the same for one device. So you can have a device that's a data source and a power sink at the same time, and you have lots of dual road devices. So they can be either power source or sink, depending on the configuration you plug things in. And then they can do swaps. You know, mid session, they can swap who's been the source and who's been the sync. And of course, the connector is the same on both sides. So you can't just design it as like, oh, always ground the shield for Taipei's. I've never gotten the shield for type B's or something like that. So yeah, it's it's a very complicated and yeah, I think the best option is if you're designing a new product is just put the pads there. So you can change the bill of materials later and fix it. And that just gives you the most flexibility unless you know exactly what you need to do based on something else

less free spins down the road.

Yeah, yeah, I'm a big fan of putting in extra stuffing options. The the copper pads are free. So take advantage of those.

Well, and Parker, don't you? Don't you do a ferrite bead to ground?

Yeah, I generally do a ferrite bead ground like 100 megahertz one. So it seems to get around every single time I've done FCC stuff that seems to because I always put, you know, a zero ohm resistor there. And then you're like, okay, there are cables radiating. So I've just popped this one beat down, and that usually gets rid of it. But even then it's like, you know, some people do that. Some people say that's the wrong way of doing it. Some people think it needs to be connected. Some people like it's all over the map. I think you'd basically it's what Jason's saying is, you need to do what works for your application. And I think that's why the USB spec says that basically,

just pepper your board with FCC pads everywhere. Things that you can adjust in testing.

Yeah, not a bad strategy. Yep. So

one of your older projects was a USB type C, Easy Bake Oven, which I thought was a really cool application of power delivery. Do you have any more projects like that coming down the line that you'd like to spill the beans on?

No, nothing, nothing on USPC right now. That was more when I was talking to people than they didn't really appreciate how much power USBC had like literal power. Yeah, it was like what else is about 100 watts. And it's like the old Easy Bake ovens just used a light bulb and turns out the new ones just use a nichrome wire. So it's even easier to modify. But yeah, these days actually, I'm working on another project. That's more related to sensor monitoring. So I have a new board I've designed just about to get into the reading stage, I've got placement figured out pretty much now. And that board will, it'll be able to monitor a variety of sensors. So things like thermocouples pressure transducers, strain gauges, and then it's all based on the new particle, Gen three hardware, the mesh hardware. And we've done some tests on our own, for what range you can get on the mesh network. And it's usually like 100 meters or so assuming you have line of sight is a pretty reliable. So the idea is you could put a bunch of these sensors around and put the mesh only PCB on most of them, and then just have one that connects to the cellular network or out to Wi Fi. We even looked at doing a satellite link, so it get like a satellite to Wi Fi bridge, and then hit the Inmarsat beacon network. And so you could deploy these things anywhere that you had power, and then monitor whatever you needed to. And then Pair that with software that's running on a server that's monitoring the values coming in. And can send alerts, hook into pager duty, or send you a text message and email, something gets out of spec. And I think one of the first applications we're going to have is our own brewing rig.

And this is gonna be on Reclaimer labs.

Yeah, I haven't decided exactly how I'm going to be doing this. But yeah, it'll be through Reclaimer labs. The board that I designed is very capable, it's got a bunch of sensors built right into it, and it uses some high quality connectors. So that means that the cost is pretty high as well. So I might not be able to put it on tindy for as less money as some of the other projects. So I might be able to put some of the extras up there and just have to put them in a little bit of a higher price.

Is this gonna be for like industrial applications? As I'm zooming and brewing? Yeah,

that's a target right now. Yeah, that yeah, industrial applications, and home brewing, or maybe even something like a micro brewery. Yeah, this all started with a friend of mine who was in grad school, doing material science research. And sometimes they have to run samples in a furnace for three months at a time to do an A Neil. And it needs to be at that temperature that whole time. And so anything that let him monitor the temperature, and you know, he could check it from his bedroom and not have to like walk all the way across campus to double check on his experiment was very helpful. And then we were looking at the board design and are like, this is pretty close to being able to do a bunch of other sensors. So we just added some extra biasing network options. So we can set it up to interface with a lot of widely available sensors. So

you can do like four to 20. Differential, although major industrial players.

Yep. Yeah. So the three ones that we're targeting mostly as thermal couple, four to 20 and millivolt per volt.

I think we did you have anything else even? I'm looking over our list? I think we're done?

Well, first I what I'd like to hear about is, well, can you let people know where they can find more about you?

Yeah. So like we talked about the Reclaimer, labs website, Reclaimer, labs.com. Front page, there's the blog posts. So keep an eye on that for new updates. I'm also on Twitter, at Jason Spindler. And we'll put a link to that in the show notes. I'm sure

we got that for you. Also, you have a you have a GitHub, too, right?

Yeah, I do have a GitHub on also under Reclaimer labs. And so we'll be putting all of all the designs I work on at home, go up there with an MIT license. So like the USBC explorer, if you have a USB C project that you're working on, and you want to just copy the relevant page on my schematic, go right ahead and get that in your design. All the firmware I work on is also open source as well.

Oh, cool. So Jason, do you want to sign us out this week?

Sure. That was the macro fab engineering podcast. I was your guest Jason Servando.

And we are your hosts Parker Dolman and Steven Craig. See you let everyone take it easy Thank you. Yes, you are listener for downloading our show. If you have a cool idea project topic. Let's Steven and I know Tweet us at macro hub at Longhorn engineer or at analog E and G, or email us at podcast at macro app.com. Also, check out our Slack channel. If you're not subscribed to the podcast yet, click that subscribe button. That way you get the list episode right when it releases and please review us wherever you listen as it helps the show stay visible and helps new listeners find us