Embedded - 292: Now I Feel Less Cool

Episode Date: June 20, 2019

Christopher (@stoneymonster) and Elecia (@logicalelegance) babble about their current projects involving ants, guitars, machine learning, and party planning. A video of Christopher’s ants Some tweet threads about our tour of Santa Cruz Guitar Company. Arrival (also: very old wood) Wood is awesome Adding science to precision craft Elecia has been reading  Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Aurélien Géron. While the 2nd edition preview is on O’Reilly’s electronic library (formerly Safari Online), it will be available via Amazon on July 5th. Or pick up the first edition. Phillip Johnston of Embedded Artistry (290: Rule of Thumbs) is looking for blog posts, exchanging editing and exposure for posts that make sense on the site. Contact him with a topic idea before jumping in. For the Embedded blog, related to the show with Phillip, Elecia wrote a post about learning to give feedback. Listener Brian asked about a CS degree for going into firmware. We mentioned our show with Dennis Jackson (211: 4 Weeks, 3 Days). Listener Craig asked about PICs. We suggested taking a look at Jay Carlson’s Amazing $1 Microcontroller. We talked to Jay in 226: Camp AVR vs. Camp Microchip. Listener Happyday asked about UL testing. We added FCC testing then asked if any of you could help us. Hit the contact link on Embedded.fm. Embedded has a Patreon. There are new sponsorship levels! Nothing has changed though.

Hello and welcome to Embedded. I am Elysia White, here with Christopher White, and this week we are just going to chat. It's probably going to be short because somebody had way too much coffee this morning. It wasn't me. You're only here with most of Christopher White. Only half of my brain is active. The bad half. So you've been up to a number of interesting things.

I have? Well, I hear you had a birthday sometime in the last six weeks. And that the best wife possible got you the best gift possible. This whole show is going to be... Shaming me into publicly thanking you as much as possible? No, not that. You did get me the best gift. You got me a wonderful acoustic guitar from a local acoustic guitar company.

Yeah. The Santa Cruz Guitar Company, which makes artisan guitars. So I had to buy Christopher a used one. Are they artisan or artisanal? I don't know. Artisanal, I guess. But you don't eat it. Boutique?

Boutique. Yeah. They make like 400 a year. So it's not a lot of guitars. And we got to take a tour. Do you want to describe the tour? Yeah, well, it was funny because we, and sorry everybody who wants to hear about embedded systems. That's not what the show's about anymore.

No, this is related, actually. We'll get to the related part. You know, I don't even work on them anymore, so it's fine. You had the address of the place, and it was out next to, you know, Costco or something weird in an industrial park area. And we drive out there, and we get to where it's supposed to be, and there's some, you know, bay kind of offices. Industrial park. Industrial park. Industrial park. No sign.

There's no sign anywhere for the guitar place. It doesn't look like it's big enough to have a factory of any kind. There's no sign on the window of the suite that it's supposed to be in. Nothing. So we kind of wandered around for a minute until somebody came out. We must have noticed from inside that there were confused people standing out front. But you go inside and it's a little bit of a messy, messy entryway. But sitting there in the entryway, there's some chairs and stuff.

Yeah. But sitting there on the coffee table is a piece of wood. There was a block of wood and then there was a bowl. And then there was a bowl. And they were both made out of the same wood. And there's a label on the wood that says 40,000 years old. 45,000 years old. I pulled up the picture, so I have it.

And so, I mean, that's amazing, right? And it's because it got buried and it didn't um actually petrify it actually was buried in an anaerobic environment so it was still wood after it and you could i touched wood that was 45 000 years old it was amazing i mean rocks are a lot older you're not amazed by that huh some rocks uh so yeah so the the uh this guy comes out who's an older gentleman and uh starts asking us questions you know where we're from and stuff and and why we're there and after that he starts leading us on this tour through the through the factory and turns out he's the founder of the company, founded in the mid-70s.

And yeah, it was really interesting to watch their process. It's all handmade. There are very few machines. And where they used machines, they were in places where it was like there's nothing the machine's taking away from our artisanal process. It was, we want to prevent repetitive stress injury so things where it's like we're just taking away lots of wood or we're trying to you know very

precisely replicate something in a particular case that we could do by hand but it's very bad for people's wrists and stuff um so i guess they, they started at the beginning, right? They started with, uh, the, you know, the top of the guitars and the, and how, which woods they choose. So yeah, he showed us the woods they chose first. And he's sort of a, uh, how did he put it? He had a addiction to wood or something like that. Yeah, he definitely did too.

Um, but they try to use all... Reclaimed. Reclaimed wood, yeah. And that means either it was used in something previously, like office furniture or some other high-end wood-based product, or they find it, or people find it and notify him that it's available. And when they find it or people find it and and um notify him that it's available and where they find it it's like oh we were putting in an apartment building and we dug up this ancient river that was nearby and we found this huge log and it's 10 000 years old um you know and

he'll buy that up and make guitars out of it and And the thing about wood aging, which I didn't really understand, and I was really skeptical about some of this stuff because you read about instrument making and they talk about how important tone wood is and different woods sound different. And, you know, in the back, and I've been playing music for years, but in the back of my mind, I was always like, yeah.

That's nice. Do you want a gold-plated Ethernet cable? I don't really believe you. Somehow in my mind was always like yeah that's nice do you want a gold plated don't really ethernet cable believe you somehow in my mind it was like the construction matters more than the i mean it's nice to have good wood it looks nice and i'm sure it sounds subtly better but i was always thinking well my ears aren't good enough to tell the difference um which was just false in 30 seconds he pulled out i think he had a piece of new cedar three years old three years old um and he kind of wrapped on it with knuckles holding it in one hand and to me it sounded like when you get kids wood toys yeah and and they it doesn't chunk but it kind of so satisfying nicely. It's a satisfying kind of mid-high-pitch clunk like you'd hear on marimbas or

woodblocks. Not a lot of sustain, just kind of a clunk. And he said, okay. And then he picked up another piece of wood, which it wasn't an apples-to-apples comparison, but it was close enough that... They were both cedar. I think. But one was slightly different dimensions, too. Yeah. It was not apples-to-apples. But he

picked up this other piece of wood, and he hit it with his knuckles and it sounded like a bell and it rang i was like that that's weird i've never wood can do that uh and that was the three thousand year old wood i think twelve hundred something it was a couple thousand years old yeah and he talked about how it isn't just about drawing the air out which is what happens when you age wood for three or four years. It's about truly drawing all of the resins out. Yeah, the moisture. All of the moisture. The resin.

The resin has some moisture in it that even after you dry out the wood. So he went through the process of how they construct them after that, which was this very strange process of shaping. Let me describe how an acoustic guitar is constructed a little bit. It's like an open box, you know, in a nice rounded shape. It has back and it has sides, and the sides are kind of bent thin wood to make that rounded shape. And then it has a flat top of thin wood. So you can imagine just, you know, thin ply.

It's not plywood, but thin plies of wood on the top and back and then thin on the sides. And then the neck attaches to the top and the side where it goes, and that gets glued in, and then the strings are put under tension. But to keep things from collapsing, there has to be some bracing under that top piece because it's very thin. And it turns out that bracing

is a big part of how the instrument sounds, which I didn't know either. So he walked us through that, and it's a very strange process of shaping the bracing in these weird little curves. Uh, they do it by ear, uh, wrapping on the wood as they go and seeing how it changes things.

And it, it alters how the, the, uh, the resonance of the top piece of wood, when the sound propagates through it, whether it self interferes or whether it boosts certain frequencies and it has to do with also where the bracing is thick in relation to where the thick strings are. So all sorts of stuff like that, very complicated process. And it, it all sounded very mumbo jumbo.

And then again, they demonstrated, you know, okay, well this one we've, we've made it thick over here. And if you, if you wrap it near where the thick string strings were, it's got, you know, more base. And've made it thick over here and if you if you wrap it near where

the thick strings were it's got you know more bass and then we're trying to emphasize so they go through and they methodically figure out how they want to emphasize certain frequencies and things uh in that wood resonance and then they you know put it all together and more complicated things happen but almost every guitar that they make is custom yeah so you can buy some in some stores but right that aren't custom for you but most of the guitars like they're building it for someone for a purpose yeah yeah and so when chris was saying you emphasize bass or whatever it's it's because that style of music the person's going to play, that's what they want. Yeah. And then you started talking to him about Fourier and stuff.

And he said, oh, well, I have a Stanford, was it a grad student? Grad student. Stanford grad student over in the back room. And he's got this laser apparatus we're working with and this sampling thing. We're trying to figure out how all this works and to codify it. And that was what was really interesting to me, or one of the things that was really interesting to me, is because a lot of the maker, not maker,

a lot of artisanal kind of things have a lot of, we do it this way because this is the way it works, and it has for 500 years, and it's all it's true but rarely does somebody go back and say well why does this work what's what's the scientific principle behind this and i just love that they were doing that it's like okay well we've been doing this for 30 or 40 years violin makers have been using similar techniques for hundreds and hundreds of years uh why does it work and if we can figure out how it works, would it make it easier for us to replicate ourselves by hand? They're not looking to take that research and turn it into a machine.

They had some taking that research and turning it into a machine. That was a different thing, though. When they build a guitar for a specific famous person, sometimes the fans want a similar guitar. And so they can build the guitar for a specific famous person sometimes the fans want a similar guitar and so they can build the guitar that was the next the next it was just the next it was just the next and um and then they can you know scan it laser scan it all three dimensions and then they had some nice uh cncs done one guitar myself, not an acoustic, but an electric, the neck is really hard. You start out with, you know, a two by three by three, I think.

Well, yeah, something like that. A Planca, Planca Maple, usually, or mahogany. And you basically shave it down until it's the right shape. And it's very time consuming. And it's great to do it by hand um once but it's kind of a the neck shape matters a lot to a player and there's not a lot of advantage to having a hand-shaped neck over it doesn't affect the tone as much it affects the playability and how you interact with the instrument but it doesn't really affect how it

sounds yeah so that that's one of those places where it's like okay let's take rsi out of the and how you interact with the instrument, but it doesn't really affect how it sounds. Yeah, so that's one of those places where it's like, okay, let's take RSI out of the equation and make a reproducible thing. But yeah, so I mean, and then they took us through the whole, the finishing procedures and how all that works and a little bit of the chemistry behind the nitrocellulose finishes that a lot of guitar makers use. Trying to think if there was anything else that was spectacular.

Well, there were two things that I wanted, that were one of the reasons this is on my list. Yeah, sure. You mentioned RSI and repetitive stress injury, and Richard Hoover, who was our tour guide and the founder. Yes, I should have named him to start with. You know, he was very passionate about this. He was very passionate about his workers shouldn't get hurt

in order to build these beautiful instruments. And at the beginning, you're like, oh, yeah, that's very cool. That's good. That's very enlightened. Welcome to Santa Cruz. But then you look at Richard's hands and you know why he cares about RSI. Because here's a guy who builds guitars

and I don't think he can play as much as he wants to. I mean, his hands looked like they hurt. And so, you know, having him out there telling the story of, I want to build something beautiful, but I don't want to, I don't, I want to use trees that are already down. I want to use, I want to make sure people leave better than they came. And then the other thing that I really was happy with is that throughout this tour, he told us anything we wanted to know. Yes. And he talked about open source. He even said the words open source, yeah.

I mean, all of the information about the guitars he talks about. And I took pictures of everything. We took pictures, yeah. I bet we could have videoed the whole thing if we wanted. And it's sort of like Alicia Gibbs was talking about on our open hardware episode. A lot of the open source difficulty with this is the manufacturing. You're not going to get an untrained... Well, in this particular case, the extreme years of skill. I mean, it's why there's apprenticeships and journeymen and masters.

And historically, it wasn't, oh, if I only knew the secret behind this, I could put it in the machine for refinement, or they'd put it in their machine for initial, and then they would finish it by hand. And then they would put it back in the machine for measurement and send that data off so that other people could be improved with the machine, and so that the machine makers could make it so that fretting worked better for everyone. It was just a very open, super nice tour, but a very interesting, super nice manufacturing environment. Yeah.

So I took a bunch of pictures and I put them on Twitter. I think I may make a blog post, even though it's not that related to Embedded. It was really neat, not only to see the guitars, not only to hear the woods, but to have a story about employees and employers. Cool. Okay. So let's see. You've also been working on ants. I've been working on ants? Ants.

I got bored Sunday afternoon for two hours. That's not working on anything. I made an ant simulator. This is a very popular thing to do at least one point in your childhood or career. It is? Isn't it? No, I hope not because now I feel less cool. Well, I mean, it's always, I think it's kind of like Conway's Game of Life, but then you start making your own rules up.

And I like the idea of ants and moving around and food. Yeah, anyway, I just wanted to play around with some of the Swift frameworks I haven't used much and probably will never get a chance to use in my job. I want a sprite kit, and it's kind of like a 2D game engine, but it's less of an engine than, you know, the code framework to have sprites move around and the physics stuff. I guess when I think of Unity, it's got all that infrastructure behind it that you script, whereas this is part of the language and, you know, you're writing code for all of it. But yeah, I just wanted to play around with that.

And so I made this little thing where it was something from high school, actually. I remember doing it in high school. I had an assignment in my CS class and I think it was from a Scientific American article where the teacher originally got it. I'll have to dig it up. Maybe, if that's even possible.

It was, you know, before the flood. The flood that's the internet. Anyway, the idea was you have these little creatures, and they could be ants, whatever you want to call them. And they're all independent agents. And their behavior is governed by sets of genes. And instead of having a bunch of heuristics that say, oh, I'm walking this way and food is this way, so turn this way, you just have these genes that say, okay, there's a X percent probability that it's going to turn left at any given time step

or right at any given time step or change its speed. So it's a kind of a motion-controlled genes. And then you just throw a bunch of them in there and watch them walk around and they can eat food to stay alive. But if they move, they lose some energy so they can die if they don't eat enough. And then they can also reproduce. So if they crash into each other, then you have a little function that takes their genes and shuffles them together and gives them to the new offspring. And the idea is you're supposed to see some emergent behaviors like from natural selection, like the ones that tend to move in a certain way, tend to live longer given a certain food distribution. But then you can just start adding lots of little heuristics if you want or new genes

or behaviors or adversaries or whatever just to experiment but it was less about that and more about it i just need an afternoon project while we're watching soccer you didn't admit to watching soccer football football um so yeah that was cool i i don't know if i'll do anything further with it maybe i'll find another project for another corner of swift um the neat thing about the neat thing about swift is uh they have these playground things so you can load up the ide and you can either build a mac app or an ios app uh or you can load up the IDE, and you can either build a Mac app or an iOS app, or you can load up what's called a Playground, and it's just an empty editor. But it's more like an interpreter in that mode, and so it compiles as you type, like really. And it'll give you, it's like the debugger's integrated, so if you click on any particular line, it'll show you the value of whatever's being accessed at that line in real time.

So it runs pretty slowly compared to compiled, but for certain things it's fun to watch, and you can see the changes you make in the code and how they affect the results immediately. So I just threw that up there and started typing. Wrote some very bad code, which I made a little cleaner. But then i threw it in a mac app because it was running way too slow when i had hundreds of ants it was fun because

we don't always do programming things on the weekends but it was fun to watch him you know first there are little dots that are pretendants and little other dots that are food. And then they become smarter and better. And it was just really fun. And while he was doing that, I was reading a machine learning book. Hands-on Machine Learning with Scikit-learn, Keras, and TensorFlow. Wow, that's a boring title. But I have been doing more machine learning stuff. And one of the hard parts is taking a class, you watch a video, and then at the end, you're like, oh, I should use

some of my stuff. And then you go back to your stuff and you look at it and you go, huh, I wonder how I'd start. I don't know how to apply this. I don't know what to do. That must have been useless. Machine learning as a whole is useless. And this book really tries to help you go through the process. And I think it's like 700 pages and it's really dense. That's long. Dense 700 pages. But it also comes with the Jupyter notebooks, which are those Python things, which are like half code and half written documentation, not comments. And their code you can run in little pieces. I love Jupyter notebooks.

And GitHub now has a way that you can see them more easily instead of having to ask everybody to output HTMLs. I didn't realize that's similar to the Playground thing. Some of their documentation in Xcode actually have text and then a snippet of code and it runs. So it's very similar to Jupyter Notebooks. Jupyter Notebooks, I'm sure, are much, much better, but I'm just reminded that they did a similar thing. But that's super helpful. So they have the exercises and the things per chapter that way? They have every chapter. Or is every chapter kind of a walk through one of the notebooks?

It's more like the notebooks show all the code that was used in the chapter. Okay. As well as having some connecting bits. But I think if you just tried to read the notebooks without the book, it would be a little hard. Although I keep going back and forth, like I'll read the book and then I'll go to the notebook and then I'll get to where the end of where I read in the book. And then I'll continue on until I don't understand it. And then I'll go back to the book. I mean, I go back and forth. And it's, it's been good. I mean, it's a lot of math, but it's also a lot of, okay, you have a data set.

What do you do first? What do you do next? What are the processes? And I keep seeing that it's using a lot of the things I've seen in other machine learning contexts. There's the minced data set that is all of the handwritten digits. And it's a very popular data set. And it talks about that and it talks about what you do with it and how you, but it also talks about Kaggle, which is this place where you get these giant data sets and you have contests. And he

talks about other places where you can get large data sets, which led me onto this whole adventure with another one of my clients. And maybe we should look at this data set to see how things are working for them and figure out how to make this sort of similar sort of data set for us. And it's just been really kind of good. I don't always read that kind of book when I don't have to, but this was a good one to read ahead of time because I think I'm getting tools I can apply. But it's not out yet. Oh, actually, there's a first edition that's out and a second edition

that's on O'Reilly Safari, and the second edition should be out on paper soon. Okay. And since I have looked at both, the second edition, I think that's the one you should get if you can, but it's not like the first 400 pages in the old one are any different. I think it's only later when you get into neural nets. Okay. Let's see. I wanted to talk about, oh artistry uh philip johnston they are looking for guest posters for that blog that we mentioned a couple of times

and the idea would be that the guests would write a post on par with the current posts, and they would help edit it. And the guest poster can also cross-post for their own blog. Of course, you should contact them ahead of time to ensure the topic aligns with the blog. And you aren't getting paid for this, which makes it a terrible deal if you are just getting established, or I'm sorry, a terrible deal if you are already established and are a good writer. But if you are uncertain of your writing skills or want to get more into that area, this is a good way to

practice. Being edited is hard and learning how to do the process is really useful. So that's Embedded Artistry. And they have a very big following. So, you know, if that floats your boat, try it out. Let's see. I wrote a blog post on our blog, Embedded FM slash blog, about giving feedback, partially because of our talk with Embedded Artistry and my writing group. I really hope they don't read that or listen to this. I don't think so. Okay, listener questions. Oh, poor Craig.

Hi, Craig. Poor Craig. What? Oh, poor Craig. Hi, Craig. Poor Craig. What? I was hired straight out of college by a shop that uses exclusively PIC controllers. Sorry. After using several microcontrollers in my classes that were frustrating to set up and to cajole into doing what I asked of them, I found PIX to be a welcome change. Since then, I have acquired a great appreciation for them, finding them to be feature-rich, intuitive, and easy to use. As the old saying goes, the greatest indicator of what processor a designer will use in their next project is what they used on the last one.

Craig continues, I have found myself embarrassingly only having used pics. I enjoy using them a lot, but I'm a little concerned that I'm typecasting my career by not expanding. One thing that has me concerned is a polarization in the industry. Namely, between those who love pics and those who are really bothered by them. I haven't been able to find anything online about why they are vilified and was wondering if you were able to shine light on some of this. Hang on, let me see here. What am I missing? I don't believe in pics suck.

Have pics ruined me or am I missing something non-intuitive? Okay, don't search for that. That was a huge mistake. That was really a huge mistake, Christopher. Do you need me to find some eye bleach? Did I mention the half a brain? Yeah, PICS. PICS.

Stand for Programmable Intelligent. You're just making this up. No, no, no, no, no, no, no, no, no, no, no. Programmable, programmable... Well, it's not only PIX. Programmable intelligent computer. Intelligent computer. Initially, it was called peripheral interface controller,

but that was too boring. It's the programmable Intelligent Computer. That's not what I would call them. Look, they're old, old microcontrollers. There's PIC32s now. Those are MIPSs. PICs splatted on their name, or on their label. They're pretty old, and so architecturally they're a little behind the times.

And for what they do, they're kind of expensive in one issue, right? They're not that bad. I mean, they're not as bad as the, like, Atmega series, which, of course, now is owned by microchip so i can hate on everything okay that wasn't a helpful answer no you're the one who's used all this stuff more than i have you you recently did a pic project so i still have a pic project going on you got that fresh in your mind so you've greg you've been PIX and you're accustomed to them. And some of the processors are difficult to set up. And with PIX, you don't, you can't, you don't get the

option. Should I use Kyle or IAR or GCC? You only get one option. And, you know, GCC is free. And MP Lab is sometimes free. I just, you know, some of it is that people are trying to use tools that can be made industry standard. And PIC has no, they aren't headed that direction. They don't care about that direction. There are a lot of good things about picks. The way that they boot up makes it so that their power doesn't have to be as stable. And so you need fewer chips on your board. And they can be very low power. And they can be smart about their low power. So again, fewer chips on the board. It's one of the reasons that when you talk about PIC versus other chips,

it's often a hardware versus software argument. Hardware engineers love the PICs. There are fewer chips on the board, fewer things for them to worry about. As a software engineer, the PIC documentation is not great. The PIC compiler having to pay for the optimizer is not great. It's more like not paying for the non-optimizer. I mean, do you think it's really the optimizer? Do you think they're producing less optimal code in the cheap one on purpose?

Yeah. Oh, yeah. Less about not including dash 03 and more about... Including the be an idiot. This is Dash be an idiot. Yeah. I mean, yeah, a compiler costs money to be a full-featured compiler. It's not just speed improvements, it's space improvements. Your tiny chip that you spent maybe got less expensive because you didn't think you needed all the space is easily filled up.

I think that flash programmers are not great either, right? It's not like you can buy a discovery board. You have to buy a programmer, a separate programmer. Yeah, you do. Yeah, you do. Yeah, you do. An ICD. I don't know how expensive those are. I'm looking at a couple hundred bucks here.

Yeah, and so there aren't as many $4 boards or $12 boards that you just get and you can treat it like a serial port and a thumb drive, like you can with some of the arms. You need more tools as a software engineer. However, once you have those tools, yeah, the picks are going to look a lot easier to you. Once you have a license for MPLAB, yeah, I don't see why you would want to change. Well, I mean, there's still price performance, and I don't think they perform well compared to like a Cortex-M0.

But as I said, they have some other advantages. They have a lot of peripherals and stuff. And because of their power and low power situations, you can have fewer chips on the board. So it's not terrible. It's kind of like the 8051 at this point. To me, it's becoming the new 8051. It's fine.

People use them. They have reasons to use them, but there's a lot of people who use them because that's what they know. And meanwhile, over here, technology is continuing. 10, 15, 20 years past it. And if you spend a little bit of time learning that stuff,

I think you'll find the cost benefit is way on the other side. But, you know, whatever. You shouldn't have allegiance to parts. Yeah. And while I'm all for making fun of P, or I are, let's be serious here, it doesn't matter that much. Yeah, it probably would be good to get a little bit of experience somewhere else just so you can compare and contrast yourself. I'd say if you're working with micros, you should have some arm experience. I think it would look much better on your resume if you could have a little bit of arm experience, since those are the most popular. Okay, so do you think we've answered that enough?

I don't know. I was looking for what Andre said the other day. Oh, man. Andre from the Great White North. I kind of posed this to him. Where was that? I guess we should have asked him to just record it it was like asking someone can you please just go off on an entire rant

i think it was something about how hating yourself is really not what you should do i it was it was pretty funny yeah but you know if you if you if you want to kind of get a sense for where PIX fit in with the overall ecosystem, there's that Jay Carlson article. More like a giant interactive website. We talked to him a long time ago on the show. The amazing $1 microcontroller. And he compares and contrasts in great detail all of the sub-dollar microcontrollers that are out there, PixM SP430, Cortex M0s. And he compares them based on power, price.

Power, price, capabilities, tools. Performance. He did a fantastic job on that. So read that. Search for $1 microcontrollerroller and you will find the article. Let's see. Do you want announcements or another listener email? Do announcements last. Okay. From Brian. I am just getting going at age 46, but I'm coming along quickly.

I've been reading about how to become a firmware engineer, and I'm learning there's no direct path there. That is so true. Probably because firmware engineer is not well-defined. Yeah, yeah. Where are you going? Brian is currently working on a computer science degree, trying to decide whether EE or CE would be better in order to get to a firmware route. Oh, CE, computer engineering.

Computer engineering. And those seem like more work since he's already partway through CS. He's not scared of the work, but come on, when you're 46, you don't quite get as much time to do all the other stuff. Usually there's life responsibilities.

He apologizes for being vague, wants to know if it's worth going EE or CE instead, or if there's something close to firmware that will put him in a better position to go into firmware. Well, it's interesting because I'm about the same age and I'm thinking of getting out of this business forever. I don't know the answer, EE or CE. My gut feeling is CE is a little better

because that's going to be more processor oriented. But I think you could tailor an EE degree toward embedded systems and firmware better. I think you could tailor a CS degree. Yeah.

I'm not sure you need those other things. I have none of them. Christopher has a math degree. So, you know, I think we had this question a couple shows back where somebody asked, or maybe it's a recurring question, where it's like, what kinds of things can I do to learn better? I'm in school and they're not teaching. Yeah,

it was how to improve the educational system with regards to embedded systems. And I think this is basically the same question. And I think it's going to have similar answers. It's going to be kind of what you put into it and firmware is such a hardware dependent and i don't mean you need to know about hardware i mean it's there's specific systems to kind of get familiar with right there's the microcontroller plus peripherals, that basic concept. And whether it's an Atmel chip or a PIC or a Cortex M3 or whatever, they're all sort of the same thing.

So it's hard to find a course that teaches that. Well, there's often microcontroller courses. Microprocessors. There are more microcontroller. There are more microcontroller courses. Microprocessors. There are more microcontroller courses. Well, if you've got one of those, take that. But an entire degree to try to adjunct your CS stuff, I don't see

that that's necessary. I think what's necessary is starting to use them and getting experience with what all this means. I don't think you need to leave a CS degree to go to a different one. No, I wouldn't. I wouldn't at all. I think the preponderance of... Maybe not the preponderance,

but a huge fraction of people doing embedded software and firmware work are CS people. Yeah. And you do have to go in with the idea that after college, you were getting a piece of paper that says, you know, a few things, but when you go into embedded, the chances are you will only know half of what you need to know and you'll pick up the other half. You just have to be kind of humble about learning more. Yeah. So we did a show with Dennis Jackson, let's see, four weeks, three days. Yeah.

And it's about moving from pure software to embedded. So some of that might be useful. Also, we always suggest that if there's a robotics course or club, that's a good way to get real hands-on experience about working in an interdisciplinary team, which is what you do with embedded systems. You know, if you have electives... I was going to say, take them out, them out, look in the CE course catalog and the EE course catalog and just pick things that look most applicable. So if there's a microcontroller course, like you mentioned.

Or if there's a CS course about things that are coming in Embedded. I would take a CS course in Machine Learning because I think that is coming to embedded. If there's anything about computer vision, if there's anything about sensors, networking is important. Compiler design is weirdly important because we often do have to look at assembly. You're disagreeing

on that one? I think that would be if there's nothing else, maybe. With limited course time, that would not be high on my list. They don't teach Linux as a course, do they? No. No, I don't... Well, I don't know.

System administration, that would be... Might be part of the CS department somehow. That's right. Or operating systems. Operating systems, yeah. There's an operating systems course. Operating systems course, that would actually be higher on the CS department somehow. That's right. Or operating systems. Operating systems, yeah. There's an operating systems course. Operating systems course, that would actually be higher on the list to me

because you're going to get concepts. They're not going to teach you RTOSs necessarily, but they'll teach you things like synchronization primitives and schedulers and things that apply to real-time operating systems. And they'll probably touch on real-time a little bit. It depends on the OS course. But that's a really good course

for anyone who's thinking about doing embedded and thinking about doing something bigger than a flashing light. You know, anytime an R test comes into play, those OS concepts really trip people up and are a huge source of problems and bugs if you don't know what you're doing. Because if you're doing embedded and you're, you know, you get into a good role, you're going to be writing device drivers, you're going to be

writing tasks and interrupt handlers and things, and all that synchronization stuff comes into play there. And if you don't know how that works, it's really hard to learn on the job. You can do it, but it's much better to have seen them before so you at least know, oh, that's a mutex. I know what that's for. It's the sort of thing that you take a class, it makes sense,

you forget it, and then when you have to relearn it, it isn't as hard. That's what college is for. That's what college is for. I'm sorry to tell you that that you're not going to come out of most people don't come out of college experts they come out of college knowing okay here's what there is to know and here's what i don't know well if you come out of college

knowing what you don't know you're well ahead of a lot of people because i certainly didn't like what's that oh they didn't teach me that. Oh, God. Okay, moving to the next question, which I have to admit we're not going to be answering. This is from Happy Day, who would like a discussion, a walkthrough on taking hardware through UL testing. What are the steps? How much will it cost? What can we do to optimize acceptance? And as long as we're talking about that, I want to talk about walking our hardware through FCC

testing. But we can't do that, actually. So I am... Why not? Because, I mean, I've done FCC testing, but... So have I. Only to the extent that I... You go to the place in Milpitas, they put your thing in a room with some funny stuff on the walls, they push a button and they look at a graph.

They tell you to turn your thing on and make it do some stuff and they push a button and they look at a graph. And then later they come in with like this wand thing and they stick it in the product you're looking on. Alien probes. The thing you've worked hours and hours to build, and they just zap the shit out of it. And then they hope that it breaks or doesn't break in the right way.

And then they give you a certificate, and you move on. That's the deal. Yeah. So if anyone has experience in these areas... Come on the show. Yeah, come on the show. Yeah. Come on the show.

And tell me why I'm wrong. That's why we always have guests. Oh, and then you go to lunch at the lousy place next door. I think it was a Greek restaurant. It wasn't very good. So many people in the Bay area are just nodding right now. Let's see. Now I'm down to announcements. Do you have anything

you would like to talk about before announcements? I don't think so. Okay, so first this is show 293? 292. Are you sure? I am reasonably certain. Okay, so this is show 292. which means we're coming up to a round number.

294 is the round number? No. 296 is more round. It's got the 9 and the 6. 298 would be probably even rounder. We're coming up to 300. Oh, that's pretty round. It's got the two O's. The O's, yeah.

And the three, which is sort of round, but it's got those sharp edges that you don't want to stick on. Like claws. Yeah. Actually, that... You want to approach 300 from the back, not the front. Which is what we're doing. So it's time to have a party.

I know, I know. It's been like 150 episodes since we last had a party. And I know that many of you won't be able to come because you're all over the world, which is just amazing and awesome. But we're going to have a party. It's going to be in Aptos or thereabouts, Aptos, California, which is near Santa Cruz, California. I think it's going to be mid to late August.

And I think... Oh, the cold season. The cold season, yes. I think the theme last time was hats and hacks. We gave out prizes for people who had the best hats, which were jellyfish. And we gave out prizes for people who... the best hats, which were jellyfish. And we gave out prizes for people who... Jellyfish? I thought they were squid. And then we gave out prizes for the best hacks.

And then we gave out one prize to Ben Henke, who had a hack that lived on his hat. I wanted to warn you that that is not going to be our theme this time. This time I think our theme is cats and hacks. Cats and hats? Cat, cats. You want to reconsider this? Yes, I really do right now. Cats and hacks.

You should not bring an actual cat. Do not bring any actual cats unless they're tigers, in which case tell me ahead of time because I've always wanted to pet a tiger. Don't bring a tiger. Yeah, we'll meet somewhere else, maybe the beach or something. Don't bring a tiger anywhere. Don't bring a cat, but do either come garbed with cat-like wear, or bring a hack, or possibly bring a hack that is related to cats. Don't bring anything a cat has hacked. Yes. Oh my goodness, I didn't think about that. No, no, no, no.

Our cat will not be joining us, so I just want to make sure you all understand no cats should come to the party. They're not invited. You are invited, not your cat. There will be more details as we go along, but I wanted to give you plenty of time to consider your cat costume. What are you going to wear, Christopher? Or are you going to hack something? Are you going to change your ants into cats and make all of the little food into mice? I could do that. I could do that.

They're just sprites. Who cares? Projected on the wall for the whole time. Oh, yeah. If it's at the place we're going to, it's got a nice wall. Yeah, so I don't have confirmation on the location. I don't have confirmation or anything.

But I'm telling you, we're going to do it. Making you a promise. All right. Let's see. Oh, look, all the rest is just very boring. Well, let's go through the boring. Many of you know that we have a Patreon thing. We have a Patreon account and you can support us, support the show, support the party.

And if you want to, you know, a buck a month is great. Five bucks a month is great. Five bucks a month is great. 30 bucks a month is great. And we'll tell your boss that this is totally an expense necessary for learning. I'll write you a note. Yeah, I really don't think that's how anything works. And you have an expense that's anything in years. It's not that much fun. No, I haven't expensed anything in years. It's not that much fun.

No, I haven't expensed anything in years. Sticking on invoices. A lot easier. Consultants. Yes, I am both the purchaser and the approver for our company. Let's see. So, yes, we have the Patreon.

The only benefit it gets you besides the show. Wait a minute. No, you get the show anyway. The only benefit it gets you besides the show. Wait a minute. No, you get the show anyway. The only thing it gets you is an invite to our Slack channel, which is pretty fun. People there are neat. Very helpful to some people, I think. I think it's been helpful to people.

People ask questions and other people answer. Sometimes we answer. Sometimes we're even there. And there's the job board in the Slack. And I really like it when people talk about going to interview for jobs. And everybody jumps in and says, oh, you should be sure to ask him about this. Or they may ask you about that. Maybe you should study it.

I think that company's fantastic. It just is so fun. And there's quite a few people on there. A couple hundred now. It's not you know, overwhelming amount of traffic. So you shouldn't feel like if you join that you'll be distracted forever.

But it's fun and I think people find it useful. I find it useful. If you think of some other rewards that we could conceivably do, I mean, we're open to them. Well, they do also get to go to the front of the line for when I send out stickers, which realistically is like every six months.

Right. So if you ask me for stickers, don't expect to get them soon. And be sure to tell me when your address changes, and I haven't sent them out yet. Let's see. If you don't have the money for Patreon, that's fine. If you could write a review on your podcast. iTunes is the only place, really. All right. Write a review on iTunes. As part of this whole 300 episodes,

I decided we should have a press kit and I should write a PR press release for the party. I don't know what I'm doing. Part of me says, you should make sure everybody knows about the podcast because people might like it if they only heard about it. And the other part of me is like, yeah, you know, I just want to read about machine learning and occasionally

talk to interesting people on the show. And I don't really want to do all this press crap. But in prepping the stuff for the press release, I went and read the iTunes reviews, which I have never done because I just, I'm like an ostrich with my head in the sand. You guys have said some really nice things. Thank you. And finally, if you don't want to do Patreon, you don't want to do iTunes reviews, you can support the show just by telling someone you like it. Preferably somebody who listens to podcasts, but doesn't already listen to our show. But we don't, we're not really, we're not really, what's the word? Picky?

Picky. Yeah, we're not really that picky. If you like the show, tell somebody about it, and we appreciate that very much. That's all I had. Did you, did you have anything else? Um, no. Okay. Shall we go watch soccer? No. Oh my God. You're going to work? Maybe. Ah, I don't know what we're going to do with you. Thank you all for listening and thank you to Christopher for producing. Now. And co-hosting. And co-hosting, of course. Just drop that when I wasn't there and you never brought it back. Oh, really? No, I'm just kidding. I changed my notes.

Let's see. It's Winnie the Pooh. When we last saw Winnie the Pooh, Pooh, he had eaten all of Rabbit's snacks, and the honey, the condensed milk, the bread, and then Rabbit was leaving. And so Pooh started to climb out of the hole. He pulled with his front paws and he pushed with his back paws and in a little while his nose was out in the open again and then his ears and then his front paws and then his shoulders and then oh help said poo i had better go back oh bother said poo i shall have to go on i can't do either said poo

oh help and bother now this entire time rabbit wanted to go for a walk too and finding the front door full he went out by the back door and came around to poo and looked at him hello are you stuck he asked no said poo carelessly just resting and thinking and humming to myself here give me a paw poo bear stretched out a paw and rabbit pulled and pulled and pulled. Ow, said Pooh. You're hurting me. The fact is, said Rabbit, you're stuck. It all comes, said Pooh crossly, of not having front doors big enough. Embedded is an independently produced radio show that focuses on the many aspects of engineering. It is a production of Logical Elegance, an embedded software consulting company in California.

If there are advertisements in the show, we did not put them there and do not receive money from them. At this time, our sponsors are Logical Elegance and listeners like you.