Embedded - 182: Sorry Little Diodes

Episode Date: January 4, 2017

Chris (@stoneymonster) and Elecia (@logicalelegance) talk with each other about about a party, listener emails, and assorted questions. RSVP for the Embedded.fm party! The Embedded Blog is at embedded.fm/blog. Chris Svec wrote a post about picking a processor platform. Don’t Panic Geocast episode with Elecia Elecia’a book: Making Embedded Systems Compiler explorer is GodBolt.org Imposter Syndrome: episode #24 is all about. And you might find #78 with Chris Svecrelevant. Also: Adam Savage talking about overcoming self-doubt. The RSS feed for all of our shows (not only the most recent 100) is http://makingembeddedsystems.libsyn.com/rss We have a Patreon fund that buys mics for guests (plus the occasional goodie for us and our blogging team). Crunchy Frog RTL-SDR: Software defined radio BaoFeng’s unusable Ham radio ESA investigation on the Schiaparelli landing

Welcome to Embedded. I'm Elysia White. This week, Christopher White and I will be chatting about emails from people we don't know, failing to answer questions from people we do know, and generally discussing parties. One party. A party in particular. Shall we do the party first?

Sure. Let's see. We're having a party. Embedded FM listeners, you are invited. You may note that I have not tweeted this. I have not posted it on the blog. It is for listeners primarily. I may tweet it before we actually have the party. Anyway, January 28th, it's the last Saturday in January, and it's in Aptos, which is near Santa Cruz, and 2 to 5 p.m. there is an RSVP Eventbrite thing. You should check it out. Where? Aptos, Little Owl in Aptos.

It's a sort of pizza Italian place. They have a nice separate room. We're saying party, but there won't be dancing. No, no, there won't be dancing. Oh, but there will be hats and hacks. You're supposed to bring a hat or a hack. You're supposed to be able to tell us about your hat or hack for like two minutes, but not for like 20 minutes.

It's not going to be formal presentations. It's just going to be a people wander around and say, hi, I have this interesting hat. Your hack looks really great. Maybe we should merge them into a single hat hack. So people can come and hawk their hats or hacks. We need to keep adding things to that, aren't we? Let me look up all the words that start with H.

And end with AK. Right. Okay, come to our party. And end with AK. Right. Okay. Come to our party. Come to our party. Meet us. Meet us. Meet a few past guests.

Meet other people. Meet interesting other people. And then afterwards at five o'clock we'll kick you out and you can all go wander the tide pools. It doesn't get dark until 630, so you'll have plenty of time. It's low tide. It'll be great. Assuming it doesn't rain. Assuming it doesn't rain. Assuming it doesn't rain.

Yeah. There's that. Next. Hello to Alex and Steph, South African listeners who were biking from San Francisco to Los Angeles. Oh, not from South Africa. And they stopped, and we had a lovely coffee, and it was nice to meet you, and I hope you had a brilliant coffee and it was nice to meet you. And I hope you had a brilliant time and it didn't rain too much. You could do that.

You'd get one of those float bikes. One could do that. Yeah. Yeah. Let's make sure we have pronouns properly. Somebody gave you a new moniker? Yes. I was on the Don't Panic Geocast with John Lehman.

And he, well well it was great i had a lot of fun but one of their listeners jenna uh called me the bob ross of electronics and i absolutely adore that the hair doesn't work the hair could work you, I actually have to straighten it to get it down. So, it could, yeah, it could work. Happy little what? Happy little capacitors. Happy little capacitors? Yeah. Cheery little inductors. Sorry little diodes. Nice reflective Arduinos.

I liked it. It was such a nice thing to say. We have a lot of topics on this list. People are going to think that we're kind of already out of material. This is it. But this is a long list for us. It is a long list for us. And we may cut some.

What? No. All right. Well, should we should keep going then all right what's next andre andre's complaining i talk to him all the time well i i just wanted people to know that ah andre is not sure if he's got readers for the blog if they're very quiet or if they're busy or uncommunicative or just they don't exist. Has he considered the possibility that his blog is perfect in every way and requires no feedback? Well, after I finished editing it, it should be like that, right? Wow. Yeah, not so much.

Anyway, the blog has been on hiatus for the last couple weeks so i think that's why it's quiet there's this holiday thing that happened and the blog will be back next week it is embedded.fm slash blog so it's really well hidden but if you would like to give people encouragement uh spec or andre or me you give me encouragement too i don't write much on there but go ahead encourage me i won't get you anywhere an encouragement for christopher and four bucks i'll get you a cup of coffee four bucks for you buying coffee some of the coffee right really expensive but yes uh let's see more postings this week next week whatever uh on that same vein for the people who email me shocked and surprised that i have a book uh you crack me up when you send me that

email i just want you to know that your surprise and your shocked email telling me that I have written a book is funny. Did you know you wrote a book? I actually was aware of it. Yes. It's called Making Embedded Systems. This is just an indication that you haven't been doing enough promotion. Yes. Well, now I'm just going to make fun of it.

You haven't reached the point of saturation yet. I actually have two books. Yeah. But you should only buy Making Embedded Systems. So you just have a line item in here last week's show with Chris Gamble we did a show

with Chris Gamble last week of the Amp Hour of the embedded.fm yes yes and it was cross posted and it was sort of fun

yeah and it was really long yeah but it was really fun yeah okay yeah Andre had a comment

we were talking about PWM circuits. Yes, making a PWM circuit, which was kind of a loosey-goosey discussion. And apparently he was yelling at his iPad about using a 555 timer IC. Which is totally an obvious solution. Right. I had it in my head that there would be no ICs involved. I had it in my head that I would look up the solution.

That's not fun. I know, and it's not something I think you should do for everything. We've been watching Man vs. Wild. There must be a way to construct a PWM circuit out of a dead sheep, some plastic found on the ground, a rubber band, and something truly

disgusting that he finds. Yeah. For those of you who don't know, Man vs. Wild is a Discovery show or a BBC show that he goes off. He's an ex-SAS, ex-military special forces type. He goes off and he does fairly crazy survivalist things and it's amusing.

I think we should do a show that's all electronics like that and you're in one of those surplus stores but you can't go to the front section where they have actually finished electronic. You have to go to the back corner to get all your parts or the stuff has spider webs and rust. You have to piece stuff together

transformers that are as big as your head and try and dials and VU meters and so you're saying we should spend a weekend with john shuck no no no you make it a show it's a survival show like the oxygen is going to run out so you have to make an oxygen generator in a half an hour and if you don't you die or or uh or there Or you're going to be electrocuted by something that's attached to you and you have to figure

out a way to shunt the electricity off of you to survive. These are good ideas. Yeah, I miss Junkyard Wars too. No, no, no. That was different. They had parts and that was just mechanical. That was sadly too mechanical. okay uh also on the things that you're changing the subject going back to popping the

stack uh back to chris gammill uh having him on the show we talked about christmas presents one of the ones that we didn't talk about much was your radio, your new radio. I got two radio things. Yes. My parents got me two radio things. One is the RTL-SDR, which is this little USB stick. Looks like a thumb drive, a little bigger than a thumb drive. It's a big thumb drive, and it's got USB on one end and a BNC.

That's not BNC. That's BNC. That's not BNC. That's BNC. An antenna connector. Antenna connector, an SMA connector on the other end for antennas. And it's actually a hacked, I think a hacked TV broadcast receiver. But they've hacked it up and it turns out these can receive, I think it's anywhere down to like, I'm misremembering, but I know it goes from like tens of megahertz up to 1800 megahertz. So it covers a huge wide band, and it's a software-defined radio receiver.

And you plug this into your computer, and you load up any of 15 open-source software packages that are available and you can just scan through stuff and get these beautiful waterfall plots and you can go look at different frequencies and knock yourself out receiving all kinds of weird stuff i don't know what to do with it yet but apparently there's i mean there's stuff you can do like uh well you can listen to fm radio if you're really boring but you can receive the broadcasts from airplanes the the position broadcasts. I forget the acronym for that. And decode them and get the GPS coordinates so you can make your own map of all the airplanes flying around you. 1800 megahertz?

Yeah. That stops just short of truly interesting. Truly interesting for people who want to use this to look at bluetooth and wi-fi sure well i mean those were the ones that there's cell phone yeah that would be kind of neat to look at no there's a ton of stuff in there uh and you can you can go further down uh so if you want to listen to like uh amateur radio hf bands uh you can get up converters, which go in between, and they'll shift that stuff up

into the receiving range of this by a known factor. So that's pretty cool. I only played with it for a little while. I'll show it to you. You'll like it better. I'm sorry it doesn't receive. I'm sure you can get ones that go higher.

Oh, yeah, of course. And most of the ones that go higher and are in the ranges I want are more likely to be... What are you going to do with that anyway? I mean, yeah, of course. And most of the ones that go higher and are in the ranges I want are more likely to be... What are you going to do with that anyway? I mean, you're not going to... That's all complicated stuff. Well, BLE actually debugs sometimes, so...

Can you debug it from the wire? From the air? You can sniff it. Especially if you're writing the software and you can make it unencrypted to be sniffable. Anyway, that was cool. I also got a little handheld radio to start playing around with

2 meter stuff but it turns out that according to listeners and twitter friends twitter friends it's not that high quality and sometimes has spurious emissions so I may not be pushing the transmit button.

It would probably cost me more than new radio to get it checked out. I was surprised they could sell things like this. Yeah, I didn't think you could. It's Chinese, so there's no import restriction. There's no way to stop somebody from making a transmitter on any frequency. So there's really no way to stop them from selling it. I don't know. I guess it is the person who has the ham radio license is supposed to understand enough of this to be culpable.

Yeah. And so they, they are responsible for knowing both the radio and the regulations and whether or not they can push that transmit button. Exactly. So you are being good, but your radio is not. Yes. All right.

So finished technologies. You put this one in here. I did? Oh, it was a thought I've had repeatedly, and I opened it up on Twitter because I wondered what other people thought. But it was the idea that what things should we not really be spending any time on as a culture, as a technological culture? And you have toothbrushes and razors as your examples.

The obvious examples, but I mean, what more innovation can we really do with toothbrushes and razors? You know, they keep adding another blade. But they kind of work. Toothbrushes work. Razors work. Do we really need to spend any time on them? I have to admit, getting an electric toothbrush really made a huge difference.

Yeah, but that kind of electric toothbrush has been around for decades. Yes, if only I'd had it then it wasn't toothbrushes beside the point it was just it's interesting thought about where we put our resources and capitalism being what it is you put resources where there's demand um and there must be demand for advanced toothbrushes which is fine i have no problem with that it's just passing thought. Other people had some things they weighed in with like somebody mentioned toilets especially after being in Japan.

Somebody brought up razors. Vapor compression refrigerators beyond adding better insulation and using more efficient motors. I don't know. Engineering isn't just about making something energy efficient. And it isn't, I mean, there's energy.

You said anything about energy efficient. Well, the reason these things are continuing development is because there is some number of sellable and which is based on features and price and... I think there is some transition. You're optimizing for different things. But I think there is some transition layer where it goes from being actual technology improvements

to fashion, right? Like toothbrush, plastic handle toothbrush. I'm not talking about electric. Plastic handle toothbrushes. Yeah, there's... The big differentiator there are superhero character shape handles. Right? I mean, I don't know. I could be wrong.

It was just a passing thought. I'm not pushing any agenda. All right. Well, that is... And of course, Saar mentioned forks. But then he backtracked and said, except for people with disabilities. And there's the...

The steady fork. The steady fork and things like that. That's amazing. So I wasn't thinking about, you know, advancements that help people that have to interface with the world differently. I was just thinking about bog standard, you know, stuff. But yeah, obviously there's lots of advancement to do for accessibility. the world differently. I was just thinking about bog standard, you know, stuff. But, yeah.

Obviously, there's lots of advancement to do for accessibility. So maybe we should be putting some effort there instead of different bristles. Okay, I'm done. That does kind of bring up a question we got from Elizabeth that is completely unanswerable. How much do you think about the ethical

issues involved with the product you're working on? For example, she works on something that involves quantified self people, and some of those people can get a little obsessed. She may be feeding the anxiety of others. And I think what she's worried about is not super critical, but I do think the ethical issues, how much time do you spend worrying about them? I think what she's getting at is kind of almost a utilitarian cost-benefit thing are the number of people being helped far outweighing the number of people who might not be i think in that case probably so but i don't know how much attention i think i think it's a deeper question when you're actually working on something that is borderline right i, medical devices are kind of murky. Weapons are not murky.

Things that could be used in weapons but have applications, primary applications elsewhere, that gets a little trickier. I think we should all be questioning these things and thinking about them. But at the same time, I don't think we should necessarily spiral into the problematic nature of all things. Yes. And worrying about the fringes sometimes is a little too hard. If you're a confectioner working on the crunchy frog from Monty Python, you probably should be.

Remember crunchy frog? All right. We'll put a link in the show notes. All right. I don't even know where you were going with that um i tend to worry about the ethics at the beginning of a project when i'm trying to decide if i want to work on the application and the project and then after that it's mostly a matter of staying on the right side of privacy and security issues and i don't worry about the ethics of the whole thing anymore.

Right. It's what could you do wrong with this perfectly innocuous product to cause somebody a problem? I think that's a greater question usually than is the product in and of itself going to cause problems or be misused. You can release something that's completely innocent that has the potential for great harm, right?

The webcams that took over the internet a few months ago. So there's kind of tactical ethics that have to do with how you do your job, regardless of what product you're working on, right? Yeah, and that is security and privacy are the two big ones there. Quality is one that is important, often depending on your application. Quality in medical and transportation is, of course, super important. Quality in children's toys, software you're not gonna hurt anyone you may teach them the letter n says puh which would be bad but it's or fun or fun sort of fun you could just advance the entire language it's sort of sad you can't i mean the leapfrog toys i worked on

it would have been very hard to hack them that way. Unless you actually rewired, but even rewiring was hard because it was a row matrix keyboard. Anyway, yeah. I don't know. That's too hard. Okay, let's go on to other hard questions. From Svek. What changes do you think our industry will see in the next one to five years

do you think we'll get an actually good ide in that time no me neither i think more people will start shifting to open source tools those get better yeah and they are getting better um i think tools electronic tools will get cheaper continually. Yeah. I'm already seeing that with... Oscilloscopes and logic analyzers are super cheap now. And debug tools and things. And debug tools, USB analyzers are still expensive. Sniffers are expensive, but they're getting cheaper.

I wouldn't bet huge money on it, but I would like to see more automated testing facilities for embedded devices that are so that you don't have to roll your own things that could observe I2C, observe different buses. I brought this up before, but rather than just being an analyzer,

something that has more intelligence to say, okay, your rise times are bad. Yeah. They could go through and, you know, if you have the right right test points you could put it in a harness and it could just do a complete signal integrity check of your entire board without you having to find out that something's wrong in software and then go follow it all the way down to the wire that is okay so i i have noticed in the last i i would say, five years, we've gone from chips having very limited amounts of software to chips being shipped with huge drivers and whole RTOSs and even applications. I mean, you look at Nordic's tutorials and it has, you know, you just add this sensor and it's a heartbeat monitor.

And it's complete. I mean, it's totally finished. And if you go on an STM product, you have that cube thing where you can rewire things and then you push the button. And it will do DMA and interrupts. And it's much better drivers than you used to get from the old HAL. So I feel like we're headed towards writing less software. But I wish that those tools and that software they gave us was of a higher quality. It was MISRA.

It had unit tests. It had places for our unit tests to add in. I think we're getting faster, and we need to be getting better next. Yeah, I think it's a problem. I don't have high confidence we're going to get better. But I'm not sure. Whoever gets better first?

Maybe. I mean, I look at some of the code I've gotten from vendors, and I can see how they were trying to be helpful, and I can see how they spent a lot of time on it. But in the end, I couldn't use it. I think it's a matter of incentives, right? Do they measurably sell more units the better that stuff gets? Can they attract talent to go work there?

Because you need to write good, solid, embedded code. Example code. It's not just solid, embedded code. It is example code, which is an order of magnitude harder. And just speaking for myself, that wouldn't be a job I'd really be thrilled to do. So I'm not sure you can attract really senior people to go in there and do a good job. Maybe that's neither here nor there, but... I think it has to be in conjunction with some education outreach.

Yeah. That's possible. I do think, you know, the Arduino way is pointing that way. But that hasn't quite merged with, that hasn't moved reliably outside the hobbyist space. So as soon as something makes that leap, that will be large. But I still think there's going to be a big call for custom stuff. Oh, yeah. And no matter how good it is, there's going to be a point in many projects where you you're going to end up, no matter how good it is,

there's going to be a point in many projects where you throw all that out and somebody has to write stuff again. Because I need 5% more battery life and there's no way I can get it out of this example. And with the examples, that has been very true. With the Cuba Mac stuff, while I didn't really enjoy using it very much,

it had all the flexibility I needed to be able to do the low-power things I wanted to do. I'd like to see them break up parts of those things. Because a lot of the problem with developing embedded software is all the flags and things and registers you have to set to get a particular configuration. Well, that was cute to do that graphically.

Or to configure a bus. It's very complicated sometimes to set up iSquared C or SPY, especially when you have DMA involved and you might be changing some sense of the high and low times and things. And that can be very complicated, and it's different on a per-chip basis and a per-peripheral basis. It would be nice if that stuff was separate from, okay, I do that, and I have to take all this other crap, you know, the whole driver or middleware above it.

Is it possible to just say, hey, CubeMX, I want to set up my ports this way, I want to set up my clock this way, I want to set up my I2C, but you stay out of the rest of it? I think so. I think things that weren't set up weren't disabled. That's not what I meant. Yes, I think if you don't add their RTOS lighter

or any of their middleware layers, like their USB driver system, you can do what you want. Just do configuration and init. I don't want their driver. I want the part of their driver that sets up the peripheral, but I don't want any of the rest of it. I don't know if you could do that without going

into modify and delete their code or by never calling their functions yeah all right uh okay so i think i think we'll see a move towards more quality as well as a move towards more software provided um and i do think that it must be selling chips because the amount of software provided in the last five years has grown immensely. I mean, it's huge. They're spending all their money on that. What else? Well, so CPUs are going to get more and more powerful.

Yes. Which means that we're going to get RTOSes truly everywhere. A drive. I think at some point there'll be a drive away from Cortex-style MCUs to A-series-style arms with lots of peripherals. 8051 forever, man. And then the question ceases to become RTOSs

and becomes Linux everywhere. That would be something. You wouldn't be wasting your time as an embedded developer if you haven't done any linux programming to start playing around with linux yeah that's on my list of things especially embedded linux and figuring out low power linux you that's a good job skill and i know there's probably a lot of people cringing out there and i internally cringe a little bit but i'm just gonna hit my head on this microphone okay but uh yeah i don't know i mean i don't know how that'll work i don't know if they'll it seems like the m series they kind of slowly bring features over piecemeal i don't

know if it'll be like that or if at some point there'll just be a cut over and say look now you've got an MMU and caching and a bunch of other stuff that we can put on this silicon without costing anything. On the other hand, 8051s are still out there, so it's not like the Cortex-Ms are going away in a year. No, I think they'll still exist, but the Cortex-Ms will become the 8051s.

Yeah. You'll use them for, they will continue to get better in more and more low power, right? So you might have things that, you might have a cost-benefit analysis that changes from today where it's like, well, I have to use a Cortex-M. It's the only way I can get low enough power and low enough cost. It might be between that now and, okay i use a cortex m i can get 30 days battery life or if i use a cortex a i can get 10 days battery life or five days battery life but all these but my development time goes down yeah so that i think that's where things are going to move around yeah all right uh how do you think our consulting jobs will change in the next one to five years?

No comment. Yeah, I didn't think you had a comment on that. I have been doing more proper consulting, where people come, they say, I have this problem. And I say, well, you should try that. And they go away and they try that. I did that this year, last year, whatever year. It's sort of fun. I really enjoy it. I could only do it given the amount of experience I have and my continuing interest.

I mean, I do go out and look for new things. I couldn't just do it on its own and stop learning. But it is pretty fun. I would like to do it on its own and stop learning. But it is pretty fun. I would like to do more of that. It's hard to bill for because once I hear the problem, either I know the solution or I don't. Yeah, mine was not such a great experience. Yours would have been a great experience

if the company hadn't tried to shoot itself in the foot. Well, I saved them tons of time and money and they decided not to do that. That's their prerogative. Yep. You don't still have any stock with them, do you? I do. It's not their prerogative.

Shareholders should decide. I don't own the stock, so... I have no prerogative. How have your consulting jobs changed in the last one to five years? Not a whole heck of a lot. 2012? Well, I'm doing more of the consulting and less programming, but I'm still doing plenty of programming and architecting and design.

We get a lot of the same kinds of requests in the last year or two. I am so sick of sensors on a BLE stick. I mean, I have this sensor, I want it to be on BLE, and I want a light. Good. Have fun with that. Please, please, please.

How many has it been? It's been like five or six things you've done where it's some... That I've actually worked on? It's been more than five. It's an accelerometer and a light and BLE. Actually, we've gotten approached for more than 10. Yeah.

So, yeah. BLE on a... Sensors on a stick. Just choose a different sensor. Yeah, choose a sensor I've never heard of. And maybe not a light. Maybe a beep. A motor.

But then that sort of defeats the purpose of B maybe not ble maybe zigbee no or ant no or ants actual ants we can lay lay a pheromone trail and ants carry no no uh okay uh meaning the type of work you've wait a minute oh that question actually had more to it oh well sorry spec you're the sort of person who didn't read all the directions before starting a test right oh and when i make recipes it's always such a disaster cumin you Cuman? You wanted Cuman? From Andre. It's sensing a theme here. What events will make the embedded systems industry grow up? That's right. Smart devices being used as denial-of-service drones.

Car companies getting caught cheating on their emission tests. Fame-seeking hackers taking control of medical devices. How do we blow the hell up? All of the above? I mean, those have all happened. Don't plug medical devices into networks? Don't design medical devices that go into networks? Those, I mean, yes.

And yet, the guest we have next week, he talks about doing peer reviews, and while I agree with him, I still was sitting there through most of our talk with him thinking, Resisting. How can I possibly convince somebody else that this is the right way? I'm sorry for people who are confused. That is the next show which we recorded before this show. I asked Andre for examples of other industries that had to grow up.

And so that's something to think about as an analogy. It's like, well, okay, what other industries had problems? The auto industry is obvious, right? There was the whole safety period of time of learning about safety and how to do safety. I think the medical device industry went through that as well. I think things were a lot more fast and loose in the 60s and 70s. Certainly talking to people at companies I worked at who were around and involved in developing medical devices in the 60s and 70s, boy oh boy, you could try all sorts of stuff on people. And you know, it takes a few high-profile,

unfortunately, it seems like it takes a few high-profile incidents before things change. And I don't know, I don't like to say the government has to step in because that should always be the last resort in my mind. But it's different because embedded, it's not cars, right? It's not medical devices. It's not one industry. It's a huge umbrella.

So, you know, some parts of the industry don't need to grow up. I don't think the toy industry, unless they're, you know, making internet connected things and caught up in the whole internet of things problems. But, you know, there's certain classes that are fine. And then there's other classes that, yeah, they certainly have made headlines. And yeah, I don't know. I'm open to other people's suggestions. It's an interesting topic of thought. Well, it's hard because most of these things do sound more difficult. Unit testing, peer review, writing requirements documents, user testing. That's all tactical, right? That's not an ethos of you can do all those things and pretend you feel good about what you're doing and still leave huge

problems, right? So you think that growing up is more about the individual's own ethics and morality? And culture, company culture. Company ethics and morality. Yeah. I've certainly, and you've been at places that have strong testing procedures and processes that don't end up actually helping.

It has to be combined with a desire to produce quality. There's always ways to game those things. Or to have them set up so that they give the appearance of quality without actually producing actual quality. I mean, the peer reviews. This is pre-chat about something we've already talked about

on a show that is previous but not previous will be the next show. I'm so confused. But the peer reviews, he talked about having checklists.

I can develop a checklist that's awful. Yeah. During this peer review, I will make sure that you have spaces in your code. Well, not even that. At least three. Not even that.

You could, you know, you could have... You could have a checklist that said you have to have a comment for every line. That kills me. But those are all silly things. But you could still have meaningful holes in your checklist while having things that seemed right. Like making sure that your code matches your requirements document,

but never checking the requirements document has anything useful. Or the specification matches the requirements. So, you know, there's a broken link there. But it's stuff like that. People have to want to do a good job and that there's a little bit of fear of of having an incident that helps but that's that's not motivating either um not positively motivating anyway no it's it's kind of like public executions decrease recidivism uh maybe but stats say most people don't believe they'll ever get caught, so not so much.

Anyway, it's a difficult question, and I do think, I'm not sure embedded has to grow up, but I think that parts of it do. And I encourage myself and you and our listeners and pretty much everyone we know to learn stuff. It doesn't have to be learn software, learn Python, learn blah, blah, blah. Just keep an open mind and keep learning things, whether it's how the water moves, because I found a really neat book about water or calculus or anything, because that muscle of learning, I think, is connected to the muscle of being able to talk about ethics and say, no, I have good reasons for feeling like this is a bad idea. And whether they're larger, overarching social reasons and whether they're larger overarching social reasons or whether they're technical reasons,

you, I don't know. Well, I think some of it's, I think it's a little bit unfair sometimes to say everyone needs to grow up because I do think, I think things change quickly. Yes.

I think people are bad at anticipating how things might be misused. So the example of DDoS attacks from IoT devices. Don't think anybody intended for that to happen. No, they just thought this isn't important enough to protect. What difference does it make? And they produce a lot of them and they probably produced them based on stuff from 2005 or six, you know, very, very old that was working. And so, yes, they shouldn't have done that. But the people who originally made it probably lived in a different time and universe. Yeah. Possibly. I mean, that happens a lot, right?

Where you have something that's five or six years old and couldn't possibly necessarily anticipate the changes that happened in half a decade and how it could be misused at that point. Old stuff sitting out there and being used past its lifetime. It's a very hard question. You can't even, I mean, for example, encryption. Okay, we're going to encrypt this stuff. We're going to do encryption you know okay we're going to

encrypt this stuff we're going to do a great job we're going to protect everybody's data we're going to sell 3 million devices based on this new standard 5 years later somebody cracks it okay whose fault was that does that matter yeah

there was no way to I'd like to separate those kinds of problems from problems of genuine, you know, blind spots. Malfeasance? Well, not necessarily malfeasance, but carelessness. Anyway. Well, then there's the problem of growing up often is bashing your head against a wall over and over again until... It doesn't sound like it would make me grow up. But that's what it feels like it is.

It's like, okay, I want you to do peer reviews. Here's why. I want you to do good unit tests. Here's how that's going to work and here are the stats that prove it. And then nobody listens to you. And then you get bitter and cynical. And is that growing up?

Is giving up growing up no absolutely not absolutely not it's a little easy to give up and to get cynical yeah you know that brings up a a cute little phrase i saw um that i have been using which is I am very optimistic about the future as a whole, but cynical about the future in the short term. Yeah. Well, yes. The great curve of many things is trending upward. It's the noise that's the problem a lot of times.

And of course, the noise is what gets reported. And it's what's that's the problem a lot of times and of course the noise is what gets reported and it's what's right in front of you but yes i i really am quite optimistic about the future just not about you know the next three months or the next year but overall i mean we have a vr we're gonna have self-driving cars we're to have phones that talk to satellites to tell us where we are on the world. I already have that. I know, and nobody remembers that it is cool. All right, shall we go on? Yes.

Because we have a long list. Can we stop asking hard questions. Do you want to ask about if you had two solid weeks to dedicate to learning a new embedded skill topic area, or even non-embedded, something tech-ish, what would you choose? This one's from Svek again. He had some really hard ones. Maybe he should get a podcast. Answer his own damn questions. I don't know. Actually, he asked me about linear algebra recently that was partially you're doing um because you were saying how linear algebra was

one of those things that was super useful yeah it is why um well if you're designing any sort of system that's the kind of math that generally is used to analyze it or solve things. Graphics, it's used everywhere in graphics. Machine learning, it's the core of a whole bunch of computer science and mathematical computing. Fourier transforms, that's all linear stuff, linear algebra stuff. You don't need linearity for it. No.

But I guess it does make more sense to talk about the bins and a vector and then you start matrix and blah, blah, blah and linear algebra. All right. Yeah. So would that be one of your answers to what would you... No.

All right. What would you spend time learning? Well, the original question was learn embedded skills. So things I haven't done much, I haven't done any BTLE. So that would be a skill I probably could use at some point to make one of those lights. Sensor on a stick. Sensor on a stick. What was the other thing?

I've done a lot of low power stuff so probably I don't know how I would learn that exactly that's a harder one to learn than BTLE you just have to do it I mean yeah

we'd have a few dinner conversations and I should learn Morse code yeah I was learning that but it's one of those things you really have to stick with

yeah I don't know. As far as embedded skills, those are the BTLA and the low power are the two things I have not had to deal with directly in a while. I already mentioned I was working on calculus. I'm continuing to work on calculus. That's not very embedded-ish, but it has been a tech thing that is nice. Let's see. I did spend some time working on

more statistics and forecasting in Bayesian, and that was fun and useful. But not in an embedded way yet. Doesn't that have linear algebra? Yes, it does. I was only hassling you about linear algebra. It is really cool. The problem with linear algebra is it's often taught divorced from applications, which makes it really boring. Yes, and it is neat when it isn't divorced from applications.

I think you're right that Linux is what I should be looking at. But I'm not sure I'm willing to go there yet. And I always think I need to know more about electronics. And yet, instead of picking up an electronics book, I picked up a calculus book. When I finish calculus, I'm more likely to go into differential than to go to electrical, which just says something about what interests me so i i don't have a lot of embedded skills right now if you ask me this question to learn to learn um but if you ask

me this question in six months i'm pretty sure i'll have a different set of answers i am sick of btle so there's a new one coming out yes yes there is new one coming out. Yes, there is a new one coming out. We should talk about it, but not now. You had a compiler to talk about. Oh, yeah, yeah, not a compiler. Well, it is a compiler, but not a compiler. It came across on Twitter this morning, and it's something I mentioned. I don't know if I mentioned it on a show months ago

or just on Slack, but if you go to godbolt.org. Godbolt. Godbolt.org.bolt.org you'll see on the left-hand side of your screen there will be a little source editor, and on the right side of the screen, you'll see the assembly for the source that you've created. And that's not the cool thing.

The cool thing is, above the assembly, there's a little pull-down menu, and if you click on that, you will see that it has every compiler that could possibly stick on the web from ARM GCC, MSP430 GCC, PowerPC GCC, Clang, x86. So you can select which compiler and which version of the compiler you want. For example, here's x86-64-GCC-51-52 all the way up to 6.3. And you click on that, and it recompiles instantly

and shows you the change in the assembly on the right side. So it's a really interesting little tool. I'm not quite sure what to do with it. But you were looking at... Optimization. Optimization, and how it changes the assembly for the optimizations. You can put in the command line arguments for the compiler in a little box. You can put dash 03 or whatever and just change that and it instantly changes the assembly.

You can see if there's big changes happening. The amazing thing, the bigger processors don't have that much change between versions, but I was looking at the MSP430, and if you do the same code on GCC453 versus 5.3, it changes completely. So it's kind of interesting. Compilers are a neat thing. And I tried a bunch of stuff with it that I didn't expect to work,

like you can put standard includes, so I wanted to see what square root looked like, for example, in various, in assembly in various processors. But I needed to include math.h and that worked. So it's, yeah, really cool. Weird tool. Yeah. That's pretty weird, but pretty neat to be able to see what's happening.

Oh, and you click on the bottom and it'll tell you if there are any warnings too. So you can get the compiler output. Yeah. Wow, look at this. Sorry. And we have lost Christopher. Sorry, there are buttons. I pushed the buttons.

From Elizabeth again, what comes after gamification? I don't know what that means. Next. You don't even what that means. Next. You don't even want to talk about that? I don't... So... Gamification. Gamification is where you take a problem that you don't necessarily want to do, whether it's learning to code...

Or brushing your teeth. Or brushing your teeth. And you turn it into a game and people then do it. Think about, you should go out and go for a walk. Everybody knows they should get some exercise, and walking is a fine exercise, especially, it's low impact, you can do it for a long time. Humans are built to walk. And yet, not everybody goes out for a walk, or sometimes we'll go for a walk around the block take the dogs whatever but then pokemon comes out and suddenly people are racking up 10 mile days and just insane walking around with their head on their their phone right in front of their face in front of cars and whatever.

That's gamification done right. All right. And it means that you can learn assembly code because you're playing a game. You can brush your teeth because you're keeping alive some Tamagotchi-like thing in your toothbrush, which I find very odd. I don't have that, but i have heard of it and that is good because we like games most people like enjoy playing games depending on the game of course so what's after gamification? Derealization. Where the game bleeds into reality. Yes.

Yes, I saw a Pokemon. Realification. Where you are playing a game, but... Wait, I've almost got this. I've almost got it. It's the opposite of gamification. Is this like the game you were playing in VR where you had to do things that you kept having to put on more VR glasses?

That was... yeah, exactly. Inception, I think. Yeah, yeah, yeah. No, but AR stuff where not only things, it's less gamification and more fake things bleeding into the real. Instead of taking something real and making it a fake game, taking something fake and bleeding it into the real. Indicators that, you know, in front of your eyes and things. Okay, so you went for the technology after gamification.

That was the question. Was it? Was it? Wasn't it? I was going for the social side. Once you have tricked people into walking long distances, how do you convince them that they should be doing that without the stupid game? Oh, well, I think you're misreading the question.

Oh, alright. You're thinking what literally comes after gamification. After I've gamified something, how do I de-gamify it? How do I convince people to do this all the time without requiring the little dopamine hit? It's not a convince thing, it's a habit thing.

You can't convince somebody to do something. You have to make it so that they're used to doing it regularly until they make it a habit. And so if gamification can do that, and then you take the game out and they still have a habit, then you're done. Yeah. But I don't know if that works.

It shows people that you can walk long distances, so why don't you? It takes away some of the excuses um but yeah okay well if that's not a technology all right all right let's let's let's go let's let's let's rapid fire through some of these we got we got lots of other questions to do do we okay what's next um i don't know who this is from that's fine when starting fresh on a project that is going to use a common part like a plc that has nearly infinite flexibility i suppose I don't know who this is from. That's fine. When starting fresh on a project that is going to use a common part like a PLC

that has nearly infinite flexibility, I suppose FPGA would work with that or even microcontroller. That was my parenthetical. How do you strike the balance between just using what you're used to, even if it isn't the best tool for job, ordering the best part out there,

even if it's overkill, and spending hours not producing anything, going over catalogs, comparing different options. Fear. Fear helps you not spend hours going over schedules. Yeah, applicable to microcontrollers as well. Often I'm overwhelmed when having to specify a module for a new function and always question

if the part I bought is the right one for the job. Chris Zweck had an awesome flowch chart for how to choose an external. Yeah, so we should link that. It started with, what are you using now? Yeah. Get the one with more RAM and slightly more processing speed. The truth is there aren't too many wrong choices for certain applications. I mean, you know, there's various big class bins

of when you're talking about microcontrollers. There's, you know, the little ancient things, the 8051s, and then you move up to MSP430s, and then you move up to Cortex-Ms, and then maybe somewhere in there is Atmel in between. No? You're making faces at me. I'm making faces, but that's okay's okay go on but you have a you know a half dozen choices maybe half or three quarters of them get eliminated by how big the thing is you

need yeah the flash ram and um cpu power cpu power are of course going to give you a target and then cost and power will eliminate some If you need Bluetooth, that narrows it even further. But it still isn't a fully constrained problem. I agree with the listener. At some point, you

make a choice and that is what you have to live with. I tend to agree with the fact you should be making your choice in large part based on the choice you have already made. As horrible as that is, there still is a lot of consistency between vendors. If you don't have a current vendor, then I would suggest choosing for whatever you think you would make the choice in three years. So don't choose the processor you want now.

Choose the processor that has enough headroom for three years, which is a really long time. That depends on the product, though. Of course it does. If you do that, you may over-specify. But yeah, that's true. And I think it's fair to choose, and I think Chris talked about this,

choose what you know too, because it's that whole balance between having the right part might mean your development time takes a lot longer. So what's the cost of that versus the cost of perhaps choosing not the optimal part, right? If I'm able to finish something in two months

because it's a part I've worked with several times in the past versus I have to ramp up on a new part and now it takes me six months, that could be the difference between shipping and not shipping. And it depends on how many you're shipping. You're going to spend a little more time looking for the perfect part when you are shipping a half million or a million of something because you can't waste the three pennies extra for the ram you don't need anyway but i think we're we're we're taking the part of his question that uh that is most applicable to us rather than the

first part which was something with you know truly infinite flexibility like a programmable logic part that's more difficult, right? Because then you truly have a sizing issue. If you haven't started your project, then you have to be really good at estimating how many gates you need and that sort of thing. Well, that's when those examples are really useful. If you buy one of those that's too big,

yeah, you really are completely wasting money. But if you buy one that's too small, you may... Then you've got to bump up to the next one somewhere in the middle of the project. But there isn't a learning curve jump from that, right? It's like buying a chip with more RAM when you hit the wall. I think there are still families. Yeah. And so that is one thing.

You don't ever want to be at the top of your family. Yes, yes. If you're choosing something, choose middle to low of the family. Actually, if you can, choose middle of the family, then you can go low if you want to eliminate stuff. I think you just have to do it and you have to get experience. You have to know what you need. But you can also look at their application notes. If they have an application note that says washing machine and you're making a dishwashing machine, you might think

okay, they have certain modes, certain controls, motors, blah, blah, blah. That's reasonable. Check to see what else other people are using this chip for. That's a tough question, though. There's no easy answer. Let's continue to not answer people. Roland says, after listening to

a rerun of our show with James Grinning, one of my personal favorite shows, and hearing about the analysis of the ESA's

Schiaparelli Schiaparelli Schiaparelli Lander, he was wondering how such a product is done regarding the software the esa article mentioned that that it was possible to recreate the failure in their simulations

uh shouldn't they have seen it before they actually got there shouldn't they have seen it in simulation shouldn't they have fixed it up front how in the world did it pass quality control yeah just because they could recreate it in the simulation doesn't mean necessarily that the conditions were set correctly or that they anticipated the right conditions right the simulation can be right but if you haven't fed it that scenario, you won't see it. So I can see reasons why that would fail. I think the short answer is these are really hard and complex problems. And one of the failures was a saturation of the inertial measurement unit. And so something happened that it must have spun too fast. And because it's a feedback system, having something out of range means that it all goes to crap.

And so that suggests perhaps their simulations should err more on randomized inputs and fuzzing of situations. But then the other, the quality control person in my in my head says but that's not real life real life is yes international measurement unit does this and then this and then this you can't have non-continuous data except you can have maximum measurements you can go out of range um well the problem is is, you know, he mentions software. How is the software done? But this is not software, right? An IMU is a piece of hardware. So a simulation of a system that

involves hardware is necessarily going to be artificial. So you may not be able to get into that situation without explicitly forcing it to be. Right? I could recreate it in the simulation after the fact because I went into the thing that was simulating the IMU and gave it a... you know, and forced it to saturate. Yes.

And you should have. They should have. Maybe. I don't know. I don't know the details. Well, we do have a link to some of the details. Oh, I didn't read it. I will make sure to include the link. Yes.

But, you know, these are difficult things. And they're conglomerations of systems. It's always integration that's the difficult thing. You can do as many unit tests as you want of individual parts, but when you plug the parts together, you have combinatorics working against you. There could be

a thousand things that could go wrong with part A and you test for all of them. There could be a thousand things that go wrong with part B and you test for all of them. But there could be a million things that go wrong between part A and part B. Simply by multiplication. And and now you add part C and then you add Mars yes and Mars

has clearly got the same field that I have that causes all electronics to break it's Marvin Marvin and his ray gun alright that next thing in the outline I don't know what it's there for. What, the NRF?

There's a... It's a new chip. There's a new chip. It's exciting. It's got Bluetooth 5. It's got 32-bit M4F. It's got a megabyte of flash and 256K of... This is your next chip, people.

I don't really actually know how that got there. It's clearly just an ad, but we should know. All right. Nordic is a new chip. Yay. Yeah. It's got lots of numbers that are large. It's got one more Bluetooth than usual.

Yeah. Usually they come with Bluetooth, four Bluetooth. This has five Bluetooths and that's one better. It's got a thing, an IEEE number that's 802.15.4, which I believe is 4.4 better than 802.11. This is sort of how I see these numbers sometimes. It's got something called Ant. Which, again, I hope involves actual ants.

It's got a million flashes. which again, I hope involves actual ants. Uh, it's got a million flashes. Yeah. It sounds like a great, a great thing. You can, you can, you can,

fella could have a pretty good weekend in Vegas with a lot of stuff. I just tripped. Uh, Jonathan said, hello. I, he didn't have any questions so hey how you doing that's great um and then i think the last thing that's actually on my list is from lars uh which is funny we can spend a whole hour talking about this one because we've done it in the past. He asked if we'd done an episode of our podcast discussing the so-called imposter syndrome

and overcoming self-doubt in our technical fields. So-called imposter syndrome. It's sort of an imposter of an imposter syndrome it can't even be a good syndrome it's a fraud of a syndrome i'm totally faking it you can tell yes we did a show we did a show a very long time ago it was number 24 24 and uh we also talked about it in number 78 which was chris ve's about happy cows. I'll have both of those links in the show notes. But it hasn't been a while.

Yeah. Any new advice, thoughts on imposter syndrome? Do you think we need to define it? No, I don't really want to do another whole show about that right now. Okay. But any new thoughts on it? Still happens to me some, but I'm getting old and jaded and crotchety. So I'm caring less.

I'm seeing other people make mistakes more. You're pointing and laughing. What? No. There's no mistakes you make except Except with, you know, Windows. Oh, wow. That's just because you get mad and click on everything until it explodes. I really like my new Windows computer, but I like it better than what the competitor was offering. You don't like it?

You just don't like Windows. Yeah, I just don't like Windows. Well, you know, you could put Linux on it. Could, but it won't. What software on there do you need Windows for besides all the stuff you use?

Yeah, all the stuff I use. Compilers. They have those for Linux. No, I'm going to be using Windows only, but I can't talk about it. Ooh. Okay, so imposter syndrome.

What you are feeling is incredibly common. Yep. Lots of people feel that way. Not everybody, but lots of people. Most of them are wrong. That was mean.

That was mean. The thing with imposter syndrome is that it can get a lot worse. That'll help. And so you have to take care of it. Yeah, all right. You can't just feel awful about work and go home and think it won't affect your home life. You can't feel awful about work today and think that we'll all get better tomorrow.

And then after a month, you realize it hasn't gotten better. Last summer, I talked to a friend who left a job that sounded perfect for him, and he just was not happy there. Good reasons. He liked the people, it just was not a good fit and he said that he had heard uh that a past colleague had actually committed suicide at his desk and we don't talk that much about how the isolation the constant need to learn new things, the constant fear of being late and fired.

These things can seriously screw you up. Well, and there's, I don't want to get into psychology and stuff because I'm not qualified. But, you know, sometimes imposter syndrome is not the primary problem. It is a symptom. It is a symptom. It is a symptom of other depression and anxiety. And so don't assume that it is the problem. But it is a symptom that you have options for helping yourself. You don't have to see someone if all it is is that today you're feeling like,

oh my God, I can't believe they let me have this job. Someday they're going to take away my computer. It may be in an hour because they're going to realize that I have no idea what I'm doing. And so that feeling, yeah. It's really common and it's really common even at the highest levels of uh lots of disciplines i mean it is not i've read interviews with musicians yes musicians who you know you know very well and you'll go on youtube and you watch them you know give a talk or something and or play some example and you'll look at that going i have no idea what that is

that's impossible and then you'll read an interview with them, how they think they're the worst ever. And it's just, you know, part of it sometimes is your brain's way of motivating you to do better when you already do really good things, I think. I think it's a, you know, it can be, and this may be controversial, it can be healthy a little bit. Well, if the opposite is total arrogance and narcissism, I'm okay with feeling like an imposter sometimes. If those are the opposite ends of the spectrum,

I'm okay being on this side. Probably feeling like an imposter is not healthy. Feeling like a fraud is not healthy, but feeling like you could do a better job. It's a slippery slope though. It is. Yeah, I mean, I think we did a better job discussing that a long time ago

well yes because we did a whole show about it almost a whole show about it poor Jim GF he wrote in for one question we just answered a totally different one well also you were talking about musicians not a musician

but Adam Savage somebody who's fairly well known oh that's a good link has a video about how he sometimes feels full of self-doubt and uh i'll have a blog post up which is great for somebody who often sets himself up in mechanical machines that could kill him well Well, yes, exactly. But I'll have a blog post up with some of the strategies I find helpful. So if you're looking at imposter syndrome and wondering, how can I feel better today or next week? I do have some suggestions, but I think we are out of time.

We're as out of time as you want to be. Well, I have a few other things I need to get done today, and this is the third podcast we have recorded in less than five days. Which means that next weekend we got nothing. Should we be getting pretty good at this point? No, no, the book I've been reading about

practice says we're not doing it right. What? Yeah. You're supposed to practice the items. You're not supposed to like just do the stuff. So what would a practice podcast look like? We just say certain words over and over again? I think I would do the intro over and over again until I said it slow enough that it was intelligible.

With drumming, all the exercises, you're supposed to start unnaturally slow and then increase very, very slowly. And that's the best way to build speed. So to build speed for the podcast, we should start lightning round and just do lightning round over and over and over again very slowly and then bring up the speed. Because I think people would tell us that lightning round is not very lightning-y.

So that would be something we could practice. People would tell us that lightning round is not very lightning-y. So if we want to bring it. It has not been lightning round. So that would be something we could practice. We could just do, you know, lightning round people off the street. Lightning them. Hi, I'd like to lightning you. So back to the party. Thank you for that segue.

We are going to have lightning rounds. We're going, I'm going to take all the lightning round questions. I'm going to have lightning rounds. We're going, I'm going to take all the lightning round questions. I'm going to make them neat. I'm going to pass them out. You can all ask each other lightning round questions. All right. It's a party game.

And then we'll retire it. You think so? Eh. We have gone through the questions a lot. I do enjoy it because it tends to loosen up our guests. But it does, I feel like we've asked all those questions. We can just instruct

them to take a shot of whiskey or something. Your sister gave us a game. It was a word game. Maybe we'll ask them for that. Watch them call it. Yeah. What would you call a purple chicken who flies over the ocean? Fred.

Of course. That made no sense. See, I've stopped making sense. We should stop recording. Wrap it up. I don't have to ask you for final questions. I have final questions. I have tons of them.

Okay, go ahead. Oh. No. All right. Thank you to Christopher for co-hosting producing and being my sweetie did it hurt to have your eyes roll that far back in your head no the eyes didn't hurt but i think i pulled something in my neck thank you for listening if you would like to see blog, it is embedded.fm slash blog.

There's all sorts of cool stuff there. On embedded.fm, you can also see the contact link, which will let you say hello. You can see the show notes for this episode, which will indicate where you should go to RSVP for our super fun party. And if you would like to encourage us to continue making fine highbrow content, such as this particular episode, where can they go to encourage us in a fiduciary fashion? I meant to have something about Patrion. What is with your mic? I don't know.

You just keep punching it. I'm sorry. Sorry, Mike. Sorry, listeners. I didn't mean to punch you yeah Patreon that was supposed to be high on my list Patreon supporters

thank you we really appreciate it it makes it so much easier to send guests microphones and I admit that some of your money might be going to the party so you should definitely come and what was the other thing it's all fungible it's fungible but admit that some of your money might be going to the party, so you should definitely come. And what was the other thing?

Well, it's all fungible. It's our money. It's fungible, but it's been really nice to have the podcast pay for itself enough that I can buy the podcast things. Maybe I'll buy more stickers for it. And so thank you to everyone who has supported us. If you would like to support us, it's patreon.com slash embedded. And there'll be, of course, a link in the show notes. And there are no special benefits.

You don't get to come to the party earlier. You don't get extra beer. You don't get any special benefits other than our thank yous, our thanks. And maybe at some point we'll do something. We still have thoughts. Yes, there are thoughts. Sometime this year we'll do something.

We have some community building thoughts. I learned about Gitter. Yeah. Maybe we'll talk about that next time. That's it. Thank you. Thank you to everybody. Thank you to it. Thank you. Thank you to everybody.

Thank you. Thank you to listeners. Thank you to supporters. Thank you to Christopher's. Thank you to- That's all, folks. And you hit your mic again. I'm sorry.

You've never done this before. I put it in a different place. Okay, so final thought from Winnie the Pooh, continuing on our adventure here. This is, of course, Pooh, who is the he in this next sentence. He crawled out of the gorse bush, brushed the prickles from his nose, and began to think again.

The first person he thought of was Christopher Robin. Was that me? said Christopher Robin in an odd voice, hardly daring to believe it. That was you. Christopher Robin said nothing, but his eyes got larger and larger and his face got pinker and pinker. 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.