Embedded - 90: Stick It in a Pumpkin

Episode Date: February 26, 2015

The Linker post for this Episode: Solving a Different Problem  ThingM's Tod Kurt (@todbot) joined us to talk about the most important part of every embedded system: blinking lights. ThingM has been making I2C lights (BlinkM, MinM and MaxM) since 2006.  The newer, more productized USB light is the Blink(1) (there is a coupon near the end of the show). Blink(1) had two successful kickstarters (second one). The BlinkMs have an ATTiny85 (which is also on the Adafruit Trinket). The Blink(1)s have a PIC processor that is small, cheap, and supports USB quite well (PIC16F1455-I/ML and dev kit). Other smart LEDs include WS28xx  (aka NeoPixel) and APA102 (aka DotStar) Seeed Studio was discussed as a way to get boards built, assembled, even housed. Elecia mentioned Tindie's new CM review site. Tod is cofounder of Crash Space (@CrashSpaceLA), a Los Angeles based hackspace. They (including Tod) were on the short-lived Mythbusters-hosted Rube Goldberg devices show called Unchained Reaction. Tod has worked on some neat art projects, including the Crystal Monster and the Cash Machine. Tod's blog. Speaking of blogs, Chris and Elecia are going to start writing after (podcast) action reports forElement 14. More announcements (and actual links) soon. Don't forget the Chris Savage (Parallax) call for assistance!

Welcome to Embedded, the show for people who love gadgets. I'm Alicia White, here with Christopher White. Our guest today is Todd Kurt. We'll be discussing the most important piece of an embedded system, the blinking light. Before we get started, I want to remind you that Chris Savage and his wife could use your help. I know you may be in the car, so think of it when you get your coffee or tea or diet Mountain Dew, and then click the links. Now back to Todd. Hi, Todd. Thanks for being on the show today.

Hi, thanks. Could you tell us a bit about yourself? I'm a CEO and co-founder of Thingum, which is a small ubiquitous computing, or now I guess we'd call it Internet of Things company, that's been around since 2006. We make the pretty popular Blinkum smart LED product that's mostly used with Arduino and more recently the BlinkOne USB notification light. And you say it, Thingum. Thingum, yeah. I like that. I have always said Think M and blink M.

Well, see, this is how the company came to be as a friend of ours was asking, what should we name our company? And I'm like, I don't know. We make these little thingums. And he's like, well, there's your company name. Okay, so I got into Blinkums a while ago, and I want to say like six years ago, with Halloween pumpkins, because you can program the little I squared C lights with an Arduino to do like a script. And then you can then put it in your pumpkin with a little battery and it lasts all night. Is that how most people get started? Or do people actually use these for non-holiday oriented things? Yeah, well, the funny thing is, is for the longest time,

we really didn't know what people were using them for. Like we, for the longest time, they were one of the better, or I should say more immediately graspable examples of how to use I2C on the Arduino. But as to what you'd actually use it for, we were in the dark, but we were selling thousands of them. And we slowly got word that they're used in a lot of industrial design and prototyping houses whenever they need to mock up some indicator light. It doesn't necessarily need to be hooked up to a computer, but could be. So we've heard about them used in car dashboards, in various set designers for movies and TVs to create a fake candlelight

or various little sci-fi console displays blinking. But of course, no one really talks about these. So we don't have a good resume of where Blinkums are being used, just that they're kind of every once in a while we hear about them from Maker Faire, that, oh, hey, I used your thing in my thing. Yeah. And I use them, I actually have one in a little proof of concept I'm building right now, because it's I squared C.

And sir, I could put a wire out there and figure out PWM, but my client right now just wants to know if the other stuff works and I want the LED for debugging. So it's an easy way to go ahead and get an RGB LED on there without bringing up another subsystem.

Exactly, yeah. That was kind of the whole point. It actually came about because when we were doing a bunch of different ubiquitous computing prototypes, we found that over and over again, we needed to have a little indicator. So it was kind of one of the interesting aspects of what we now call Internet of Things is you have these computational devices, but don't have displays. And so how do you indicate state? How do you indicate the process that the thing is going through?

And a light is kind of the first thing. And so me and my co-founder, Mike, we often considered the LED or the smart LED really being sort of the atomic sort of base unit of Internet of Things or ubiquitous computing. It's like the thing you get working first. I remember a few years ago, even before I think I'd heard of Thingum, there was a big push for kind of ambient indicators.

You'd have this glowing thing next to your computer that would tell you if your stocks were doing well, it would be green, or if your stocks were doing poorly, it would turn red. It seems like that was kind of where this started. Well, yeah. We're friends with David Rose, the guy who did that. It's Ambient Devices. I think that's what their company name was.

And the product was called the Ambient Orb. Okay, yeah. And you could buy it at Brookstone. And it sold horribly. Even though it was a great product, because time you could buy it at Brookstone and it sold horribly even though it was a great product because all you had to do was plug it into the wall and it hooked up via the pager network I think

and you would log into a website and you would hook it up to some data feed and then it would all just work I think part of the problem though is that that was over 10 years ago and people weren't used to having their lives as instrumented as we are now. And so I think if that exact same product was launched today, it'd do a lot better. And in fact, that's kind of what

Blink-1 is. It's kind of the USB tethered version of an ambient orb, honestly. And so Blink-1 is your new product? New-ish, yeah. It's been around for about two years now. And it is pretty much came about because we time and time again saw people that were saying, hey, can I hook a Blink-1, sorry, can I hook a Blinkum up to my computer? And it's like, well, yeah, you get an Arduino

and you plug it in. But that becomes a pretty expensive proposition. And so we're like, well, we can make a cheaper version of that. And let's use this as an excuse to learn about more retail-oriented products because we originally weren't set out to be making products to sell. We were initially going to make our own sort of interesting designs and take them really big on a retail level. We only kind of accidentally fell into the maker slash Arduino market

just because that was a thing to do. So let's see if these things sell, the Blinkums, and it turned out they did. But we didn't really get the mass production experience on the retail level from that because turning out a circuit board is actually pretty easy. Now there are turnkey companies where you just send a bill of materials and a schematic and a Gerber to, and they'll send you in a couple of months a set of ESD bags full of

fully assembled product. It's not going to be long before you can just send them the Gerber file and they'll tell you which shelf on target it appears on. Well, that's the thing. I mean, it's like certain companies like SparkFun and C-Studio and a few others are doing this thing where if you submit them a design that uses their standard part library, then their assembly process is kind of like so you can turn out a board that uses standard like 18 megas and 0402 size resistors and capacitors in just a few days, honestly. But the BlinkM is not like that. It's an actual self-contained. I got one from your co-founder, Mike.

Oh, yeah. self-contained i got one from your co-founder mike oh yeah and um i don't know it's smaller than a quarter and it sticks into the usb port and i understand that there's software that i have to install and then i can control it programmatically yeah it's it is it is the spiritual successor in many ways to the the blink one is the spiritual spiritual successor in many ways to the blinkum in that it's it's an RGB LED with a computer hooked up to it that you can control. The main difference is that it's got a USB port instead of I2C, and it's in a retail package. I mean, it's in an enclosure, and then that's in a retail package. And they both have this scripting engine inside of them so they can run standalone without having any sort of controller commanding them. And I personally think that's one of the cool features. I like that a lot.

Well, with the USB version, your computer sort of commands it. If you want it to, you could also load up your own little color pattern onto it and then plug it into a USB battery. You know, those little batteries

you use to recharge your phone. And then it'll run through its little color pattern that you set into it. So you can stick it in a pumpkin. Exactly, yeah. In fact,

that's kind of how I cheated this year with some of our pumpkins is I just quickly slammed a bunch of orange, like random orange colors onto a Blink-1, stuck that onto one of the USB batteries I use to recharge my phone, and then just tossed that into a pumpkin. Oh, see, I go for the blue ones. I find it much creepier to have the blue lights in my pumpkins. Very good.

But I know the Blink-1s are being used by server rooms now. You just plug them in all of your blades and run a program, and you can walk in and see which of your blades is sick. Oh, totally, yeah. In fact, thanks to a colleague up in, I think, Montreal, it's now in the mainline Linux kernel, the ability to control the Blink-1. And the reason why he wanted that is because

he builds a custom version of the kernel that is really lightweight and boots up very fast. And he wanted the ability to indicate the state of the system as it's booting up, but without having to load a bunch of normal Linux driver subsystems. And so now he can have these Blink-1s plugged into the servers. The system comes up, the kernel itself is talking to the Blink-1, indicating the state of the system. And he can kind of tell at a

glance if there's any problems, like if, oh, this one hung on boot or something like that. So now that i understand you can just make it play whatever you want and if i plug it into one of our wall chargers i can have a nightlight that would tell me what time it is without necessarily having a clock uh see unfortunately no this is kind of one of the this is the problem that we've run into with with the blinkum the previous product as well in that the microcontroller has in it the ability to keep a type of time, but it's not a very accurate time. And so it's enough to know how long maybe 10 seconds is, but it gets really inaccurate after about, you know, a minute. And so it's good enough for plain little color patterns, but you can't plug it in with the understanding that, oh,

in 12 hours it will flash a color. It'll, I mean, it could do it, but it's idea of 12 hours will be different than your idea of 12 hours. You know, that totally makes sense. And if you had asked me, could you do that with a BlinkM or BlinkM, I would have said no, because the time basis is not that precise. You're using a crystal and an ATtiny, it's just not going to be that good. And yet, taking this USB stick out of a box and has seen that it's enclosed

convinced me that this is a, I don't know, a magic device. No, it's- Somehow it should have, you know, network time. Yeah, this has actually been a recurring sort of, I don't know if I could call it complaint, but sort of expectation on these products. Like yours with BlinkOne echoes similar issues that people have been wondering about Blinkum back in 2007 when we were first thinking about

this idea, I wanted to collapse it all down to a die an actual IC die and have it encapsulated in a plastic package so the physical LED would have all the electronics inside of it and then that would be a lot more obvious, I think,

that, oh, this is just a light. That's all it does. But it turned out we could do that only if we bought a million units. Because it turns out working with bare dies and bonding wires to bare dies and the process to make LEDs is a different process than the one to make microcontrollers and so they have to run on different fab lines and it's just a lot of complication that we as a very small company did not really have the ability to deal with. So the smallest version of the let's embed logic inside of an LED ended up being the size of what

Blinkum is which is like about I think 0.6 inches on a side or so, or 0.4 inches for the smaller Blinkum Minim. But once it becomes that, people start to think, oh, it's got a computer in it. Because it does. It's got a little AT Tiny. You're running at 8 megahertz, eight times faster than the Apple II.

You think that there would be a lot more that it could do beyond just talk I squared C and drive an RGB LED. But it's hard to do anything more interesting without extra stuff like in the case of a clock, having a real time clock chip in there to and a battery backup. And and exactly. Yeah. And so and so in the case of blink1, it looks a lot like a USB fob. So people are like, oh, is this the 8-gig version or the 4-gig version? Yes. It's like, well, we couldn't fit memory in it also, so it's not a thumbstick. You play MP3s on this, right?

Yeah, you know. Again, if we had the secret knowledge that some of these larger manufacturers have that can put, like in the case of, say, the iFi SD card that is a full Linux computer in the size of an SD card, and you stick it in your camera, and when you take a picture, it uploads it over Wi-Fi to Flickr or Instagram or whatever. Yeah, I have one of those. I had no idea it was a full Linux computer on there. Yeah, it's brilliant. It's like this super special purpose sock from Broadcom and you can't

get any information on it until you sign an NDA. And so I'm like, ah, I like open hardware. I'm not really going to go that to go that route but you do have you do have a processor in there it's the at tiny 45 it was the at tiny 45 and now it's the at tiny 85 the at tiny 85 is for the usb uh no well oh oh, so, okay. So the Blinkum has the ATtiny85. The original BlinkOne also had an ATtiny85 running a software USB stack called VUSB.

But the current version of BlinkOne uses this PIC from Microchip, the PIC16F1455. You changed vendors? Oh, yeah. If you want to do really small USB with no crystal

oscillator, there are very few options out there. And the PICs are better. The PICs are better, yeah. It's really frustrating because the PIC16 architecture is really weird. And you have to basically buy their compiler because you can't get GCC for it.

I'm so much more in the Atmel camp. I just think their software is so much better. Yeah, me too. Yeah, plus Atmel has been really good to the open source community. They've been actively supporting the development of AVRGCC. And lately, the last couple of years, I've been actually helping out Arduino. I suspect you thought long and hard about switching over to microchip and PIC

because it is a different environment. Yeah, it's totally, totally different. So going back to the Blinkum, Blink-M, I'm not going to get that in this hour. USB light. No, no, the other one, the I2C light. I2C light, yeah. To some extent, it is basically an I2C to PWM converter. Yes, at its most minimal.

And people have reprogrammed your chips to do different things. I've met a few people who've said, oh yeah, I have one. I reprogrammed the ATtiny and I look at them like, it was an LED. What did you program it to do? Have you heard stories about these? Oh, no, totally. And we've actively helped people to do that. Yeah, the first thing we heard of is that we had an open source implementation of the

firmware on Blinkum, because at the time the firmware was closed because of a license agreement that we had with another company who had some IP on I2C LEDs. And so someone else created a white box implementation of the Blinkum firmware. That was kind of the first example we heard of. But then when Dave Mellis started doing some work at MIT on extending the Arduino IDE to the ATtinies, I jumped on that and created a quick little set of library files that let you reprogram your Blinkum with the Arduino IDE. And at that

point, people started making them into little robots because you could, you know, yes, the three IO pins that are mostly free are hooked up to the LEDs, but they could also be driving, say, a servo without even any disconnection of any of the internal wiring on the Lincoln board. Or you could unsolder a couple of LEDs or clip the LED off. I've seen people do that because it's a nice little 0.5-inch package you could plug into and do something with. Wow, I didn't realize they've gone that far. That's neat. Yeah, it's pretty good. Now, unfortunately, we didn't capitalize on it as well as possible.

And so about maybe I think a year ago or two years ago, Adafruit came out with the Trinket, which is an ATtiny-based little development board that's beautiful because they have a bootloader for the ATtiny that lets you program it over USB using the software USB. And it's great. It's really cheap, too. It's like $10 or something.

Well, they actually give one to you if you order enough, which I found out recently. Yeah, I think so. Yeah, and there's the Trinket which is an at tiny and then the trinket pro which is only a couple bucks more but it is it is the arduino ad mega processor yeah and they're both really great and setting up the at tiny in the arduino interface is a little bit of a pain but it's not i mean you just follow the directions and crank. Yeah. It's not as simple as a normal Arduino,

but it's a lot easier than hooking up the similar sort of setup with the Blinkum, because the Blinkum wasn't really designed to be used as an Arduino. You can do it, but it's not designed to be one. So is the Blink-1 as hackable as the Blinkums have turned out

to be? So unfortunately, for the first version of the Blink-1 that had the ATtiny, I was working really hard to get a bootloader that I found working with it. But I wasn't able to get it done in time for the Kickstarter that kind of bootstrapped the whole process. And I was also a little bit leery because the ATtiny doesn't have any protected ROM or protected flash storage. So there was the chance that the bootloader could overwrite the bootloader and brick the unit. And so I didn't want to give that potentiality a chance

on a product that we were really targeting towards retail users. So I didn't have any sort of bootloader in it, which meant that if you wanted to reprogram it, you had to use one of those little chip clips that would clip onto the chip, which is a whole different level of skill for people who want to reprogram stuff. It was possible, but I haven't really heard of anybody doing it. And then in the case of the current Blink one, which uses the PIC, there's so many weird barriers to entry because it's a PIC,

it needs a compiler that's, while free to download, they have this free versus pro mode, and in order for it to compile, you have to use the pro mode. And because the space is so small on the board, we weren't able to break out any of the ISP header pins. And so in order to reprogram that, you have to do a little bit of kind of pogo pin stuff to actually get it to work. And you do have to open it,

which I guess it isn't that hard. Am I going to break it if I continue to have to open it, which I guess isn't that hard. Am I going to break it if I continue to try to open it? Oh, no, no. So one of the things we did do is one of the ways you can hack the Blink-1, the current Blink-1, is that we used the WS2812 LEDs on it because there's two LEDs on it, one on the top and one on the bottom. And we broke out the continuation of that serial line that

the WS2812 LEDs use, so you can wire up to 16 more LEDs to your Blink-1. But you can't drive them from your USB or your computer will fry. Oh, no, no. You can drive them just fine. Really? I thought they took a lot more current than that.

No, up to... So if you drive more than eight at a time, I think at full brightness, you might have a problem. But a lot of times, you can just kind of blink them or only run them on a single color if you're doing 16 of them. I personally have never had any problems on my Mac, but then I think the Mac kind of cheats on its power available on the USB port because it lets you charge your iPad or iPhone or something from their USB ports. So I think the Mac USB ports puts out more power than the USB spec allows. I'm not sure, though. No, they do. It's either one or two amps versus 500 milliamps.

Yeah. Well, you said there's a WS2812, and that's the same chip that's used for the adafruit neopixels that's the one i thought that would be competitors for you i mean there's that's what that was how can they not be your natural enemy well for one um the the people at adafruit are awesome like they like like like'Amour and PT have done more for the maker community than almost anyone in the, in the world. Like, they've, they just continuously make awesome stuff.

And they, their branding of the WS2012s as NeoPixels was brilliant because it gave a very friendly name to a very unwieldy part number. Well, and to a fairly unwieldy device. Yeah, and they created some good libraries, and while the NeoPixel slash W2812

chip is kind of suboptimal in a lot of ways, like its refresh rate is pretty slow, so you can't use it for persistence of vision, it has issues with uh you can like fry the led pretty easily if you hook it up hook it up wrong and things like that um and it's got very fussy timing requirements because it's a single wire bus and so you have

to be right on the timing yeah super fussy in fact that that delayed our kickstarter the our most recent kickstarter by about two months because I was doing my timing test because I had to rewrite some PIC assembly language stuff for the timing and I wrote it wrong because I was using an early batch of the chips. And when I got then the production samples of the Blink-1 back and some production WS2812s, I found that they didn't match. And so the colors were all wrong.

It was flashy. I had to basically have a whole 2,000 units reprogrammed by hand by me and then fix the firmware and send it off for production. It was just a big mess. But to get back to your original question, the NeoPixels are awesome. They fulfill my original desire back in 2007 for logic embedded into an LED. And basically making the LED smart.

I think that's a brilliant thing. I wish that I would have been able to be the one to do it, but I'm glad it exists. I'm sitting here watching Elysia disassemble and destroy your product and drop it on the floor. Yeah, yeah. It's pretty hardy. It should be able to take it.

I only dropped the outside clip on the floor. So, yeah, I am surprised that the NeoPixels happened without you because you were on that path. You were in that direction and people were really becoming fans of the Blinkums. And then, you know, I was like, oh, if you want to use a simple I2C LED, use a Blinkum, and then got blindsided by the NeoPixels. I was happy they came out, but I went from, oh, well, you obviously use this to, oh, now you have a choice. Now I have to learn both options. Yeah, and pretty much for most things, if people want to add LEDs to their project,

the solution is a NeoPixel or something like it. There's some better ones out there. They're still not the best because there's all this, like they draw like a milliamp per LED even if they're not doing anything. There's all these little gotchas beyond the timing requirements. Well, the timing requirements,

I find a lot of people saying, well, now I have to put an ATtiny on there as a coprocessor to just play with the NeoPixels, especially if you have an operating system and you can't do anything on true real-time boundaries. One of the solutions is to use a different

chip or different LED. There's this new one out, one of the solutions is to use a different chip or different LED. Like there's this new and out, well, kind of new called APA 102, I think. And I think Adafruit's got a term for those as well. And it uses a two-wire protocol, so data and clock. And it can be used by standard SPI, or you can just bit bang it with, you know, doing GPIO high and low. And so you can get around, you can use really slow, even a basic stamp could do it to drive those LEDs. It doesn't look like they've got a fancy name for these. Oh, it's part of their dot star strips. Yeah, dot star, yeah.

And those aren't all RGB. Some of them are single color, which means they're cheaper, which is nice if you just want to be able to have white Blinkin' lights. Yeah, and that's also good, too, because one of the problems we've had is when people asked us, usually set designer types, like, hey, we want to have a nice white light we can control with the Blinkin' protocol. What do you recommend? Like, can we just drive to have a nice white light we can control with the Blinkum protocol. What do you recommend? Like, can we just drive these and set them to white? It's like, well, the white that comes out of RGB LEDs looks pretty poor, especially on camera, compared to real white LEDs, which don't look as good as like incandescence, but look better. And so one of the things that we would say is like, oh, you should use our Blinkum Maxim, which can drive white LED strips, just like dumb strips, not the NeoPixel strips.

And that's the way you'd solve that problem. But now you can have individually controllable white LEDs via these APA 102 dot stars. And I think those are pretty awesome. So the Maxim is one of the products in the same line as the blinkums it's not usb it is i squared c correct and it's big i mean it's physically large i keep wanting to use them um but it has three leds and i never managed to blend all three leds probably because i put them in little boxes and wonder why this side's red and that side's blue.

Yeah. But they do other stuff. They have input lines. Yeah, and the original, so the genesis of the Maxim was that we wanted to make something that drove LED strips, which at the time were kind of the bleeding edge of LED tech. These LED's, service mount LED's on flexible circuit substrate wound up into spools and with stickers on the back you can put them anywhere. But driving them meant you needed to have some sort of

12 volts MOSFET drive sort of stuff. And so, okay, let's just make a Blinkum, but make a wake one that drives these LED strips. But one of the things we found when we were testing it out is the standard users of our products are pretty new. And having that extra step of hooking up LEDs with 12 volts was a little bit extra, and people were doing it wrong. And so we're like, okay, let's just add an LED cluster to the top.

And it'll just come with the product and you can use it all the time if you want. But if you want to use the maximum as it was intended, you can just pull that off and hook up your LED strips. You're saying I'm using the wrong one of the two boards. All the time I have thought it was all about the LEDs and it was all about the MOSFETs, which I have needed for other things and I never really think about taking apart my Maxim to get them. Yeah, see, again, that's probably a failure in our ability to advertise the Maxim really well. Well, it's also I go to you for blinking lights and I don't really think about the whole thing necessarily. Yeah, I have heard of one guy who

reprogrammed his Maxim to be a robot controller for like, you know, steering and forward and back and something else. I forget, like an arm or something. Because it's got a pretty powerful I mean, it sure sucks a lot of juice. So I assume it's powerful.

Yeah. Yeah. I think, I think each MOSFET is rated for something like six amps or something. So like we tell people you, they can run three amps per MOSFET if they wanted. And that should be, be within the safety margin.

I have so many more things. I mean, I do not blink. I'm now many more things I'm going to do in BlinkM now. All these ideas. So one of the reasons why we didn't get into the whole NeoPixel sort of space early on, and we were watching it, we were seeing these LED strips go from being dumb strips to being the smarter strips that had a a little soic on it that would control like three of the leds um but but then being them being very expensive

like they were like i think three hundred dollars for a for a five meter spool and they only had a like six bit color resolution or something um this was back like four or five years ago but one of the problems is it was kind of one of these like innovator dilemma things is that we were so invested in this I squared C LED thing that going to something else was problematic. And the market, sorry, not the market, but the different LEDs were changing so fast. Like there was all these different kinds of LED strips out there. And if it wasn't for Adafruit calling them NeoPixel, I think the WS2012s would have gone away a lot longer.

Like I think they would have faded away and we would have seen something else replace it. But they kind of stuck a stick in the ground saying like, this is what we're going to rally around for a while. And unfortunately, we didn't do that. And that was kind of a failure of us thinking too shallowly about the problem, I think. It's easy to see what's in front of you and hard to see what's all around you no totally right so you went for blink one instead and and went for your stayed on the path of

indicators instead of trying to go the path of the neopixel yeah, and part of that is because the maker market has really changed in the last five years or so. It's like SparkFun and Adafruit, they really kind of own that market. They're able to move a lot faster than we can. Because while Thingum is my main thing, Thingum is mostly just me. I've got a couple of helpers that help me with some of the things

I'm really bad at, like finance and certain types of customer service. But for the most part, it's just me doing the electronics design. Whereas both of those companies have engineers on staff besides the heads of the company. And so with BlinkOne,

BlinkOne was really more of us wanting to explore the retail sort of side. More of like, instead of exploring a cool engineering problem, this is more exploring a cool business problem. Like, we've done PCB assemblies forever now. What does it take to make an enclosure? Like, oh, what if we want to make the enclosure not out of plastic? Like, what if we want to make it out of metal or glass? Like, how do we deal with the logistics of ordering 10,000 of something and getting it shipped from China and then to all over the world? And so those are kind of the problems we were kind of interested in a couple years ago, trying to learn how to explore.

And what did you learn? It's all pretty darn easy. As long as you've got time. one of the more famous where you can just send them a set of Gerber files, a set of SolidWorks files for the enclosure, an Illustrator file for the package design, and they will, within a couple of weeks, send you samples of all of that. And you're like, okay, this needs to change, that needs to change. They'll give you inputs. They'll be like, oh, this enclosure is not moldable because of X, Y, and Z. And you work with them and depending on your level of experience, they'll,

like you'll hire some of their engineers to help tweak the design. And then you just wire them money and then they ship you units and then you sell them. It's interesting to hear you frame it as wanting to learn to solve a business problem. I don't think I've heard anybody really say that explicitly before, because especially from an engineering driven company, it seems like a lot of people, like you said, they want to solve an engineering problem and figuring out how to make a product is an afterthought and and and realizing hey we don't necessarily know how to do this let's let's treat this as a business problem let's build a retail product and and focus on that aspect of it rather than the let's just you know always be trying to to do the hardest

coolest thing from an engineering perspective. That's pretty interesting. And I think a lot of people and a lot of companies could do well to mimic that. Oh, thanks. The neat thing in the last couple of years that's been happening is there's a lot of companies that are out there that'll help you handle a lot of the or at least they're attempting to handle a lot of the business aspects. So like there's Highway One for instance that will I think you guys have probably heard about them

or hang out with them even they will give you something like $40,000 and a team of advisors to help turn your napkin sketch into a real idea and so they've got electrical engineers, they've got to help turn your napkin sketch into a real idea. And so they've got electrical engineers,

they've got product design people, they've got mold engineers, and so they can kind of walk you through, handhold you through the whole process. They'll even arrange, I think, a trip to Shenzhen if you want to go hang out there for a week and see all the cool robots and people that make the products that everyone carries in their pocket well there's those hardware incubators and hardware accelerators

yeah they aren't cheap i mean yeah they give you money and it still isn't cheap no it's true and and for us you know the the, the, the 40,000 that, uh, that highway one gives us, it's like, it's, it's kind of not worth it in the sense of, um, that's to me, that would be so much more worth it to me several years ago when I didn't know very much, uh, about, about manufacturing. But now I know enough about manufacturing that I'm like, well, what if I want to go really big, you know, highway one is the company, maybe their partner, PCH International, is who I should be talking to if I want to go to 100,000 units. But I'm not quite personally or psychologically

ready to go there yet myself, perhaps. So this is where, for me, Kickstarter and other crowdfunding things are really a great solution because they let you gauge interest in your idea beforehand and get money up front to help you pay for the production. Yeah, I was a little interested that you said Kickstarter because you already had a pretty big install base. So why do you go to Kickstarter to get funds like that? But this taking the temperature makes a lot of sense. At our hackerspace, Crashspace, there's a lot of Kickstarters

that have happened. And people, I think, from judging how we've been talking about it, people misinterpret what Kickstarter is for, or at least for hardware projects. To me, Kickstarter is a way of gauging interest and doing a bit of PR, because if you are lucky enough to catch a wave of interest,

it'll get retweeted and blogged about and all this kind of stuff. And Kickstarter gives you a bit of cachet that just some random web page on the internet doesn't. Because there's like, oh, it's on Kickstarter. Like the new Pebble watch that came out today, it's like 7 million now or something in only a couple hours. It's crazy.

Yeah, I think they have, the last time I looked, it was over 40,000 backers this afternoon. And some people have, some people do question an established company going back to the well like that because maybe they don't understand what Kickstarter is for. So it is kind of a difficult problem company going back to the well like that because maybe they don't understand what kickstarter is

for uh so it is kind of a it is kind of a difficult problem because a lot of people have a notion that kickstarter is for truly neophyte companies that are trying to get their first initial funding and have some safety nets so that they know that there's a market for their product yeah um and then pebble comes along they've already had a successful Kickstarter, they've already had two successful products, or I don't know how successful the steal was. And so they come along and then they have another blowout. Clearly it works for them.

And it's good for PR. And it's good for PR. I don't particularly personally have any problem with it, but I do see that angle of, you know, your, no, it's only for new things. And I,

I think, I think personally, I think that having a massively successful Kickstarter makes it easier for other people because people, you know, associate Kickstarter with success. Yeah. And,

and judging, judging from how the stats, from what I've seen of the stats in Kickstarter, because when you have a Kickstarter project, you can kind of see where people are coming from, both outside of Kickstarter and within Kickstarter. I think there's sort of a rising tide lifts all the boats issue. Like, yes, a lot of people have come to Kickstarter to back Pebble,

but they then see the sort of related projects. And there might be other cool things that are on there. Or you'll just start, like, I've got a bad Kickstarter habit. I'm always looking at Kickstarter, finding cool things. And I don't know, to me, it's... For our second Kickstarter, I actually was considering not doing Kickstarter. Sorry, for our second version, I actually was considering not doing... Sorry, for our second version of BlinkOne, I was considering not doing Kickstarter.

But I was talking to one of the guys who works at Kickstarter, and he was like, you know, you've got these people who love your first product. Tell them about the new product. And the best way to do that is to have a Kickstarter, honestly. And also, I think the format of the Kickstarter is quite good because you have a single page that sort of makes the case in a very standardized way. So you get a very consistent description what what the product is all about you know there's always the video there's always a bunch of text with pictures and such yeah i know it's it's a good story and it's an understood methodology yeah i think that's it yeah and so you did two different

kickstarters for blank one based on the different versions. Yep. And you both times went significantly over your goal. Mm-hmm. So, but you had some trouble with the second version. Is that right? Oh, we've had trouble with every, yeah. Tell me about manufacturing in China. Oh, man. Yeah.

So, making atoms do what you want can be difficult that's a broad statement it basically describes everything yeah so i mean pretty much any pretty much every uh problem that one can have with a uh hardware based kickstarter or really a hardware-based Kickstarter, or really a hardware-based product we had. Like, there was the aforementioned firmware problem for this latest Kickstarter. There was a silkscreen, well, I should say a PCB layout issue that could have potentially caused shorting with this most recent Kickstarter,

which is why the metal case had to be modified a little bit. Let's see. There was a, for the first Kickstarter, the metal top part, there was a bunch of issues with the machining, because it was all CNC, and sometimes the CNC didn't, I think they didn't line up the part very well, and so the machining was all off. What else do we have?

We didn't have any typos on the cases, I think, on the packaging. But what else? Yeah, it's like almost every problem we had. Oh, we were late on the software that drives the thing on your computer because of some contractors we hired not delivering quite what we wanted. And that's partly because it's really hard to describe very accurately the software you want. You can make a software design document until you're blue in the face, but until you actually have an application running in front of you,

you don't know that, oh, well, of course, it should do this and this and this and this. And it wasn't written down because it's an assumption that you just kind of assume should exist. And so writing down all those assumptions is very difficult. And so, yeah, we had pretty much all the problems. And we're mostly over them. But yeah, so we just recently had another run of 10,000.

And what problems did we have there? So I think the only problem we had is that for the metal metal injection molding for the for the metal frame they're using a different process so they had to add a slight uh a slight hole to allow the the air to escape when they were they were assembling it but the hole is in a place that is not visible when it's assembled so it wasn't a problem but it was it was again one of these things where the produced thing differs from the original design document so how when if you know the next time you go back and do that how how are you planning to kind of cut these off at the pass not stepping these minds again yeah yeah that's that's a good question it's really. Some of it, in up being in Eastern Europe,

and there was a lot of differences just in culture. I think just computers are set up differently sometimes. And also there's Mac versus Windows differences that come up. Culture of a different kind. I suspect if our accountant is listening this is the time when i'm supposed to say and that's why you should choose companies like ours but this and i are fully booked please don't ask us to work for you and tori if you're listening i i didn't mean that last part of course well that's the thing is is it seems it seems to me that like the thing it seems to me that the better

that you can describe what you want in a in a computer in a computer driven document the better your output is so like Gerber files coming from your schematic capture program

you almost never have problems with the PCBs that are created from those. And then to me, one step up from that is like, oh, the CAD files, the SOLIDWORKS files of your design, of your enclosure design, those, then those will come out exactly as you want. But if you have any kind of fuzziness in your description, you're going to get fuzzy results. Well, yes, of course. I think of Gerber files more like object files. It's after code.

You really shouldn't mess up at that point. Whereas you were talking about writing a software design spec, and that's similar to a hardware design spec where there's lots of leeway. And if you find the right contractors, it can work out. But it is hard.

Well, it is definitely a challenge if you're crossing time zones and language barriers. That makes everything so much worse. Yeah, and also my inexperience with hiring software contractors was borne out. I would love to be able to hire a really good software contractor that just knows everything

and then can help teach me, basically manage upward, on how to be a good client to them. You know, like give them the documents they need to create good stuff. And ask the right questions. Yeah, totally. But of course those people are pretty expensive so you have to balance that. Yeah. Have you seen Tindy's new Yelp for Contract Manufacturers? Yes, I've looked at it a little bit. Would it have helped you with any of these problems?

It sounded like your problems were more design unless your CM ran off and left you with a pile of parts that don't go together. I don't think it would help us now. It might have helped us a while back, but for the most part, we're not making

too many new designs that we don't know all the answers to. We have a whole bunch of different contract manufacturers that we've worked well with before, and we know that if we're doing this kind of product, we'll send out this to them, and if we're doing this other thing, we'll send this other thing out to this other company. And so we kind of know how to manufacture PCB enclosures, things like that, and sourcing accessories like cables and stuff like that, then I might be into it.

But I've not looked too deeply into what they are selling. I think they want to get to the point where they're doing everything. But right now, Tindy is only doing, the ones on there are the assembled boards and the PCAs. So, what is next for Thingam? That's a good question. I've been looking into making a Bluetooth version or a Wi-Fi version of the Blink-1 because we get requested that. A lot of questions about that.

Yeah. Are you deep into the NRF Nordic, NRF 51? Oh, yeah, yeah, the 51822. Yeah, yeah. I've been playing with that for probably over a year now. And what gets me is the FCC certification. Just there's this huge barrier of like $10,000 to $20,000 of getting your RF device certified. If you want it to be small, you have to do that.

There are a few people selling modules that are pre-certified. But you still have to get the entire thing tested. It was simpler. Yeah, it is quicker because you're able to sort of, I think, draw a dotted line around the parts they don't have to care about so much. Yeah. But yeah, and so I've been vacillating between using an existing module versus doing my own layout. And the problem there is that the modules are like 10 times the cost of the,

or I'm sorry, like five times the cost of the bill of materials. And so you're like, how's it really worth it? You know? So, so yeah, so,

and, and, but more importantly, there's been a few other wireless notification lights out there on the market. If you go to Kickstarter or Indiegogo, you'll see a couple of them, and almost all of them aren't successful. I'm wondering if maybe it's the ambient orb issue all over again.

People just don't know what to do with this disconnected light. I wonder if it's hard to set up like it's notoriously hard to get things on the internet um if they don't have a display so you have to have to have some sort of process to for having them join your wi-fi network i believe i had a long rant about that last year yeah yeah it's it's it's a it's a problem that we have yet to solve yet in the whole IoT industry. And so it's just this level of complexity. And also the expense. I think one of the problems, we price the Blink-1 kind of at the most it could be.

It's $30 retail. And one of the ways we rationalize that is it's a retail product. But it's also trying to recoup some of the like multiple years of engineering time we put into it. And it's all open source. And so we can't really kind of lean on the IP aspects of it so much. And so, I think the, if we were to do something that was wireless, it would increase the cost to a little bit around,

like the $100 range at least. And at that point, it becomes not as much of a no-brainer to buy. You have to really kind of think about it. You don't want to spend $100 on a light,

a little blinky light. Whereas $30, you're like, ah, okay, $30, that's not so bad. A lot of it is confusing, too, because there are a lot of internet-connected lights happening right now. Even from light bulbs and stuff.

They're leveraging massive volume production, so it's a different product. There's all these Philips Hue and things like that that are proper lights that go into normal light sockets and hook up to your phone and stuff like that. I don't know if we really want to compete in that space, but one thing that we could do

that would be pretty easy and not too expensive is a desk based BlinkOne. One of the things we've had a couple of people have asked us who have call centers is that they want a light they can sit on a desk or sit at the top of a cube so it mounts better than

this little disembodied fob that the current Blink-1 is. And so if we made a larger Blink-1 that had a base and that sat on a table or on a desk, that wouldn't be that much more expensive. The engineering time is mostly in the case redesign. We could add more LEDs to make it brighter. It would be kind of cool. But it would still be functionally

almost the same circuit as the current Blink one, and it would be backwards compatible with all the existing Blink one software that's out there. And that would be pretty easy, and that existing BlinkOne software that's out there. And that would be pretty easy and that we might do as a Kickstarter in a couple of months. And again, we is mostly just you. Yeah, so

there's me and I'm pretty much the only full-time Thingam person. There is our designer, Abe, who does all the industrial design stuff. He's part-time. And then I've got Leah and Alex and Janie, who are three assistants who help me with finance and customer service and bookkeeping and stuff.

It turns out that getting QuickBooks running for our company was... Don't talk to us about QuickBooks. So I didn't even know that we were going to have to level up in QuickBooks because if you manufacture things, you have to have multiple... See, I don't even really understand it. I'm so glad I have people to help me understand this.

Or I should say, I have some people... I hire people to run it, so I don't have to understand it. Because there is... You have to do with... You bring in raw materials that then get turned into products that then get sold, then you bill against and I don't even know. But it meant that we leveled up in our QuickBooks complexity quite a bit because we were a manufacturer rather than just a retailer. Ours is so simple and we're always getting it into states where. We. We. i was being nice thanks

yeah i i i don't even touch it because i'm like you know just tell me what i need to what information i need to feed you guys to get it all working well not only is it fragile the consequences of making a mistake seem so dire. Totally. It's a very unpleasant experience. So moving along, you were involved with other things. I know you mentioned Crash Space, and you're a co-founder of that, which is LA-based hackerspace. Yep.

But you also do art installations. Yes. Tell me about the art so um me and my okay so one of the first things i did as a sort of a big art thing was this uh installation that was in downtown la for a couple years that was called the crystal monster and it used 18 uh blink of maxims running something like 200 feet of RGB LED strip. And it was made of several hundred pounds of clear plexi with the LED strips on the edges arranged in sort of this kind of skeleton, sort of like a spinal column for some weird space sea monster. It was, it was crazy and it was beautiful. And I was mostly just the technical person on it. But, um, the artist Beverly Tang is the one who was responsible for it all, but it was, it was a lot of fun and, uh, we built it and installed it and it lasted um in kind of this corner gallery

in downtown la for a couple years and and it was very very nice uh lately in the last couple years my girlfriend carlin and i and this uh the guy who runs machine project in la which is sort of a tech art gallery we um installed this thing called the cash machine down at a museum in san diego which is a sort of a pneumatic tube system that takes that sucks money out of your hands to help donate to the to the museum and um as the uh your dollar bills fly through the tubes these little bill detectors will sense the bill flying by and it will trigger bells and lights and spinning kinetic art sculptures and then end up in this spinning vessel where all the money collects. And it's a great sort of visual feast.

And it's for a children's museum and the kids love to swipe dollars from their parents and stick them in the machine and watch the dollars fly around. And so we've been getting some interest from some other museums around the country to install something like that for them and that requires us to build uh even more kinetic art and more little bill detectors maybe better types of bill detectors maybe uh ways of diverting the the bills around to have them have alternate paths and so on. It's sort of a, who is the guy, Rube Goldberg machine for bills? It's a little bit like that, yeah. That would be a great way to get money because I would totally want to put bills in and try it.

Yeah, that would totally work, wouldn't it? Well, it's funny. You know how most museums have a little donation box kind of near the door, and it's a little sad. People don't usually stick much money in it. Apparently, the cash machine brought in more in the first week than the donation box brought in in two years. Yeah, so it's a... It's a toy yeah no and so as a as an art installation um it actually pays for itself which is which is pretty interesting most art doesn't pay for itself well yeah you know it would even pay for itself even more if instead of treating all bills the same, you treated different bills

differently. Oh, man. If we could only do that. Unfortunately, when you feed the bills up through the system, you have to slightly crumple them because otherwise they just laminate against the side of the wall of the tube because of turbulence or whatever. And so, you can't really detect what the bill is at this point because it's kind of in a little ball

I'm thinking of horrible variants of this right now flames, shredding shredding in flames really? I was going with kittens through the tubes? you're going to put the kittens through the tubes? and then they go and

they come out a little sick I don't think that's a good idea. Sorry. No animals were hurt during the recording of this podcast. But yeah, the let's see,

Crash Face itself is full of lots of arty folk that do a lot of more interesting stuff than I usually do. But one of the neat things that happened maybe three years ago is that many of us, I think like eight of us from Crash Space, were on a TV show that didn't end up getting picked up, but it was an offshoot from the Mythbusters. And it was essentially a contest to build Rube Goldberg-like machines.

We saw that show. We saw that show. Oh, really? Yeah. We did. You were on that? Yeah, it was. I'm going to have to take that back up.

I was on the first episode. It was called, the show was called Unchained Reaction. Yes, yes. That was a really cool idea. I'm sad that it didn't get picked up. Yeah, well, it's, so part of the problem, apparently,. I'm sad that it didn't get picked up. Yeah, well, it's part of the problem, apparently, I'm not sure if this is true or not, but

so the machines that we made in our contest didn't have a specific goal. They just were like, do something really cool, and it has to be this level of complexity, like have 20 different things that it does. Whereas, but that's not a Rube Goldberg machine, because a Rube

Goldberg machine is something that accomplishes something very simple in a really complex, crazy way. So there's a specific goal in mind. And I think this was one of the problems, is that it was hard to tell what was going on. Why is this doing all this crazy stuff like what's the point and um and i think they couldn't use the rube goldberg term and the rube goldberg concept because the rube goldberg like like they couldn't license it from the rube goldberg estate or something which i think it's kind of weird that that is a thing you have to license but

there you go well then you were on the premiere episode and that one's free so everybody can watch it yeah we'll put a link in the show notes yeah yeah we we did pretty well but we were up against a team of uh set builders and so they knew how to actually use a pneumatic nailer. And it actually knew how to saw things. And we're a bunch of electronics hackers, mostly. And they're expecting us to build things with fire and pneumatics and tall. And I'm like, the things I build fit in my hand. None of this really works on camera.

Yeah, that's a problem from the producers. They should balance things better. Well, I mean, we came up with some pretty good stuff, and a couple of our people had some experience with some of these things, but yeah, overall, it was, I mean, I found it very personally fulfilling because I got to learn how to use a bunch of pneumatic tools, which I'd

never ever touched before. So, that was, I'm like, I'm going to get an air compressor now, and I want to get a nail gun, and I want to get a... Yeah, I'm going to get an air compressor now, and I want to get a nail gun. Yeah, I could see that. Well, we are about out of time. Chris, do you have any last questions? It's okay to say no.

No. All right. I mean, you could ask Todd about his toast. His toast. And his toaster. What? That's been a huge Twitter storm today. What?

Toast is important. Todd, do you have any last thoughts you'd like to leave us with? Let's see. See, it's hard. It's hard. It is hard, yeah. So how much of this is going to be edited down? Some, not all. Okay.

Mostly only places where we actively say something wrong or super long pauses like that awkward one you probably still got left in and this meta discussion about editing well I don't know we could leave this in but yeah any last comments

well yeah I don't know I think people should try building stuff at a production scale more often because I think it's not as hard as people might think it is. So you're bucking the hardware is hard trend? No, no, hardware is hard, but it's so much easier than it used to be. Good, cool. Yeah, yeah.

Well, then you have a coupon code for listeners. Yes, for Embedded FM listeners, if they would like to get a BlinkOne for themselves, there is a coupon code called EmbeddedBlinkOne. And if you use that when you check out on our website, which is buy.thingam.com, they can get a BlinkOne for 25% off. And that is EmbeddedBlink, all one word, and then the numeral one. Yes. Without a space, just embedded Blink, smashed it, one smashed them together.

Yes. Cool. Our guest has been Todd Kurt, co-founder of Thingum, and so much more. Art and all sorts of stuff. So before I go into the whole thank yous that usually happen now, I have one little announcement. We've decided to monetize the podcast. No, no, no, it'll be all right. I promise. We're going to write about the podcast for Element 14 and it's going to be a blog. There's going to be some synopsis, you know, who's on the show and what we talked about,

but it's going to be more than that. There's going to be some analysis, maybe something we think is especially interesting about this guest or some topic that's tangentially related that we decided to write about and just used this guest as an example. Anyway, we'll write something there and you can check it out as our after analysis and maybe hang out at Element 14 a little bit because they're paying us for it. And this lets us, you know, keep paying our ever expanding Libsyn bills.

They're not expanding, are they? They could. All right. Well, in the meantime, that's the plan. And we will let you know when it happens,

but be excited, you know, thank you for listening. And we will let you know when it happens. But be excited, you know. Thank you for listening. Also, thank you to Christopher White for co-hosting and producing and hating the word monetization. If you'd like to say hello, hit the contact link on embedded.fm or email us show at embedded.fm. If you'd like a sticker, send us a note. Tell us you told somebody about the podcast and include your address.

And don't forget about that coupon code, embeddedblink1, and you can use that on the Blink One website to buy one. Thank you. Thank you for listening. And I do have a final thought for you, this one from Joss Whedon. He had something very important to say. As we all know, blinking lights means science.