Embedded - 210: The Glass Hour

Episode Date: August 10, 2017

Alan Yates (@vk2zay) told us about his entries to the 2017 Flashing Light Prize. Alan's entries involved making a light bulb and dripping charge. Alan works at Valve. He told us about making virtual reality hardware in Embedded episode 162: I Am a Boomerang Enthusiast. Hackaday SuperCon is Nov 11-12, 2017 in Pasadena.

Welcome to Embedded. I'm Elysia White. My co-host is Christopher White. I remember last summer when we talked about how Valve's VR system worked and got into boomerangs and reverse engineering. That was a good show. So I am happy to say that we have Alan Yates back to speak with us again about none of those things.

Hi, Alan. Thanks for joining us. Oh, it's great to be back. Thanks for inviting me. For people who didn't catch that show, which was 162, can you tell us about your history, your background, you know, however you introduce yourself at a technical dinner party? Technical dinner party? Technical dinner party? I guess I call myself a hardware engineer nowadays. But yeah, I've been working here at Valve for almost five, is it six years now? It's been a while, but it's been a lot of fun. Here I work on virtual reality technology on the tracking system known as Lighthouse.

Talked about that a lot last time. Most of my career has actually been in software and I still primarily sort of do firmware here, but I do dabble in hardware on the side and I enjoy hacking on analog and old tech and all kinds of things like that. All right. Cool. So now lightning round where we ask you short questions, want short answers and probably won't succeed yeah i think i might have expressed my everlasting desire for tweaky or something last time that that's right all right no questions about buck rogers this time okay so buck rogers uh favorite book or movie which you encountered for the first time in the last year?

Okay. The whole Wonder Woman thing, I guess, I used to watch it as a kid, but I've got back into it from the movie. The original show? No, not so much the original show. I mean, I love the original show, but it kind of hasn't aged well. Yeah, we're watching the original show. I mean, I love the original show, but it kind of hasn't aged well. Yeah, we're watching the original Batman series, but it's aged really well. No, no, it has not.

Preferred voltage. Preferred voltage. Ooh. 240 AC. Favorite cartoon character. Oh, wow. This isn't very lightning, is it? No, it's okay.

We'll speed it up later. Okay. Either that or the listen on 4X. You can pass. You can pass. You can pass. I'm going to pass on that one because I have too many. If you could be a superhero, would you be in the marvel universe dc manga or other are you getting

these questions like no question okay here's here's one you probably will feel more comfortable answering uh technical tip you think everyone should know don't be afraid to touch your circuit as long as it isn't high voltage. That's a good one. Yeah. I remember being a kid and playing with electronics and thinking I was going

to destroy everything all the time. You can find all kinds of loose things or things that are oscillating or temperature dependence. It's great. Hot things.

As long as it's not more than 48 volts, go for it. This is the current that kills. Beer or wine? You you know even though beer tastes terrible beer okay good because my next question was about beer on a scale of one being no hops and 10 being entirely hops how hoppy should an ipa be ipa should have no hops in them i mean

i thought those were the ones they were like double hopping. I know, but I hate hops. Okay, good. Good, good. And how hoppy should a rabbit be? Oh, very hoppy. They should jump higher than me. All right.

We should maybe get this show started. That was pretty bizarre. I thought this was the show. You have been tweeting some about this flashing light contest that I've seen Ben Krasnow and Micah Scott tweet about. And it's a contest where you try to have a flashing light. Yes. I found out about it from someone's entry on YouTube I stumbled across. I can't remember who's now.

But, yeah, the general thing is you have to flash an incandescent light bulb at somewhere between 0.5 and 2 hertz. Other than that, it's pretty freeform. You can do whatever you like, but it must be incandescent and it must flash somewhere within that range. So I actually came up with a bunch of different ideas to do this. So I ended up only making two of them and one of them at the last moment. But I've been playing around with vacuum technology

and glassblowing for a little while now, and I've gotten okay at it, like well enough that I could make a light bulb from scratch, and that was kind of my first entry. I wanted originally to use the Edison effect, which is the vacuum diode kind of effect, to make an oscillator, but I didn't manage to make a tube in time that had a third element in it. So I just ended up using a little relaxation relay oscillator for that one.

And then a couple of days after that, it was right at the end of the competition, I'm like, well, I've got all this stuff just sitting on my table. I mean, my bench is legendary and it's strata of discarded stuff that I haven't cleaned up. So I thought, oh, I'm going to just try and build stuff out of whatever's sitting on my table. So with the exception of a syringe that I used to drop some water, I made a thing that used dripping water to carry charge from a old-fashioned Leyden jar and into the front end of basically a lightning detector that I built that would then flash through some modern electronics through a MOSFET and just blink an LED and as well a sort of a little torch globe. And that was my second entry.

And, yeah, I don't know who started this competition or prize, whatever they're calling it but it's really cool like just all the different things that i've seen and how people have approached it it's it's like the triple five contest again it's just really interesting to to watch people um be creative with you know electronics i think i like that about it it's that the point of it is you know okay it okay, it's a flashing light that, you know, that's not the thing it's, but it opens up, uh, all this creativity because if you were solving a specific, you know, really, you know, world beating problem, you tend to gravitate toward the same solutions. Whereas this is, well, we know how to do this. So pick up the weirdest

way you can think of to do it. Indeed. Yeah. There's been some weird and wonderful stuff out there. I actually really like, you mentioned Ben Krasnow, who made that tiny, tiny little one. I'm still not entirely sure how he managed to get that battery to actually deliver enough current to flash it. Micah's was quite inventive as well with the induction swing. That was pretty cool. He had the whole seconds pendulum and some coils.

I like it. I saw another guy's who, did he make an entire ASIC in Inkscape? Yeah. Yeah, I can't remember exactly how. I don't know if he was doing a run of this, and he just used, like, doing the chip for something else. But, yeah, that's pretty hardcore and so there was no the light should flash for the longest amount of time you weren't trying to limit i mean it had to flash for five minutes but you weren't maximizing that

or maximizing brightness it was about solving this problem creatively. Yeah, creatively. So I guess in some ways my entries were the least creative possible. I mean, a relay oscillator is not exactly interesting, but I just had to make the light bulb because that's something I've been doing. There was a bunch of other ideas. Actually, I shouldn't tell people because apparently they're going to do it again next year, although it may not be incandescent light bulbs I hear. It may be something completely different.

Fire. One guy's idea was really cool. He had an electric drill turning two cups, and the cups were plugged into a power supply, and he had one of those little linear bulbs like you have in automotive, and he insulated half of the wheel, and that would turn it on and off as the wheel kind of rotated around. I thought that was pretty cool. The rules did say you can't just have a light on and flash it by having a cup go around it.

You actually had to turn on and off the incandescent bulb. Yeah. You couldn't have vanes. You couldn't have anything. It just, yeah, occluded the bulb. Why was that hard? Why is it an interesting problem? I mean, the creativity aside, why?

I guess that makes it more of an electrical rather than a mechanical or, you know, electromechanical project. It limits people from just making, like, a wind-up mechanism or something, although that would be pretty cool. But there were some interesting mechanical solutions to actually toggling basically a contactor on and off. I mean, there was one guy that built like a little mechanical hysteresis system with like a linear actuator, so his motor would drive a screw one way

and then it would flip with a spring and then it would flip back the other way. I thought that was really interesting. They were all really interesting. I mean, there's a long list. I don't think I've actually seen them all yet. I've been going through a couple each night and commenting on them on YouTube. But it was a great contest. I'm eager to see who won. But in many ways, I mean, the prize money, which I think was 200 pounds. Yeah, it's very minimal. Yeah, I would have actually been cool if it was just like a beer or like, you know,

something completely random, like a bag of peanuts. Yeah, I didn't really feel like people were doing it for the prize money. Well, since most of them cost way more than the prize money to implement. I mean, it seems like a Rube Goldberg contest without calling. I mean, the temptation is to make it just the most complicated, ridiculous thing ever. If I were to do it, I'd have a room full of stuff that had no bearing on the actual problem. Yeah, there was a couple like that. There was a couple that sent messages around the world on use the internet.

It's a bunch of really interesting things. Okay, so you said three elements and Edison and light. And I nodded to that, but I didn't know what you meant. And the Leyden jar. Well, the Leyden jar I kind of understood but that was the second one. I was going to ask about that separately. Okay. Okay. Yeah, so back in the early days of like the incandescent light bulb,

one of the problems was blackening the inside wall of the light bulb. They use carbon filaments and the particles would evaporate off the filament and coat the inside of. They used carbon filaments and the particles would evaporate off the filament and coat the inside of the bulb and darken it and eventually reduce its life. And even if the filament didn't fail, the thing just got dim because the bulb got slightly opaque. One of the solutions that Edison or one of his engineers, it's not completely clear, was doing was to put an additional

electrode inside the bulb and then polarize it, connect it to his DC power supply because he was the DC guy, to try and attract the particles that were sticking to the glass to the plate to keep the glass nice and clean. And during the investigations in doing this, he found that if you connected it to the positive pole of the battery, you got a current. If you connected it to the negative pole, you didn't. And that was the first rectifying device. That was the diode.

And that's called the Edison effect, or basically thermo-ionic emission. So a hot filament heats up, electrons boil off the surface. In a low-pressure environment, they have enough mean-free path that they can make it to something else like another electrode. And that's really the beginning of vacuum chip technology and everything that we're using today, sort of in that solid state now, is derived from our ability to manipulate electrons in that fashion.

So my idea was to put an additional electrode in there and basically use the thermionic emitted current to charge up a cap that would then, through some other gain device, essentially oscillate that way. So the filament had to emitted current to charge up a cap that would then through some other gain device you know essentially oscillate that way so the filament had to come up to temperature and actually start emitting electrons and producing a space charge and having some current and use that to oscillate it should work fine the only thing I'm not too sure about is the time constants and whether because I had some other interesting designs

that were thermal that I discussed with actually a few other people who were doing thermal-based, like the four candles guy. I think he was doing a Peltier thermoelectric generator-based design, and he had to use an oscillator that was faster than the thermal because the thermal time constant really means that you couldn't hit that 0.5 hertz limit, unfortunately. And I was kind of worried that I might have that problem but I don't know I'm definitely going to build it and see what happens at some point but I haven't sat down and you know actually built the internal structure of the thing and tried to

pump it out. I had another design as well that was based on a gas discharge tube. So it sort of charged up a cap and then dumped that current through a transformer into the light bulb. I think someone actually did something very similar in the competition, but they had a discharge in space and just break down a spark gap. And they were using a transformer to transform it down. And I think there was another person that was using a science lab adjustable gap discharge tube to do something very similar. So obviously a bunch of us came up, you know, convergent engineering had kind of the same idea. It seems like this is a good way to learn about analog electronics

as directed towards incandescent bulbs, but so many things are like that that it makes sense. Yeah, and people have talked about, you know, doing a 7400 logic or a 4000 series logic thing. You could basically build anything which is interesting and you could do some super creative stuff, but in many ways it would be extremely varied. By having, you know, one common target that's super simple,

I think that puts the creativity on the other end of things, like in the weird and wonderful, not just grabbing, you know, 1990s, 1980s technology and sticking it on a breadboard. I find that more interesting. But that said, I mean, if there was a 7400 series logic or a 741 only rather than a triple five contest, I'd be there. I'd totally want to do it. Back to the Edison story. Did he know he'd made a rectifier? Did he know how important that was going to be?

I don't know if he completely understood how important it was. I think I'm not a huge historian on Edison. I think he kind of like just considered it a curiosity at the time and didn't really understand how important it was until someone put a grid in there, basically, and discovered that they could control the current. And that was, you know, there was a bunch of patent disputes and about who actually came up with that. Several people, you know, there was a lot of people in the world

were all trying this because it was relatively accessible technology. Once you had decent vacuum pumps, there was plenty of glass blowers in the world and there was plenty of people who were interested in this. It was a rich man's hobby, but there was a lot of people that really, you know, were playing around with it. And no doubt someone somewhere else in the world, probably more than one person tried this. But history records, you know, that the Audion, kind of the first tube that was probably fairly gassy, simple kind of tube, was built by one guy and then he tried to sue the heck out of everyone. It didn't end well for him. What filament did you put in your bulb? I used molybdenum. I was going to use tungsten, and I have some tungsten wire.

But molybdenum, it melts at a much lower temperature, but it's cheap and available. It has a couple of other advantages as well, but it just happened that I had a bunch of it. They use it for EDM, electro-discharge machining. You can get fairly fine molybdenum wire. And that meant that I couldn't run it at a ridiculously high temperature, so I had to kind of cut it off at orange heat rather than like full-on white incandescence.

And I babied the tube quite a bit. Like my actual relay version of the project had a biasing rheostat and a bunch of things to make sure that it didn't thermally shock the filament too much. That said, I mean, it still works. I've been operating it now for quite a while and I haven't destroyed it yet, but I wanted to be super careful because, you know, it's about

an hour or two's work to carefully assemble that and then you've got to like anneal it in the oven and you've got to come back and make sure that it doesn't have any leaks. It's a protracted process, but mainly because I'm just not set up for it. Once you had some fixturing or something, you could actually turn them out pretty quickly. They're not a particularly difficult thing to make.

I was thinking about making a video about this because a lot of people have asked me how I did it, and I'll probably sit down and do it with virtually no tooling, like just with a blowtorch and just with hand tools. It's a bit more challenging to do it that way because mainly just holding stuff, because when the glass warms up, it gets soft and stuff kind of moves around. But it's totally accessible to anyone with, you know, just buying some stuff off eBay and Amazon and a blowtorch, really, and you could do it.

I was going to ask, were you lamp working with the blowtorch or were you doing what I think of as real glassblowing with the furnace and all of that? Yeah, no, I wish I had access to that. Actually, living in Seattle, I've always been interested in, you know, real hot shop glasswork. Not so much, I guess for me, scientific glassblowing is more interesting, so I'm more of a lamp worker. But I've been to the hot shop, you know, I've made the the traditional paperweight kind of you know things that you you kind of do when they they you pay your $50 and you go in there and you get a little time and you can take a dip from the you know the collection from the tank and go and make some stuff um and it's a lot of fun but yeah that kind of equipment is not something you're going to have at home unless you're pretty set up.

But around here, there are a lot of shops where you can rent time once they know that you're not going to hurt yourself or hurt the equipment. And I should probably take up that. I mean, Chihuly, the famous glass artist, this is kind of the area that he came from. And yeah, there's a lot of art glass around here. Less scientific glassblowing, though. Yeah, I actually take a glassblowing class and have been for a month or two. And it's a lot of fun, but I would actually rather do the scientific work, and I don't know how to find a class for that.

It sounds like what I should do is watch your video and buy myself a blowtorch. Great. Chris is really pleased with you right now. Uh-oh, more stuff. Stringy as a blowtorch. Great. Chris is really pleased with you right now. Uh-oh, more stuff. He's doing his blowtorch. Yeah, it's actually not that bad of a hobby. It doesn't take up that much room. I guess the big thing, if you want to do it properly, particularly with borosilicate, is you need oxygen.

And that can be expensive by the tank. You can get oxygen concentrators that just pull it out of the air, and they can run a small torch unless you get multiple of them and run them in parallel. I think Christopher's going to insert explosion sounds here. So your other entry, a Leyden jar, which is the old-style batteries. That's how batteries were initially. It's basically a capacitor, yeah. This one is

a cup with a metal cup and then inside that is a, traditionally it was glass and they used to foil the inside or the outside and they'd put, so the insulator, the dielectric used to be the glass in this case. I think it's like a piece of polyethylene or something I had lying around. And then the inside is another cup that sandwiches it. So it's just a capacitor, right? Metal, plastic, metal, or metal, glass, metal.

In this case, I was charging it up to about 8,000 volts, something like that. I was using one of those piezoelectric static guns, zero-stat guns. They come from, I guess it was the 1970s 1980s when lp records was still pretty common and you'd use them to spray charge onto your lp record to get rid of the dust i actually bought them for you know electrostatic stuff with the weather here can be pretty dry and we have static problems and also getting dust off optics and it's kind of

useful to just spray some charge on something to kind of neutralize it and then hit it with the air duster or whatever to get rid of any dust that's sitting on the surface. But they spray out ions with each squeeze, and as you squeeze them, they spray out one polarity. I can't remember which way around it is. And then when you release them, they kind of spray out ions of the other polarity so by sort of pointing them at something while you squeeze

and then pointing it away and releasing it you can you know or vice versa you can kind of control how you're charging something up in the case of the laden jar i could like charge one plate with one set of ions and then i could charge the other plate with the other polarity of ions and sort of double pump it. You just didn't hook it up to a DC power source. Yeah, I didn't happen to have one of them. I also got a piece of PVC pipe and started rubbing the PVC pipe, you know, with the hope of generating some static.

And the weather has not been particularly dry here and it wasn't cooperating so uh yeah the zero stat gun it was but you could totally have used a dc power supply where's the fun in that yeah exactly it's more fun to use something a little bit crazier so are you preparing to travel back in time 120 years and take over the world? Because it sounds like you'd do pretty well. It's funny, isn't it? All the easily accessible technology. Yeah, if you went back now, you could be Edison or Tesla. Yeah, totally.

I'd probably want to be Edison rather than Tesla. Yes, he did a little better. He did well for him, yeah. I think about that sometimes. It's kind of an interesting fantasy, isn't it? Although rebooting the world, if anything bad ever happens, I think it's good to have a couple people around that know how everything at least used to work and could slowly bring things back.

The lightbulb thing, I think people take for granted. It was kind of the simplest, cheapest artifact of technology before they banned incandescence. And everybody thought they were very simple, but a lot goes into making one. And it took a long time for them to become reliable and figure out the physics of, like you said, oh, look, this stuff's discharging on the inside. Well, let's do this and accidentally invent all of electronics. So, yeah, I think it's good that people are, that you're doing this and sharing how hard this is actually and what goes into it.

Yeah, I mean, I'm definitely not the first person to do this, that's for sure. There are plenty of people on YouTube that are way more, this is all very new to me and, you know, I'm just sort of starting my journey down this path. I don't have a particularly good vacuum pump. I don't really have a lot of the equipment required of just being kind of collecting it piece by piece.

Fortunately, eBay makes that a lot easier than it used to be. But there are still a lot of challenges. The biggest one for me for a long time was glass-to-metal seals. It took me a long time to be able to make a good one. Boros silica, in particular, kind of seals to tungsten pretty well and also to titanium and a few other things. But you've got to match the coefficient of the glass and the metal. Otherwise, when it cools down, it will crack. And that's the biggest problem.

You've also got to have the glass actually wet the metal oxide layer really well, otherwise it'll leak very slowly and ruin your device over a period of time. And even then, once you've got all the seals all nicely done, if there's any fingerprints or oils or your material, can sort of absorb gases and things and release them slowly and ruin your vacuum. Fortunately, most of the devices that I've been making at the moment aren't actually that picky. And I've had the opposite problem in a few cases

where I sealed the thing up pretty well and it was a discharge device that was actually, you know, had to have a couple of tore of gas in it. And the electrodes would sputter slowly and actually pump down the gas that was in there to the point where the thing was starting to get too low of a pressure and it was uh yeah it can it can happen in uh in discharge device it's actually one of the things that limits the lifetime of um of many things like neon tubes that even though the gas is noble if the early days when they didn't really have the metallurgy and the understanding of the electrodes,

the electrodes would slowly like sputter and take out the gas, basically, that was in there. It's interesting. You learn a lot about the general physics because, obviously, the study of the vacuum in a lot of ways is what started much of our technology, not just electronics. But the understanding of emptiness or empty space and particle accelerators and a lot of experiments

that were done in the early days of nuclear physics like proving that alpha particles or helium, for example, all come from our ability to produce low-pressure environments and a bunch of thermodynamic stuff like Crookes radiometers, and all of this kind of thing you can totally build with relatively hacked-together technology for basically the cost of a vacuum pump and a blowtorch and some glass. See? We're going to need a blowtorch,

and then a vacuum pump, and then some glass. It's a fun hobby. you should totally do it but you also need neon other things i mean moving on from the the flashing light contest you have been also making light bulbs and uh neon what are you making with the neon tubes? Science? Yeah, just playing around, really. Like, trying to understand, and just really mostly just practicing glasswork to actually get good enough at stuff so I can build whatever I want. Noble gases like neon and argon. I mean, argon's not too bad. You can get it for welding.

It's pretty easy to get. It's relatively cheap. Neon and then progressively, you know, krypton, xenon, they start to get. It's relatively cheap. Neon and then progressively Krypton, Xenon, they start to get pretty darn expensive. You can get them, but yeah, they can get very expensive. So I'm kind of practicing with the cheap gases like air, carbon dioxide, and argon until I get good enough to actually graduate to the noble, expensive gases.

And the different gases give you different colours? Yeah, each one has its own characteristic emissions and it's not a simple, I mean, if you put them through a spectrometer, you can look at the spectra and, you know, that's well understood science. But the pressure kind of changes which kind of emissions you'll get and some of the interesting things are at fairly high pressures where you get convection going on in there and the plasma is getting pulled out

by the actual heating of the plasma causing convection currents. Your traditional plasma globe kind of thing has fairly high pressure in there. In some cases, you know half an atmosphere or something even closer to one atmosphere and the heating of the gas is why it moves around you know it's a it's got a fairly low breakdown voltage so the you know 10 000 volts or something 5 000 volts whatever they operate at high frequency ac can ionize the gas but then once it's ionized those filaments of plasma are getting moved around by the convection currents that are generated by the heating of the actual gas.

Yeah, it's a lot of really interesting stuff, and I could see why people who are more of an artist than I am would be very interested in doing this, because you can make some pretty cool looking stuff. I mean, I think it's pretty cool, and I'm not an artist at all, but I really enjoy some of the just beautiful things you can do with it. What got you interested in it to start with? I guess I've always been interested in it. Glass blowing in itself is its own end. Glass is a very interesting artistic

and engineering material. It's transparent and it's got a lot of fascinating properties, but it can contain vacuum and it can do all kinds of wonderful things that we don't really think about. I've got a glass of water sitting here on the table and that's kind of a really cool piece of science. For a long time, we couldn't make stuff. We used to use an animal skull or a horn or something

as a drinking utensil. I used to go to the rocks area in Sydney, which is kind of like the early part where the ships used to land. It was a very early part of Sydney. And they would have glass blowers down there, you know, professional glass blowers that would make art glass for tourists. And, you know, I'd sit there in the window and watch them because glass blowing is one. It is fun, isn't it? Yeah. Yeah, it's really fun.

And people would stand around like people used to stand around and watch TV, I guess, when it was first put in the windows of the store and just be amazed at some of the things they could do by manipulating this really interesting material with nothing but heat and gravity. And it's pretty. I didn't ever like the color orange until I started to learn more about glass blowing. And that color orange is so different than, I don't know, other oranges in the world.

You mean the glow or the cadmium orange colors? The glow. Yeah. Yeah, and you get sort of used to it. You know what temperature it's at, and you kind of get a feeling for the viscosity as a function of how much light it's emitting, which is, yeah, it's interesting. And those skills, they take time,

and you can develop some bad habits, because I've never had any formal training with glassblowing. I mean, I've watched a bunch of videos, and I've watched a bunch of people do it for real. And I probably do everything wrong, so as soon as I'm going to put a video up on there, I can imagine a bunch of videos and I've watched a bunch of people do it for real. And I probably do everything wrong. So as soon as I'm going to put a video up on there, I can imagine what the comments will be like. There'll be people screaming at their computers. But yeah, it's relatively quick to develop basic skills.

But then being able to do exactly what you want in a consistent way, that takes a lot of practice. So much practice. And there's so many ways to do it wrong. And then at the end, you're like, oh, no, I totally meant to make my glass look like a shot glass. I wasn't intending to pipe glass at all. No, no, I swear. Yeah, yeah.

Well, the thing that really gets me is that you make something really good and then, like, you don't get it to the annealer fast enough. And it shatters. Yeah, and it's just that little tink and it's like, oh, the thing that really gets me is that you make something really good, and then, like, you don't get it to the annealer fast enough. It shatters. Yeah, and it was just that little tink, and it's like, oh, great. Yeah, I know what that means. Sometimes you can recover. But I've actually, the light bulb that was in the, that I entered in the contest, I actually cracked the glass to metal seal on it because I didn't, I cooled it down a little bit too fast. But I was able to warm it back up and get it to melt together and then brought it down slowly enough that it stayed together to get it into the annealer

anyway. That was all made out of soft glass and soft glass is a very different beast to borosilicate. Now borosilicate in fairly small sizes, you can really abuse, you can stick it straight in the fire and it won't explode. You do that with soft glass and it will pretty much just explode instantly. You have to be far more careful with warming it up. You've probably experienced this yourself. If you're doing shop glass, though, I guess everything's kind of coming straight out of the furnace, so it's all pretty hot to start with.

Yes, very much, but then we have canes, which is where all the color comes from. You get clear glass from the furnace, and so you have to be careful with the canes not to slop them in when it's super screaming hot. Yeah, I've always wanted to do cane work. I've bought little pieces of cane and stuff and stuck them in things, mille-feuille kind of things.

Another thing that I'm not good at is I don't know any of these terms of glassblowing. I mean, a lot of them are Italian and I have no idea how to pronounce them. But, yeah, it's always fascinated me how they managed to make, you know, you look at a final piece of art glass like one of those, you know those sea pictures where there's all these really complicated things embedded inside clear glass and you're just like, wow, how the heck did they do that?

But then when you kind of learn about like implosion pendants and all the kind of interesting things that glass sort of just does by itself by virtue of its surface tension, if you get it in the right configuration to start with, you slowly sort of get an understanding for how to make something. I kind of find the same thing with lamp work, with scientific glassblowing as well. Like you look at something that's got, you know,

particularly things that have spheres and stuff embedded deep inside something else and you wonder how the heck did they put all that together? But then as you get a little bit more practiced at it you get an understanding of they sort of build it from the inside out or whatever and a glass lathe would help me enormously and one of these days I'm going to have to get a glass lathe

I think to make some more interesting devices and maybe some art objects too Well I hope you all are enjoying the glass hour with Alan and Alicia. Yeah, this is an electronics podcast. What the heck are we talking about? Let me ask you one of the other questions you, or one of the other things you've tweeted about recently,

which really caught my imagination in a totally different way. You were mentioning implementing trigonomic... Trigonometry functions. You just keep trying, and I'll just edit it out until you get it right. You're going to leave that in, I know. Trig functions, sine, cosine, you know, those happy things. And you were going to implement them in analog hardware. Yeah.

I have a penchant for making particularly bizarre implementations of things in analog that should probably be done in digital. There's always make-a-fair and other things to come and bring your crazy project to and then explain it to a bunch of people for two days straight until you can't talk anymore. I love it. And that's kind of where that comes from the the trigonometric function implemented in analog hardware i would i want to build an analog computer basically that

can do trig a couple of reasons one was i wanted to actually build like a ray tracer that would directly drive video like old-fashioned television, analog TV out. Oh, that kind of ray tracer. Okay, good. Yeah. I was thinking 3D graphics with analog computers. Well, basically, instead of driving a digital display, you'd be driving an analog one.

Yeah, yeah. That was one thing. I also wanted to make a 2D lighthouse implementation completely in analog, which kind of lends itself to in many ways. And I'm sure if I'd invented it back in the 50s, it's probably what I would have had to do to make it work. So the first thing, there were some really cool chips,

which some people on Twitter reminded me of as soon as I mentioned that I was doing that. Unfortunately, they're not available anymore, and I kind of like to build things discreet when I go crazy rather than use chips. Not that I'm against using ICs or anything. I certainly make your life a lot easier but yeah, again, it's like all the flashing light and the triple five contest and all that. I just like doing things from scratch to learn stuff along the way. Sorry, go ahead.

I was back to the trigonomic functions, and sine and cosine are just oscillators. Those are easy, right? Yeah, I mean, generating the sine or cosine part is pretty straightforward. Taking what you basically want, as we want a voltage in, linear voltage, 0 to 1 volt or whatever,

and out of that you want, that would represent angle, and you want the output being sine or cosine. So there's a couple of different ways to do it. One way is you could use a sine shaper where you develop a bunch of analog electronics that takes that input signal and produces an output voltage that follows the sine or the cosine of that input voltage. So the Taylor, okay. Every function can be kind of expressed as a Taylor expansion.

This is just a mathematical thing where you have an infinite sequence of increasing powers of a variable and some coefficient associated with each one of those powers. And you can use that. It's a super useful tool in mathematics and many, even in digital computers, you know, quite often that is how you'll generate a function. You'll just evaluate, you just evaluate a couple of the terms

until you're kind of happy with the precision. So what you could do is you could kind of do the first couple of terms of the Taylor series in analog electronics because multipliers and squares and logs and all that kind of thing aren't too difficult to do. So you can do each of the X to the N kind of things in analog, and then you've got some resistors that mixes all of that together and produce an arbitrary waveform, in this case sine or cosine.

So that would be one way to do it. Wait a minute. I think we need to actually go a little bit slower on this because I'm really curious. So the Taylor expansion of sine x starts with x and so the very first thing you could do would be to just say

if my input is between 0 and 1 my output is just my input and that is the first step well in fact that's where you get the small angle approximation for sine x anything near 0 is 0 and then so the next step in the Taylor expansion,

which is just, I looked it up on Wikipedia. There's no magic here. I don't have this stuff memorized. Wikipedia is pretty magic. It is. So it's x minus x cubed over three factorial. Now, the three factorial is just a constant. So I imagine you put in a resistor of the right size or something or another.

Yeah. Vectorial is just a constant, so I imagine you put in a resistor of the right size or something or another. Yeah, so the hard part is really just each of those polynomial x to the n terms. And then each of the coefficients is just a resistor or an inverter if it's negative. That one is negative. How do you make x cubed given x with analog hardware? There's a couple of ways to do it i mean x cubed is a little bit more annoying than x squared x squared you could do with a you know a multiplier um we just multiply x by x basically i mean you could do x cubed that way as well you could multiply x by x and then multiply by x again so they have you know four quadrant multipliers where it's sort of

use a current to control a current. A Gilbert salt is a classic example, right, which is used as a mixer in RF tech, where you basically multiply one signal by the other to generate products. Same deal. You could also do it by taking the logarithm of the signal and then scaling it and taking the antilog of the signal, which you can do with translinear electronics. So things that have a naturally exponential or logarithmic response, like diode junctions, are particularly good at doing this.

So logarithmic and exponential amplifiers generally have a diode in a feedback loop. So you have an op-amp and you have a diode and you can use that to generate log and exponential functions and then just scale and do the opposite. So that's actually a trick you can use in FPGAs as well. You can take a signal converted into its logarithm and use logarithms the way Napier intended, basically, but in hardware.

What is a multiplier in electronics? Is this resistors, capacitors, inductors, or are you using some sort of IC? Yeah, it's an inherently nonlinear thing. So, I mean, the very simplest version, I guess, would just be like a cascode where you have some kind of gain device. Let's say, let's use a bipolar transistor. So you have a bipolar transistor.

It has some inherent gain. Probably would have been easier to start with a FET. But anyway, you can control, obviously, the gain of that by controlling its emitter current. So you can stack it on another transistor, which is basically a current source, and you can modulate that current source with its base voltage, for example. So you've got one

signal that's controlling the gain of another amplifier. And so there's obviously a bunch of constants and offsets here, but one signal is multiplying the other signal. And there are various arrangements of this. The Gilbertson mixer is the most classic example, which is a big translinear combination of two, like, differential amplifiers that all the kind of, you know,

square terms cancel out because you have a pair of them. So you can, they're useful, you know, because they reject the local oscillator in an RF system. But at its most fundamental, it means that you're using one signal to multiply or control the amplitude of another signal. So it's really a variable gain amplifier. The Gilbert cell has got a lot more symmetry and canceling properties. But a multiplier is basically just that, a variable gain amplifier. And a four quadrant one can go negative as well.

So you can invert the signal electrically if the voltage is one polarity versus the other and it'll do the full four quadrants. So you've got X being one signal and Y the other signal and the output is X times Y. How do you know what functions you can implement this way hmm um well i mean because of the taylor expansion i guess you can do pretty much anything that can that has a taylor expansion but it wouldn't be terribly practical when they start

getting big and ugly um there's a bunch of things like integrators, differentiators, any kind of scaling offsets. They're all super easy to do, logarithms, anti-logarithms. Multiplication, all of those basic electronics does pretty easily. The more complicated ones, the transcendental or trigonometric functions, they're a lot more complicated. And you have to do either the approach that we were just talking about with the Taylor expansion, or you could also just simply,

what you talked about earlier, you've got an oscillator producing a nice, clean wave, which you can, you know, you just filter the thing aggressively. So you've only got the original signal, the first, you know, primary. Alan, I'm really sorry. I'm going to have to cut you off. Our puppy needs to go to the vet, or dog. I guess she's not really a puppy. I had questions about eclipse planning and what to do on August 21st in Seattle and all of that, but I have to go.

So any quick things you want to leave us with? Well, I hope everything works out well for your dog, and it's been great talking to you. Do you think we can put in a record scratch sound right there or something? Because I really did cut Alan off mid-sentence. See, during the show, Chris had been running upstairs to check on the dog, and at this point, he came running downstairs looking very worried. Actually, I guess I should start closer to the beginning.

We have two dogs, a cat, and several fish. The fish are all named Henry and have no place in the story. The cat, who has been a guest on the show in the past, acts only in a supervisory role, naturally. The poofy white dog, the little one, Bear, has been sick since we got back from Southern California. That was one of the many reasons that when this week's planned podcast fell apart, I waited until the very last minute to find a replacement. And Alan had said if we ever were in a bind to call him, I did. And he came through. It was wonderful. Anyway, it was the beagle who was feeling poorly before the show. We hadn't paid enough attention to her because Bear was sick.

And so during the show, she started having some scarier, let's go to the vet right now issues. And that was where I cut Alan off. So Chris and I could take Zoe to the vet right now issues. And that was where I cut Alan off so Kristen and I could take Zoe to the vet. Transpires, that was not a trivial thing to do. But once we got there, they looked pityingly at the neurotic owners, threw antibiotics at us, and Zoe is fine, 100%. And Bear's okay, more like 80%. But he's getting better, so don't worry. So that's the story. Alan not only

filled in at the very last minute, when I pretty much hung up on him, he offered to sort out a time to finish, and it was totally my bad that we didn't. And with that, I will say it was an absolute pleasure to have Alan Yates on the show again. He is a hardware engineer at Valve and certainly the sort of person I would want to have coffee with every day. He was an honorable mention in the Flashing Light Prize, and there'll be links to his entry video in the show notes and on Embedded.fm. I'd like to take this opportunity to thank the Pacific Veterinary Specialists for fixing up our beagle, Christopher for producing and co-hosting the show, A1 Overhead Doors for the garage door that was smashed up later in the story,

and of course, thank you for listening all the way through this nonsense. You all are the best. And now a thought from me, which could only this week come from James Harriot. When all the world goes one road, I go the other. Embedded is an independently produced radio show that focuses on the many aspects of engineering. It is a production of Logical Elegance, an embedded software consulting company in California. If there are advertisements in the show, we did not put them there and do not receive money from them. At this time, our sponsors are Logical Elegance and listeners like you.