#428 – Setting Fire To The Tracks

Chris Gammell: This is The Amp Hour Podcast. Released February 3rd, 2019. Episode 428. Setting Fire to the Tracks.

Dave Jones: Welcome to the Amp Hour. I'm Dave Jones from the EEV Blog. And I'm Chris Gammell of Contextual Electronics. Happy New January almost. Sorry, February. February.

Chris Gammell: February, right. Welcome back. Welcome back, friend.

Dave Jones: Thank you.

Chris Gammell: Had a heck of a holiday, huh?

Dave Jones: I did. I did. I was on walkabout.

Chris Gammell: Nice, nice. The Aussies do... You guys do that right, I got to say. I'm envious. We do indeed. Yeah. It's good. It's good. But now you're back and it's still beautiful weather out there anyway. So it's... Who cares, right? I mean, still a...

Dave Jones: That's right. And you're just shoveling snow and they're setting fire to the train tracks. That's right. Oh my God. You saw that? I saw that.

Chris Gammell: What the hell? Yeah, so it is a historic cold here. I was outside earlier. It was minus 20 Fahrenheit. So that's about minus 30, I think, Celsius. Before wind chill.

Dave Jones: I don't understand how and why people populated these locations. They get like this. Why?

Chris Gammell: I don't know. I didn't think it was going to be this bad.

Dave Jones: I don't know, as we were talking about before the show, I don't know anyone who actually enjoys living like that, where they have to shovel their snow every morning before they can get their car out or whatever.

Chris Gammell: It's just unbelievable. I don't know what to tell you, man. I just live here. Right.

Dave Jones: Goodness. All the more for working from home, I guess, right? You can just lock yourself inside and just piles up outside. Who cares if you can't open your front door?

Chris Gammell: That is today for sure. Yeah. Yeah.

Dave Jones: Yeah. Because I've actually heard of overseas, especially when we have subsidiaries overseas that are in these locations and stuff. Oh, the companies, no one's at work today because they couldn't get in the front door. It's piled up with snow. Literally, they couldn't get into the building.

Chris Gammell: Right. Well, and usually that's the thing, though. So the places I used to work in Cleveland, it was like they wouldn't shut down unless it got really bad because they were always prepared for it.

Dave Jones: Right. Oh, okay. Yeah, yeah. Of course. How do you prepare for it? Do you have like a company bulldozer or something to...

Chris Gammell: Yeah, yeah. Snowblowers are a big thing and yeah, you have plowing service and it's a whole big thing. I'm sure our listeners, some of the ones in the north can relate. And it gets a lot older than it gets here too. So like, you know, I know I have some friends who listen to the show who are up in Minneapolis and some people further north up in Canada, like Alaska, I'm sure. So yeah.

Dave Jones: Well, all you need is some solar freaking roadways to melt all your snow. You're right, exactly. You're right. Yeah.

Chris Gammell: Yeah. The solar panels probably are really efficient in the cold too. So yeah. That helps. That helps convert that road, that sweet, sweet road sun into energy. Sorry. Anywho. The more things change, Dave, the more they stay the same.

Dave Jones: Speaking of which, has anything changed with you? Are we still doing the same, working on the same stuff?

Chris Gammell: A lot of the same stuff. I started doing that, I think that part I mentioned, that Laura chip, I started working with that and that's been good. I've been doing actually some dev boards and getting a couple different dev boards up and running. I've been fighting the good fight with a lot of getting started guides and stuff like that. It's still as bad as I remember.

Dave Jones: But they shouldn't be. I mean, you know, that's the whole idea of these dev kits and getting started guides. You're supposed to just be opening them up and any dummy's supposed to get it working.

Chris Gammell: Right. Well, you know, I'm a special breed of dummy, Dave. Join the club. Yeah, right. Right. I think one of my problems is that I always kind of get psyched out as well. So I'll be like, ah, it's something I'm doing. It's something I'm doing. And so this latest one, I was doing a Bluetooth dev kit and it was like an NRF 52 dev kit or something like that. Yeah. And Alicia from the Embedded Podcast was helping me actually troubleshoot. It was really great. And at the end of the day, it was like I was struggling and struggling and struggling and trying all these different things, trying way too many things, honestly. And at the end of the day, it was because I had the board was labeled in the wrong bag, like the wrong bag. And I didn't actually look at what the part number was on the board. And so basically, I was just trying the wrong setup the entire time. But the whole time-

Dave Jones: You were trying the entire wrong board?

Chris Gammell: Yeah, exactly. So it was like a- How does that work? Well, because it was like an NRF 52 832, right? Yeah. And I thought it was the 840, or 52 840. Oh, okay. Oh, okay. And the idea is basically because they compile differently, I kept trying and trying and trying. And I kept- The thing, the problem was that my self-talk was, I'm doing something wrong. I'm doing something wrong. Yep, yep. And the problem was, it was just like much more foundational than that. You know? It's like that needing to take a deep breath and step back and be like, okay, what's- What the hell is actually going on here?

Dave Jones: Make sure I've got the right device to start off with. Yeah.

Chris Gammell: And of course, once that happened, it was like, yeah, then it worked fine. You know? But it was because of that self-talk, you know? And for people listening, too, I don't know if other people experience this stuff. I psych myself out so often where I'm like, it's something I'm doing. It's something I'm doing. Yeah. I mean, I guess I didn't check the board, so it was something I was doing. But it was not, I don't know how to follow a guy.

Dave Jones: No, but it was more, yeah, it was more fundamental. Yeah, exactly. Right. You know? Yep.

Chris Gammell: Yeah.

Dave Jones: So, I can remember. Once that, I was on a website, can't remember what it was, you know, and you've got to enter in all your details. And it had, you know, date of birth or whatever, you know? And you put in all your details. And I couldn't figure out why this website wasn't working. It would just say, incorrect information, incorrect information. And I'm trying it like 50 times. Like, and I had to do this. I couldn't abandon it. It was absolutely essential to do it. Yep. And, like, it literally took me like half an hour to realize, and through all this other troubleshooting and trying different browsers and doing all sorts of stuff, that I realized I was entering today's date instead of my birthday.

Chris Gammell: I've done that.

Dave Jones: Like, my brain just got in that mode.

Chris Gammell: Yep.

Dave Jones: You know? And, like...

Chris Gammell: Dave, come on. We were born yesterday. Right. I'll pay that.

Dave Jones: Yeah. Yeah. Oh, man. Yeah. And it's just, yeah, something fundamental like that. But then, you know, my engineer brain kicked in and just started thinking about all the other things.

Chris Gammell: I wonder if the server's down. I wonder if it's this. Yeah, yeah. You know, like... Yeah, yeah, yeah. Yeah, exactly. You go over the top, and sometimes it's just that needing to step back and... Yep. Wow. But, you know, you always learn something out of it. And, uh... So, I definitely learn a lot. I do.

Dave Jones: I always remember that. I always... Yeah. It sinks into your brain.

Chris Gammell: Right. Well, and I think when you do struggle... I mean, like, you know, I'm not a huge fan of the struggle, right? But, like, of, like, going through a... Yeah, yeah, yeah. A dev platform and learning... Versory and... Yeah, well, and, like, just, like, learning all the different... Like, because, you know, you can read the cursory overview of, like, a dev kit and be like, all right, click this, click this, click this. It works. And then, at the end, you have a blinking light, and that's great, and you've, like, understood it. But it's kind of like what you talk about with, like, the I hope your project doesn't work. You know, it's like, if it doesn't work at first, you're definitely... You're going to read the documentation a lot closer, because you're like, all right, where did I mess this up? You're going to try a lot more things. You're going to... And then, you know, at the end, you're probably going to have a little bit more dopamine, because you finally did get it working, even if it's something... Of course. ...kind of stupid. But out of that, you know, so I was asking around then, because I'm curious, too. Like, so the reason I've been doing a lot of different platforms is I've been... And maybe I've talked about it here, is, like, I'm just trying to get tools in my toolbox. Like, basically... Sure. I'm trying to build it up to the point where, like, I need Bluetooth, it's this one solution. I need, you know, Wi-Fi, it's this other solution. Like, I don't want to have to think about it. And, like, I think we've talked about before, too, of just having that one... Having that platform for that thing, and then you pull it out of your back pocket, like, yep, here we go, ready to go, no questions asked, because it's rock solid or whatever. Or I'm just trying to build up that mental model for myself.

Dave Jones: So that's why I'm trying all these things. Because you need to, because you're in the contract design business.

Chris Gammell: Exactly. Yep. So, yep. And, but I was asking other people, like, what their favorite stuff is, favorite part families and dev boards are these days. Let me send you this. I think it was a tweet. I'll send you this link. But it was great, because, like, a lot of people were responding about what they're working on. And, you know, it's good to kind of just see what's new and what's out there. So there's a lot of people who are really excited about the ECP5, which is that new FPGA family from a lot of the opened FPGA tools. There's a lot of people talking about that. Hang on.

Dave Jones: Who's the FPGA vendor?

Chris Gammell: That's Lattice.

Dave Jones: Oh, oh, it's Lattice. Okay, right.

Chris Gammell: Yeah, so this is when we had Peter, who did the remote correspondent thing for us over the holiday. So he was talking to Clifford and Dave, and they were talking about a lot of these. So basically, the new tools that Clifford Wolf was working on, and Dave and a bunch of people are working on, basically, there's new tool chains that enable these FPGAs to go through. I still haven't looked into the MyGen, M-I-G-E-N. But I think that's like a generator tool for FPGA stuff, right? So it's kind of like a higher level, yeah, Python toolbox for building complex digital hardware. So basically, it's...

Dave Jones: So what's a Python to VHDL compiler or something?

Chris Gammell: I think Verilog. Yeah, I think it's basically, it's like a scripting thing that allows you to do that. Honestly, I literally just looked it up, but it's FHDL, whatever that means, functional HDL.

Dave Jones: Well, there's C to Verilog compilers as well.

Chris Gammell: Yes, that's right.

Dave Jones: That allow you... Altium has one of those that allows you to just write your C code and then literally convert to hardware. Yes. You push a button, convert to hardware, and it may not be optimized, but it's...

Chris Gammell: Right. I think this is even higher level than that because it's just like, I want to do like a math function and put it into... Make it hardware, right? Sure. To really accelerate this thing. So a lot of people talk about that.

Dave Jones: And some... But it doesn't work for everything. Some things lend themselves to the parallelism nature of FPGA architecture. Other things don't. Right.

Dave Jones: You're better off just slamming it through a fast pipeline CPU.

Chris Gammell: Mm-hmm. Yep.

Dave Jones: So...

Chris Gammell: Mm-hmm. Yeah. So let's see what else was on this list. So a lot of FPGA stuff. A lot of the... You know, there's a lot of all-winner type PCBs these days. There's actually one that's on the list that I had for us to talk about because it's like... There's a new one that's out. It's actually using RISC-V 64-bit processors in it. So really crazy. Another super cheap processor, CH554. That's another one that's kind of like that three... What was the three-cent microcontroller you did?

Dave Jones: Yes. The... Paduk.

Chris Gammell: Paduk. Paduk. Okay.

Dave Jones: Yeah. If I'm pronouncing it correctly, I don't think so. Yeah. Yes. Paduk. Yeah. There's actually an active thread on the EEV blog forum, actually. Oh, yeah? Hacking... Actually reverse engineering their programmer.

Chris Gammell: Mm-hmm.

Dave Jones: Because they... I actually spoke with the manufacturer. Oh, that's right. They reached out to you. And they said no. Yeah. Yeah. We... No. Sorry. We aren't going to release our tools. We aren't going to release the programming algorithm for our chips. You know? So there's a lot of people out there trying to reverse engineer that just for kicks. You know? Because they like the concept of a three-cent microcontroller.

Chris Gammell: Right. It's... I mean, that is crazy cheap.

Dave Jones: Yeah. Yeah. Yeah. But, you know, realistically, for most of the stuff that most people work on, there's not much difference between a three-cent micro and a 20-cent, you know, or a 30-cent pick or apmell one. Right. Right. Right. Yeah. Like, yes, it does cost 10 times more. But in the scheme of a typical BOM, if you've only got one or two of them in your project, it doesn't matter. You might as well just use the 30 or 40-cent pick. Right. Right. Right. Yeah.

Chris Gammell: Unless you're making enough that you're... The upfront cost of, like, figuring that out. Yeah. Makes sense.

Dave Jones: Yeah.

Chris Gammell: Yeah.

Dave Jones: But, you know, how many designs have 100 microcontrollers on them? There's not many, you know?

Chris Gammell: Is that what people are doing with... Like, they're doing... Yeah. Oh, so they're, like, throwing a micro at everything.

Dave Jones: Oh, well... Well, no. They... Well, most of them are just working on it because they think it's cool. Sure. Most of them don't have an application for 100 microcontrollers in the one project, you know? So, I'll tell you what's an interesting thing that we were just working on the other day, and we might do a video on this. It's on our list of videos to do, is that we're looking at the ARM ST Micros, because we're using that for the new micro supply project. And we're looking at the aspect of... We just realized that the chip we chose didn't have internal E2 Prom.

Chris Gammell: Oh, really? Okay.

Dave Jones: And, yeah, like, you know, people, I guess, kind of take it for granted that a lot of micros, in addition to the flash program memory, they have internal EEPROM, where, you know, like a few hundred bytes or a few kilobytes.

Chris Gammell: Like for storing constants and stuff.

Dave Jones: Storing, you know, program defaults when you power off and stuff like that. In this case, it programs which mode you're in. For example, you know, when you change modes or you change voltage on your power supply, you want it to power up at the same voltage, right? So, unless you've got a battery-backed SRAM in... Yeah. ...which is often the case if you've got a separate real-time clock chip, for example, that actually has a, like, has a battery, often they have, like, a few tens of bytes in there that you can use for storing a few little power-on default things. So you don't lose contents when you power off. But most of the time, yeah, it's completely volatile. And some micros, yes, you can write to the internal flash memory, right? So you can use the program memory to store data. But they've got a limited number of write cycles. And it's like, you know, 10,000 write cycles or something. And that might sound like a lot. But if you calculate it, well, if you've got a power supply and you're always changing the voltage, and you're always saving that voltage just in case they turn it off. Yeah. Right? It can be gone in six months, a year, you know? Right. So that's the issue. Huh. And we're actually trying to think of all... You know, if you were stuck with that, if you were really stuck with using the internal flash memory, what, you know, little tricks could you do to get around that? One of them would be to have a roll-in code... Yeah, like a rotating thing.

Chris Gammell: ...which...

Dave Jones: A rotating code, either based on... No, just on the number of write cycles. So if you know it's 10,000 write cycles, and you've got enough program memory, then you can actually have a counter in there as you could write a counter as well. And when it gets to a certain, say it gets to 9,000 or something, or, you know, 10,000, you can say, right, I'm not going to use that section of flash memory anymore because all these things are, all these ware-based things are based on the block of flash memory, right? So you can write to these different... So you don't kill the entire flash memory. You only kill that particular block. So you can actually ware-level, like a lot of SD cards do this. They'll have ware-leveling in them, and you don't know that these SD cards have these little ware-leveling controllers in them that will decide where your data goes on the chip. Every time you download it, it'll, like, spread it across different parts of the NAND flash memory. Wow. And stuff like that. So, you know, which is not... I don't believe that's in the SD card standard, but a lot of ones like Sandisk, they will actually implement these ware-leveling algorithms. And I'm sure you'd get the same thing on solid-state disk drives, you know.

Chris Gammell: Does that mean you need to have, like, more capacity than they actually publicize them? So, like, if it's a 2-gig card, they have 4-gig?

Dave Jones: No, I don't believe so. I know. But you need to, like, I don't know how the algorithms exactly do the ware-leveling. I'm not sure, but I'm fairly sure that they don't actually require any significant amount of extra memory. Maybe there's a small overhead or something.

Chris Gammell: Yeah, okay. To actually do it, you know. Right.

Dave Jones: Yeah, just for, you know, storing where things are or something like that. But I don't believe it's large. It's not like your 2-gig card's going to go to a gig and a half or something. Right. Because it's got ware-leveling or whatever. So, yeah, so we thought about, you know, different things about implementing that and stuff like that. But the thing that, the reason I brought this up is because, right, it doesn't have internal E-squared problems. So we went, oh, look, can we upgrade to another model of micro that has internal E-squared problem? Because ST, these STR micros, right, they make, like, 200 different variants, right? Yeah, right. It's ridiculous. And sure enough, even in the same family, if you get the slash L, like the low-power version, it's got E-squared problem. But the regular one doesn't. They're identical chips. Yeah, yeah. And the low-power one has E-squared problem, probably because low-power devices have to store stuff, maybe. You know. I don't know. Anyway, it seems silly. But the price difference is, like, 40 cents.

Chris Gammell: Yeah.

Dave Jones: Between the exact same chip, but one that has E-squared problem and one doesn't, right? That's a lot. Amen. 40 cents. And we went.

Chris Gammell: I bet you're the only person doing this, though, right? I mean, like, that's probably they figured out where that pain point was. Like, yep.

Dave Jones: Right. That's worth it. Yeah, exactly. And then, sure enough, like, we didn't even have to go to the Shenzhen market to get a cheap external E-squared problem. You just go to DigiKey, you can buy them for four cents.

Chris Gammell: Oh, really? That cheap?

Dave Jones: Yeah, that cheap. Oh, nice. Like, as in Atmel ones. Atmel slash microchip ones. They've made E-squared problems forever, you know? And, yeah, even those on DigiKey are only four cents for an external E-squared problem. So it's cheaper to just get an external E-squared problem chip than it is to upgrade the micro.

Chris Gammell: Yeah, that makes sense.

Dave Jones: Bloody ridiculous.

Chris Gammell: I was going to ask if there was other chips out there that might do it, too. I know, like, I remember I always thought about FRAM, but that seemed like that never actually came down in price.

Dave Jones: Oh, well, no, no, no, you can buy FRAM. David's a big fanboy of FRAM. Right, but it's not cheap. And, yeah, they've got essentially unlimited write cycles. And if you had to, if you were writing, you know, a hundred or a thousand times a second for some particular system reason, then you'd be using those. Right. You'd be using FRAM. You wouldn't use E-squared problems. E-squared problems still has, even the best, still has, you know, a million write cycles or something like that.

Chris Gammell: Yeah.

Dave Jones: But it's certainly better than 10,000, you know, that you might get in a micro. So, in the program flash memory. Right. So, yeah, that's, and, of course, micros don't need a lot of write cycles. They just haven't optimized them because how many times do you have to reprogram your microcontroller? Not many, right? Not if you're doing it right, yeah. A few times in its, yeah, exactly. And, yes.

Chris Gammell: Cool. That's great. So, there you go.

Dave Jones: I thought that was interesting that it was 10 times, I thought, oh, no, just buy the micro.

Chris Gammell: Right.

Dave Jones: That has a built-in, but, you know.

Chris Gammell: How much are you guys playing on price with that thing?

Dave Jones: And this isn't.

Chris Gammell: Like, is price a. What, final price? Is, like, well, is, like, the 40 cents, would that have been, is that enough where you were, like, rather the expensive cycles?

Dave Jones: Oh, yeah, yeah, we're trying. Sure.

Chris Gammell: Yeah.

Dave Jones: Yeah, yeah, yeah. We're trying to trim. You know, we're already adding so many bills and whistles, it's just getting ridiculous. You know, and everything adds up. Yeah. So, yeah. Yeah. You know, the last thing I wanted was another 40 cent bomb item.

Chris Gammell: Yeah.

Dave Jones: You know, so, well, extra 40 cents on the bomb. So, I'd rather have another part, physically another part, and then the extra pick and place cost, which is almost trivial, you know. Mm-hmm. Yeah. Or it may not be. Imagine, just imagine if we were at, say, you know, 60 bomb pick and place items on our board, right? And the pick and place machine at the assembler only has 60 real slots available. And then we added one more part and took us to 61. Right. Yeah. Yeah. Well, eh.

Chris Gammell: Yep.

Dave Jones: You need a bigger machine, go to a different manufacturer, or they've got to put the run through twice.

Chris Gammell: Right. That's really expensive. Right. Double setup. Yeah.

Dave Jones: And they'll charge you for that. Yep. And they usually won't tell you any of that.

Chris Gammell: No, they'll just quote a higher price. They won't tell you.

Dave Jones: They'll simply charge you. Yeah. Yeah. Exactly. So, how are your pick and place adventures going?

Chris Gammell: They are okay. I'm trying to think the last time we spoke about this stuff. I did use them for another high-density LED board, and that definitely is the, that's definitely the best use case for me. I did, I did actually.

Dave Jones: Yeah, where you've got like 100 of the same part, you know.

Chris Gammell: Right. I have been talking to a couple. Is that? Yes, that's right. Yeah. So, like 100, yeah. So, that was like, I think 300 of the exact same LED being placed. And then I had like a panel of four. Exactly. So, it was like 1,200 placements in a row. Classic use case. Yeah. Yeah, yeah. And so, in that case, it was like, I think I had three pick heads going, and yeah, it took about, I think, probably 10, you know, five to 10 minutes per board. So, like a total of like 20 minutes per panel. So, not terrible, but like that was also more of an academic thing.

Dave Jones: How long did it take you to set up, do you think, estimate, like all up, realistically?

Chris Gammell: Yeah.

Dave Jones: To set up that, as opposed to just going, bugger it, I'm sitting down, and I'm placing 1,000 LEDs.

Chris Gammell: Well, that one is actually really bad, because I kept having all these issues. So, like the actual pick and place, like the files, like modifying the file, exporting the file, the position file, modifying to get the format that the Neoden wants, and remembering how to do all that stuff. That was probably two or three hours, right, realistically, just because I was, you know, it's not like I'm doing it often enough. The real problem was the first run that I did, I was flinging parts. It was pew, pew, pew, like parts were just flying off everywhere. Yeah, yeah, yeah. And I was like, what the hell is going on? Because the first time I had used that machine, it was fine. And it turned out it was because... Didn't optimize the pressure? Not even the pressure. Not the height, the Z axis? It's the height, yeah, exactly. So, I was...

Dave Jones: Yeah, Z axis, yeah. I didn't realize... Which actually translates into pressure. That's right. If your height's not set correctly, you push down harder.

Chris Gammell: Yeah, you're trying to basically push through the board, and there's spring in both the board and in the nozzle, right? They're both there to basically say, okay, you want a little bit of pressure so that it releases right. I just hadn't set any of that stuff up right. And basically, I had to go back and like set up from scratch almost at that point. And I'm obviously learning more limits of the Neoden machine. It's not, you know, it's not a commercially viable machine in any way, shape, or form. But I did finally figure that out. There is actually a new... For people that are interested, there is a new Neoden 4 video that's out from... Apparently, there's a distributor in the United States now. And so, they made a pretty decent video. They still actually don't show one of the methods properly. So, like, they keep showing, like, how to place for a single board, but they don't actually show how to panelize in the proper way. And so, that's something. All right. But the video is much more inclusive than the prior videos. So, yeah. Got it. I also started a forum. I had a couple people I had email threads with. So, I started a section on Contextual Electronics Forum about that machine. And so, if people want to talk about it, there's a section there now. There's probably about six or seven people that are kind of talking through the particulars of their own machine, how to get it working well. So, that kind of thing. So. Cool. I also learned there's a new Neodent coming out, like a Neodent 7, it's called. So, it looks like it's probably... They actually use commercial feeders. And so, that's a big change and definitely going to be more expensive.

Dave Jones: Right. What? Some other brand feeder?

Chris Gammell: Yeah. I think it was like... It was someone... It was like Casio, but it wasn't Casio. It was someone who's like a commercial brand who was like, oh, maybe Sanyo? Something like that. Maybe Sanyo feeders.

Dave Jones: Wow. Well, is it like a company that just specializes in feeders? Or another, like a Yamaha kind of?

Chris Gammell: Maybe it was... Maybe. I don't know. I don't remember who it was. But the idea was this was a slightly better machine. Definitely more expensive. And I think it has six heads. It says here somewhere. But basically, it's not included anymore. So, basically, it would be another cost. So, I think the eBay price that I'm looking at right now is $12,000 US. So, without any feeders. So, probably closer to $20,000. Oh, actually, no. This one actually contains feeders. So, for $12,000 for like another 12 feeders or something like that. So, it doesn't have a ton of them, but... And it is Yamaha. Sorry. My bad. So, track and handle with Yamaha feeders. So, okay. Got it. Yeah. So, you know, it's... But again, it's not going to be something... Like, you or I might be interested in this kind of thing. It's definitely like still accessible. Or like something like mHub, like the space I work out of. They might be interested, but I can't imagine many individuals would be.

Speaker ?: And...

Dave Jones: No. No. You'd have to be in a unique situation. Exactly. Very niche kind of stuff. Your own personal 20K pick and place machine.

Chris Gammell: Yes, exactly.

Dave Jones: Yeah.

Chris Gammell: So, interesting. Still interesting, I guess. It's... Yeah. So... Yeah. So, yeah. That's my current state of pick and placeness. It kind of... It ebbs and flows with like... So, I've got another version of that board coming in. And I'm just going to hand place it. Because it's not really worth the setup time. I think... I did finally... With the 300 odd LEDs.

Dave Jones: Yeah.

Chris Gammell: I mean, what does that take? Like an hour? Two hours? No.

Dave Jones: Yeah. It doesn't take long. Yeah. Put on a podcast. Especially like a... Because you would paste stencil it, right? Of course. You would paste it. Yes. Yeah. Yeah. Yeah.

Chris Gammell: And... Yeah. That... You know, it always is that balance point. I don't know where that happens. But...

Dave Jones: Although it doesn't take you that long to hand solder either. Right? You know, all you've got to do is go along and tin each row. Like, I've done this. I've done like, you know, hundreds of LEDs. And you just go down and like tin one side of each one and then boop, place, boop. And you can get... You know, you can place one every two seconds or something. Yeah, exactly. It's annoying, you know. It's the main thing. It's like your eyes will get real tight. It's pretty annoying. Your hand gets cramped up and, you know, all sorts of like... You know. Yeah. But you can certainly do it.

Chris Gammell: Yeah. So, it's... Hmm. It's worth thinking about. I don't know. It's... I don't know where that crossover... I don't know where that crossover point is for you. Like, are you guys... When you guys do prototypes, are you going to hand assemble? Are you going to send out? Or what are you doing for the...

Dave Jones: Oh, we always hand assemble.

Chris Gammell: Yeah. Okay.

Dave Jones: You know, they're not that complex. It's, you know... It just isn't worth the grief. But we're actually going to do... With the micro supply, what we plan on doing is actually doing a... Sort of like a pre-production run of like 10 or 20 of them. And then, you know, send them out to people. So, they can use them and give us feedback and stuff like that. And we wouldn't do those by hand. We'd obviously, you know, get a small run done.

Chris Gammell: You know, I've been thinking more and more. Like, obviously, I've been using a lot of the low-cost PCB houses that are overseas. And I've just been thinking about it. Like, their costs are worth looking at, at least. I mean, not right now. If you're listening to this, it is Chinese New Year. Happy New Year to anyone in China listening. Yeah. Don't expect anything for another couple weeks. I definitely was, like, racing to get a board in at the end of last week. So, we'll see if I actually get it in the next two weeks. But, yeah. I mean, I need to... That's on my to-do list as well is to just kind of give that a shot. Because, like, the costs aren't that bad. And there's, you know, there's a lot of the online ones like PCBNG and Macrofab and Circuitab and all the ones we talk to.

Dave Jones: Yeah, see PCB do assembly. Yeah. A lot of them do. All the big prototype houses have assembly services.

Chris Gammell: Right. So, yeah. It's just more of, like, that control freak inside of me, you know. Yep. And, again, that calculation, like, you're talking about of, like, what's the... You know, where's the balance point of, like, is it just faster for me to order it and do it myself or whatever? So...

Dave Jones: Right. Yeah. It's probably faster to do it yourself. You know, once you've got the parts, you just sit down for even if it takes you a whole day, it doesn't matter. Right? It's still going to be faster.

Chris Gammell: Right. Right. Right. And an hour... Guarantee. A day of my time is, like, how much is that worth in terms of, like, the assembly cost? Eh, you know, maybe the assembly would be faster. But then it's, like, even if you're doing the assembly, you've got to, you know, make sure all your files are good. There's no, like... There's no backsies. Right? But if I order the wrong... If I order a 10K resistor instead of a 100K resistor, I can fix that on my bench by digging through my parts and being like, oh, yeah, of course I got a 10K. You know? Or, you know, bodging something or whatever. If that happens elsewhere, it's like, oh, well, let's have a meeting or let's do an email.

Dave Jones: They'll just stop and they'll email you. Yeah, exactly. And then it's in the middle of the night so you won't get the email until the morning. And it's like, yeah, no.

Chris Gammell: Right. So, yeah, maybe it's good for those 10 and above kind of things. But it's still that valley of death, whatever they call it. Yeah. Right.

Dave Jones: Generally, any such issue that comes up, you can add two days to the time. Oh, yeah. Yeah. You know, that's like, you know, that's just the way it is. With communication turnaround in different time zones and stuff, you can, you know, easily lose, you know, two days.

Chris Gammell: Yep. Yep. So... It's good to get it right that first time. That's for sure. Yep. As I will find out if my small change on my board, if it shows up before or after Chinese New Year. We'll see.

Dave Jones: Because often, you know, they don't... The assembler, for example, may not, or the PCB manufacturer or whoever, whatever, or assembly service, wherever you're getting done. They may not, like, they may not encounter the issue and then immediately email you within minutes, right?

Dave Jones: They do it at the end of the day or whatever. And they'll do it at the end of the day. They'll move on and, you know, because they're a production house. They've got to keep the production going. That's their number one requirement. So, yeah, at the end of the day, when they do their housekeeping, oh, yeah, we had an issue with that board and we had to stop. And then, you know, and then you add in there. That's why I'm saying two days minimum. Right. Sort of turnaround.

Chris Gammell: I'm sure that I'm at the top of their list, but maybe not, you know. Right. Yeah. I have nothing. Oh, well. Oh, dear, I do. Anyway. I did want to bring up one other board from this list of the people tweeted back at me. I've lost it now, of course. Of course you have. Yeah. But I also posted it to the forum. Basically, it was, yeah, here we go. The Cypede. Oh, there's one I mentioned. The RISC-V64. RISC-V, whatever. Sorry. Okay, here we go. It's got, this is a dev board. It's got a camera. It's got a dual core 64-bit RISC-V processor on there. It's got a 2.4H LCD and a bunch of inputs and stuff. $18. $19. $18.

Speaker ?: $18.

Chris Gammell: $18.

Dave Jones: $18.

Chris Gammell: Holy crap. So this is on Seed Studios. All right. But the reason I've been interested in this is actually not related to electronics as much as my holiday reading list. I've been reading a bunch of Yuval Noah Harari books. I don't know if you know who that is. He wrote like Sapiens and Homo Deus and stuff like that. But he keeps harping on and on and on about AI. And so I got real, I got real worked up, Dave. Let me tell you. I got real worked up. Tell us why. Because it's, you know, it's, you know, like obviously it's a futurism book and it's like, you know, nothing's going to be as bad as they say, but also it's not going to be as rosy as people, as naysayers are going to be. You know what I mean? Like, so it's going to be somewhere in the middle. And I just like started thinking about like, oh, well, maybe I should learn about this stuff. And like from a, you know, a good place, I think, like I actually do want to learn about this. And this is what I wanted to say here, because I think it actually has a lot of relevance to people in the electronics industry of like, love it or hate it, like machine learning and whatever AI version of stuff that's out there, it all wants like sensor inputs. And that's kind of like the future, like basically any big data platform or anything like that, a lot of what we're going to be doing in the future, I think, of the electronics world is making, making data that's digestible by the systems that then makes decisions based on them. You could say that that's always been the case of what we're doing, but I was like, all right, well, I'm going to learn about what's actually happening in these sensors, or sorry, in these systems so that I can make better sensors for it. And then that kind of put me down this rabbit hole. So one thing, this, this system was very interesting to me, the CIPED, because it's basically on board processing with like GPUs and high end chips and stuff like that. But then the other thing was I started taking a course and I thought that was kind of interesting too, about machine learning. And it's actually very accessible. It's called fast.ai. Have you ever heard of that?

Dave Jones: Right. No.

Chris Gammell: It's basically like learning how to code up algorithm, how to code up algorithms, I guess, but not really, but like how to do, how to do machine learning really, really easily. And it's pretty, it's pretty cool. And I, so I would, what I'm basically getting at is I would recommend that people that are interested in doing the electronics industry side of stuff start to learn this, maybe alongside me, if you're interested, because I think it's important to understand how, how a lot of the machine learning is happening and then how it's going to be better to, how you can make a better product that then can be digested by a lot of these algorithms.

Dave Jones: Maybe, but I like, I wouldn't recommend everyone just go learn it because it like the, the number of practical everyday applications where it's going to be practical to implement machine learning is, is still going to be small, right? It's not like, oh, your latest little pick widget is going to use, you know, you're going to put machine learning in it. Oh, that's true.

Chris Gammell: Yes, that's right. Yes. No, yeah. And then this is actually, so what I'm really talking about too, is like, um, thinking about how, how you collect and bundle up sensor data and then probably send it back somewhere. So that's really what I really mean. So yeah, you're right. You're not going to put anything. I totally agree that like putting machine learning down or like putting like high, high levels amount of high amounts of processing down onto like a battery powered platform. You're kidding yourself. You know, like that's, it's maybe possible, but you're going to have a big battery and it's going to suck. Right. But, um, I'm just generally interested in, uh, how the sensor side of things is going to end up impacting all this stuff. So one of the things I was thinking about is like, um, is it going to be possible to maybe take like an SDR or something similar? Maybe this is already happening, but take like an SDR, put, uh, put like a software defined radio for those playing long time. And put it in front of like, so you take like a hacker ref, basically you have a, uh, you know, a 2.4 gigahertz device in front of it. And then you basically start to teach algorithms to detect what's going on there. That may already be happening, but like basically the idea that this course is starting to talk about is like even just taking visual processing stuff. A lot of that can be translated into, um, into learning patterns using a lot of this high level processing stuff. It's all happening on like servers and, you know, like GPUs and all that other crap. But I don't know.

Dave Jones: I, this brings me to a point where you've got to be careful in engineering that you don't implement technologies and stuff just for the sake of it. Sure.

Chris Gammell: Yep.

Dave Jones: You know, like sometimes like just keep it simple, stupid.

Chris Gammell: Yeah. I mean, if you're turning us, yeah. If you're like flipping a relay, right. It's not like you need some of this stuff, but I think that there's going to be more, I think there will eventually pop out some practical applications. So this is kind of more of a gathering, gathering tools now for, you know, potential implementation later, that kind of thing.

Dave Jones: And people will do this in software too. You know, all they need is like a hundred lines of code to, you know, for their little pick based widget. Right. Oh, yeah. But then they implement the full software engineering cycle of, you know, so that it's, you know, like a thousand lines long and they spend, you know, two weeks on it instead of just writing the hundred lines of straight code. Right. You know.

Chris Gammell: Yeah.

Dave Jones: Yeah. They, they do it all software engineering. If you know what I mean.

Chris Gammell: Yeah. Like a lot of overhead and, and a lot of.

Dave Jones: Yeah. There's a lot of overhead as the, yeah, is the term I'm thinking of. Yeah. Yeah. So anyways, I was interested. Oh, no doubt. Get criticized for that. Oh, but what's wrong with that? It's going to troubleshoot. Extensible. Whatever. It's extensible. Right. Sometimes it's not just write your bloody hundred lines of code and inline code and be done with it.

Chris Gammell: Yeah.

Dave Jones: You know. Yeah.

Chris Gammell: So, I mean, I obviously got, I was a bit secure just talking about it, but I think this board's interesting because it is kind of like, I think it's going to basically push in both directions, right? Some people are going to be like taking sensor data and then immediately piping it up to a server and doing stuff on the server. And then they're going to try and push stuff down into the devices. And that's kind of like what this, this, this, this side Pede thing is. And there was actually, I saw that Espressif has a new board too. They have a camera based board as well. So, basically the same kind of, it's like really simple processing, but it's basically you can figure out where a face is and then it can kind of track it. It has a camera built in. You've seen this thing?

Dave Jones: I've seen, there's lots of people doing this. Yeah. Oh, this is like a dev board though. Yeah.

Chris Gammell: So like, yeah, this is a $10 ESP32 development board. So.

Dave Jones: Yeah.

Chris Gammell: That's kind of cool too.

Dave Jones: Hmm. And a lot of them are targeted around running the, what's the, God, I forget the name. Cause I was going to work on this years ago. I was looking into it. Um, the open vision system. What is it? The open CV. God, there's a name for it. Yeah. Open CV. Is it? Yeah. Yeah. Yeah. Computer vision. Open computer vision. Yeah. Yes. Yeah. Exactly. Yeah. Yeah. Cause I was a years ago, quite many years ago, I was going to like, I was going to do one of these, like a little module that had cameras and FPGA and, or whatever, some super whiz bang micro that was running open, uh, open CV. Right. And, and then it would, you know, it'd be pre-programmed so that it could just do generic things that you needed to do. And it'd give you a simple output, you know, does this, does this recognize motion?

Chris Gammell: Right.

Dave Jones: Output pin goes high. You know, does this recognize a circle? Right. Output pin goes high, you know, stuff like that.

Chris Gammell: Is this a hot dog? So. Is this not a hot dog?

Dave Jones: Yeah, exactly. You know. And yeah, that was the dream, you know, but yeah. Every man and his dog's working on that now.

Chris Gammell: Yeah.

Dave Jones: Yeah.

Chris Gammell: Well, I think. Which is good. Yeah. Yeah. Yeah. I think it's, I think it's interesting. I think it's interesting to see where this is going to all go. It might go newer. I don't know. But I think the cheapness of it is usually like a harbinger of like, oh, well, it's really accessible now. So people will probably try it for a lot of things and then a couple will stick and yada, yada, yada.

Dave Jones: But the hardware is cheap, but the software isn't. Right. The software pain is not cheap. Right. Yeah. You may be able to buy your $10 module, but if you've got to do months of software development to get it to recognize a freaking banana. Yeah. Right. Whereas if you can buy a pre-programmed module that you can just like set a bit, which says recognize banana and the output pin one goes high when it recognizes the banana. Right.

Chris Gammell: Right. Right.

Dave Jones: Right. Exactly. Right. That's what I want. If I'm developing a system, I don't want to spend.

Chris Gammell: Their branding is going to go really sideways when they call it the banana eye and they're like, oh, the banana pie. Right. Yeah. Oh, I love it. I think the big setup stuff like you're talking about, I think one of the differences is going to be like, so like comparing the CYPED one and the ESP one. So one's actually an embedded version and one's an actual Linux version. And I think the Linux version are going to win. I think that obviously the power differences are going to be significant. Right. But basically being-

Dave Jones: But you just think the added flexibility is going to be the killer. Exactly.

Chris Gammell: And like having like pre-compiled solutions where it's like, you know, people basically being able to step in and say, all right, well, I know how to work in Linux environments elsewhere. Now I can do this on an embedded basis. Right. I think honestly, just like the idea of having more, having platforms that are more accessible for software people is going to, that always kind of changes things. I just-

Speaker ?: Sure.

Chris Gammell: Yep. Yep. Yep. Yep. Yep. So totally agree. So it'll be interesting. You know, 2019 will be an interesting version of that. I still haven't figured out the frigging battery thing though, right? I mean, like that's always what it comes down to is like- Power. Yeah. Density. Exactly. Power's a killer. Yeah. It really is. Batteries still aren't getting better, you know? But-

Dave Jones: Well, people forget how much power a Raspberry Pi takes. Yeah.

Chris Gammell: Right.

Dave Jones: You know, like you think, oh, a Raspberry Pi, use that for any embedded application. You know, five volts or two and a half amps, thank you very much.

Chris Gammell: Exactly.

Dave Jones: You know? Right. That's freaking 10 watts. Yeah. Right?

Chris Gammell: Yep.

Dave Jones: It's nuts.

Chris Gammell: Yep.

Dave Jones: That's a ton. That's a metric buttload of power for an embedded system. Yeah. That's just nuts.

Chris Gammell: It's not bad for a day on a battery, you know? But a week, a month, a year, you know? Like, yeah. Good luck. It's not going to happen. Right. Yep. Well, should we go to the list?

Dave Jones: We have to.

Chris Gammell: We have to.

Dave Jones: Now, I'm jealous of this Hackaday article. Yeah, which one? On circuit sculptures.

Chris Gammell: Oh, yeah. That wasn't that great.

Dave Jones: I've always wanted to do this. Yeah. I've always wanted to do this. And I wish I had the talent to do it. Yeah. If there's one thing I wish I had the talent for is to do some sort of sculpting. Yes. Something like, you know, arty kind of stuff.

Chris Gammell: Yep. Yeah, this is a visual one. People have to go click over to it. Yeah, yeah.

Dave Jones: You've got to click.

Chris Gammell: I will also give a shout. I don't know if Mohit did it, but Mohit is, I think I'm saying his name right. He's one of the, he actually works for Particle, and he does a lot of their board design. But he, if people don't follow him on Instagram or Twitter, he does, I think that actually inspired a lot of this stuff. And some of his creations are just unbelievable. So that's another one.

Dave Jones: Does anyone in engineering use Instagram?

Chris Gammell: I used to. I've kind of cut it out of my life. You used to. Right. Okay. But I think some people do. Why?

Dave Jones: Right.

Chris Gammell: You do, right? No?

Dave Jones: Well, I used to. Like, I set up an Instagram account. I've got like, I don't know, 2,000 or 3,000 followers or something. And I used to post there because then it used to cross-post over to Twitter. But then I realized it was cross-posting lower resolution photos to Twitter automatically. So I thought like, because I'm always posting photos to Twitter, right? It's one of the big things I do on Twitter is I'm posting photos all the time of everything. And yeah, I thought, well, I might as well post to Instagram instead if it doesn't take any extra step, right? And it automatically cross-posts. But it just, yeah, it just wasn't great.

Chris Gammell: Didn't deliver for you, huh?

Dave Jones: It didn't. It didn't actually deliver the, you know, the image quality. You know, if I'm going to post an image, I want it to post a bloody image. Not some 640 by 480 bloody, you know, res stupid thing, right? I want it to post the original quality that I uploaded in or at least near enough to it. Okay, it might compress it a bit, but I don't want it to resize it down to bloody nothing. So, yep. So, oh, yeah, I see. Oh, I like that. Yeah, okay. Sorry, I'm just looking at some of the circuit art. Yeah. Yeah, yeah, very nice. I can appreciate that. Excellent. It's beautiful.

Chris Gammell: Yeah, there's a lot of good stuff out there. And, oh, I guess we should probably describe what it is. So, basically, it's, like, people using the connectivity points of, like, you know, either, I guess, gold bars. These are all through-hole bars. None of this service-bound rubbish. Well, some of it's, like, modified, but, yeah. But, basically, using the interconnects to actually form artistic links between them.

Dave Jones: So, yeah. Yeah. They look fantastic. You know, the wiring becomes part of the art.

Speaker ?: Mm-hmm.

Dave Jones: You know? Yeah. And, of course, and none of this, you know, PVC-covered wire crap. Right, right, right. You know, these are, like, all exposed conductors running over each other and just making sure they don't touch because these aren't insulated. Mm-hmm. So, they just, you know, yep. Just running over the top of each other. Just beautiful. Mm-hmm. Yep.

Chris Gammell: Yeah. So, speaking of, well, not speaking of, I saw your discussion with your brother-in-law as well about patent stuff. Oh, yeah? I was just going down the list and my brain froze. Right. Are you guys thinking about patenting something? Where did that come from?

Dave Jones: No, no. He just happened to be here. We were just hanging out, you know, because the wives booted us out. So, we went, oh, yeah, we'll go hang out at the lab. Yeah. And I just, I was going to, I had just shot, I was about to shoot a little, you know, which was going to be like just a two-minute thing on the difference, because I did a teardown. And the difference between a utility patent and a regular patent.

Chris Gammell: Mm-hmm. Yeah.

Dave Jones: Right? And then I thought, well, I've got a patent attorney here, may as well sit down, and then half an hour video later, you know, we just started rambling on about.

Chris Gammell: As one does, yeah, yep, yep.

Dave Jones: Yeah, yeah, exactly. So, I just uploaded that as a separate video. So, there was no, there was no intention, you know, it just started out, what is the difference between utility? I thought I'd include that as a quick segment in my video, you know, like five minutes worth, and nah, nah, half an hour later, we're talking about all sorts of crap.

Chris Gammell: Yep. So. That'll happen.

Dave Jones: Yeah.

Chris Gammell: Yeah, I liked your video about the, I was watching the one about how to be, it was kind of like a recap video, but like how to get into engineering. That was good, too.

Dave Jones: Oh, right, yes, because I was stunned. I did a five-hour live show the other day.

Chris Gammell: That's crazy.

Dave Jones: Which is, which is nuts, I, I, like, I just blinked, and like five hours had gone by, I was wondering why I was a bit hungry.

Chris Gammell: I said it was, you know. Pent up, pent up, you know, video energy after your long holiday, but. Nah, nah, I've got that energy every day. Sure, sure, sure. Nah, I can just do that every day.

Dave Jones: What? You don't believe me? You want me to do a five-hour live show?

Chris Gammell: Five hours for a week.

Dave Jones: Five hours a day for a week?

Chris Gammell: Five hours a day for a week. You do that. That'll be my entire. And I will, I will solemnly swear. What's, what's the bet? I will solemnly swear that the chip printer will never become a thing.

Dave Jones: Jeez, that's tempting. Because it's on the bloody list this week.

Chris Gammell: Is it really?

Dave Jones: Yes, the chip printer's on the bloody list.

Chris Gammell: I didn't see it.

Dave Jones: It's down the, did you add it? Where is it?

Chris Gammell: Oh, Zglue. Yeah, this is someone else. Grendel, Grendel Key posted. Someone else added it. Yeah, this is, so this is like a new, someone else posted about this. So Zglue is kind of like a, it's like building blocks for, for doing chip design. It's not a chip printer though. It's basically a, it's a way to piece together blocks. I mean, this is kind of how they do a lot of design anyways, I think.

Dave Jones: Yeah.

Chris Gammell: But you still need to make it at some point. It's not going to happen. Yeah.

Dave Jones: Yeah. Now that was going to be one, I was going to do a video on why Mars, a Mars planet settlement is not going to, is not a thing. Like as in. Oh, like a terraform? Like as in, no, no, no, not, not terraform. But why we, why if your goal to set up a colony on Mars is to just in case this planet destroys itself.

Chris Gammell: Oh yeah.

Dave Jones: There is no way that's going to work. They'll, they'll just die out in 50 years because they can't make. People just don't realize the amount of resources and energy that goes into making even the most trivial product here on earth. Oh yeah. Yeah. Yeah. Right. And then once that thing fails, like, like a chip, for example, you don't know how much it just stuff that goes into making one chip. You can't replicate that on Mars. So if your chips fail.

Chris Gammell: Right. It's a supply chain. That's it. That's it. Right.

Dave Jones: Yeah, exactly. A long distance.

Chris Gammell: You think you have FedEx is in a new business now, right?

Dave Jones: We, we have a billion people on this planet working towards producing stuff, you know? And, and it's just like, you just can't replicate that on Mars. So it's not going to happen. Well, until, until we can replicate it, I guess. You can't set up a colony on Mars. Yeah, exactly. It's, it's going to be 500 years minimum. Wow. Like it's, it's hundreds of years. If you think it's going to happen in 50 years, you are, you haven't thought about the problem hard enough.

Chris Gammell: Well, there is that whole exponential thing, but I think it's a, yeah.

Dave Jones: No, it can't. If you set up a colony, even with the most advanced technology and you left it there, how, when they, when stuff fails in 20 years, 30 years, 50 years, how are they going to replace it? They can't.

Chris Gammell: Yeah.

Dave Jones: They, they literally cannot. It'll devolve. They will devolve into animals, you know, fighting to survive, right? That will devolve back into aliens, you know? It's like.

Chris Gammell: You know, this actually is a, there's actually a good, a good link that actually is tied to this. So someone, someone linked it to me on Twitter the other day, but the history of Japan's electronic industry, basically. Oh. So this is from, these are like 1980s, like the late 1980s.

Dave Jones: When Japan made all the best stuff. Exactly. Yeah.

Chris Gammell: And basically it was like the history though. And it's showing a lot of the, you know, so it's basically still early days of semiconductor processing, you know, relatively, it's not like, you know, but you know, just showing the energy inputs and like, and specifically showing how Japan struggled in post-war days of like, there wasn't as much knowledge sharing. I thought there was a lot more knowledge sharing going on, but basically they were kind of reinventing a lot of stuff based on papers they saw after the fact. And so like, they were doing like crystalline structure. So like the first thing they showed, like they were like four or five years, I think, behind the first transistor where they were trying to do point contact transistors, but they were doing it based on the small snippets of information they got through the ministry.

Dave Jones: Oh, so they just had to reinvent everything.

Chris Gammell: So this is, this is amazing because these videos are actually interviewing the guys that were doing these experiments over and over and over again. It's so good. Yeah. Yeah. And then the thing you'll love is part three is the calculator war. So it starts talking about that and how they, oh yes. So this is, these are really, really good videos.

Dave Jones: I think somebody linked that. I've got to watch it. Yeah.

Chris Gammell: Really good videos. Um, but like, so like to your point about like the BN Mars, you know, you don't really realize another thing they talk about in those videos is just like the, the purity needs for silicon based processing. Right. And, and, and as they struggled doing that, um, and even the most rudimentary ones, I'm

Dave Jones: talking like 1970s chips, right? 1970s density, you know, five microns. Oh yeah. But I'm, this is, this is that, that era of like getting, getting high purity planar.

Dave Jones: But I'm talking like a Mars colony, right? If you go there and you use your new arm processes with their 10 nanometer.

Chris Gammell: Oh, sure. Sure.

Dave Jones: You know, how are you going to, like you can't even replicate five microns fab on Mars, let

Chris Gammell: alone a bloody, you know, 10 micron. Yeah. You'd be, you'd be more worried about like breathing and eating and all the other stuff rather than making a clean room and processing stuff. Yeah. No, I do agree with that. And I think that, um, I was thinking just like, so say you wanted to do, you wanted to do new, new, new silicon chips, right? How would you, how would you refine silicon in, on Mars? Right? Like, how would you do that? It would be very, very difficult. There's like, you were talking about, like, there's just so much energy input, even the amount of energy input. So say you wanted to make, let's say that you wanted to make more full of voltaic cells, right? So basically big PN junctions.

Dave Jones: But that's solvable. That's solvable. We could get a fusion reactor in 50 years. Right. That, that, that could be a thing.

Speaker ?: Yeah.

Chris Gammell: I believe they've been saying it's 50 years away for the past 50 years. So yes.

Dave Jones: Hey, but no, seriously, that is orders of magnitude more viable than a semiconductor fab.

Chris Gammell: You're saying from the energy input perspective?

Dave Jones: Yeah. No, no, no. Just, just from the, just from the technology needed, right? No, no. It's just from a problem to solve, a problem to solve, right? How do you, you know, set up a semiconductor fab on Mars as opposed to designing a fusion energy reactor, right? I think the fusion energy reactor is easier.

Chris Gammell: Yeah.

Dave Jones: Well, I don't know how to do, I've. I think it's potentially easier.

Chris Gammell: I've worked in one. I haven't worked in the other day, so, you know.

Dave Jones: No, no, no. But if I, I, I had to bet which one was easier, I'd be betting on the fusion reactor.

Chris Gammell: Yeah, but I don't understand why, because you're saying that the, the, the constraints of complexity?

Dave Jones: Just because it's, yeah, no, it's like, no, the, the energy, I'm talking about the energy problem, right? The energy production problem is easily, is more easily solved than the ridiculously orders of magnitude more complicated.

Chris Gammell: Okay, it is complexity.

Dave Jones: The energy aspect of production supply chain for something like a chip or even something like the clothing you wear.

Chris Gammell: Sure, yeah, that's true. Or anything else, right? It's just. How about, like, you're not going to be able to ship deionized water, and so you'd have to make deionized water, you'd have to make, you know, even just stuff like alcohol, rubbing alcohol for cleaning things, right? So if you needed a clean environment, you'd have to make every single thing.

Dave Jones: There's hundreds of different things, hundreds of different things you need.

Chris Gammell: Yeah, that's a good, that's a really good point, yeah. I hope you make that video. That actually would be really, watch this video. Yeah, yeah. This might be good fodder into your video, because this was a very well done video.

Dave Jones: Well, I think this, like, if I do the video on that, it's just got to be a rant. If I try and research it, it's a week-long rabbit hole. Oh, sure. You know, it's like, yeah.

Chris Gammell: Yeah.

Dave Jones: No. So either you just rant about it and let the people find out in the comments, or you, if you want to try and explain it with data and facts to back it up, it's a week's worth research. Yeah, yeah. You know, it's just crazy. So I'm not going to spend a week on a video, on a rant video. Anyway. Yeah, back to the thing, back to the video before we tangent it, the video about getting into engineering. That's the one you talked about, wasn't it? Yeah. Yeah. Yeah. I was doing the live stream and I couldn't believe, like, over five hours, people just kept asking the same question over and over again. How do I get into engineering? You know, how do I become an electronics engineer? Not just engineering in general, but how do I become an electronics engineer? So I thought, this is, you know, I just got sick of repeating myself over and over during the five-hour live stream. So I just decided to do a video.

Chris Gammell: That's great.

Dave Jones: On it. And yeah, I think it's things that we've talked about here too, right?

Chris Gammell: Of like, you know, bringing stuff to the interview and I mean, obviously people should go watch it. It was a good condensation of knowledge. I like that. I like that a lot.

Dave Jones: Well, it's more food for thought, really. It's like encouraging people that, no, you know, you don't have to go down some formal route to become an electronics design engineer. Oh, I agree.

Chris Gammell: Yeah, I agree.

Dave Jones: You can get into the business any way you like. Yeah. You know, there's more than one way in.

Chris Gammell: Right.

Dave Jones: So, yeah, it doesn't necessarily mean that's the best way. You know, it's like whichever way works for you. Just to give people, like, there's a lot of people who just felt stuck, you know, that, oh, I can't do my degree. It's too expensive or it's too hard or I'm too old or I'm too, like, I can't do this and that. And it's like, well, you don't have to. There's other ways in. Yeah, I think it's that.

Chris Gammell: I think it's the idea of the first job. And I think that, like, getting that first gig, especially if you don't have, you know, like, it's always that chicken and egg of, like, well, how do I get a job if I don't have experience? How do I get experience? You know, like, it's always that hard part of that. And I think that you're right, though. I think it's about taking on a lot of the work, you know, maybe even pro bono, maybe, you know, maybe, but maybe low cost. Maybe you're just trying to find anything out there. But basically taking on some work. Yep. And, you know, stretching yourself a little bit. And then hopefully that turns into something else.

Dave Jones: Yep. And there's a lot of people out there who made the argument, but you need to get the degree to get your first job. And it's easier if you go through the formal process, get your formal degree, and then you get it. It's easy to get a job. Well, I know a lot of degree qualified engineers who can't get a job, right, who have trouble getting their first job, whereas others who don't breeze straight through and get a job straight away without even trying. You know, it's horses for courses. It's a matter of how much work you're willing to put in and which and how are you going to actually approach it.

Chris Gammell: Well, I always try and encourage people that are, you know, like some people are in it for the science and they want to get like all the way through. They want a PhD. If they know that at the beginning, it's like, okay, well, that's a different thing.

Dave Jones: Of course. There's only one way to do it.

Chris Gammell: But I think that if like people want to be an engineer, I think that like focusing on getting practical experience, getting hands-on, regardless if you're doing like the way you're talking about or doing it through a co-op or an internship or whatever, I think that's really. Yep. Um, you just got to have a focus on like getting, getting real experience as soon as possible. Just even just to know if you like it. Right. I mean, like, you know, or what you, or figuring out what you don't like, because there was a lot of jobs I had where I was like, Oh, don't want to do that for the rest of my life. You know?

Dave Jones: Right. Yeah. Yeah. Of course. Yeah. Some things are like, well, it's a job. That's the thing. That's a reason it's called a job. Right. Because that's something that somebody else didn't want to do. And that's why they're paying you to do it. You know? Right. Yep. Yep. Yep. But, but there's no, like, the thing I hate is when people make an excuse. Oh, but I can't get my first job because I don't have the experience. And it's like, it's a chicken. I hate the chicken and egg arguments. Bullshit. You go out, you do the work and do your own work and get your, make your own experience. Make your own egg. Damn it. I don't know if it's true. Exactly. You know? And it's trivial in today's, you know, well, you know, connected global e-commerce world to do this. Yeah.

Chris Gammell: Right.

Dave Jones: Right? I think, well, I think, you know. It's so incredibly easy to go work on your own project and you don't need thousands of dollars.

Chris Gammell: I agree with that. But I think.

Dave Jones: You know, or you can go help out on some other open source project.

Chris Gammell: I think the important thing, though, is, is like something like you were talking about, though, is like giving, giving like the almost like permission. It's like, no, it's okay to just do your own thing first. Right? I think a lot of people are, I think a lot of the traditional educational model is, you know, there's gatekeepers and you, you know, you apply to get past the gatekeeper. Right. You know, in doing so, you also learn some things. But like, but that's always been, you know, sometimes there are real reasons for that kind of thing. And I, and I think there's some in place for engineering, but I think not as many as people perceive. And I think ultimately that's what you were talking about. You're saying that like, there are gatekeepers. That's what I'm talking about. You don't need to go through them. There are other avenues. Right. Exactly. There's a way around.

Dave Jones: You don't need to go through them.

Chris Gammell: And I totally agree with that.

Dave Jones: There are benefits to going through them.

Chris Gammell: Yeah. I think, I think there are some opportunities that are not available unless you go through them. Right. So if you want to go, if you want to go work in, and I think like, you know, the majority of people that are in those roles will have gone through that, that main path. Right. So like, if you want to go be like a, you know, a chip designer, right. Most of the time, you know, you're going to go to a university, you're going to go through, you know, VSI classes and do that stuff. But sometimes you're going to be Jerry Ellsworth and you're going to, you know, like, but Jerry is the exception in that case. Now, both are possible and, you know, but, you know, they're definitely just, they're just different paths. Right.

Dave Jones: Yep.

Chris Gammell: So, yeah.

Dave Jones: Agreed.

Chris Gammell: Because I did a video.

Dave Jones: That's right. Yeah. Of course.

Chris Gammell: Yeah. Of course you agree. I agree with you agreeing. Excellent. Um, there was an, there was another, uh, uh, there's a link to one of your videos. I didn't actually see the link, but the, uh, but I really liked the, the original thing for it. So basically at the bottom of this PCB checklist.com, um, there's a link to your number 1129 creating nice readable schematics. So that's the one. Um, but this is a new, a new website that just popped up. Um, yeah, PCB checklist.com. Basically it's, you can go and contribute to it. So if you have stuff that you're like, Hey, this should be on the list. This, this person, uh, I think it's Heinrich Heinrich, Heinrich, uh, is, uh, maintaining this list of, uh, you know, what should you be doing before you send out your PCB? And I think this is great. I've, I've actually, I wrote about this back in the day. Uh, and I was looking for something like this and I'm really glad this is out there. And some of the stuff is, uh, simple, right? So like, are there pull-ups on an open collector? Right. I mean, like, are there, are you within your, like sometimes just like that there's a book that, that was a couple of years ago. It was like the power of checklists or something like that. But basically, you know, like when a, when an airplane pilot, they don't need to know that the flaps, you know, that they assume that the flaps are working. They know how flaps work, but the reason they're doing is because the power of it is to check each and every time. And you have, you know, people, multiple people checking in as part of the process. It's part of like the routine. And that's, that's the really important thing. And I can tell you the amount of times that I've forgotten, like, uh, putting a power LED on the board, really stupid stuff. But if it's on a checklist, you know, and I go past it and, and I get through, I get through to my prototype and I'm like, where the hell is the power LED? That's on me. You know? Yeah.

Dave Jones: Yeah. The, the problem with a generic list like this, and it's already starting to show is that it's so massive that you can't possibly go through the checklist because it's, it's virtually, you know, it, it becomes infinite.

Chris Gammell: Right. It becomes infinite. But I think this is a good starting point. So people don't want to, you know, they want to take this. So it's on GitHub. You can go and fork it if you want. You could start your own, that kind of thing. Um, but I think that's a good start.

Dave Jones: It'd be nice if there was some way, I don't know if this is possible. I'm just, I'm just floating this idea of that. You put in like, maybe you could sort it by like a, like what type of design are you doing? What's it? Oh yeah. What's important to you? Is this a, you know, like, yeah. How, how, how critical is this board?

Chris Gammell: Right.

Dave Jones: Like, you know, I've, I've worked on boards that are like, you know, 16 layer gigantic, like 400 millimeter by 400 millimeter boards that literally are the entire, the entire $10 million project hinges on that one board.

Chris Gammell: That's fun.

Dave Jones: Right. Like, I am not kidding. Right. Yeah. It's terrifying. Right. That has a much different set of requirements than. Yeah.

Chris Gammell: Two layer nothing board that you send out to.

Dave Jones: Duino hat board that, you know. Yeah, yeah, exactly. Right. But so there's, if there's some way to sort of like sort the importance, but I don't think there is. Once again, I think there's, once again, a million permutations of that. Of course. Right. So.

Chris Gammell: I just like it because it's a good starting place for, like you could probably go through and cross that. Of course. You've got to have something. I'm not doing a high speed design. I don't need to do, you know, 3.4. Yeah. Yeah. Or whatever through 0.6 or something. Yeah.

Dave Jones: So, eh, it's interesting. Yeah.

Chris Gammell: Yeah.

Dave Jones: All right. Well, I think that's our show, is not. I think so. I mean, there's. Oh, speaking of which. No, speaking of which, there's a last minute one. Bill, a friend of the Amp Hour, Bill Hurt. Oh, yeah. He did a video for Hackaday. High speed design. Testing myths of high speed, PCP design.

Chris Gammell: That's great.

Dave Jones: I think he was wrong on one point, though. And I did post in the comments. Sorry, Bill. He said that there's a difference in the impedance whether or not the trace goes horizontal or vertical on the board because of the glass weave. And I don't think that's correct. Because there is a difference. There is a difference in the positional location of the trace. Because if you've got a loose glass weave board, because they're cheaper, right? If you don't know, a typical, you know, FR4 glass PCB. FR4 is the flammability rating, but, you know, that's the generic term we use. Anyway, a board is made of basically glass and epoxy, right? Which is, you know, a glue. So they combine this glass weave, which gives it its rigidity and its, you know, strength and its dimensional stability and blah, blah, blah. Right? And its warp, you know, its inability to warp and all that sort of stuff. So this glass weave, so it's like a fabric and they weave it in both directions. And then they encase the whole thing in epoxy, basically. And that's how they make your board. Well, if, you know, the glass weave costs money, right? It'd be cheaper if they just poured the epoxy, right? Without the glass in it, it'd be cheaper. But, of course, you add glass, adds all the stuff. So what they do is they do, like, on cheap boards, cheap FR4 quality boards, they'll do, like, a really loose weave. So there's actually gaps between, like, so it'll be like a checkerboard type pattern. You can call up his, there's a screenshot in his video. It's a checkerboard type pattern. And if your trace, if your controlled impedance trace happens to be above the glass weave, then it's going to have a slightly different impedance than if it's above the one that has all the gaps. But the gaps between there.

Chris Gammell: Right, because it's just a different coefficient, basically.

Dave Jones: It's a different dielectric coefficient, exactly. So, and it varies, and it's a real thing. And Bill actually demonstrates in the video using a TDR. Cool. Yeah, and that's cool. But he claimed that it was a horizontal vertical thing. But if you actually look at how the weave is manufactured in the board, it's actually a symmetrical layout. So I think what he was seeing is the difference between the slightly offset trace. It wasn't the fact that it was horizontal or vertical. So anyway, I put that in a comment. So please jump in there. And that was my initial thought when I watched. I didn't watch the entire video right through. I was kind of, like, skipping through, and I just saw that he mentioned that. Cool.

Chris Gammell: Yeah, I mean, I like this kind of stuff, too, because this is something where it's a vanishing number of people that are going to get into this level with their designs. But I think that number grows over time. It's just boards get more and more high speed.

Dave Jones: Everyone's hooking DDR after there. But then if you're doing that, most people aren't at the board. Like, they're going to be using, like, a module. Right. They're going to be using a pre-dump module where it's all laid out for them. You know, so, yeah, the technology is getting more accessible, but most people won't be laying out DDR4 memory onto their microcontroller.

Chris Gammell: If you're coming from the software world right now, don't freak out. Don't freak out. This is what me and Dave do when we get to, you know, software. We freak out. Don't freak out about the hardware. Yeah. Exactly. Yeah. Yeah.

Dave Jones: Because you just don't have to. You know, there's so many options to just avoid doing that. Yeah. That's not funny. So, yeah. Anyway, yes, that was a really good video. I like that. Yeah. Yeah. And maybe we're talking in the background about a possible collab. Ooh, nice. Okay. So, we'll see what happens. That's great. All right. All right. That might happen.

Chris Gammell: Well, there are a lot of other links and many more links to Dave's stuff, too. There's a Roden Schwartz thing. Yeah. So, lots of links from the past six weeks, month and a half. I don't know. Since 2018. So, we'll talk about more of them in a couple weeks here. Cool bananas. All right, man. Talk to you soon.

Dave Jones: Catch you next time. Bye.

Speaker ?: administered administered administered administered administered