Electronic Wear Items

Hello, and welcome to the Mac fat engineering podcast. We are your hosts, Parker, Dolman and Steven Craig. This is episode 264. So this is going to be a very interesting episode because we just have a whole bunch of random topics that we're just going to rip right on through. Let's do it. Okay, so Tesla, are they actually okay, let me just do the title I guess for it. Tesla recalls cars with E mc failures, calls Part A where item. And this was on Hackaday. Today that we're recording. So we're recording an episode like a week ahead, just so that we can have like, a day off. Actually, I think it's worth saying this week is five years of the Mac Feb engineering podcast. Oh, yeah, it is. Yeah. So congratulations. We did. Yeah. But yeah. So instead of like skipping a week's we have that. So in five years, we have not missed a single episode going out for a week. We're pre recording an episode so that we can have a day off? Yeah. I'm okay with this. Yes, this is fine. So yeah, Tesla. So are they actually recalling cars for this?

You know, I have to admit, I saw I posted this, and I skimmed it, and skimmed it for like, the funny stuff. Oh, so like, I don't know, the full extent of like, are they? Like it says they're doing a recall. But I don't know exactly how much

that is. Because I think what it is, is the NHTSA, which is the National Highway, transit safety, something sociation maybe, anyways, that's like the, the, the, I was gonna say three letter, but it's actually a five letter or agency here in United States that does, like safety and patient make sure cars aren't going to explode while driving down the freeway. So they were requesting Tesla to do a recall on this component that was failing in their Tesla vehicles, which was a memory a flash memory chip, an eMMC chip. And Tesla responded that it is a rare item. And so they don't have to, they don't have to do a recall or do any warranty issues. Because it is a quote, where item like brake pads. Right, right. So so basically, the car's operating system is Linux, and they have been dumping logs to this flash. So it's getting written to a lot and often, yes, and I remember this coming up. I don't know if we ever mentioned it on the podcast. But I remember reading about this issue, where someone was hacking the Tesla operating system. And notice, hey, there's a bunch of logs going into this flash memory thing. And that's probably not a good idea for longevity of this component, flashed forward three years later, you know, like person was right. You know, it's kind of funny, because I sent I sent this article to my boss, and there was a product that was designed years and years ago, that inadvertently the code wrote to an EE prom, more often than was expected. And the company saw some premature failures, where like, it would get stuck and locked in in a in a situation and you'd have to reset the processor, and then it would access a different location in RAM. And so like, it could just freeze at certain point and the product was actually discontinued. And there was a design effort for another one, mainly because the Flash was just getting written to too often and you end up like bit rot actually does happen in flash memory, they have a specific number of rights that they're allowed to accept. And yes, ran into that. And it sounds like this is a very similar situation. Here's the thing that sucks though. According to this article, they said the problem leads to widespread widespread failures. That could put it in limp mode, which limp mode from modern cars in a lot of ways is like, it limits your your speed to like 10 miles an hour. And it like basically barks at you to get it to some kind of service center. Correct? Yeah. So this is the thing with them calling it a wear item. So this is this is coming from someone who works on a lot of cars. cars. So where item one is, quote, easily serviceable, unquote, some cars are more easily serviced and others, but unlike, let's say like a oil filter, okay, that's something that gets changed every oil filter oil change which is depending on the vehicle 3000 to 10,000 miles some are like doing 15,000 miles now. And some of them are in weird spots that are like you have to like triple joint your arm to get to, but you can get to them and replace them usually without having to remove any other components in the car. This E and MC components is inside the dash, you had to disassemble the entire dash of your Tesla. Or I guess you wouldn't be doing it, it'd be the service center, or the poor schmuck who is going to get paid like half a day's of fixed pay to do like a two day job for the warranty. To replace this components into your car, so it is not an easily quote, easily serviceable item. And to this item doesn't appear anywhere in the recommended service interval manual for these cars. So if it is a wear item, it has a how many mile usually it's by miles or hours, depending on the type of equipment of when these were items should be serviced. That's not in here. So it was never even considered a wear item when they were building their documentation. This is just a straight up cop out to avoid warranty issues and recall issues, you know, and okay, so if this was aware item, it's this is a ball grid array memory I see. That's not something that an any normal person would be able to replace that. So your only hope of replacing that would be replacing the entire board. Like, I know, this might be a stupid analogy. But that's like saying the coils in your alternator are aware item and you should be able to replace the coils. Right? Like it's ridiculous. Exactly. They could have made it like if it was a component. Let's say it was just a circuit board, you had to replace every I mean, in this case, it was like it's like five years, four years. So like pluggable flash or something like that. Yeah, yeah. Like, you know, pop off a panel off the like, like pop a panel down. Oh, there's the board right there, you just kind of pop it out with like a like a memory stick or like a PCI Express card. Then I would be like, totally they design that to be replaced. Sure, maybe that component, it's not the most cheapest component because like you have to replace the whole board and said the just that chip, but again, that's not too big of a deal. And then you could say, hey, every five years you need to replace this. Because half your car's not going to work if you don't, right. Yeah, that's actually the surprise thing. It's running out of, it's wearing out because of they're putting in a lot of Linux, like OS logs there. But so much of the car uses it in terms of like your, like the example given is like the defroster, yeah, to run the defroster on your car, it has to use that component. And that component goes out. You can't use your defroster.

Well, it also controls according to this the heater and warning systems. So there's some potential safety behind that.

Say, Yeah, so a component. That's another thing is a component that's designed or that is part of the safety system of your car is now aware item. Which boggles my mind. I don't know. I mean, like, I guess brake pads. I was about to say. You're right. You're right. You're right. You're right. Okay. You're 100% Correct. There. Yeah. Yeah. But but something like that. You can't design it to not be aware item. Like brake pads. Yes. correctly. Braking your car has to be well, it has to wear in some way. Correct? Yeah. Thanks. It's friction. Right? Right. So yeah, it's just one of those well, okay, we can borrow that but you could build brake pads that last the lifespan of the vehicle. That's what a lot of manufacturers Well, manufacturers aren't doing that. But doing other things like lifetime fills, I'm doing air quotes, lifetime fills of like transmission fluid and stuff. Yeah. Which are like the transmission fluid last forever in quotes, because the car falls apart and for years. So what is the solution to this? Not use eMMC. For for the logs Yeah. For things that you write to on a very regular basis. Very, very good. Would races Yeah, or make it something that's easily replaceable. But I would say, just instead of just spend a little bit more money on the storage medium, like or actually use like DRAM or something like something, another kind of technology that you can, you can write to a lot and doesn't have a problem with it.

And I think someone in the Slack channel was mentioning also split it up. And use use smaller chunk of DRAM for just logs, and use the eMMC. For other things that don't get written as often.

Like turning your defroster on and off. Sure,

yeah. I mean, honestly, yeah. If you needed to log, okay, defroster was turned on at this time, and then turned off at that time. That's not a regular, right, that's very minimal. And that should last the span of the of the car. Well, and I mean, you should be able to predict all of these things, too, because the data sheets for memory typically tells you the average amount of writes that you can do to it. But then again, perhaps they perhaps the engineers at Tesla don't have like a full grasp on how much Linux is trying to write to this thing.

And I'm saying that because perhaps nobody does. You know, the thing is, this was brought up years ago. And Tesla responded by saying, it's not a big deal.

It's aware of them, don't worry about it. But it's like they could just fix it with a software update. Well, but But you see, they're actually telling the truth, because it did wear

it did wear out, because it says it says in the datasheet they were out. Yeah, so it must be. Yeah. Well, I'm calling that a cya right there. Yeah. It just, it's like, it's like calling a, an internal combustion engine. It's like calling your rings that seal your pistons wear item? Yes, they were. But they and they last a long time. But no manufacturer calls it a wear item. Because you don't regularly replace them. They lost the quote, lifespan in the vehicle, unquote.

I wonder if there's like an official definition of where item like to which look, like does aware item have to be something that the owner should, with reasonable competency should be able to replace themselves?

Like, how do you define aware item? I mean, obviously, define it however you want a good that's that's according to this. I mean, that's what Tesla is doing. But what the auto industry says is a wear item is something that's designed to be easily serviced in quotes easily. Or he's got to do that nowadays. And it has a regular interval like procedure to to change it or inspect it. Yeah, that's like a wear item, I would I would consider and that's what I've seen other like, if you open up your like your service manual for your car, it will have all the what they consider were items in there, right. And they're generally not covered under the warranty, like wearing out your brake pads is not covered under warranty. Right, because they considered wear on tires, tires are aware item did not under your warranty. Your piston rings, on the other hand, are not considered wearing even though they do wear. And so if they prematurely fail or wear out. You're you're under warranty that will replace them. Right? Yeah. Because everything wears out. Yeah, everything eventually. Interesting. Yeah, it's, um, I know that Tesla bugs me a bit. So

you know, sort of in parallel to this just yesterday, Derek on the Slack channel wrote. So it finally happened. After all these years of using SD cards on Raspberry Pi's. I've never had an SD card fail on me until today. And it's just kind of serendipitous that, that like a flash memory issue happened to him. Then which I you know, I was playing around with PI's the, the other day and thinking about some server applications. I'm like, man, if I'm running anything on this, that I want to maintain data, like it's ridiculous to run it on an SD card, you would want to boot it on some other medium that has a bit more reliability, like SD cards are not intended to run an operating system and pass data continuously.

Correct. Yeah. I mean, you're gonna lucky with the Raspberry Pi that it doesn't well.

Yep. Now Tesla can fix this by making that not log a whole bunch of stuff and then actually own up to the fact that this was a design mistake, and then actually go fix it. Like and have they will have to either replace everyone is a chip or part under warranty because it's is not aware item. But we are equal opportunity. Talking about car companies. So GM to extend vehicle plant shutdowns for more than a month. So this is continuing our holy crap the supply chain is screwed up right now for electronics like train we've been writing for like the past two months on the podcast. So if you're sick and tired of listening to this too bad, we're gonna talk about from like the next 10 minutes, probably, or at least five minutes. So GM expects this to get worse. They they temporarily shut down to this GM slash Chevy. They temporarily shut down three factories. And they were like, oh, yeah, it's only gonna be for like a week or two, because we don't have components to put into our cars. Well, they didn't get their components. So now they're extending it to more than a month of shutdown. But the big thing is, I took away from this article, this is article from fierce electronics is that Ford is having to cut pickup truck production. Oh, and this is America. I was about to say you don't really love our pickup trucks. Yeah, it's so it's ridiculous. So if you're not from the United States, you just don't realize how many pickup trucks Americans buy? Oh, yeah. It's ridiculous. Especially Texas, especially Texas. But like the Ford F 150 is like the most sold car in the world, and is only sold in America. Just think about that.

It's just ridiculous. How can they survive that? I mean, that's a huge downtime.

Yeah. And so and this is, since that happens is Ford is having to cut down their pickup truck production, which is their breadwinner, by the way, like that line probably makes most of their money is Ford's earnings are going to be predicted to be $2.5 billion less this year, because they can't make pickup trucks. Oh, geez. Yeah. Components suppliers are saying by mid June there will be caught up with all the automotive orders. So there is a light at the end of the tunnel. Unfortunately, it is in like, month five or six of this year. I wonder, like they got they had, I mean, they have to be doing something during like, this must be like, okay, spring cleaning time or you know, something like that. Yeah, let's optimize everything while we have four weeks of downtime. And then Texas Instruments said that, we we saw this ahead. So we built a bunch of production. So apparently T is good. In terms of their volume, I'm sure they have an entire divisions that just sit and look and try to predict this kind of stuff. Yeah. But remember, it was like it was two or three years ago? Yeah, it's 2018 that ti had a bunch of supply problems. But they fixed it by buying a bunch of cheap fabs but didn't actually really have to spin up any new ones. Yeah, when you got money like that, you can just buy fads. It's like, last episode, we were talking to Chris Carter, about, you know, preparing for production on large scale things. Like one of the simplest solution is just buy all the parts you need when you can, as opposed to, like, you know, doing it just in time or whatnot. Like if you have all the parts then you can make it right. So just buy everything. Just buy planet Earth. Yeah. By all the mineral rights in the world wouldn't be surprised me if that someone has tried to do that.

So get this done a few weeks ago on episode 260. We talked about or I brought up the the idea of a different form of component searching for large component vendors. And today, Digi key released a solution to what I was complaining about. So yeah, thank you so much Digi key,

which is pretty cool. So the idea had or the concept behind this new search form was to basically be able to zero win on whatever part you want faster. And go back to Episode 260. If you want to hear a little bit more details, but effectively all it boiled down to is I want to be able to search look at all the The items in my search and as I go through and read data sheets, just be able to peel those items off my search. And Digi key now has a solution for that. So there's a little search bar up at the top of Digi key that that is called Search within results. So after you've already done your initial search to get your whole list of items inside that box that says Search within results, you can just type till day, and then put whatever characters you want to delete from your search. And it'll just dump those parts from the or active search box. So you can just leave your window open and keep peeling off parts as you look for them. Or look through them and, and then zero down to whichever one you want to look for. I'm super excited about this, because I have, I have a very simple thing that I want to try that this for I'm in need right now have a pin header, just a regular Joe Schmo, 2.5 millimeter pin header, but I need a pin header where the body, the actual plastic part is one millimeter in height, which normal pin headers are 2.5 millimeters. So I'm going to test this in one of like the crappiest situations because if you go search for pin headers, you get bazillions of copies of stuff where it's like, you know, based on pin count, or whatever, blah, blah, blah. So I'm gonna I'm gonna give it a shot and see how fast I can narrow down to a pin header that has the right body size that I'm looking for. Because that's not a normal thing that's going to be part of the search fields that I could initially go for. So I need to go get a filter for it. Right? I have to I can't filter for that. That's not typical. So I'm gonna have to get a whole list of pin headers, and then go search through data sheets.

And I'm sort of if you can't tell I'm kind of excited about. But yeah, thanks. DigiKey. Like, I'm super happy that this exists now. I'm totally going to use Digi key a lot more now.

Yeah, I can't wait to start using that because I think it was Oh, who is in our Slack channel? bumbled monk, right. Yeah, I think he works for Digi key. So he hangs out in our Slack channel and released that. That changed note, I guess today to us. Digi key has spies everywhere. So I can't wait to start using it. There's also that's like only one of the features that got released. There's like a Oh, yeah, there's a boatload. Yeah, there's like you can do like, basically, operations like and not and stuff like that. And or in the search fields as well. From what I'm privy to, Digi key is doing a lot of fancy stuff on the back end. And there's a lot more beyond that. In fact, bumbled monk from our Slack channel, what went on?

What, what's the Gosh, why can't I remember it? Now the other the other podcast, electronic or the software podcast, embedded FM, embedded it went on and talked about some Digi key stuff there. So go check that episode out. There's a lot of cool information in there.

All right. So this week, I've been working on a really random project, and it is done minus testing. So I can present it Oh, you're past 50% I've passed but I'm not like 90% Because I want to test it is I was thinking I want to be able to reseal 12 ounce cans. Okay. And you're like, let's say it or something, a beer soda or whatever. They make like lids that you can like snap on that kind of keep the pressure. But they don't seal well. Like if you spill if you tip it over, it's going to leak out. I wanted something that like you actually can seal it, and then toss it in your backpack and then you don't have to worry about it. In theory, okay. So there's a couple different products out there that like might work and I tried a couple of them. Like one actually, like turns a can into like a screw top. Which worked great until like it like if you throw it in your backpack, it would pressurize up and then pop that piece of plastic off. So anything that attached to the can with like a friction fit was out of the question like it just can't support the pressure. And so I decided okay, I need to build something that goes around and holds it from the top and bottom right, and then compresses a seal on top and like the lid lip. Right the lip on a 12 ounce can Yeah. So I Add a Yeti, a Yeti which is a brand of like, third most their most style coolers like that for drinks. They make one that fits a 12 ounce can and so I started with that as the base because that gives me like most of the design, right? And I just needed something that would thread on the top and seal like a cap. Right. And so I designed a cap show and Stephen. Wow. Okay, take a pic. I'll take a picture for the podcast. Yeah. Did you borrow the side? It's printed. knurled Wow, that's a turned out amazing. Machine neural. Yeah, it looks the machine. It's amazing. So this is one I that came off the printer last night. Here's one with a seal in it. So it's got a a FDA material approved seal which is like a silicone sheet. They I bought on like Amazon, and I just cut it out with a I've removed like the center Arbor from a hole saw. And then like very carefully on the drill press like cut into it. So because you know once the walk because you don't have a senator Arbor Yeah, for sure. But I was able to get it cut pretty cleanly. Okay, and then i i It's a, a sheet that had adhesive already on it that was FDA approved. So it just kind of like stuck in there. And you can actually see the ring of the lid over the lip of the of the can as you screw down. Oh, look at that. That's cute. And it doesn't rattle anything. So I haven't tested it yet. Because I don't want to test it with like a beer or anything. I want to get like club soda, something that's Sticky, sticky. And basically, I want like the worst case scenario is like put a cat in here, open it up, and then reseal it. So it's fully charged ready to go and toss that thing in your backpack and see if at the end of the day. Does it have pressure? Oh, yeah. At the same time. So crack. Actually don't crack a beer. If you can't. Well, okay, so I'm looking at at a chart right now, soda is typically pressurized at 60 degrees Fahrenheit to about 40 psi. That's more than a beer.

Beer Beer is a little bit lower than that. So if you and I don't know about club soda, I would think it would be closer to the 40 psi.

Yeah, it's pretty bubbly right there.

So crack open a club soda, put it in that thing, close it up as quickly as you can't shake the living bejesus out of it. Yeah. And I don't

think it's going to explode. I'm more worried about like the plastic. It's probably carbon at this thing is tough as nails. I'm more worried about that seal. Being able to hold that pressure. That's why I'm saying like, if you shake it, like see it, see if it jets out, you know? Yeah. And I know there's someone out there to me like, why don't you just like pour it into some other container that's like designed to be sealed with like a screw top water. That defeats the point. The point is to not pour anything. And have you ever tried, like, when you pour a beer or anything, it's actually really foamy into one of those, it foams up like crazy. Yeah. So you it's it, you have to like, like, barely pour it in slowly. I'm like, Man, I got I got too much, too much stuff to do, I just want to like slam the beer in there, open it up and be good to go.

You know, if you wanted to get really fancy, you could, you could make your little flask thing that you have here and then have a side port, that you could pressurize the internal of that to whatever pressure is in the can and then you kind of equalize the pressure of everything

I can where you're going with that. I don't worry about it. I was just making this making up something that you could equalize the pressure such that you wouldn't have the chance of explosion. Oh, I see what you're saying. Yeah. So But yours is cheaper and easier. Yeah. Well, the original idea is I started designing like machining my, like, a lot of aluminum and like the base and stuff. And I was like, you know, it's gonna suck trying to keep that cold if you make like basically you're strapping a heatsink to your can. Sure. Yeah, that makes sense. So I went with a thermos style Yeti. And amazingly, I got the threads on the first try, or almost the first try. And then I tried to improve the threads and I have like a huge pile of like thread tests, like 20 of them. The first one fit the best Okay, The threads on those things are not like, national, thermo standard threads. No, they're not. They're just some custom thing, right? There's an actually this is the newer style lid, or whatever you want to call it. It's like one circle of tooth, right? No, it's It's, it's all on horse start thread. Oh, okay. So it's called their Lock and load, trademarked thread design, the only so you can get the drinking faster, because only a quarter turn to lock it. Instead of having like, screw it on. Which I do, easy to form for them. Yeah, the, the older style is like a multi turn single start thread. And I do have a lot of those. And I was gonna use those. But I was thinking like, Okay, if you've got your gloves on, like you're snowboarding or skiing or whatever, and you got your gloves on, you don't want to be like fumbling and screwing this thing open and close. And like, man, you just want to walk and load that thing into your mouth.

I like how like, they've gotten to the point where they've, I shall we say, perfected that the heat side of things like they've made like, you're not gonna get significantly better than the thermos. So they're like, What, what's the next thing we need to do to make a luxury brand? And they focus really hard on like, the feel of the threads?

Yeah. And actually, I will put it this way, whatever I printed fits better than the thing that came with.

Wow. I don't know why. How did you how did you end up doing it? Did you just guess or like, not guess but you know,

I took the ring that it came with? It's like, it's like a locking ring. And I actually I cut it in between the threads and then flattened one out and then measured it. Okay, kind of got like a just a profile of it. And then yeah, I got profile. And I, I did it as a, an fusion 360. I did it as a because you can't do it as a thread because you can't do multi start threads and fusion. So I did it as a coil. So you have to do four coils, basically that only go like point two revolutions or something like that. Right? Like, if this works, I'll share the design. Yeah, because like, just getting the threads right was like 90% of the design. The rest of it is just like, oh, yeah, just extrude it out and round some things and print it out. How'd you end up doing the knurling in fusion? That is also a coil? Or I'll tell you do a what was that? Oh, did you do like a mirror? A circular mirror pattern? Yeah. So you do you do A a coil one direction. That's like a point to revolution. But at a steeper angle. I got a 45. And then you mirror that. So you got one going the other way at the same base. And then you take both those and then circular pattern those around like 100 times. Yeah, yeah, no. And that worked the opposite way. It looks like a machine neural.

No, on the image here, it looks like you have black anodized. Aluminum. Yeah, it's

like you actually can't tell it's a printed part unless you look at the top. Because that's where it was sitting on the bed. And you can see like some of the print lines, you need

to come up with some goofy logo and slap that up on the top. Oh, yeah, yeah, just put like a circular like

plate and glue it on the top should Well, this is the prototypes I brought to make them out of aluminum. So they have Lock and load threads, you need to come up with some dumb name, like, I don't know, crack and slam or something like that. But it's mainly for like, so if you are, what I want to do with it is when you're out skiing and snowboarding, I usually bring a couple beers in my bag, because I'm not going to pay $8 Actually, it's on the cheap side, like $12 A beer on the side of the slope. Right? So I just, you basically just like, sit down somewhere and you crack a beer and drink the whole beer for sometimes you don't want to hold beer. You want to like drink couple sips and then like get back going right? So that was the idea of this was like Okay, put your beer in here. That way you can have you know, a quarter of a beer loss of a backup, throw it back in your bag and then next couple runs later. Pull it back out drink couple more. Engineers always have solutions for it can either be the most amazing idea I've ever had. Someone's gonna take that and make a bazillion dollars or, or it explodes all over my backpack. Yeah, next time you go skiing and it'll be trial by fire. Yeah. Well, I expected them to A five minutes. Topic and that took 15

Yeah, that wasn't one of the longer ones today. No, that's super great. Just out of curiosity, I'm gonna extend it just a second. How long did that take to print? Like one lid? Two hours, two hours. Okay. Not bad. Set them up. So just a lot of 40 bucks apiece for the lid.

That's more than the goosy. Okay. And the tar revision for it's been a while since I've talked about this. Prototypes are ordered for rev four. I maybe I'm confused. I thought the last rev was Rev. Four. No, no. Last rev was three. Well, I mean, am I confusing myself? Might be I thought I thought you were already on four. I'm looking at the last board right now. It says rev three.

Okay. I'm the idiot. Alright, so what's special? And what's new about rev four.

So we had a couple of episodes, we were discussing relays for the predator. And so we're basically I'm putting on it like the most badass relay that we found that fits the specifications is like 500 volts 10 amps. And it's actually rated for that as a switching voltage. And it actually has like two contacts and you have to in DC applications, it has to be wired up the correct way. Because the how the contact shapes are. Oh, god, it's expensive. Yeah, but I mean,

it's gonna work in quantity. 1000, these relays are 11 bucks apiece. Yeah. But hey,

it's a safety device. Yeah. You can't complain too much. You can't complain too much about a safety device. Yeah.

It's not aware item is. You don't want to you don't want it to just be

aware. Replay unsoldered the relay after 100 plays?

No, no, no, no, no 100 Put that in the manual. Just if it goes bad, then tell someone that it's aware.

Yeah, exactly. We had to change the current sense I see on the relay on the high voltage, I would say it's really on the high voltage side as well. The old one, it's just a higher spec rating. amperage rating, not voltage rating that we're currently using, mainly for In Rev on the rev three board, we started using the the fault line on the chip to like actually fit that into an AND gate. And that gate drove the relay circuit, then drive it directly drove it through a fence, all that good stuff, but not an gate. So the AND gate got a signal from the microcontroller, that everything was good. And it got it from this, this current chip. So basically, the current chip, if it over amperage, then it would hit the alert to flag and then it would shut down the relay, which is basically a whatever. So basically is like is like a faster alert than what the the microcontroller could respond with. And we were actually spiking that relay that that that sensor to its tripping point. And so we just bumped it to like the 30 amp version, and now we're safe again. Okay, basically, as the inrush current was tripping it?

Well, I'm almost wondering like could would it be possible to instead just put like time delay, so you allow transients to pass through?

You can't control the signal like that in the in that chip? Well, but you could tell the output you can like the output the AND gate you could you could time delay that? No, because the the fault line only resets on power. Got it. So on power that chip when it's when that current sense chip sets its fault line, that fault line cannot be reset unless it's power cycle. Okay, that makes it it's not programmable, so you can't say hey, ignore spikes less than 20 nanoseconds it's an all or nothing pin. It's all nothing. Okay, and so the way to get around it is just go okay, it was going to have less resolution on the amperage which in reality doesn't really matter because we only care about really is current flowing which should be flowing or not flowing winds.

Isn't it just like a one bit system? Zero is okay and one is Oh shit.

On the Yep, well, if we actually are reading the voltage amperage, but we really only care like it's really only there to see if there's act like a bad MOSFET or bad diode on the wall. In the board, or on our in the pinball machine, and probably show up as a big spike. So it's pretty easy to spike. Yeah. And so this is we added basically the safety of gate logic to flip the relay off, just in case like, there could be a situation like, hey, the microcontroller actually locked up, and a FET went bad. What else can we do to prevent that from happening, and like, you know, a five cent AND gate adds adds more protection to the system. So we did that. And then we added a second RGB channel for zero LEDs. And we did this because not because of like, how many LEDs we needed or anything like that, but actually the current rating of the connectors we had to use for the LEDs, so we're using actual jst 2.0 millimeter connectors. So jst is actually the brand jst I'm talking about. And we're actually under the current limit of the connectors, but it's the wire that we have to use, because I think we're limited to, I want to say 24 gauge, but I think that's I think it's 22 Either pull up, let me pull it up. Oh, based on the pin contacts, yeah, like your crimps. It actually, you're limited to a certain amperage or amperage gauge of your wire, which then limits your, your current. And so we had to change over to a to baseline so we just have to power rails now. Do main PCB. There's my cable joins. It is 22 gauge. So we had because 20 gauge 22 gauge is like the largest you can fit in the terminals. And it's just like, well, that's just what we have to go with. So hopefully that's all the changes we have to do

nice for red production. Yeah, it's not too bad. Yeah, because you I mean, you had it pretty close on rev three

already. It's just just learn the washing, basically. Yeah, Rev. One is still running like prototype pinball machines. So it works fine. It's just one of those like, oh, we need to add this, we can improve that. Clean this up is basically what it was like rev two was like just cleaning traces up and changing some silkscreen and then rev three, we decided to change a little more like the relays and stuff, because we just weren't not satisfied with the, how they were performing.

You know, whenever I do design stuff, on a larger scale like that, that's usually what I shoot for Rev. One is just get it functioning and get it such that like, try to get it on the first go to just do the general things you want to do. And then use the next revs to polish that. And you know, in my line, polishing might be like, does somebody like it? Or does somebody not? Or polishing could just be like, oh, you know, the amplitude of this signal needs to be bumped up just a little bit. So I'm going to change a resistor value or something like that. Or, you know, simple things like that. And most of the time that tends to work out and you're you're getting close to production by your third or fourth revision.

Now, I agree, that's kind of what I do. Yeah. But that's all I have on that. Hopefully, when those show up. I plug them in, it all works. And I can go onto Mac fabs website and go print 500

Nice. Windows. When do you think one of their going to be built? We had like two weeks or something?

Yeah, in two weeks. I'll have nice. And then so we're running close to our time, but I have one more thing. Something I had been waiting for. Oh my god. I will say a decade. I want to say a decade years,

many, many years. Yes. Parker made memes about this years and years ago.

Yeah. I think actually, like when you started at MACRA I was talking about this. Yeah, still. Yeah. And I made the meme this meme of like, a skeleton waiting on like a park bench. That was still waiting for a prop to to be released. And that was six years ago. At this point. Yeah. Yeah. Six years, six, five and a half years. So anyways, I and we had chip and Ken Gracie on our podcast a year or so ago, no longer than that because before COVID and We talked about the prop two, and all that good stuff. And they said it was coming. And I'm like, I will believe it when I get one. And I a couple we were talking, it was just me and you actually not on the podcast we're talking about. I've been working on spoiler alert, because it's not 50% done yet. But the the multi multi gauge like Octo prover thing. Yeah. You talked about that a handful of weeks ago. Yeah. Okay, so I did mention it image. And so I'm gonna use the I was like, oh, yeah, the prop two is supposed to have like, basically an ADC, analog digital converter on like, every pin. So you have like, 64 of those. I'm like, that's perfect. This is actually perfect for this project. Because I need a crapload of ADC like, I don't need super high resolution ones. But I need like, a lot of them. And so I'm like, Oh, I wonder it's out. So I went on the web on parallaxes websites, and they are selling them and thought, oh, my

gosh, he's holding it up. He actually has a physical chip in hand. Well, not a dev board,

Dev board. And this dev board is not the cheapest development board. This was like the most expensive one. It's it's pretty chunky, though it's got some weight to it. When you say most expensive go ahead and spill the beans. So it's 150 bucks. is now they do have some that are like 40. But I wanted one that had I wanted an all in one board. So it has like all the pin headers are broken out. It has all the supporting hardware you need. So I just have to plug USB into this. And it works. Yeah, it has it has the memory it has the switchmode supply on it. Things like that. Yeah. All that good stuff. It has a a Yeah, it has the switch mode. LTO on it. Yeah. Low dropout regulator. Oh, okay. I thought that was a switch mode. I thought I saw it is switch mode, but they called an LTO. And maybe it's both? I'm assuming it is because it has an inductor on it.

Yeah, it looks Yes. Right. Yeah, that's what I saw that big old chunky inductor. But yeah, it looks like it.

So yeah, I'm pretty excited on actually getting to use this, I powered it up and the lights turned on. And it didn't like smoke or explodes. I assume it works. I haven't done anything with it yet. But my first goal is to build up some dev boards with like all the components, I need to use the for this project, like the thermocouple readers. And then I want to build some I need to have like a isolation barrier for the analog stuff. Because I don't want to connect those pins directly into the prop. I want to have some kind of buffer there that protects it. And then I want to build up that that Analog Devices, app note on a board as well. They kind of like cook it all together with spaghetti wires, and see if I get something functioning, and then I'll spin a whole board that, you know, runs runs this guy. Nice. So prop one has its own coding language. They, what was it called spin spin spin. I'm assuming prop two runs spin as well. Yeah, they it's called spin to now configure, but it does anyone just like the prop one though, is it? It does run C, it does run actually a couple of different languages they have compilers for it's not just C, not just spin, right? I'm going to use spin. Because it's I think it's actually a super powerful language for doing the embedded systems for this style, because it's you are instead of having interrupts or anything like that you do have eight processors to wrangle. And spin actually does a really good job at managing that for you in a conceptual way. And it's very colorful, is very colorful. I know a lot of people are turned off by least were turned off by prop one because of spin. But I'm like you can use seen do it that way if you really wanted to. And really though it's like it's just syntax in the end of the day. It's just a different syntax for for the language, right, I've write it, I write it a lot like see it just in SPIN syntax. And back when I did a lot of hardware development and prototyping, I would I would do my prototyping on a parallax propeller because I could shove out firmware so much faster. And then when it came time to actually finalize the product, I would switch to a cheaper processor like an EF M eight are a tiny and I will just put my code into from spin into C for that microcontroller and it wasn't that difficult to do. Now I wasn't doing super crazy like firmware stuff that you know takes months or years to develop but it is possible

yeah

all right, I'm looking looking forward to seeing that thing rip Oh yeah, I'm I want to see how much power it consumes because as a chunk chunky chip now I want salsa. Awesome chunky salsa to put on my chip. All right, let's wrap this guy up. Cool. So that was the macro fab engineering podcast. We were your host Stephen Craig and Parker Dolman take it easy Thank you. Yes, you are listener for downloading. And I was gonna say something like eating chips or something, but I couldn't think of anything. Downloading a podcast. If you have a cool idea project or topic or salsa recipe. Let's do it. And I know Tweet us at Mac fat at Longhorn engineer at analog E and G or emails that podcast at Mac fab.com. Also check out our Slack channel. You can find it at Mack fab.com/slack