#673 – Lifelong Learning with Bitluni

Chris Gammell: This is the second episode where I've had some less than stellar audio. So I wanted to just first call out our wonderful editor, Matthew, who's been working with some garbage I've been handing him. And also to pour one out for my AT2020 USB. It was a 13-year microphone. And it has now, unfortunately, given up the ghost. That being said, this episode of Bitluni turned out great. Content continues to be king. And future episodes of The Empire will be with a shiny new microphone. So thanks for listening. This is The Amp Hour Podcast. Released July 15th, 2024. Episode 673. Sponsored by Slosh and McWashtie. Lifelong learning with Bitluni. Welcome to The Amp Hour, I'm Chris Gammell of Contextual Electronics.

Bitluni: I'm Matthias, also known as Bitluni on YouTube.

Chris Gammell: Hey, what's up, Matthias? Welcome. And sometimes Luni, you said as well. So maybe we'll go by Luni.

Bitluni: Yeah, that's my short. Yeah. Thanks for having me. I like it.

Chris Gammell: Hey, my pleasure to have you here. We've been talking about your stuff a bunch. We just had Matt Venn on a couple of shows ago talking about silicon. So we'll definitely be talking about your entrance into that market. But before we do that, I mean, you've been making electronics on YouTube for a while now. So what do you like building?

Bitluni: Yeah, it's been now 11 years on YouTube. Yeah, it's a lot. I mean, I started... My background is not in electronics, really. I'm a coder, rather. And that started as a hobby. I think one of the first projects was MIDI pedals and music instruments, synthesizers. And that were the first projects that I documented on YouTube and very poor quality. You can still rewatch them. I didn't take them off. And then it grew like with Arduino and ESPs at some point. And then some of the more popular projects were like the LED walls, the ping pong LED walls, which are very basic projects, but reaching like the crowd out there and people who do it as a hobby and looks nice always.

Chris Gammell: Yeah. No, I think that's the kind of stuff that... When people think about projects they want to build, it's stuff like that, right? It's showing people the stuff that they see it. It's maybe accessible, but then it's also... There's a technical challenge to it. And it's also something that they're like, I want that on my wall, right? That's like a great combination because you're showing people how to do it. And then you're also kind of inspiring them with these early projects. I'm always surprised that the first projects are often tough ones to find for beginners. And not beginners, even just anyone who's kind of diving into the world of electronics. It's like, what do I build first?

Bitluni: Yeah, usually it's something that's more presentable. Or yeah, I had like always... The thing was where I reach people in the beginning. Now it changed a lot, but it was because I was a beginner myself. And I was like this Dunning-Kruger effect. I was convinced, oh, now I'm a crack because I made a PCB or something. But I was documenting all the process along the way. And people... I didn't know the specific terms or the language of electronics. I barely knew Ohm's law or anything. And that was like maybe the language that most of the people who were interested in those topics were speaking and understanding. And that's why they kept following. And I think that changed a lot. I'm still in this Dunning-Kruger bubble where I think I can make ASICs at the end or...

Chris Gammell: I mean, you are making ASICs just so we're clear here. I mean, you're... But I think you're right. The mindset of like, I can do it. That's a big, bold jump, right?

Bitluni: Where it's like, yeah, you can do it. But it's bold to like, yeah, it works somehow. And now I present it to the world as, yeah, here I am. I'm the crack in electronics.

Chris Gammell: Of course I did this. Look what I have brought.

Bitluni: The rule of YouTube makers, it has to run once for the camera. No, no. Movie magic. It's a joke. But yeah, now I moved on to a little bit more complicated projects. Also like every topic that's quite interesting to me, I take on that and try to do a little bit of research. I don't know if you have seen the video about like face arrays. That's like...

Chris Gammell: The DIY Sonar Scanner one?

Bitluni: Yeah. Yeah. The Sonar Scanner. Yeah. That was something where most was more in the educational realm, let's say, where I presented how waves propagate visually. And many people like that video, even though it's not like the video with the most reach. But yeah, some people like that one. And actually, that's one of the projects I want to continue. I started with FPGAs after that to be able to take out more of that Sonar Scanner stuff. I don't know if... Yeah, we can maybe talk...

Chris Gammell: Yeah, so that was like... Yeah, so that was like you were processing. You were like basically like imaging the world. I think I remember seeing... There was stuff... I remember watching it and thinking like it was kind of like Daredevil, the Netflix version of Daredevil kind of thing where you're... And you were actually visualizing the world by doing sonar maps, right?

Bitluni: Yeah. Yeah. That was the core of the project. It was basically ultrasonic scanners, but as an array. And then you can basically reconstruct where the echoes in space, in 2D space. You can make it... You can do it in 3D space as well, but that was only like a 2D, like 1D array that was visualizing basically this space in front of you. And there is like a story to this project specifically. Yeah. So my goal for that was to... You probably know the movie Aliens.

Chris Gammell: I never watched all of it, but I know many of the hype ones.

Bitluni: The second one... Oh, the second one. Yeah. James Cameron. Yeah. They have like sonar scanners there. Like they visualize like the aliens and it has like the specific sound. And now, yeah, I was always fascinated. There was a fake on the movie. Actually, if you watch... I'm a film fan and if you watch like the making of... It was like the CRT is like way too long. There is like a color CRT and you watch... You look at from the top so you can actually... You don't see that it's sticking out of this whole contraption just to show the visuals. So it's not real there, but I wanted to make it real even with those CRT screens. So that was another thing of mine doing graphics on CRTs, generating video signals and so on.

Chris Gammell: See, that's great too because it's going to be like, you know, some younger YouTuber is going to watch your stuff. And do that and it'll be like art inspiring science, inspiring art, inspiring science. And then you just get these like leapfrog effects of just, you know... Yeah. Yeah.

Bitluni: That's actually... And fun fact is I never finished that like the actual scanner right now. I have a version that is not like... It's not film, but I have like FPGA version that I started to do on stream, which is a little bit faster, like because you can decode in parallel and stuff. And now the alien movie Romulus is coming out in six weeks, roundabout. And they... You can hear in the trailer already that they used those scanners again. So... Oh, really? Yeah, that's a hot topic. And I think...

Chris Gammell: So is that like the same... It's the same universe, but like a new version? Is it like they're kind of continuing with Doggett?

Bitluni: I think it's like a prequel or... Prequel. There's like... I mean, the whole story, Prometheus and so on. It started before the alien movies later. And it's something in the middle between the prequels and the movies from the 79 and 82... Got it. No, 82 or 88. Something like that. Right.

Chris Gammell: But same franchise. Same like... Yes. Kind of like... Yes. Yes.

Bitluni: And I hope they pick up like in the old style from James Cameron or Ridley Scott.

Chris Gammell: Yeah. I get a little frightened of those kind of movies. I'm a bit of a winch, so... Oh, really? Yeah. I don't know why. I think also when it came out, it was like I was... It came out a little bit before I was born. And then it was through my whole childhood. I was like not allowed to watch them. And by the time I did, I just wasn't... Maybe I should do it now. You know what? I'll watch your video. I'll go watch the scanner thing.

Bitluni: Oh, wow. Yeah. You can watch. I mean, the new one. I hope it's good.

Chris Gammell: Yeah.

Bitluni: Yeah. You should check it out. At least I will try to finish the projects. Most of the projects are never finished. I will try to finish it like somewhere on time to be in this trend bubble.

Chris Gammell: Right. Right. Well, what you really got to do is try and get the movie to pay you to do the content. You know, that's what you really need. Oh. Call the studio.

Bitluni: Usually that never... YouTube rarely pays. I don't know.

Chris Gammell: You have to be Mr. Beast to be... That's right. Exactly. Yeah. It's a winner take all. You know, one thing I like about looking through all of your video kind of just thumbnails and just the things you're doing there too. One thing that I think really kind of pops out to me, especially with the phased array, is beginners... Just me as a beginner, right? I always looked at those ultrasonic sensors. You know, it's like these are the ones that look like little cans that have like the screen on over them. I was looking at them. I'm like, oh, there's so much I could do with them. And then it's, you know, basically I just run the Arduino sketch where it's just like my hands getting closer and closer. And I'm doing just like the, just the distance. And that's interesting, but not that interesting. And so like the fact that you're taking this stuff and then extending it into the realm of interesting and almost past, you know, like those ultrasonic sensors are not meant for that space, but they can be used in that space. And then you're kind of testing the limits of where they are and then doing a lot of software and other kind of interesting things around it. That's really powerful, I feel like, because it shows all the things people could be doing with them.

Bitluni: Yeah, what I was, when I was doing this project, it took me like three months to do everything and record. And I think that that video is all right, popular. I think 600,000 views or something, maybe more now, but it barely paid for my upgrade. Let's say so. And it was a lot of work. And yeah, okay. That's a little bit of YouTube meta. But during that work, I learned like, this is the part of this lifelong learning or self thought things that the technology of the arrays, the phased arrays where you have like beam forming is the keyword here, opened up to me all the parallels of the technology. I mean, that's not working only in sound. It's also in electromagnetic, magnetical space, light and so on. It works in all the realms. Similarly, the funny thing is that I didn't think a lot about that. When I used those distance sensors, I actually sold out of the cans of this distance, distance sensors to, to like start doing that project before I ordered parts and so on. That ultrasound scanners for medical ultrasound scanners, basically, they work the same way. It's identical. And I worked in that area for five years as a coder, like visual, making 3D visuals for medical treatment and ultrasonic and also CT and stuff like that. And all that stuff works the same way, but a little bit smaller, microscopic crystals and so on.

Chris Gammell: Well, like better, fancier sensors, right? I mean, like more tuned, high quality, better power supplies. It really, like you're saying, the same thing applies. It's just then, so then why is the cost so much higher? It's because of all the other engineering that goes into it, right? I mean, like all the precision you require.

Bitluni: And the fancy thing is about that, like, I mean, the topic about phased arrays, they have been around for, I don't know how long, like as long as radio is around. Because you have like the array antennas and so on for basically radars, of course, like military stuff. Sure. So that's like analog stuff. And it's always amazing to see what is already possible, even though if you have no computation power at all.

Chris Gammell: Oh, yeah.

Bitluni: I mean, that's all analog. And then you just process the outcome, the result. And stuff like ultrasonic scanners or CT scanners, they work basically the same. They have been around from the 80s, around about, I don't know, maybe earlier. I don't know for sure.

Chris Gammell: Yeah, I'm not sure.

Bitluni: So that's always like mind blowing to me to see we have like C64s back then or computers like that. Yeah, right, right, right. And then you are able to visualize the internals of a human body or something or see airplanes on the radar.

Chris Gammell: I think as the costs come down to, I mean, some of it is just true unit economics too, right? So you think about the earliest CT scanners, it was all this work just to get that one thing working. So of course, it's going to be the priciest thing. But then as you start to have standardized components, as industry grows, it feeds back on itself. You have other electronics coming in, computing lower, like all of these things get lower. And even the point where like for ultrasonic, so a former guest, Greg Charvat, who was on the show a whole bunch back in the day, he worked on one that's literally like a handheld ultrasonic that talks to an iPhone. That's called Butterfly, Butterfly Networks, Butterfly, no, Butterfly Networks is the Bitcoin miner. There's a butterfly something, butterfly, whatever. And it's literally like a handheld ultrasonic so you could do remote medicine. Oh yeah, telemedicine, yeah. Yeah, it's amazing. And that's the level it's shrinking down to now. And then it just keeps feeding back on itself. And then it finds new areas like you were talking about. Yeah, it's just technology, man.

Bitluni: Yeah, yeah. But it's also amazing how analog tech, humans developed analog tech before digital, even though digital is way simpler. And analog like the physics, developing all the physics and making it work. CRT TVs. It's crazy. There's like a accelerator beam that's shooting at you and displaying images.

Chris Gammell: Yeah, right, right, right. What we do is we shoot electrons, but then we stop them with phosphorus. Yeah. And then we're going to put kids in front of this screen so they can watch TV.

Bitluni: And then you can use a high voltage, like this high voltage beam and then magnets to deflect it and to draw super fast in the image that's transferred over air. And that's barely, like that was developed barely after we made the telephone or something.

Chris Gammell: Yeah, right. Exactly. Telegraphs. Well, I mean, yeah, right. I mean, in this realm of like technological surplus, also like, you know, I can feel like cognitive surplus is like a bad term, but it's not really. It's like people who are spending free time exploring, right? I mean, it's your community. It's, you know, the stuff that you're a lifelong learner as well. One of the things that is limiting at that point is just where we're pointing this technology. Like, so again, looking at all the videos you've made, ESP32 is very common in your projects and just the amount of compute power that's in there and the things you could point it at. It really comes down to just how, how much people can dream up and actually point it at. If you think about the surplus, the technological surplus of ESP32 and what, what you like to point that stuff at, where would you, where would you say you like pointing the ESP32 these days? We're going to get into the chip stuff later, but otherwise, right, you, you kind of have focused on the ESP32 as a, as a platform and as an enabler. Where do you like pointing that stuff? I see like a lot of like visualization and video and things like that, but maybe there's other areas.

Bitluni: So my thing is always like testing the limits of, of stuff like that. It seems like the microcontrollers are getting so powerful that that's overwhelming. I mean, the possibilities are basically endless. You can run Linux, Linux computers basically, and do everything on them. Yeah. That's, that's the current state of microcontrollers. So where Raspberry Pi started at some point, there is where we end up connecting to with the, with the ESPs right now or other brands as well. Also the FPGAs, they, they are so cheap and affordable. The ones that I'm using right now, the GoWin FPGAs, like 20 bucks for the 9K. One is able to run MS-DOS emulator on that with HDMI output. So basically you have everything what use whole computer from the nineties on this 20 bucks FPGA. And you can.

Chris Gammell: Right. Like you said, I mean, like all the things that we're doing with that in the, you know, digital analog realm as well. So now you've basically. Yeah. All compute. Helmuth in your hand. Right.

Bitluni: So this is, this is like the one thing that I see and there, there are almost no limits right now or the limits, or let's say you have to spend way more time to reach the limits or exhaust the capabilities of one micro microcontroller right now. But that was one thing that I tried to, to do with the ESPs back then. We talked about the video generation like CRT control and basically the signals, color signals and coding. That was, that was one of the limits of the ESP 32, the first version. It was able to output parallel data that were like a R2R resistor, ladder, DAC, and it had like a two built-in DACs that they removed now. So there on the S3, there is no, no DACs anymore, unfortunately. Yeah. But that was good enough to generate basically a video signal with, with color. And that were limits because you have like to have the right clock and so on. There is a PLL clock on their APLL clock. They also removed that one. You could, you could output, you could use APLL clock to output the video signal at the right rate to generate the color, color video. But yeah, it's also possible on the older ones like Charles Lohr showed, I think on ASP 8266, like NTSC generation and also like transmitting from the wire.

Chris Gammell: Right. Yeah. I think Charles is in his own, you know, we're talking about like the excess of technology and Charles was like, well, there's transistors here so I can make it do something. Yeah. Yeah. Yeah.

Bitluni: So he is also one of the geniuses or he is one genius that, that also always explores the limits of certain. I agree. Yeah.

Chris Gammell: Yeah. I would put you both in the same category. Yes.

Bitluni: Yeah. We, we met also on a Hackaday Supercon twice now. I think. Yeah. Twice. So we are like-minded, but he's way ahead of me. And we all, we all got to have heroes, you know, like I look up.

Chris Gammell: Yeah.

Bitluni: Yeah. Actually, before I started my YouTube channel, I think I was subscribed to a few of other channels. Like Dave Jones was the, was the common one to, to watch, but also Charles's videos were my reference, how to make PCBs at home.

Bitluni: We had like videos early on how to add PCBs. I, I used the little different method. Sodium persulfate was mine. He was using like some chlorine.

Chris Gammell: Fluperchlorator. Yeah. And probably the. Yeah. Yeah. Which, which is. You know, EU versus US. In the US we're just like pouring it down the drain. We're like. Yeah.

Bitluni: Yeah, but, but also getting like the sodium persulfate here isn't, isn't like straightforward. So you have to. Yeah. You don't have to register, but I think you will be. You will be tracked or something. If you order that you have to specify the purpose that you're ordering for, and then you probably will be tracked or on the screen of the federal investing because you can basically build a bomb or something.

Chris Gammell: Yeah. Yeah. Yeah. Enterprising individuals. Right. Yeah.

Bitluni: PCB edging was, was once a cool, cool thing, but now really not, not necessary.

Chris Gammell: I was going to say like a necessary thing is a better way to say it. Right.

Bitluni: I mean, like, yeah. Yeah. It's now you can get bars like within two days or something delivered for a few bucks. Why would bother buying like stuff for hundreds of, of dollars or euros to, to edge it at home. Or yeah, I made, yeah. I made, yeah. Well, the most popular videos, I think milling, milling PCBs on, on like 150 bucks. That's right. Yeah. The cheap milling machine. Yeah. About that. Yeah. That, that, that is one of the videos that I made. That was the most viewed.

Chris Gammell: Most viewed. Yeah. But people should watch that video just for the first 10 seconds where Looney puts in a meme, the cat, the cat meme.

Bitluni: Oh, yes, yes, yes, yes. True. Ah, I should buy a mill. Yeah. Yeah.

Chris Gammell: That was the one. Yeah.

Bitluni: I almost forgot. Oh, that's so long ago. Oh, damn.

Chris Gammell: Deep, deep memes. Yeah.

Bitluni: Yeah.

Bitluni: Yeah, basically. And that is also, I think a good transition because when I started doing this 11 years ago, getting PCBs from China, it was possible. I think it wasn't like as straightforward or like from Germany as well. It wasn't as straightforward as, as nowadays or as common. So that went obsolete and now it's quite common, at least for me to get some PCB. You can even do PCB art and we have a lot of capabilities or assembly, even like super cheap assembly. Yeah. Yeah. That's, that's amazing. Wow. Amazing. And you can explore, you can, you have the possibility to fail. Yeah. Because without being like financially catastrophe for you. Yeah. Exactly. If you can imagine, I mean, you wanted to do a project and then you submit it. Like an assembly, you wait like months. That was once, once in a time you wait for months, you pay hundreds of bucks, maybe thousands. And then you discover like after you submitted and you're full in production, you discover errors and then devastating stuff.

Chris Gammell: Listeners here will understand that he is actually explaining, explaining the silicon process currently. Yeah. This is, this is called foreshadowing.

Bitluni: Yeah. And that's, that's, that's, that's basically what blew my mind with Tiny Tape Out. Yeah. That we went from the process of developing PCBs or making PCBs ourselves. I mean, Jerry, Jerry Ellsworth also made silicon back then, herself in her kitchen with

Chris Gammell: Yeah. Home etching, right? Home etching.

Bitluni: So there are a few, few people out there who make, make silicon at home, which is like crazy. What, what, what stuff even make a transistor alone is so much. Yeah. And they're like, Oh yeah.

Chris Gammell: I mixed together like a, you know, Drano. And I was just thinking about Jerry and Sam. And I think Ben did some of this too. All of them just. Yeah. They're like, Oh, I just bought this stuff off the shelf and you just reuse this thing. this thing and you know this is in your kitchen cabinet but uh it can be used for this other purpose it's like what first off what am i using on my countertops and second off i gotta lock the cabinets i got kids now you know it's some dangerous stuff yeah yeah yeah ben ben is also

Bitluni: amazing every time i see one of his videos that i just put everything aside and sit down and enjoy

Chris Gammell: what am i going to learn about chemistry i feel like that's that's like a whole other realm that we're talking about ben krasnow for people that i i'm always get criticized i'm like oh jerry and ben and sam so we've been talking about jerry ellsworth sam zaluf ben krasnow but ben has been doing chemical stuff like chemistry basically is the name he renamed his channel yeah that's right thank you yeah that's good yeah and that's another area where it just feels like it's very out of reach but ben shows that it's like no you you know you do measurements you read papers you have some simple equipment and yeah it's accessible and it's like wow that is that is a leap that i am not personally willing to make but i but it is possible and it's going to enable more people to try stuff and have you ever been dabbled in that realm i mean i guess you said you home etching type

Bitluni: stuff oh no not really i think the stuff that you need to to have like furnace and whatnot and to get the silicon and stuff i don't know i think my adhd wouldn't allow it to finish stuff that's a bake for how long i mean i can i can imagine myself getting all the stuff the equipment and so on but then it's sitting around it's also in some corner next to the other projects that i never finished background declaration yeah yeah so that was basically the segue that the pcbs to the asex stuff which which we see now after 10 years like we made those leaps in manufacturing actually to make it accessible to people which i wouldn't be able to even imagine that you can make your own silicon at some point and tiny tape out allowed like with like 450 bucks it's now i think the run it was i think it was cheaper in the beginning because it was really funded by efabless that's right yeah matt said he they up they up the price

Chris Gammell: because it was not calculated at the beginning but now it's more realistic with a sustainable business

Bitluni: which is like okay sure and we would like yeah that's cool i mean if you can do like what is tiny tip thing uh the that was now tiny tape out seven i think uh yeah eight is going september 6th is closing yeah it's it's open now and if you can make it sustainable on the eighth run which is it's amazing it's cool because it takes so long that that's one disadvantage right now that you have to be patient right now if they can finish it in 10 months it's all right yeah but still like for price that it's priced in the realm of to being a hobby without being a serious business right now you can you can use it for for business for like developing r&d and and stuff like that to have like some iteration maybe before you try yourself and fund like the whole run for 10k per chip the whole chip is i think the the price yeah as a show as a show like piece to to for fundings and so on

Chris Gammell: well we brought your work up on the show a couple times now most recently the best chip in the world let's talk about that that's where that was your yeah since you're talking about the importance of raising funding and you know doing important r&d what did you what did you build what half you run

Bitluni: oh damn yeah so actually that was the meme project that was the last one for that tiny tape out that was tiny tape out too and it was matt asked me i met him at the hackaday supercon and and he asked me later oh you can submit something if you like to tiny tape out too it would be nice and it was like three days left or something until that line and i was at the time i was exploring fbga so i was a little bit familiar with very lock at least for to that extent that would fit on tiny tape out and i sat down and made like one project that would show the capabilities that are possible in that run because you are limited to like the io clock is only like in kilohertz range and i wanted to make project that shows an actual practical use of that and then when once i was finished with that one that was like my last video i i thought oh well i can still submit like a meme project on top of that and then i just encoded and animated uh give uh the the rig row to to fit on there and put it as a secret file i had even a typo i think in datasheet document or something anyways i did i wouldn't think that that is actually the project that would go more or less viral but i i needed something to put out on on youtube quite quite quickly after it returned and i thought oh well i can do it right now with the meme project and yeah of course i have to go clickbaity that i i'm it's youtube meta but i can talk about this a little bit so for me at that time it was the best chip in the world especially because you have so many projects on there and also we have finally like uh rig roll etched into until eternity of time into silicon there was no way back yeah you can't erase the solution is crappy it's 32 by 32 and i think four frames or something so i had like to shrink it down and down to 450 bytes and even the gif header is already so big yeah it has like the fun thing is animated gifs usually you would have like gifs that are compressed and for them to be animated that was like a netscape extension so for the netscape browser so you have actually the word netscape in the file encoded so if you like wait really wait still yes yes yes that is the marker yeah that's the fun fact yeah that's like in the header there is like one marker that says netscape whatever that indicates that's an animated gif otherwise it wouldn't like animate it would show just the first frame

Chris Gammell: is this how was it mark andreessen horowitz or andreessen one of them did netscape stuff i forget

Bitluni: that's how they made their money i don't know i don't know who did it but i was trying like when i i wasn't very familiar with with gif i know the compression and so on but i was trying like to shrink down the file back then and there were extra headers with description and so on and i removed them manually because i there is not much software that you can use you can do use online generation and so on but i used gif construction set some older folks might might remember that was one way of generating animated gifs back in the time and yeah i tried to remove the headers and then i it stopped working when i removed like this netscape marker and then i researched a little bit and so okay yeah

Chris Gammell: that's the indicator for an animated gif to unlock the capability it's like code muncing you know like muncing is like when you like knock stuff off a board muncing that's the word yeah yeah yeah yeah it's like the code version of that yeah the last decoupling cap that's right and you had to put it back on yep exactly the netscape cap yeah yeah i i i wouldn't i wouldn't i probably the people that

Bitluni: were developing netscape couldn't imagine what what they will unleash like the dream world

Chris Gammell: oh man so and that's interesting too so you were you were in the fpga space you so you turn this around it's super fast obviously you talk about it in the video and people should watch that video we've talked about a couple times but you know to actually get that down now you have 450 bytes those are then those are individual registers that are sitting in the silicon can you just kind of like walk us through what those actually are in terms of the the data how it's how it's in the silicon

Bitluni: so just storing data that that's a i would say it's a rom okay basic rom and what it takes is just a clock signal and each clock it would just expose the next byte of the the rom file on the parallel output which is 8-bit and what it has internally is just a counter that will loop after 512 bytes so it's just padding zeros at the end and bitloony i had to put a string there as well

Chris Gammell: you got it yeah after after the end of the work right yeah and it's the new netscape you know you

Bitluni: have to add that to the silicon no i mean i it's my it was my first first like silicon project so i had i couldn't now you can you can put like art on the actual silicon for so you can see it on the microscope you you can edit the files that are generated as output it's easy i think in the and the runs that are currently because you can put analog stuff on there as well now right right

Chris Gammell: well and this is interesting too because so like matt told us about like all the changes that came i think was from three to four that's where like now the analog this stuff is enabled yeah what i'm what i'm really wondering is have you been putting new stuff like you have stuff in the pipeline that we haven't seen yet so like that means that that because all the stuff we're talking about like the the rickroll is in two and then three and now we haven't even i think five is shipping soon so like what are we gonna see bitloony come on like what are we i didn't i didn't submit i will i thought i

Bitluni: would submit something for seven but i didn't make it in time basically i'm too ambitious now i see for that so one thing that i was i was trying to develop i had like one stream where where we wrote an assembler and interpreter that runs in in the browser that was also one video that i made it was like crazy because you can do this like in four hours i i don't know if i use jgpt also for that but developing own assembly language with interpreting and so the purpose for of that would be i'm using several platforms like microcontroller platforms and i want every time i try to develop something to show how the project runs like a handheld game console or whatever and i have to rewrite a snakes game and whatever every single time and i thought okay let the people write games for that and i will make it to run it in the browser and then it will run the same way on the interpreter on the microcontroller so that was that was the thing it generates bytecode at the end that i can put on the microcontroller and then and will run i showed that i think in the video with one of my handhelds and i thought oh okay let's put that on silicon for this loony-ism it's it wasn't me i i it wasn't me to call it like that i just uh took the name afterwards and

Chris Gammell: does that mean that if you had a clock in there eventually it would be the uh it would be each each

Bitluni: clock cycle would be the loony tick oh yeah oh that was good yeah something like that but i don't i'm not sure if it like a whole process like that there are processors and fbgas on on tiny tape out runs yeah i wanted to put a whole processor on there and since matt already showed that decks are easily possible there are templates for that i would imagine to have like a project i at least i started on stream to have something like a cpu with built-in gpu for vectracks like things so you can you have like a vector scope output uh you can run it maybe on the oscilloscope or whatever with a cpu that runs that loony asm assembler that you can write in in the browser and simulate there or whatever so that was the idea but that's way too ambitious for like the time that i took for that i think i

Chris Gammell: thought oh i will make it in one week and then nothing you know you're asleep right like uh 24 times 7

Bitluni: we struggled with i i say we because my community while streaming there's only like 20 people or something or watching or 30 people watching on twitch and youtube while streaming before we say

Chris Gammell: streaming again too how do people find this because we'll say it at the end but people were like oh i want to be there for this right how do they get to this just the bit loony bit loony live is the

Bitluni: channel name okay okay yeah that's an extra channel on twitch it's bit loony so now i'm parallel streaming on both on wednesdays at eight o'clock in the evening uh central european time you have to

Chris Gammell: take your time zone yeah i think it'd be basically 2 p.m 2 p.m u.s eastern and then probably 11 a little

Bitluni: bit inconvenient but at some point people sign off in europe because they're tired and have to go to work now i think it's wednesdays because it wouldn't collide with other streams yeah anyways the people are quite helpful with catching my errors and and so on and what's tiny tape out 7 it had more gpios so you have input output pins that you can use for like connecting sram because ram is very expensive you can only store a few bytes in tiny tape out and then in d flip flops and then your space is gone so just writing the memory interface was so challenging already to synchronize that correctly because you have to break up the address addresses like 16-bit addressing it's it's like the memory management or the the accessing over like 8-bit interface is a little bit more difficult and then externally you would need like some multiplexers or whatever to enable the addressing chips so you need some buffer for the address uh 16-bit address or 10-bit address or whatever and then use the io lines to to like manage the input outputs of it's already it's already a task on its own too if you're not familiar with that stuff like i'm i'm not really familiar but to handle that correctly so you are using so you're

Chris Gammell: using external flash external sram and then you're managing your own bus yes like you're basically making

Bitluni: a bus master kind of thing yeah basically basically that that's what's necessary right now i i think if at some point if there are even more like lines available like to have more than eight address lines because that's not really enough to have uh like a lot of ram or or even flash i mean you could activate flash or switch it like that anyways don't want to dive too deep in that topic but it's interesting it's already challenging yeah i think it's it's actually really good

Chris Gammell: indicator i always get frustrated when i know this is a very different realm but when subsequent part families like a silicon company be like oh we have this new microcontroller it's so great and then the flash and ram stays the same and you know like on board right and it's yeah in a similar way like that is often like geographically on a piece of silicon like it's just so expensive you know did to upcycle through to to move up to you know from 512 to a megabyte of flash or you know 256 to 512 of sram you know like it's just like a lot it takes a lot of space and that's expensive oh yeah that's the

Bitluni: reason why it's expensive is because it's all ips basically of the foundries i believe it's the foundries i don't know the the details about that but what what happens right now is if all the the tiny table stuff is open source so you can reuse it commercially for for all your projects or all the details how it physically is shaped on on the silicon everything of that is right now open source and for the process that is 130 nanometers sky water right now there are ways of storing like implementing sram or drem or whatever that's way more compact like way more efficient which basically exploits i don't know the physics there it works but you wouldn't yeah you have to like try an arrow for a long time to to develop that but that stuff is not open source and that's the issue right now if you want to have like sram then you have to use like yeah you have to use the transistors and and you have to hold you

Chris Gammell: have to hold the state in like yeah like a footplot yeah exactly quite huge right yeah i mean like yeah if you look at like the dram structures they're often big like capacitive cups right and there's the column and row controllers but yeah like that would be that's like a yeah that would be a custom mess that i'm sure that that's like an ip block that you're dropping that you're talking about same

Bitluni: as flat it's like that's basically weird you know yeah and that's that's one of the issues that also it's all intellectual property that you can't use for for project like that but got it you have less i had an in the tiny tip out seven run or i think there was the possibility to use such ip blocks because if i had like experimental run for that yeah yeah but it's it's like an exception right now i i don't know where the future goes but it might happen that at some point there will be like a shared sram block or whatever for inside of tiny table so that's why i think it's quite difficult to put more ram on there more and more ram because you have like a block that you license from the foundry and then

Chris Gammell: yeah yeah you go with that never change it's amazing like flash ram all these things that are in there right i mean it's all the same ingredients you know it's like that scene in silicon valley where the main character is like uh take scrambled eggs right what are the scrambled eggs made out of electrons right and it's just like yeah it's all made from the same stuff but it's like it's a completely different thing right i mean it's not like you know like the the way it's assembled actually really really matters in this case and and especially the testing and the you know just knowing that it works is pretty damn critical for making something that's yeah yeah i can i can imagine i mean many many of

Bitluni: those companies that have run those experiments for such a long time which were very expensive they want just to protect their investment at some point but it would be nice to have like really efficient ram and flash that's that's open source at some point it would be yeah yeah i wonder how many how many

Chris Gammell: iterations it would take to get there right and it's like but maybe if you had a couple cells maybe it's maybe i'm overthinking it in terms of the things that would need to have maybe like if you had the structure you could have basically design of experiment where you have a thousand different versions of the same structure with small tweaks across you know across an array and then you're like oh actually it's you know it's row 25 column four that's the best one and we'll just use that from now on you know that sort of thing

Bitluni: so actually something like that is what we see on tiny tape out with other things that that has happened or like with testing the internal clock speeds so there were many experiments before the analog stuff was out there there were already tests on the tiny tables that were tested oh how how much can we run internally what is the highest clock and there were many experiments that that were just exploring those details so it's happening and it's happening in parallel because you have like 200 500 whatever experiments on there or on one chip so that might also happen with the analog stuff what once it's arrived like where when we have like documentation and the tools that make it really convenient to experiment stuff there are some some people that are really into it uh testing that and and i think something will will come out of that yeah it might take years because the iterations are slow exactly yeah yeah i

Chris Gammell: mean like that's and thinking there are test structures in silicon wafers even internally right so you know the engineers will run test structures to off the main area to run different experiments or whatever it's just now we're using the tiles that are on tiny tape out which are you know lower cost effectively so in variable yeah like you're saying yeah it's it's better to have complete control of the fab of course if if you can get it it's nice it's nice work if you can get it right yeah i am curious about your so you you mentioned a couple times the the fpgas and specifically the go-in fpgas i've seen those before but i've never used one it's been a long time since there's any fpga what are you doing on fpgas these days and where do you suggest that people start around the go-in one specifically

Bitluni: so when i started that i was reached out to by gabriel who founded like lushy labs that's like an website that basically has ex not experiments but tutorials for an fpga development and and that was based off the go-in fpgas the 9k ones now you can get 25k which are really like and as big as an arduino nano or something yeah so basically a stick and what's interesting on there is that you have like an hdmi connector on that and it's powerful enough to like the 9k is already powerful enough to generate or fast enough to generate like a 720p signal which is basically perfect for hdmi and you can do smaller smaller things with that unfortunately that uses ip blocks from go-in you can use like the ide that's that's free to use and use the ip blocks from going for that as well as like the built-in memory they have like drem stuff on there or ps ram on the 9k's and i think the 20k has like the also external drem or something and then you have uses which is basically the open source variant for generating the bit streams that doesn't support that for now i think but the support grows as we speak i don't know if it's i didn't look into the latest updates but basically what you need for hdmi is differential signal generation and i think that's like an interface that's very specific for the fbgas and it lags there but what what happens like on lushy labs you have like a visual studio code plug-in that you can basically install the free open source tools uses and stuff that's like one package one executable that you can install in your pc and then you can basically start writing in visual studio code and you have like an emulation test bench that's it's called a test bench basically where you can just flip the bits of the input signal like in a sequence with clocked or whatever and then it will simulate the outcome even before you you upload it and upload the bit stream to the fbga so that's quite convenient and i i think lushy labs has even uh tutorials how to develop a cpu on the fbga so it starts like with a blinking led the typical stuff that we had once with microcontrollers it's basically now available for the fbga stuff especially for the 9k in this case and it's like from china you get it for 20 bucks which is really like yeah you can't even

Chris Gammell: get a raspberry pi one i think 20 right exactly it's like the tang the tang nano 9k and you said it has an fbga thing on what is the other connector there's it looks like there's a ribbon connector on

Bitluni: there is that just for yes yes you can connect basically a screen uh tft screen to that okay i think the the first one was 10 tang nano 1k then we had like 4k which had also a camera interface oh cool i bought all of them to just test what what they're capable of and there there are so many variants of the of the going that had have like core like inside 9k doesn't have like a risk core additionally to that but you can make like a risk 5 core put put on soft core on there so there are many variants but the tang nano 9k has like this parallel interface where you can just connect and parallel like tft screen screen that's what i was using to like on stream i i made like an oscilloscope or started developing an oscilloscope that worked like with an external adc that was sampling at 50 megahertz or some mega samples world channel and yeah that the capabilities of the 9k are barely fast enough like to get input from that adc in 50 or 100 mega samples and then display it on the tft so you can that's great too because

Chris Gammell: you're like on the edge of where the capabilities are and those kind of those constraints are also kind of like they're it's like forced learning at that point right it's yeah yeah you gotta get around

Bitluni: to some the bottleneck at the end is the uh psram on on the uh on actually on the 9k i think the 20k is better because it has it says like like dram which is faster and the psram is somewhat serial internally it's it's on the same chip i think but it's it's very complicated and then yeah figuring it out yourself is uh difficult but the ongoing ide moon river studio is think the name oh really that has like ip blocks yeah moon river like m-o-u-n yeah oh that's the same as the yeah i think dave dave was talking no that's what wch uses as well for the sorry sorry i mixed up sorry okay it could have been though right i mean like one river it was another i think okay so it's a different like maybe ide that looks like just the generic oh wait no it's just go win okay all right no for it wait go win fpga designer yeah are you using their tools yeah are you using the other stuff i i had to use for like the things that are basically on the on the edge of the capabilities of the device i have to use their stuff because the ip blocks are like developed so they have like the if you if you need like a frame buffer so basically the memory to store a whole frame to display it on on hdmi or on tft which you need to to render like anything if you you can you can run it like in parallel on the fly you can basically

Chris Gammell: right see the scan lines and stuff like that yeah which has its own like aesthetic right you know yeah

Bitluni: it's it's it's cool yeah i i think i managed to do like a mandel broad on the fly with few iterations or something okay yeah but usually you would need like to something to run in the background and put put put everything into this frame buffer and then the print buffers scanned with like the hdmi or whatever and developing this frame buffers already quite challenging because you have different clock domains you have like the clock domain of hdmi which is way faster and yeah this is this is where it gets really complicated if you do fpga stuff so the simple things is i think tiny tape out is simpler than trying to exploit every capability or uh or explore every capability of fpga yes because you usually would work like with one clock maybe you have an experiment internal clock and then you have two d flip-flops where you synchronize stuff but it's a science on its own which i am not really familiar

Chris Gammell: with and it's not worth it as like in terms of just pure dopamine over time right like you're just trying to get people to build their first thing right and if you're like well if you start to you know i remember the first time i started doing fpga stuff right i had seen fpgas as this amazing answer and i was trying to like replace the 6502 and like a legacy product and i was gonna be like oh i'm gonna just re-implement it in the it was a 6509 i don't remember what it was it's 6509 just re-implement it just re-implement it yeah no big deal i found oh look at this i found a core on open cores.org it's of course it's gonna it's like no and then it was like a 50 megahertz domain and a 33 megahertz domain and i had to do all the clock and stuff myself and it's just like what are we doing here like yeah it was like you know like the dunning kruger like you mentioned at the beginning i was like new engineer i'm like how hard could it possibly be and i i effed around and found out you know what i mean yeah there's where physics kick in yeah exactly yeah and just some of the realities of timeline project time that's like could i have done it maybe yeah i never found out but i sure as hell did not

Bitluni: do it at some point you have to have a high frustration tolerance yeah yeah yeah yeah yeah and there well there's like uh you have ip blocks for for that usually in the going ide there there there is even like an ip block for frame buffers so you can just use it sure and don't worry about that and develop your project and propel that and then when when in terms of like those ip blocks on

Chris Gammell: a go-in type like using the going the fpj tool like you're talking about that is how that has to stay captive it's like a it's like an encrypted bit stream type of thing like yeah you pull your logic into that instead of being able to pull the block out the open source tools so so like uh it is it

Bitluni: is like you have a binary file okay so that's like in the project file you see there is a binary file with the bit stream whatever and i don't know what the licensing is about that so if you want to use it for commercial stuff you probably have to license it somehow but that it's a that's never a question for

Chris Gammell: me because i don't want to so for the frame buffer that's a good example if you wanted to choose to use the frame buffer you're like i'm just gonna you know use this as a tool i'm gonna use it whatever yeah does that mean then that that you have to operate with inside that tool or can you actually use the binary and pull it out to the open source tools and just render it that's a good question i don't know actually i can't i can't answer that but from a practical perspective when you do need that ip you just say i'm just gonna use their tool that kind of thing yes yes it will basically

Bitluni: generate everything i think you can even name that or whatever you get like a very log interface like a header and then the bitstream in one file and you can just access access the stuff good old blobs yeah yeah yeah yeah unfortunate i was i was looking at a bitstream ah maybe you can decode it somehow at some point it needs to be converted to uh to the actual bitstream or whatever that ends up on on the flash so maybe you can export it from the flash maybe yeah you know you know the thoughts

Chris Gammell: with a million monkeys and a million a million oscilloscopes we could we could decode this we could decode this and uh and reverse engineer it but why not why not just build your own at that point

Bitluni: you know i i think i think that's that might be happening like an open source tool so people just extract the bits bitstream from from the flash or whatever or from the file and then see oh okay that has to be how that works or whatever i i'm not sure but okay yeah i mean it is it is still magical to

Chris Gammell: me that it's literally just like bits floating around the edge of an fpga and it's like that bit stream determines the magic inside it's just it's just magic sand you know magic sand it's at this purest

Bitluni: yeah it was it's always scary i i don't know when i when i started with the electronics and fpga it was basically anyone who enters the realm never returns it's like a rabbit rabbit hole yeah it's like it's truly universe yeah yeah yeah you can you just open up like the dimensional portal and you disappear any fpga decoder just disappears in that realm and never comes back that's right yeah sorry

Chris Gammell: man you're getting sucked into the vortex speaking of this is usually part of the show where we talk about future stuff that you're excited about things you're working on things you're obviously people can go to your stream and look at it but what do you you know where do you feel the draw of technology

Bitluni: these days where where you where's it pulling you oh wow that's a good question oh yes i i would i would love to move more to back to art like oh really oh yeah yeah i like i like that like you have a

Chris Gammell: do you have a background so you said you have a software background you have an art yeah at some point in my

Bitluni: life when i was i think 10th grade in school i i was choosing my path because i was good in math and i was good in arts and and then i chose to go to um yeah i was coding in my free time quite early on and then i had to to choose between two schools like one was about arts and one was about programming and i chose the programming path but arts never left me so at some point like in i think 2000 or whatever in work i met a friend with whom i founded like a demo scene group and we coded a few intros and demo demo scene intros like 4k intros that that's also exploring the limits can you explain demo scene real quick uh demo scene is computer art basically the demo scene was basically it's demonstrating the capabilities of a computer in visual and acoustic way in some so some sort that developed from from the cracker scene back in a time where you had like software that was cracked by some hacker and they liked to put like a small file with the crack that basically showed off like graphics and music and the scroll uh copper bars like moving stuff on the screen and that people liked like the coders like to show off the the capabilities of coding and exploiting also hardware like the actual stuff like copper bars are basically exploiting hardware registers of uh to drive the crt and and stuff like that it's it it was fancy back then and then from that basically the demo scene developed where people like to show like animations that were calculated on the fly or pre-calculated later when there was more memory available and then demo scene parties were basically like competitions where you have like different categories c64 amiga pc and then uh different sizes so you have like pc demos are basically unlimited i don't know what maybe there is a file size limit but you can put like mp3s which they have always like super cool music and state-of-the-art graphics in some way or fashion they they like try to like excite you that's the background and there are some categories that are smaller like 60 64 k was like 64 kilobytes intro which usually used like uh generated graphics so it was pre-calculated or calculated on pc this sometimes like textures and so on if you do 3d stuff it's pre-calculated because you have enough ram on c64 it's not so much ram yeah then there are there is like the 4k category so four kilobyte intro you can imagine four kilobytes of data fits like on one like in ascii it fits on one page like a a4 page or letter it fits and and that's all the code for generating the music and animations that runs for maybe four or five up to

Chris Gammell: 10 minutes or whatever and uh yeah it's like a pure optimization exercise basically right yeah

Bitluni: what's the most you can do with this super constrained space yeah and that uh i was i was into that i got in touch with like i saw some demos in the 90s i don't know where i got them from i think they were on like a cd-rom or something that i that i bought they were just randomly put on there that had nothing to do with whatever content was on the on the cd and then i bought a book pc underground which basically explored like the techniques of writing 3d engines and stuff like that rendering graphics or playing mod music which mod files are basically you have the instruments put there and the midi notes or whatever notes and the effects and that renders the music uh that's from i think from amiga back in the time but c64 works the same way with not samples samples but with like the one sample would take up 4k with the without much uh without much thinking about it right yeah so basically usually a 4k intro has like synthesizer and some way some form of punchy graphics so like ray casters so a typical way nowadays that developed with the faster cpus is where basically you have um something like in blender but but it's rendered uh in real time and usually some some artsy artsy graphics there and uh yeah i was always interested in in those things especially trying to put like in the really small size like of 4k because my attention span isn't that long to code like one year be some some demos in groups like code for one year on on a demo to to fill all that content and i my attention span wouldn't allow that so i concentrated on the 4k competitions and basically tried to put the best tech in there that's like tech demo code so so many terms here from from the demo scene yeah and i basically participated with my friend who was more proficient in music uh than me like in the beginnings of 2000 around that time yeah so that was the art part uh that i was doing and when i heard yeah matt and matt matt said that on tiny

Chris Gammell: tape out eight there will also be a demo scene in there somehow like is that right oh really i don't

Bitluni: know i i i must admit i didn't follow all the projects i mean there's so many contributions but but they're like of course there are some old school folks there they're putting like the 80 chips on there yeah yeah yeah and stuff like that so and also uh someone put like a ray ray caster gpu on one of the tiny tape outs which was really cool so you can render something like doom or or

Chris Gammell: like wolfenstein or yeah yeah yeah oh man yeah yeah yeah that's that's really cool and i i feel like i would have to go to one of these events to actually understand you know to actually feel what it's like how does one find those sort of things is it like uh underground like you need an invite gotta know

Bitluni: somebody no no no no that's that's quite open but it's more popular in europe usually there are a few events and then usually you can find them they know they must see parties they they're all around the world and i don't know if there are some like in australia or uh there are a few i think in in us but i'm not really familiar i just lost track at some point when i started like doing the youtube thing and other interests i stopped to competing there i make my own demo scene party it's on a stream every wednesday yeah yeah yeah basically i mean there's the stuff that i'm there there i could contribute like with some stuff that i made here because there's like the white demo competitions where people sometimes do hardware stuff so basically merging like programming capabilities or or this media multimedia stuff with some hardware stuff uh is also a thing it's not very common but usually people reward that so there is some intersection with let's say the hardware or the electronics maker community with demo scene and interest like that and probably star trackers yeah probably yeah we're

Chris Gammell: all we're all in the same buckets they're all adjacent buckets at least that that's true yeah where can uh where can people find your stuff online and make sure they're they're following you so they see your next your next demo scene your next chipset your next project how should people find uh so demo scene

Bitluni: i'm not sure on which party i would go i try to stay in within germany but yeah on youtube basically on youtube uh at bit looney is basically if you look for bitlin you always find me b-i-t-l-u-n-i right yeah i don't i don't really maintain my website so if anything then there will be a link underneath a video description that points on the github that's also like basically i picked the name from like my demo scene handle was lunatic or looney and then i picked bit looney because it's unique otherwise people wouldn't find me and that's why you can find everything like github and usually the youtube videos they conclude every long project that i'm doing or show a part of it and then point to the right directions and code wise and so on awesome

Chris Gammell: well thanks so much for being here today love hearing about this stuff i can't wait to see all the other stuff you build i mean you know just keep keep making fun stuff and it's like it's a joy to watch you work and have fun with it that's it that's great thank you