 Hello, all, and welcome. The following session is about the Blinken Rocket, the nifty device that some of you might already know. It's a 100% open source do-it-yourself soldering kit that offers the best hands-on experience for beginners of all ages. Overflow and Mutsi are here to tell us all about their journey in making the Blinken Rocket fly. Let's hear some applause for Overflow and Mutsi. So, oh, well, I can hear my voice. So we're waiting for the slides to appear. And while we're waiting for the slides to appear, a warm welcome from my side as well. I Mutsi, this is Flo, but we're going to introduce ourselves later on again. So let's spend some time waiting for the slides to appear. And also a warm welcome to those who are in the streams. At least I got one hint that somebody is on the tram right now. So they probably have good connectivity to watch this talk as well. So you're going to walk with us through the journey of Blinken Rocket, which is this little thing that I have here on this band. And we're going to talk you through various topics. So we're going to start with our project goals, because you want to have goals before we start something, just like with any product we have. Then we're going to talk you through this little hardware, what we have on there, the design considerations behind that, and probably most interesting, the software running on this little device. And after all, we're going to talk you through how to actually productize a do-it-yourself soldering kit, because we know that there are plenty of ideas here in that room. And if you want to distribute your own do-it-yourself soldering kit, well, you can learn probably a lot from what we've went through and what we've learned, and also what you can do right in the first place without actually failing one, three, or four times. So shortly about us, this here on my left side is Flo. He's a dad of three and a member of the MetaLab. And he's the guy behind the Hackerspace shop. That's an online shop for various do-it-yourself kids. You created plenty of do-it-yourself kids in the past, so he's like more experienced than I am. And at daytime, I do big data topics. And at nighttime, once the sun sets, well, I'm creating do-it-yourself kids for kids. So at this point in time, I want to thank all the contributors that helped us making this little thing possible. So a big shout out and thank you. You're awesome, guys. Yeah, and you can give them a round of applause. So let's start with Project Gold. If you see this little kit, and you have probably seen it already today, because at Junghaka Talk, there are plenty of children actually soldering this. So we have a target audience. So first of all, if you create something, you have to think about who uses it or who's the, well, intentional audience behind that. So we have children, youth, and adults, basically, as our audience. And when they sort of this, they're probably at home or in workshops, in schools, universities, or even Hackspaces. Well, most of us here come from Hackspaces, so that's kind of obvious. And we have some requirements derived by that. First of all, I hate it when I have to use a special programmer to program my devices. If I have something great, I just assembled it, and then I need to assemble something to just program it. That's awful. And also, I hate it when I have to buy something because I forgot to buy it. Like with a little Arduino thing, in the past, you had to have this USB to serial converter. And I almost never thought about that, so that was kind of bad. Also, given the target audience, children, well, they don't feel small parts, but adults do. So we try to choose big parts, and they're even like little, small SMD parts here, and they can be big too. So we made the footprints wider and bigger so that if you solder them, they will flow nicely, and you cannot do it wrong, actually. And we wanted to have cheap components so that if you buy a kit, you do not have to spend too much money, and ideally, it should be free anyway because we do not care about your income and everything should be free when it comes to learning stuff. Also, the most important aspect, if you make something, it should be, well, useful in the first place. It should be fun and also educational, especially if you're targeting children, they should learn something. And with this kit, they can learn how to solder. They will see that something in the digital world can be translated to the real world, so that was our main aspect. And it's, well, kind of fun to have something that blinks your name or has nice little animations on there. So based on these considerations, the blinking bucket was born about two years ago. And on the front side, you see the 8x8 pixel dot matrix display, then the audio jack for uploading data, and that's like one thing that addresses this no programmer needed aspect because we wanted to have the ability to use your mobile basically to upload new texts or animation to that little device. So this is kind of special to that we just have soldering kit. Also, not interacting with something is kind of boring, so we added a little two push buttons so that you can move between the animations and texts that are scrolled through the display. On the back, we have, of course, a coin cell that powers the whole device, and it lasts about a day or two, so don't worry about new batteries. Also, we have a little 80 tiny microcontroller, which was the cheapest microcontroller back in the day with that much pins to drive the display directly. And also, we have a storage with 64K of storage available for your animations. And that's roughly 2000 to 3000 characters of texts, and they can be a little hard to read on that little display, so it should be sufficient for all time. And also, some additional little SMD components, well, to solder everything. On the software side of things, everything is open source, like this awesome Harold already mentioned, we wanna make everything open source. And when we talk about everything, it's not just source code or the PCB layout. We also made our documentation open source and licensed it the proper way. We made our manual that's actually printed and handed it into every kit also open source. And not only the PDF, also the source files that you can assemble it on your own or remix it based on your needs. The firmware was written in modular C++. I don't think it matters too much, actually, but I think it's worth mentioning though. Also, we have a web editor for putting little animations and text on this device, so you can use it on any device, whether it be your mobile phone or your computer, everything with a web browser and an audio device is capable of programming the Blink rocket. And it's written in JavaScript, of course, if it's a web application and the one initially created it used the awesome React.js framework because it, well, made things a little easier. The audio transmission itself is using square waves, so little square signals coming out your audio device. And you have a protocol using forward error correction so that if there is a mistake within the transmission, it will be fixed automatically. So after a while, we created a thousand Blink rockets, so a thousand kits to be distributed among children, youth, and adults. They were manufactured by lots of volunteers in mid-2016 after roughly half a year of development time in our spare time. It was initially funded by the CCC. I'm not too sure who knows the Karlsmarthschule initiative by the CCC. If you could raise your hands. Well, it's like the whole crowd. So if you don't know it, go ask your neighbor after this talk what this awesome initiative is. So it was funded to be part of these workshops taking place in the Karlsmarthschule initiative. It was packaged by volunteers at the Hexbase in Stuttgart. So it took roughly a day for 10 people to assemble a thousand kits. And it was distributed across all Europe for workshops. And here are some impressions of workshops taking place. They were awesome. The kids had plenty of fun to assemble their Blink rockets. And in the moment, they slid the battery in and everything started to blink. It was just awesome to see their faces and well, see all that enjoyment in their face. But there was a big catch and this is where I head over to Flo because we learned a lot and he's going to tell us about what we've learned. Yeah, so hello. Is this on? Do you hear me? I'm not sure if I'm on or not. Yes, okay. Hello. So yes, there was a little catch and we learned a lot. So the first thing when I got my hands on the first Blink rocket kit, I realized it was a really good one and there was potential to make this bigger and distributed to more kids. But the first thing, the first problem that occurred to me was that we used square waves to transmit the data on the Lenovo ThinkPad. For example, that worked out just fine. But the same manufacturer for another notebook, it didn't work at all. As you can see here, the data was totally distorted and we could not process that signal. So transmission from the web browser to the rocket wouldn't work reliable. Some common errors that happened were that the signal would drift over time. Like in the first picture on the left, you see a yellow curve and that's the signal that you get out of the sound card. And the blue signal, the blue curve is what the computer would make out of that. And by the voltage drifting, upside down to the y-axis, you see that this signal is really hard to process and it just wouldn't work out reliably. And also, on the right, you see the heavily distorted signal. But why? Why is that happening? Well, every sound device works a little different. We have different manufacturers of sound chips. We have different sound cards. And they all do similar things, but not exactly the very same thing. And most sound cards are troubled by square waves. So if you make a square wave signal, it goes like this, sound cards cannot reliably produce that, or some of them, most of them. And also there's various layers between the web browser and the actual sound card. For example, you generate a sound object in the web browser and it's given to the ALSA driver and that gives it to some hardware abstraction layer and that gives it to the kernel and that's handed over to the actual hardware. And in all these steps in between, the signal can be modified and is modified especially at the hardware level. The signal is modified so it sounds good to the human ear, but it's not really meant to submit data. So, Chris Weigl came up with a solution to this problem by introducing ADC, Analog Digital Converter Edge Detection, which means he sampled the signal a lot. Sometimes, and it worked much more reliable than the old way of simply detecting the edges of the square signal. But still, there were some issues. It didn't work 100% and he spent some more time digging into this problem. And he came up with this solution of frequency shift keying and that worked really good. Frequency shift keying means that you make not square waves, but sine waves. You can see that the form is like, it doesn't go like this, but more, it's more curvy, to say. And that's what sound clouds usually do. They provide waves and that kind of transmission worked out much more reliable. It's also much slower, but that's okay. We are talking about some seconds to transmit the data instead of some milliseconds. So that's totally acceptable. If you plug it in, it will take you, I don't know, two seconds and then you have the animation on the rocket. Once it was fixed, I started to take a look at the hardware level, how the kit could be improved. And one thing that we found out during workshops was that the battery holder was not the best choice because people would put in the batteries in various ways that were not intended. So this is what the original battery holder looked like and you could insert the battery like this, which is actually wrong. It looks right, but it's not because a part of it is bent. And there's also this other way of inserting it that happened a lot. And in this way, the rocket wouldn't work. So yeah, I got rid of the battery holder and introduced this one. It has some benefits. It's a little more expensive than the other one, but the battery slides in from the top and there is just one way to insert the battery the wrong way and that doesn't happen too often. Well, if the other battery holder, there were eight possibilities to insert the battery in the wrong way. So that took away a lot of problems. Yeah, then the original matrix module that was included was also very cheap, but I thought that if you make a kit that you would hang around your neck, it should look nice. It shouldn't be 10 cents cheaper, but it's worth to add a little extra and have good looks. So we introduced this new matrix module that has square dots and diffuse lights and it just looks better. And then I revised the PCB design a lot and not a lot, a little. I just shuffled around the components and I penalized the design for production, which means that I took one rocket and put it on a panel like this. So we have 14 rockets and you can feed this into a pick and place machine to automatically produce the boards. So why is that? Because pre-populated versions are targeting different users. The kit, when I got it for the first time, it was rather complex and you had to solder the MCU and the EEPROM yourself and I'm now offering three different versions. One where you solder all the stuff yourself for experienced users and one where the EEPROM and the MCU is already in place for targeting kits. And one version where all the SMD components are already on the board and you just soldered the six for whole components yourself, which targets very young people like my daughter made one, which is six years old. So this is when version two was born. Yeah, you see the new Matrix model there, you see the battery holder and the components are more far away than in the old design to make it easier to solder. And that's the thing. If you make one board, it's quite easy. You make the design, you order it from China, get your tank copies and then you assemble it. But if you make a thousand of something, you run into various challenges and problems. So for example, for the first production run, the original kits that were made in 2016, everything was done by hand. So Mutzi built this programming mass programming fixture with a zero injection force socket where you place the microcontroller inside and you have a little script that flashes the microcontroller with the actual firmware. Making thousand kits means opening the socket, inserting the MCU, closing the socket, hitting the button 1,000 times. How long did it take you? Well, roughly a day of just removing a single chip out of the packet, putting it in, closing the lid, pressing a button, opening the lid, well, moving everything out into a little plastic bag. And these were kind of nasty because they were like those Ziploc bags and well, they were hard to open. So I spent roughly eight hours flapping a thousand microcontrollers, which was awful, believe me. It's not fun. If you ever make a kit, this is something you don't wanna do. It's just... Don't do it. Don't do this. There are alternatives, I will show you later. And then, as you can see on the right, components were put on labels. So the kit consists of the actual MCU and EEPROM and eight different other components like resistors, capacitors and diodes. And they have different values. So what they did in the first production run is they took little stickers, little labels, and cut all the components up and put them in the correct position on the sticky notes and put them in the kit. Eight parts times thousand kits is 8,000 cuts. Being there with scissors, 8,000 times cutting up parts. You need a lot of people or a lot of time or both. How long did that take? Well, we had roughly 10 people helping out, thankfully. Otherwise, I would spend like weeks or months doing it. And we spent roughly a day with 10 people to create a thousand kits. And that includes also like cutting down parts. And people were really eager to help. So I was really thankful, but this wasn't feasible for the next thousand kits. Absolutely not. Yeah, it's not fun. It's not a fun activity when building kits. This part should be done by machines. That's what they're for. So when I kind of took over the production of the kits, I figured I should save time by using machines. Even if it costs a little extra, it's probably worth the time and money. And I should get pre-programmed MCUs. You can pay a little extra some cents and you get the MCU with the program in it from the manufacturer. It costs like 10 or 15 cents, something around that. And it's totally worth the effort. And I introduced color coding of S&D parts. In the old kit layout, you had the label that said R1 and then you put the part right there. In the new kit, you have a part that is labeled and there is no text on it, but it's a red part or a green part or a blue part. And you get a little card that says put the red part on R1 and the blue part there. That took away a lot of time when assembling the kits. And finally introduced a new package. It now comes in a little plastic package that you can use for transportation and also during assembly. It's very handy. So yeah, I mentioned robots before. When I was faced with the job of making a thousand kits, I thought that I would not be able to cut these parts up myself and I should make a machine for that and let the friendly robot take over. And that's what I did. I'm gonna show you a short video now where you see how that worked out. Yeah, a little fast forward there. Yeah, so I called this machine, that's a hacker. Yeah, that's a hacker. It took me some time to build it. I thought, yeah, take some scissors in the motor and it'll work. Took me four iterations. What it does is it cuts SMD parts from the drill. It also color codes the parts while cutting them. So there's a little marker that you can see here that puts a strip, a color on the strip before cutting it up. So I can just replace that for each part so they are color encoded. And I have two buttons where I can adjust how many holes I want to move forward before cutting it up because I have some parts where I need three pieces in the kit and I have others where I just need one. And it cuts a lot fast. It's really fast. I cut up to 1,000 parts an hour which saved me days of work to make these kits. It was totally worth the effort. But it took me four iterations to get the machine right. The main point I learned is that scissors are a spare part that must be replaceable. You don't want to build a machine that has some kind of mechanic to cut up parts that is not replaceable. So I used open-ass curtain, my oldest 3D printed designs to hold the scissors in place. And you see here ball bearings from roller blades that push down the scissors. And there's one part that's called a sprocket. A sprocket is like a little wheel that has spikes and that's something I actually bought in China for $15. And that gets in there and it's used in commercial pick and place machines to move forward the tape. So PCB fabrication. Once I had all these parts cut up and the firmware issues fixed, it was time to go to the fab house and have the PCBs manufactured because I got the panels that we showed you earlier from China and I wanted the parts on there. And I made a little video of how the process works. It was really fun to be there and it was a very nice afternoon to watch all these huge robots doing the work. So I will show you a little how that turned out. So here's panels and this is when the solder paste is applied. The solder paste is now on the panels and then they go into the pick and place machine which is configured before. And this is real-time footage. That's how the robot really looks in real-time. It moves quite fast. It scans the part, adjusts it. If it's not positioned use adjustment. It goes into the oven and once it baked, it comes out of the oven and ta-da! It blinks. You just have to assemble the through-hole components yourself and then you're done. So yeah. And what caught me by surprise was the cost structure to make this kit. I did not expect that the cost would be this, how do you say? Yeah, well, we have assembly, tooling and supply chain cost of about over 50% of the actual kit costs and 50% is material. So even if you use machines to make a kit, it takes time and in this case, it took more time than I expected. So I'm currently selling those kits at my shop for 25 euros each and you buy basically two kits and one is given into a little box and if you make a workshop with kits at a school where they don't pay for it, I take the kit out of there and give it to the kits for free and you might get a little email with pictures afterwards so you can actually see what happened with your donation. Yeah, so that is my part of the talk. I will hand over to Mutzi again. So I think we learned a valuable lesson about the cost structure because that will affect all of you creating DIY kits because the cost might not be obvious if you browse through shops and see that likely your microcontroller costs less than euro but then all the other costs can add up to what we've seen there. So it can be surprising and you should be aware of that. So we all ask ourselves right now, what's next? And I had a look in the audience while Flo was talking and I've seen plenty of Blinken rockets. So that's already a good sign. So what's next? Today there are soldering workshops for children here on the Congress. So if you are, I'm not sure about the age but if you're a children, go there, have fun and get your Blinken rocket. There will be more workshops because the first kits were distributed, well, in 2016 and Flo did create another 1,000 kits. So there will be more to be distributed for those workshops. There will be you can hack workshops in Austria next year. So if you're from Austria or living nearby then you might want to consider going there or well, if you are not an adult of course and sold your own custom Blinken rocket. But even if you're an adult, that's not a problem because you could create a workshop in your community. So if you like to have a nice Blinken rocket soldering workshop, don't hesitate to ask us. We have all the material required to create your own workshop. We have guidance, we have manuals, we have best practices. So just ask us and we will try our best to make your workshop a success. And also you, as Flo already told us, you can buy a kit in his shop and there's automatically one kit being donated to such a children workshop because we do believe that if a child does attend a workshop, there should be no fee attached to that because, well, we want to treat everybody equal and money should not be a thing there. So if you buy one there, of course, it will go to one free workshop for a kid. So I think that's a good thing. If you got some time to spare and I think everybody does, especially if you're in your local hex space, you could plan a local workshop. So please do so. If you're a web developer willing to spend some time to improve everything, well, there's a GitHub repository with all the code in there. You can create a fork. You could work on the issues to make everything even better. Like for instance, better support for animations or better editing tools or create a share button for sharing animations with your friends. Or you could just spread the word. Everything is open source, so we want to encourage you guys to actually use what we've created to create your own stuff. So please be welcome to do so. This is the reason for making everything open source and we would kindly ask you to do the same with your stuff. This is our talk already. There follows a Q and A session right now. If you want to ask in German, if you're not so familiar with English, don't worry, we will translate everything to English first and then answer it in English and there will be the translation angels translating it to German on the right channel. If you want to contact us after the talk, there is an email address, there is a Twitter handle and there's also a decked phone number. It was seven... Seven, eight, five, I think? Yeah, seven, eight, eight, five. Or just come to the stage right after the talk so we can have a chat and we can show you everything in detail and, well, help you out. So thanks a lot. Thank you very much. And applause for the speakers. We have four microphones set up in this hall. Please come forward with your questions and we also take questions from online. Anybody? Don't be shy. No questions. No questions. No expectations at all. So... Oh, there was a question. Just a quick question. Do you also sell the kits on-site? Yes, I do. Actually, we do. The kits are available in the hardware hacking area right there in Hall C with the white umbrellas there. It says FPGA right there. I have some. You can just follow him after the session. So I think he's willing to give you one. Or more. Thank you. Is there anybody else? Well, I guess not. Thank you very much. Over for... You're welcome. A round of applause, please. Have a great Congress.