Embedded - 98: Figments of My Imagination

Episode Date: April 22, 2015

Chris and Elecia talk about memetics, learning, and processors.  Elecia was coy about the Pasadena party May 9th and 10th, but Hackaday announced it so you can invite yourself. She will also be speaking at the Solid conference in June in SF (email for a coupon!). She'll also be at ESC-Silicon Valley in July.  Star Wars Teaser #2 and SpaceX almost-landing BLE fun: TI's CC2640 and Nordic nRF51822 (Elecia likes the BLE Nano with the free, online mbed compiler for getting started with the nRF5122). Everything seems to be a Cortex-M0 these days (including the aforementioned CC2640 and nRF51822). The new Atmel SAM-L series is Cortex-M0 and even more low power than usual. On the other hand, the MSP432 is low power and is a more powerful Cortex-M4 (and inexpensive dev kits!), Elecia has a book: Making Embedded Systems. It makes a great gift.

Welcome to Embedded, the show for people who love gadgets. I'm Elysia White, my co-host is Christopher White, and this week it is just us, talking about processors, memes, and learning. Hi, nobody. Welcome to the show. I'm not nobody. No, I'm just saying our guest was nobody. Yeah, we've kind of gotten into some habits here, haven't we?

I have to say, welcome to the show. Well, and I keep switching it between the show for people who love gadgets and the show for people who love building gadgets. Building doesn't really fit. It doesn't scan well. I think in a few episodes it'll be the show for people who hate gadgets. I don't think so. No?

No. No? There are other podcasts for people who hate gadgets. Okay. So I have some listener mail to address. Is that where you want to start? Sure. Follow-up sounds like a good place to start.

Okay. This one is from Aaron Lohr. I've met him a couple of times, and he was having some trouble getting a job, although that's all solved and he's working on that. So that's very cool. But when he was out of work for a bit, he was looking at job listings and saying, yeah, I can do most of that, but not all of it. And if it listed a lot of things, he said he maybe hadn't used it specifically, like I2C or SPI. He knew what they were used for and how they worked, but hadn't used them directly. So he went on a little bit about different programming languages, different peripherals,

and how in the world do you pick things up so it was the original question i see this on job requests and this whole grab bag full of acronyms well there's that which i think we can that would be one question but the other is how do you learn something? Yeah, that's the bigger question. I think for job recs, if you have three-fifths of the skills, go ahead and apply. Oh, definitely. Or even a half. Because they're always...

It depends on who's writing the job descriptions, but a lot of the time they're hiring managers who may not have a full understanding of what, what they're hiring for. And so they, they kind of sent cast a wide net.

Well, this is like the time when they wanted 20 years of Java experience when Java had existed for 10. Exactly. It's not quite that bad most of the time, but yeah. And I think, um, I think if you get somebody who's maybe not familiar with embedded,

but they're a first line manager they might just say well what are all the embedded things that people do let's put all those on there because that's the person we want um when in actuality there aren't too many people who are always working with all of these things all at once well and if you can make a good story like maybe it's as i squared c but you've used spy or you've done UART. You say, it's similar. I understand it's not the same, but I think my skills will translate well. And if I was writing a job description, I would put familiarity with serial interfaces. Right.

I wouldn't say spy or I squared C because there's a whole bunch of other ones and they're all very similar. One wire is similar similar um although has its own foibles so just to you know get that out of the way i wouldn't worry too much about looking at a job rec and saying well i only know how to do a few of those things or only have experience with a few of those things because it's experience and you need to be able to apply previous experience with similar things to learn new things sometimes. If you've never seen a serial interface before, and the job has a lot of requirements and I squared C and spy, well then maybe that's something where you might be uncomfortable.

And if you're a manager, if you are the hiring manager, I think the most important thing to do is to break it into things they must have and then let them have a grab bag. You say, we need someone who can do embedded software development. They must have experience on maybe list your processor family or whatever thing is most important to you. Even that? Yeah, even that.

I mean, even processor family. I mean, there's only a few ways to... I was going to say slice the cat. Did you hear the BeagleBone cookbook came out? I saw that. Yeah. Anyway... That was an interesting title for it, but...

Where I was headed before I mixed my metaphor in a horrible, horrible way was embedded processors all are kind of the same. I mean, you have an ARM core or you have some other core and you have an array of peripherals and they have their libraries which access those peripherals. And moving between one or the other, unless you're going to be digging really deep into the architecture or needing to know the instruction set architecture for some reason,

I would say if you've used an embedded processor, you're already ahead of the game. Okay, so we're not going to put which embedded processors is a must on our list for hiring, but we're going to put... Am I out of my mind? No, no, I agree. I mean, I don't... People are like, do you only use Cortex-M3s?

And I'm like, no, I use a lot of processors. Most of the time, I couldn't care less. Oh, I care. Well, from a specs point of view. And from a debugging... And from a data sheets point of view. Some from a data sheets point of view. How's that different from specs? Oh, all right. Then never mind.

It's not different. But I mean, from a debug point of view, eventually you're going to have to understand something about the architecture because you might need to know how your memories work or the stacks work or what have you. How to debug the exception handler. Right.

But all those things are different flavors of kind of similar stuff. But what I was really headed for is you need an area of high-priority must-haves, and it should be short. And then you should go ahead and list the 15 skills you want them to have and then say, if you have more than three of these, let them go ahead and choose from your needed stuff. Must-haves and nice-to-haves. It's just so much better for people reading job descriptions.

I know they say that for, like, if you want to get more women to apply, to be sure to really let them say that they have all the skills that are required because like Aaron was saying he didn't have everything on that list and so he didn't necessarily want to apply and I think that hits a lot of people and they say it hits women more but I don't know if that's true no I've back in the day when i was applying to jobs um regularly uh when the dinosaurs roamed uh yeah no it would be intimidating and you you would you would

occasionally say hey you're interested in this contract even and i and i look at that and go well i haven't done any of that so no but that more of a, I'm lazy and don't feel like sticking my neck out real far into an area I'm not familiar with. That's a different feeling than, I need this job, and well, should I really apply? Because they may not hire me because I may not match. I would say apply to everything you think you could possibly be interested in and don't worry about that match because sometimes when you go in, they may find out you're a match for something else, depending on the company.

It's like, oh, you know about automated test scripting? Would you be interested in talking to the QA department? That kind of thing. So his other question that was bigger, though, was how do you learn new things? How do you learn new things? Which I think is a really interesting question. I mean, last week with Professor Paul Fishwick, he talked about using podcasts to learn new things.

And okay, yeah, I guess people listening to this probably do use podcasts to get some information. But when you're talking about things that may show up on job descriptions, how do you learn those things? Well, I know how not to learn them. When I was a kid, I had been doing a lot of programming. And I had an Apple II and programmed a BASIC on that and a little bit of assembly language kind of stuff. But it didn't really have a compiler or an assembler or anything like that. So assembly language programs would be like 20 bytes long and hand machine coded.

And yay, it did something, right? Or BASIC. And I did that for years until around high school i guess when took a cs class in high school and i think that was when pascal was like the not quite beginner language to learn but i'd always wanted to learn c because c was what real programmers used right still is i say so. So I think at some point I got a C compiler for my Apple IIgs, which was really cool. But I didn't have anything to write.

So I'd get these books. I'd get books on C, and I'd read them. And then I'd wonder after reading them why I couldn't write programs in C. And I think a lot of people do that. They get a technical book and they read it, and then they wonder why they don't know anything. It's funny, my linker post for this week, to follow up with Fishwick's episode,

is about using models to learn and teach, because that's what we talked about. And I think it solves that which you just said. You don't just learn to program from the book. You learn to program and do something else, whether it's show off some physical model or whether it's play lemonade, which is a nice economic model, or whether it's just to draw pretty fractals on the screen.

You can't learn to program by just reading a book that gives you programming syntax. Yeah, you have to learn by doing for anything technical. I mean, you wouldn't take a math book and read it cover to cover and wonder why you can't do abstract algebra, right? That's why they have exercises. Yeah. So there's that part of it.

And I think, you know, just to take it back to the question about resumes, my first job that I would consider really embedded, I worked at a couple of companies that, you know, yes, it was embedded, but that was only because it wasn't a computer, but it was a computer without a monitor. Well, that's like Cisco.

Cisco and Procket were very, very big, high-powered systems for their day, and we were writing code that ran on Unix-style operating systems. Yes, it was embedded, but it really wasn't. My first embedded job, I didn't know any of that stuff, and I was required

to debug a serial interface the first week I was there. We had some problems with a one-wire bus. I had no idea what the hell the thing was. And I had a few weeks to make it work. And yeah, that was definitely learning by doing. It's much easier to remember when you learn by doing. Pretty much everything like that that shows up on a resume, I've at jobs not not in a project not from a book so i guess for me the answer would

be i i learn best when i'm forced to um well that's that's one reason to do projects at home is to learn some of these things outside your skill set. Yeah. But that is... But it sure helps to have the CEO of the company breathing down your neck and asking why this thing that they wanted to ship two weeks ago isn't working. That's all your fault, even though it's not all your fault because you didn't even work there when it started not working. Aaron also mentioned that I had tweeted about python which i play with sometimes and he asked

how did i learn it and how often do i use it and really i learned it because someone at one job that was what they used a script i that was what they used to write a script i needed to modify and then once i modified it somehow i became owner of the script, you know, in that sticky way. Is this the same client that I learned Python from? Yes, yes. And in the same way? Yeah.

And now, you know, I wanted a MATLAB-like interface, and I know NumPy has been around for a while, and everybody loves it. Whenever I've seen it, I've been super excited, because MATLAB islike interface, and I know NumPy has been around for a while and everybody loves it. Whenever I've seen it, I've been super excited because MATLAB is just so expensive. So when I needed to do a graphing display, I went ahead and did that. Yeah. And I suspect that a real Python programmer would look at my code and laugh and laugh and laugh. I don't think there are any real Python programmers. Well, you know, there are all these ways of slicing and dicing the data,

and I use for loops. You just need to find the right modules in your set. Yeah, exactly. But no, I learned it the same way. There were scripts that the client needed to modify them. And then actually that's an interesting way to learn, is to take something that exists and modify it. Yeah.

Rather than starting from scratch and trying to write a program, take something that you know you've used and you wanted to do something else. That can be instructive, too, because it's not quite as big a mountain to climb as starting from scratch. On the other hand, when I interview people for jobs, one of the things I'm usually looking for is, can this person write it from scratch? Or does he always need to modify it?

Sure, but it's a first step. Exactly. And so that modification is a great first step for learning these things. You don't write an I2C driver from scratch. You take the one in the examples code and then you modify it to do what you need to based on what you read in the data sheet it's not, wow, how do I start this

because what you should do is search the internet and then you should understand that code and then you should go on so for learning oh, this is the time when I'm supposed to plug my book isn't it what well learning in my book you know um for those of you who may not know i wrote a book it's called making embedded systems it was from o'reilly it was not from it was with o'reilly it was from you it was from me published by o'reilly and And apparently it's been doing pretty well for the last few years,

but it kind of went down in January, so I would like to remind you it makes an excellent Mother's Day gift. And if you already have one, you should get another one for the other eye, because then you can read twice as fast. Also, there will be many new graduates who need it, whether they graduated in humanities or whatever. But in that book, one of the things that I have been told it lacks is a section that talks about how Bluetooth works, how Wi-Fi works, and that my I2C and SPI section is just not in-depth enough. And I think that's okay with me.

I wanted more to talk about how the whole system worked, how an embedded system works, and not how an embedded system works. I see. So that, you know, I2C and SPI, you can look up, you can find. Bluetooth was not as important then as it is now. So, you know, look it up. But how do you do PWM and how do you access registers?

That is important. How do you go low power? There are some strategies that work across platforms. And so I was really trying not to make a book that would age itself out right away. I think a lot of the O'Reilly books do, though. Not all of them. Sometimes. Well, it depends.

I mean, if you're making something about a specific product, then you can't help but do that. Yeah, I understand the criticism. I also understand the way you approached it. I think talking at a slightly higher level about these are the tools available to you, these are the serial buses, this is the differences between them, maybe. Yeah, I mean, I did a brief, you know,

I2C is great because it's two wires, and SPI is great because it's much easier to debug. And you can put more things than i squared c but i tried to give the brief overview i tried to give the brief overview of tcpip but these were like half of what you would find in the wikipedia article not not in-depth things like i did for how do you do floating point math without any floating points. Yeah, because your book was mostly about strategies, not about... Specific tactics, yeah.

Specific tactics are just descriptions. Yeah. So yes, buy my book. Buy it for everybody. If you want to buy it from O'Reilly, there's a coupon. A-U-T-H-D. We'll get you

50% off the e-book and 40% off the hard copy but Amazon is probably still cheaper Going back to the spy and iceberg teeth thing the other thing I just realized is one thing that helped me learn when I was struggling with the problem

was to have a good logic analyzer. Yeah. And to actually watch an existing system functioning and decode it yourself and understanding what the waveforms mean. Because if you don't know what they look like when it's working, you're going to have a really bad time

if you're trying to figure out when it's not working. And now that a lot of the O-scopes or like our friends at CVA, the O-scope logic analyzers, they'll really show you what those things look like and not just hide the details. So it's really nice. Yeah. And it was really tough to get a logic analyzer that worked. At the time I was trying to do it, it was a big HP thing. And now it's the equivalent of a tiny puck that you plug into your USB port.

Actually, those modern things are so much better. They'll decode properly automatically and all sorts of stuff. It even autobotted for me. That was nice. Then I have to look up what my serial port speed was supposed to be. Should I go on to the next email on my list? Did we answer that one, you think? Um, I don't know. Let's see.

How do you keep things you learned years ago fresh? That's a good part that we didn't really talk about. Well, the truth is i don't i can't yes i i figure like at least 60 of my job is learning new things that's the thing and my goal is to be really good at learning things yeah i think that's i don't bother to keep up with stuff like okay if i've used SPI five years ago and all my other clients use I2C, I don't every year do take a refresher course

in whichever one I said I didn't do. You know, you'll pick it up. I feel like if you have a fundamental understanding of the basic underpinnings of these things, it's not hard to pick up a specific version of them. But I don't think you'll get that fundamental understanding until you have had to deal with two or three. Two or three? Until they start looking

alike. It took me a while to understand the difference between I2C and SPY. Really? Because I had to do both and figure out which was which and how they were different, how were they the same. Going from I2C to one wire was trivially easy. Those two are very similar.

But I2C has the pull-ups, and if you didn't know that you have to look at I2C on an oscilloscope as well as a logic analyzer to get the whole story, and you only ever used SPI, SPI to OneWire is a far trip, but I squared C to one wire is between them. SPI and UART, those are pretty much the same. I mean, not exactly the same, but close enough that I don't care.

It's all just wiggling voltages. Yeah. Although the difference between Bluetooth low energy and Wi-Fi has been bigger than I expected. I know less than nothing about those things. I know about TCP IP, but that's a few layers of the stack away from Wi-Fi and Bluetooth. I guess Aaron did also ask if we focus on one specific technology or language and learn it very well and then move on, or do you learn pieces of everything as you go? I learn whatever is required for the job I'm doing.

And I will learn things that are not required for the job I'm doing, but they also aren't usually super technical right there. I mean, if I read a technical book, it doesn't necessarily have to do with my current job. Or at least that's how I usually start it. Occasionally, I realize, oh, this is my problem, and I should have done this when I'm reading a technical book and having problems at work. Things like good code and pragmatic programmer. Oh, sure. Those things will trickle back into my job, but I don't read them for my job. Yeah, no, and I'll read lots. You know, I read that stuff, too, but I don't read them for my job. Yeah, no, and I read that stuff too,

and I read lots of blog posts and things, and I try to keep up with what's going on out there. But if I see an article about how great Rust is as a language, I don't go out and learn Rust. I remember that Rust exists and that someday it might be important. There's a language called Rust? Yeah. It's keeping abreast of what's going on

and then being able to, on a job, say, you know, I was reading about this particular thing and it might work well for our project. And then you learn it. Yeah. Well, and then you try to figure out, will it work for our project? And then you decide it probably will.

And then you learn it. Yeah. And then you realize at the end of the product a project that you really did a very bad job and need to do it now that you've actually learned it you're never going to know everything yeah and you really can't be expected to okay we can wrap that up okay kevin fitch emailed and told me that my idea about educational Darwinism, where ideas that are taught well get passed on better, both the idea and the way that they're taught, that that came from The Selfish Gene by Dawkins. And so I wrote back and said, well, that had been on my reading list

but fell off when I OD'd on all those genetics books. And his response was hilarious. To be honest, I haven't read The Selfish Gene either, but I have skimmed the too-long-didn't-read versions. And then he linked it to the Wikipedia pages. And there was a little more, but what I wanted to share with was keep up the good work. I love the show,

but I did fast forward through most of the cat interview. And the best part about that, most of the cat interview. Most. Don't worry, so did I. I listened to the whole thing. Yeah, I listened to the whole thing about 87 times. Well, yeah, it was not easy to put together. The number of people who listened to the entire thing was frightening.

I'll just have to say that. It was impressive, yes. I don't know what's wrong with you people. It was funny. It was really funny. I have to say that. So I went to go look at the selfish gene, and it wasn't quite what I had in mind.

Although the idea of memes as units of ideas does sort of work for me. And it's analogous to genes and how they... Yeah, but that's still ideas, not how they're taught. I think that inside this meme, I don't remember what the name was, memology. Memetics. Memetics, thank you.

It did talk a little bit about education, but not as much as I really wanted them to. Because I really find that to be quite interesting, that how you teach something forever will change how it's viewed. Yeah, I'm not sure I agree with your premise that good ways of teaching persist because there's so much bad teaching out there and it persists.

I think, yeah, I don't know. I'd have to ponder that some more. But I'm just thinking back to enough classes where just things were taught so poorly and just it seems like that continues to be the case a lot of times. But I know there are people who are, I don't know. I think things stagnate. And I think things like massive online courses are going to change things.

I think the inverted classroom is going to change things. So yes, there is experimentation going on, and maybe that will lead to what you're talking about. But I do think teaching was stagnant for a long time. Yeah, but it's exploding with change right now. And I'm excited because it is so cool. But I'm a little nervous because there are ways to screw this up. And how are we going to deal with a generation who learned?

Somebody tried to explain new math to me, and I laughed and laughed. Well, yeah. New math was something that was being made fun of in the 70s, so I'm not sure. Is this new new math? It is. It's new new math. It's a new way of teaching math,

and it sounded like it had to do with more estimates and I don't know. I just think it is important to know your multiplication tables just straight. You shouldn't have to think about them. Yeah, but on the other hand, people are taking specific single examples of homework problems or something. And are they not teaching multiplication tables in lieu of...

The one I'm thinking of is the extremely weird way of doing subtraction, I think. Or, yeah. And it boils down to how a checker would count up your change, which I've never really understood either. But it was really a stand-on-your-head-backwards, very complicated way of subtracting, rather than the normal borrow and cross-off and all that stuff.

But I think it's unfair to look at that and say, oh, look at this nonsense they're teaching when I think that's an additional thing. Some hit has been taken to simple arithmetic in an effort to teach some of this other stuff. That's good for us. That's right. We'll always have jobs.

We can divide a check, man. As far as I'm concerned, the next generation, they can just teach them Spongebob. We'll make ever more money. I won't have to worry about looking for a job when I'm 65. It's hard to have ageism when none of the young kids know anything.

I think I saw this Star Trek and it didn't turn out well. None of the Star Treks turned out well. Twilight zones, though. Those always ended well. Speaking of jobs. Really? You weren't going to segue for me?

I don't have the thing in front of me. It doesn't say much. It says the linker colon writing is hard. That's my note for the next segment here. Okay. Speaking of jobs. Writing is hard. Yeah.

Okay. So we have been writing for element 14, swapping back and forth with the Linker blog, and they're paying us to talk about the podcast and things we learn about and think about on the podcast. And the thing we've found, writing is hard. It's a lot harder

than talking. It really is. The temptation sometimes... Because the difference is you have to think. Right. We try not to do that when we have the show. What did you want to say about writing is hard? Only that I am surprised. I mean, as someone who wrote a book, people think, oh, well, writing should be easy. And certainly when I do it for clients, when I have to do requirements or I have to do design documents,

it is much easier than it was when I first started my career. The words just kind of go together easily. But now, as I'm trying to write something witty and educational and related to the podcast, but not a word-for-word transcription of the podcast. It's really hard. It is really hard, and it's hard because, A, it's a little tricky to find what you want to say, and B, it's...

Somehow written word is much more dense, and so you want to convey your idea in a professional, formal kind of way. It's writing, whereas speaking, especially on a podcast, is extemporaneous. You're kind of speaking off the cuff. And we're not fully developing ideas here. That was kind of the point of the blog,

was to further develop some of the ideas that come up on the podcast and go a little further in depth than we can when we're just speaking to someone and doing an interview. Because you don't always think of, oh, I wish I'd asked that, or I wished I'd, you know, this made me think of this, and here's this whole riff on this idea, which you can't necessarily do on the fly. So it's good on the one hand because it gives us an opportunity to do that, but it's also rather challenging. It's been more work than I expected.

And I liked your last one, Be Excellent to Each Other, where it was kind of a riff on Dennis' show about simple code and Chris Feck on empathy and Jack Gansel on being a grown-up. And just, it was a nice little engineering rant. I wouldn't characterize it as a rant exactly. A rant implies anger. Oh, no, it wasn't angry.

But it was a, my previous one had been Device Internationalization Part 2. So, you know, it had been all full of technical details and an information dump. And yours was more op-ed. Yeah. My next one is probably op-ed. But that doesn't make it any easier. Okay, so, after writing is hard. I wish we should actually plan

what we're going to write about when we do the show. Before the show? Yeah, I mean we could plant in little sound bites. If we planned this better it would be much easier. If we planned things then that would be extra work. You are correct. Speaking of planning things, going along, have I told you how much I hate public speaking?

You have mentioned it. It's funny. To me. We do this show, and some people are like, oh, well, she can talk and after many many years of speech class i was pretty good at public speaking but now i have found i don't like it so that whole second weekend in pasadena where i have to talk yeah i'll be talking.

But I won't be liking it. It's funny. I was listening to another podcast just last week, and one of the hosts also had a bunch of conferences that he just did. And he was saying much the same thing. It makes him very nervous. And people he talked to didn't understand because he had a podcast where he talks every week. And they broadcast theirs live, which is a bit different.

But he said, yeah, the podcast doesn't really make me nervous, but public speaking makes me nervous. And it is a different thing. Well, when we're recording the show, I'm talking to you, maybe one, maybe two other people. But it's supposed to be like a dinner conversation, and so it's not that hard. I can hold up a conversation, especially if I can prep it with notes ahead of time. Well, we don't see all the people listening. Oh, I like to pretend they aren't there. Very nice.

You're all figments of my imagination, just so you know. But, you know, it's a different, when you're doing a public speech kind of thing, you have, it's like writing, right? You have something planned that you want to convey. It's a performance more than just an impromptu thing. You have your slides, you have all this stuff that you want to go write. And you have a limited amount of time, and it's live, right? So you can't stop. You can't say, I have to go write and you have a limited amount of time and it's live right so you can't stop you can't say i have to go take a break um so there's a lot of differences well and on the show we don't often take a break but the knowledge that i can is nice you know if i want to go get

another cup of tea or i want to go to the bathroom, it's not a big deal. We just push the off button. I think the big difference is writing. You want to say something meaningful and impactful and you need it to be concise. So there's more artistry involved with crafting what you're going to say.

All the slides are a gigantic time suck. And writing the are a gigantic time suck. Yeah. And writing the speech is a time suck. But I enjoy that. I mean, that is sort of like writing for me that is good. So what's your problem? Other people.

I like it better when they're figments of my imagination. I like it a lot better when they don't have to be real. Anyway, I will be speaking. So after that build-up and advertisement. Isn't that great, advertisement? For your things. Where are you speaking uncomfortably? Okay, so there's a party thing in Pasadena

the second weekend in May, which still requires a private invite, which you can still email, and I'm happy to share it with you. There's Solid in San Francisco in June and ESC Silicon Valley in July. And I have, I'm talking about inertial sensors at solid and about making at ESC and that

thing in Pasadena, I am smashing three separate talks together for a 20 minute talk. Well, that'll be easy. Well, the three separate talks are each 45 minutes, so it's just going to be fast, which I love. I usually talk too fast anyway, and now I have an excuse for it.

How are you smashing three 45 minute talks into 20 minutes? Quickly. I can't even get past the intro of anything in 20 minutes. Well, like the inertial talk, I'm only going to be covering the first quarter of the inertial talk. And even then, I'm going to be doing it fast because I want to get to a particular part

where I'm discussing. Okay. All right. Well, good luck with that. Well, and that one involves alcohol, which I don't think it's easier. I think you should be worried about that one. I'm not so worried about that one.

And the solid and ESC, I'm a little more nervous about. But I will make the slides, and then I will practice and practice and practice and practice. And then you'll have somebody else give it. I only win. And then I will pretend that I never gave it, and I will just write it up in a blog form, and then it will be an article

and then I'll be done with it. Yay. Having accomplished what? I don't know. That's the other problem is that I sort of hate public speaking. I have no idea why I do it. I sign up for these things and then I'm like, why? The Pasadena thing, I am excited about. That one I would do. But ESC and Solid, it's like, it's advertising for our company. We can't take any more work on. It's advertising for the podcast. We don't make any money from a podcast, so why are we advertising it? It's advertising for my book. That one's sort of a plus, but I don't know. That's it. I'm sure there's another reason.

Otherwise, you wouldn't be doing it. Well, I do it for some women in tech reasons. I want to be out there and people should know that there are women in tech who talk about technical things and who really make an effort not to talk about being a woman in tech because that's just boring. So, Star Wars trailer. What does that have to do with anything?

I don't know. I needed a segue. You know, there are figments of my imagination and you aren't a proper guest, so. I'm not a proper guest. You're a co-host. I'm an improper guest. Sometimes. Yes, there is a new Star Wars trailer.

Andre is a listener, Andre from the Great White North, who actually will be joining us as a guest next week, emailed me the trailer about two hours after it came out, which I thought was pretty funny, as though I hadn't already seen it six times by then. Only six? I know, you were up to like 25 at that point, weren't you? Something like that.

Yeah. Yeah. something like that. It's good. I'm excited. There. Chris and I are both sort of Star Wars nerds. I'm whispering because I want them to know it's a secret. If I shouted it, Chris and I are both Star Wars nerds,

then they would think I was yelling at them. You think this is why people listen to our show? I hope so. I wanted to talk about the Star Wars trailer, not because of the trailer-ness, but because of that segment you showed me with BB-8. Oh, right. This week was a big Star Wars convention.

That's where they unveiled the trailer. But the first panel where they did that, they had R2-D2 come out, which is no big deal. Everybody's seen a remote-controlled R2-D2. But then they had the... Oh, I've never seen one in person. I'm sure you have. I think even at the dorky Star Wars events we've been to, I've never seen one in person.

The dorky Star Wars events. Yeah. Why? Why? Why do you have to add that word? Okay, I'm sorry. At the Star Wars events where I had my picture taken with Chewbacca. By the Associated Press. I know. Anyway.

I'm just saying, you don't need to negativize things. We're talking to nerds. Some of them are going to find the soul of Star Wars. Anyway, so R2-D2 came out. But then there was the other droid, which if you've seen the trailers is this utterly bizarre ball with a like an r2d2 dome on top that kind of floats around and it rolls around on its own and everybody thought it was computer graphics on uh when the first trailer came out and then they just they wheeled it out it's a real thing well and hackaday's been talking yeah about all the

different ways you can make one i just just saw, they just posted a... Sphero. Somebody turned a Sphero into an Airsats. The droid is called BB-8. BB-8. The little one. That wasn't actually why I brought that up, you know. Why?

In that panel, there were those two guys who built the R2-D2. Oh. Oh, I thought we were going to talk about how you make one of those and how cool that is an embedded thing that was. Oh, I think it is very, very cool. I also love that XKCD basically... Invented it years ago? Invented. I mean, they published the schematic in 2008.

Yeah, but, you know, a cartoon line drawing versus actually making something is a little bit different. Especially making something that could be controlled as independently as it could. It is, I mean, a ball shaped robot. Wow, why haven't we been doing this for a while? It seems so cool.

The head moves independently, they can turn it. So there's some cool stuff going on. The head moving independently is pretty cool. i don't know how important that is i guess if you're mounting to the movie it's important to the movie um but if you're mounting a webcam that's what you'd do with it you'd want it to be uh rotatable rotatable like that. But why don't we have more sphere droids? I don't know. It seems just... They can't go upstairs.

Don't they have little rockets like R2 does? No. R2 doesn't have rockets. I don't know what you're talking about. Oh, those movies didn't exist, I see. Christopher declares all of one through three not canon. I don't know what you're talking about. But the point you brought up was actually that the guys who did it

were just fans who had been making models of R2-D2. And they got discovered at one of these conventions by Kathleen Kennedy, who was the president of Lucasfilm, who was just walking by. They had brought their R2-D2, and they were checking it out, and it was so impressive that they hired them on the spot, or very quickly then after, to actually work on the movie and do the robots.

It was a neat example of, if you build it, they will come. Sort of, except they didn't just build it and wait for somebody to show up. They went to a convention and they showed their stuff off. But they didn't build it to get a job. No, I guess not. They built it because they loved it. Or because they wanted to build it. Yeah.

And so the fact that they got a job was sort of a dream come true sort of thing. Yeah. It was neat. I think a lot of makers are in that basket where they're building things because they really like them, and then they get hired to do these jobs.

And that's pretty cool. Yeah, I think it is. And they ended up as puppeteers on the film doing all the controls of all the robots. And that's pretty amazing. As a fan, that's kind of like, what? So cool. How is this possibly happening to me?

But, yeah, so I think if people really like stuff and they do an exceptional job and their passion shows through and they get it in front of somebody. Yeah, there's an element of luck here. I mean, you can't just... You do have to work for it. Of course. And they may not have... I don't know what they wanted.

I'd have to read an interview with them, whether they were just doing it for fun. But they may also have been trying to start some sort of animatronics company. Yeah, well, and that's one of the reasons that I think about doing projects is because it's a good way to show off what I can do. Let's see.

Going on. The other great video to drop this week. SpaceX. Oh, the rocket landing. The rocket sort of landed. It landed. And then it fell over. That's fine. I'm not of landed. It landed. And then it fell over. That's fine.

I'm not sure it landed, landed. It was on all of its legs. But not stably, or it wouldn't have fallen over and created that big crash. It's designed to do the big crash. There was fire. It's full of fuel. All right, if you didn't see the SpaceX video, not the crummy picture from far away, but the... No, there was a good one from far away, too.

Yeah, there was a good one from far away, but the one that was on board, that was really neat. And you can see just how close they're getting. Yeah, and the thing that's frustrating to me is a lot of the articles are like, oh, SpaceX has failed to land again. They weren't expecting that to work. They haven't been expecting that to work for a long time yet.

They just keep getting closer and closer, and that was their plan. Well, and I didn't understand, because I had seen those wonderful videos of the grasshopper landing. And I thought, well, they had it working. Why doesn't it work? But you said that wasn't the grasshopper. It Yeah. And I thought, well, they had it working. Why doesn't it work? But you said that wasn't the grasshopper. It's not quite the same. It's very similar, but the difference is that

once it's coming back from space, they have very limited fuel because they don't want to waste fuel. They want to waste weight that could be used for payload by having a whole bunch of extra fuel. So it's got very, very little fuel when it's coming back. And it can't hover. As far as I know, the engine can't actually be throttled down enough

to hover on the real rockets, which means if the engine is on, it's on enough for it to take back off. Yeah. So you can't just light it up, hover, and then slowly bring yourself in, both because the engine can't do it and you don't just light it up hover and then slowly bring yourself in both because the engine can't do it and you don't want to waste fuel so they come in with kind of a one one shot only uh approach and then fire the engine up to decelerate basically to hit zero right when they land and they come in just like a missile because yeah that's what it is but what actually went

wrong had nothing to do with the complexity of doing that. There was a mechanical problem. All that software and it was a freaking mechanical problem? Yeah, some valve was sticky so the control from the software lagged what actually happened and it couldn't compensate.

Oh. Yeah. Control loops are hard. Well well control loops are hard especially when it's broken exactly i mean they're hard to get right because you can do simulations all day long and then you get out there and error friction is not what you expected and suddenly your control loop is like not stable so the thing that amuses me i think it really amuses me is if you look at any comments on those videos which i do not recommend you do but there's just everybody's a rocket scientist youtube commenters explaining how they could just fix it if they only knew about this i i facetiously commented on twitter and said maybe if they put a giant electromagnet in the deck

but i was joking. And I think some of the people doing comments are joking, but some of them are dead serious that their ridiculous idea is the one that's going to fix it. And obviously the rocket engineers at SpaceX hadn't thought of this before. Obviously. Because it's not like they're rocket scientists. Oh, well.

Yeah, well, don't read the internet comments. I think it'll be a little easier. I think the next one or soon after they're going to try to land on land, which is a little bit easier because they're not trying to hit such a small target. Well, they're not trying to hit such a small target, and the ocean is not smooth. No, but I think the barge is supposed to be stabilizing. Oh, good.

So. I worked on some stabilization in marine equipment. Boy, I really can't say much about that project. I don't even know about that project. It was really hard. I've never heard of this. How is it I don't know about this? It had to do with race cars and tractors, and we

got cut out pretty quickly because we weren't set up for that sort of thing. You never mentioned this to me. We talked about going to Hawaii. What other things haven't you mentioned to me? Oh, what kind of neat things? You gotta keep up. Are you like an assassin? No. Although, I have to admit, I didn't take the spider outside in a

paper towel last night. I didn't even know there was a spider. See, this is yet another thing you haven't told me. He was in the bathroom. Ah, let's see let's see things that I am working on now yeah I can't talk about any of those either you can talk about the inertial thing I can't talk about any stuff I'm working on well I heard you were looking at a new processor

for the things I can't talk about? yeah I figured we could talk about the tech if we can't talk about the project sort of I mean I'm not looking for the company is looking for it, and I'm watching. It looks like Cortex-M4s are the new hot things for small embedded devices. Cortex-M4s take a lot of processing power. They take a lot of processing power?

They take a lot of power. They do? Well, Cortex-M3s are lower power, usually. They seemed okay, but these are newer ones. Only have floating point units. They have floating point units. They do have some other interesting features. Floating point units and

some of them have some other DSP kind of things, I think. Or is that part of the M4? I don't think that's part of the M4, but a lot of them have them. There's a whole bunch of those coming out um one of them that uh isn't specifically something i've been looking at but just got announced was the msp432 which is odd because the msp430 is a really cool low power right low cost ultra low power low cost micro and cost. Ultra low power, low cost, micro.

And the 432 is a 32-bit version of that based on the Cortex-M4, if I've got that right. Yeah, Cortex-M4 processor, ultra low power. And TI is claiming you can easily port stuff from the MSP430 to the 432, which is surprising. So there must be some kind of translation layer that they're doing. TI has had their BIOS thing for a while, which has been their gold translation layer

and sort of kernel operating system. I wonder if that's what they're doing. But it seems like a really neat project, especially if they keep it... Yeah, so the Launchpad dev board is $12. So it seems like it's still going to be pretty low cost. They're not really cranking up. So some of the M4s I've been looking at

are in the 100 MHz plus range. So the 432 is 48 MHz, so it's not super high power. But it does have the floating point unit and they're claiming 95 microamps per MHz when active.

Wow! And 850 nanoamps in low power mode with the RTC running. That's not terrible. That's pretty good. And awake from sleep is 10 microseconds. So it seems like a neat little processor, especially if you're interested in moving from a smaller embedded processor to the ARM space.

That would be something to look at. Well, the other one to look at is Atmel's LMCUs. They are also 32 bits, and they say power consumption 35 microamps per megahertz. That's an M0, though. And 200 nanoamps in megahertz. That's an M0, though. And 200 nanoamps in sleep mode. Right. Yeah, it's Cortex-M0.

Yeah, those are Cortex-M0s. Man, ARM really is just dominating right now. That's an M0 running at 48 megahertz, so that's not a lot of power there. They're claiming that in active mode? Yeah, 35 microamps. I suspect that means you have to have all of the peripherals

turned off and all the timers. Maybe. Actually, we're going to talk to Andreas from Atmel about comparing processor power, because I hate the fact that you can get such wildly different answers and not know. It's like any other specsmanship. There's a lot of ways to game the system

and say, yes, you can have this low power asterisk as long as you're not using the RAM. Yeah, he'll be by in June to talk talk to us about power questions so if you have questions about power then let me know mostly power in mcus and how can you compare them is my goal the other processor i've been well i've been spending a lot of time doing bluetooth low energy i mean just everybody who wants to talk to me right now wants a Bluetooth thing. I think that's, going back to Aaron asking what to learn. Oh yeah.

If you want to be marketable, learn Bluetooth low energy right now. At least for the next couple of years, that's something I think that's easily employable. I haven't done anything with it yet, except at extremely high level. Most of your products use it, i haven't worked on that section um so what's what's what's what's that like well if you were going to learn it i mean if you wanted if you were in aaron's boat and wanted to to learn something new um the ble nano from red bear was what i picked up and really liked okay uh it is the nordic nrf 21 something 22 put in the links yeah and it's uh it uses embed so it's an online compiler which is nice um if you do something else then you have to do like Kyle Compiler. And a lot of people are doing Kyle Compiler and using the free version

and just limiting the code size. Have you used Kyle Compiler? I have. How does it compare to the other one? The compiler that shall not be named. It is just as bad. Oh good, that's nice. They're really all just as bad.

Wonderful. Okay, so I'm skipping that. So you can use the embed thing, which they have a lot of libraries. I've used the embed thing. Yeah, you know, you say, you know, I'm going to use this with such and such accelerometer

and you just type it in and you find somebody else's library. You have to worry a little bit about the... Where did it come from? Where did it come from? Does it work the way I want it to work? And what are the licensing issues? Because not everything is, you know, free.

What? I know, it happens. And if you're not having, if your code is not open source, then you have to be really careful about that. But it's just a list of examples. Well, the whole idea of using the online IDE for a real product is a bit strange. I think

you could start there and then export to Kyle if you wanted. But what's cool about that unit hardware-wise? The BLE Nano? Yeah. It's very small, and so if you're developing a wearable sort of thing, it makes it pretty easy. It's also because it is involved with embed, it's a little easier to compile for. When I set up the Kyle version, I actually was going from one version of the

Nordic SDK to another version, and that process is painful. And so i had a lot of problems with the kyle compiler getting to understand that i really wanted two sets of libraries so i could go back and forth and that really frustrated it and me but in embed it was much simpler because it was more monolithic you took this ball of stuff and it all worked together. And so it was easier to find things that all worked together without having weird compiler-related things. I liked it. But they were both the Nordic chip. And that's a nice chip.

It's not horrible. They're certainly doing a lot of development on it, getting it so that you can use Ant or Bluetooth. Is that a big enough micro that you could put your application on it? It's got a Cortex-M0. You can't use all of it because some of it is being used for the other stuff. But it's not a bad little chip. I mean, you can do a lot with it.

But if you had a remote sensor or something, it might be enough. Oh, yeah. A remote sensor, a small screen, a couple of LEDs. You could definitely do that. I wouldn't want to put a big screen or multiple layers of data on a screen. Sure. Yeah, that's not bad. On the other hand, if you want something a little different,

maybe a little bigger as far as processing space goes, TI has their CCA2640. And this came out not too long ago. They used to have their 2540, but the 26 is very different. It has a Cortex M0. It also has all of its Bluetooth stuff on a different part of the die. It's not sharing the chip with you, which is really nice. But more importantly to me right now is that it also has a 16-bit processor. It seems like so many times when I use Bluetooth,

I end up putting all of my device onto the same processor using the Nordic or using the CC2640. And then I have to have an auxiliary chip, something that can be used to put it in super low power mode or something that can be used to put it in a super low power mode or something that can be used to control very timing sensitive things. Beyond iSquare Seed or Spy

maybe the NeoPixel WS2812. Those have really tweaky timing requirements. And other things that have just real time needs. So you can't so if the bluetooth's in the way if you're sharing a processor with bluetooth or you're you have

an operating system because bluetooth kind of wants you to have parts of one uh then having another processor seems like the way to go but it's expensive and sort of lame. The CC2640 actually has a little tiny, they're calling it a sensor controller, and it seems like it's sort of a little MSP430 in there, and I like the idea it's all in die. No, that's cool.

You're probably going to need it anyway, so. Well, they had ridiculous power numbers, too. One thing that bugs me about BLE projects is you can't really do anything unless you can build an app on a mobile device, right? I mean, I guess you can talk BLE to BLE. I mean, if you had two of these, could you make a sensor network?

Oh, yeah. So you could do that. But otherwise, you're going to have to write something in iOS or Android, which requires a whole other level of learning, right? The people I've been talking to, they're like, yeah, that's what we're going to do. It makes me sad. It makes me sad, too. Some of them want to do gateways to Wi-Fi up to a cloud server. I have not been as encouraging

of that as they all want me to be. Oh, that's a big thing. That's a huge thing, actually. I think if somebody could design a generic way of doing that... I think most of the routers that are coming out for home now have that. I think that's a new standard, is some sort of BLE

gateway. And then wearables won't have to talk to your phone to get to the internet. They can just go straight to the internet. But the way BLE is set up, I don't know how they're going to do the conversion layer. It's IPv6 over BLE. Oh, well, then your processor has to be able to speak IPv6. Different story, yeah.

Gag me with a spoon. Wow. What? Oh, well, I guess that is a little old, yeah. Yeah, I've made the spoon. Wow. What? Oh, well, I guess that is a little old, yeah. So, yeah, BLE is cool. It's a good skill to learn, I think. But it's changing rapidly. They're adding a lot of new stuff to it.

They're adding so much to it. So whatever processor you pick may not even be able to do what you need it to do in the BLE space. And the apps thing, I guess if you're doing Android, it's probably easiest to start with Android. whatever processor you pick may not even be able to do what you needed to do in the BLE space. But, and the apps thing, I guess if you're doing Android, it's probably easiest to start with Android.

Yeah. I don't know if a lot of laptops, there are some laptops that have BLE. I don't know how accessible it is from the normal development environment on a laptop, but that might be the easiest way to go. If you wanted to write some application that talked to a mobile or higher-end computer. Oh, see, doing this development,

I have a couple of the little Bluetooth low-energy dongles that go into my computer and work with one that works with Nordic's app and one that works with TI's app. And they act as sniffers. Yeah, yeah. And I have built a few things around them so that I can... Snarf data. Snarf data and look at it.

That's where Python comes in. So, yeah. Let's see. The only other question I have are rules to guide ethical conduct. Do you have any rules to guide ethical conduct? Be excellent to each other. That's nice. That's a topic for an entire show, probably.

I don't even know what I'd say on that show. I mean, don't do something you would regret. Don't do something people will... There's gray areas. There are gray areas. I turned down a contract for you, you know. What? It was a good contract. What? They wanted me to do... Oh, right.

They wanted you to do fitness stuff. They wanted me to do fitness stuff. And I really, really liked talking to other people. And I thought where they were going was really neat. Well, that's just because we share the company. If we had separate consulting companies, that wouldn't have been a problem. No, but I wouldn't want... But that's the kind of thing that comes up.

I mean, it's a big topic. It didn't occur to me that it would be a problem until I started burbling to you about how cool they were, and you were like, we can't have the hardware in the same room. No, it was that we can't have the company be under contract. To both, no. Yeah. Yeah, Sath ethics, big deal.

Be nice. Be nice. Don't do something you regret. Don't write anything in an email you don't want a lawyer to read. Don't start a land war in Asia. All right. That covers my list of things to talk about. I'm sure that there were other emails that i should

be looking at but i failed send us more we'll talk about them some other time if uh if you asked for a sticker and i haven't sent you one yet they should all have arrived by now so ask again i don't know sticker you had a sticker you were playing with it like a second ago it's gone it's gone do you see it it's on your forehead what that's not it will be later when you wake up great i'll sleep elsewhere so i do still have two stickers i'll just ask for them um send me your email or send me your snail mail address i can't send them through email. You can send a picture. Here's your sticker. Here's your sticker printed out on sticker paper.

Yeah. And then we got a hundredth episode coming up soon. Yeah. I'm still trying to put together a list of the five best starter episodes. Yeah. I haven't had much feedback yet. No. The figments of't had much feedback yet. No. The figments of my imagination are not responding.

So if you have a particular episode that you really enjoyed and think should be part of the Starter Pack recommendation list, send it our way. And some people listen from episode one and they go on, and other people listen from here, and then they go back occasionally to things they hear that might be good. We don't air these in chronological order. Yeah, we recorded this one

in 2006. We're going to put the chronological order list up at some point so the continuity is better. I really was going to put up a list of all 100 episodes and links to them for the 100th so that we have a good list of where they are

and who was on and what we talked about are you going to write an aux script to do that? I was going to do it in Excel I know that's really sad I should write an aux script but I know Excel better maybe I'll do it in Python. So yeah, send us an email if you would like to put your favorite in. My two favorites are the Jack Gansel one and the one about imposter syndrome.

So those will be on the list. What else should we add? What did you like? And I guess that covers it. Do you have anything you'd like to say? Do you have any final thoughts? My final thought is that the Tylenol is not working. Ah, right.

Christopher's getting old and we're all sad. Don't ever reach for a fork. Is that what we've learned? Yes. Don't move. The last time I believe we learned don't sneeze stay motionless try not to move and never twist at all even like this is not the slightest amount well thank you christopher for co-hosting yep and producing yep ow i'm really sorry my nose was stuffy. If you...

What? Yeah, well, if you listeners would like to say hello, or, you know, any of the other things, coupons for solids, coupons for books, coupons for whatever party invites, hit the contact link on Embedded.fm or email us, show at Embedded.fm or email us, show at Embedded.fm. And thank you for listening.

I know that you are out there and I do appreciate it. So my final thought for this week is from Bjorn Strusup. What? Oh, did you want the joke instead? No, the pronunciation of that was unique. Oh, how do you say it? Bjarnstrøstrup? Bjarnstrøstrup.

I think. That works. We're sorry, Bjarn or Bjarn. Yeah. Ah, C. Makes it easy to shoot yourself in the foot. C++ makes it hard, but when you do, it blows away your whole leg. And he did. He's the creator of C++, so he should know.

Yeah, I think so. Do you want to do the joke too? Sure. What do you call the cat that was caught by the police? Uh... The purr betrayer alright I'm gonna edit that out laughing