#169 – An Interview with Vincent Himpe - Escaped Electron Elocution

Chris Gammell: This is The Amp Hour Podcast, recorded October 28th, 2013. Episode 169, with guests Vincent Hempey, escaped electron elocution.

Dave Jones: Welcome to the Amp Hour. I'm Dave Jones from the EEV blog.

Chris Gammell: And I'm Chris Gammell of Contextual Electronics.

Free Electron: And from Silicon Valley, this is Free Electron, or as I'm known in the real world, Vincent Hempey. Hey, Vincent.

Chris Gammell: Hey, Vincent.

Free Electron: Thanks for joining us.

Chris Gammell: Hi, guys. The famous forum member and even more famouser master of silicon.

Dave Jones: Well, I'm still learning. I think we're all still learning, aren't we? I don't think there's a day that goes by where I don't learn something new.

Free Electron: It's a never-ending. Electronics is a never-ending quest. I mean, there's no – everything keeps evolving, and it evolves so quickly these days that you just have to keep up with stuff. Because it is a lifelong learning experience. And the sad thing is that you have to realize that you will never know at all. Exactly.

Chris Gammell: I figured that one out already. Dave tells me all the time. You've been an analog guy.

Dave Jones: What's all this newfangled digital stuff? Yeah, that's right. It's a severely constrained set of analog.

Chris Gammell: Yeah, exactly. Exactly.

Dave Jones: Right, yes. So what's this accent we hear here, Vincent?

Free Electron: That is Belgium. I was born and raised in one of the smallest countries of Europe, which happens to be one of the rainiest and coldest, miserable, foggy countries. It's like Cleveland. It's more like the – more like crossbreed between the Windy City and – Melbourne. I don't know what Melbourne is like. But so, yeah, it's – you know, you get this – I mean, I'm exaggerating. But it's – we have nice summers. But in the wintertime, it's cold and dreary and, you know, gray overcast sky and this miserable, damp, cold. You know, you walk outside and you –

Chris Gammell: So Cleveland, yeah.

Free Electron: If you say so. So that's why we had to come up with chocolate waffles and french fries and all that good stuff.

Chris Gammell: You guys did it right, yeah.

Free Electron: It was that on Invent Prozac.

Dave Jones: All right. So let's start out with your story. When did you start working on electronics and – Okay. And where – what was your first job and – Whoa, we'll go from there.

Free Electron: All right. So, well, one of my earliest memories – I don't know exactly what age I was. Probably six or seven. My dad had a broken little transistor radio. I don't know what brand it was. It was one of those cheap Japanese – probably like a Sharp or something like that. Sharp. You know, it ran off a nine-volt battery. And all I would do is produce static. So one day my parents were gone and I figured, I'll take this apart. I wonder what makes this thing tick. And this is all I remember about it. I opened it up and there was these beautiful colored screws, which were – now that I have evolved in electronics – probably the medium frequency transformers, you know, the little metal cans with the adjustable – And you thought that was screws in them.

Dave Jones: Well, they are screws.

Free Electron: So I decided – Slugs. Yeah. So I figured out, hey, all these screws are loose. Let's tighten them. And I got it to the point where there was no more static, but smoke was escaping. Solid starting point.

Chris Gammell: That's good. That's good. Yeah.

Free Electron: So that was my – that's my earliest memory. So my parents at that point, I think they decided let's redirect his interests. And they got me one of those Radio Shack or 101 wooden boxes with all the various parts. And I think I was about eight years old when I got that one. So that's my earliest.

Dave Jones: Yep. Yeah.

Free Electron: And I had an uncle that's also an electronics engineer. And he used to work for a company that made fire alarm systems. And my memory is that he was always, you know, in my grandparents' kitchen. He was always tinkering on some TV set trying to repair it or he actually built. We had a very prolific kit or a number of kit manufacturers in Belgium. And they're still around today. Right. And Valaman is a Belgian company. Oh, of course. Valaman. Yes. Yep. So they've got hundreds of kits. And then there was Philips from Holland that had a whole bunch of kits to make your own FM radio receiver turntable. And so those are some of my earliest memories is my uncle was assembling one of these kits and he made an FM stereo receiver amplifier. So that kind of was like I saw all these beautiful colored parts and he would sit there with the hot poker, you know, and put some metal on it. And look, this stuff sticks together. Wow. Magic. Yeah. Magic. And then for a while it kind of dwindled because at a certain point, I don't know, maybe 10 years, 11 years old. I wanted to actually become a biologist. Oh. My interest was, you know, you want, I watched a lot of documentaries from Sir David Attenborough. Yeah. I mean, we had BBC was broadcasting across the channel and he had all the, on the French television, he had all the Jacques Cousteau films, the underwater stuff. Yeah. Right. So that kind of became interesting to me. So when I was, you know, when it's time to, when you go from 12 year to 13 year, you change from base school to second, I don't know what you call that. Um, um, and then I picked, okay, well, you know, biology is kind of, you know, they call, they stuff that together with the general sciences. So I started my first, my first year of, of, of high school. Um, well, high school, 13 years old, um, doing, you know, Latin and Greek and mathematics and that kind of stuff. And that, you know, it was kind of fun, but eventually. You lie. You lie. The Latin and the Greek was actually fun. The mathematics, not so much. Yeah. Um, but the, the, the thing is that, um, I always, I still was always like falling back into the electronics a little bit. There was a, there was a point where we didn't explain. I remember we had a biology class and they were talking about the salinity of water and how you could measure this. You know, you had a pH meter to, to detect the, the acidity or, or to find out if it's an alkaline. And, and then there was a conductivity meter. And, uh, you know, you, you see these things in biology class. And I thought, Hey, I don't know electronics. Let's see if I could, you know, cobble something together that would do that. So I was always kind of, you know, being dragged back. Somehow I always kind of fell back into electronics. And then the big, the big change came in that, that second year, we had one course. We, we had a, um, one of the courses was physics. So we learned about basic mechanics and there was a half of the year was spent with basic electrics. So we had these breadboard systems with switches and light bulbs and big variable power resistors. And so that peaked it again. And then we also had another, another thing, which was called technological education, where you would learn how to, you know, do welds on, on water pipes for plumbing. And you would install how you learn how to install a light, a light bulb on the wall. That's technology. Yeah. Yeah.

Dave Jones: From the 1800s.

Chris Gammell: Vocational.

Free Electron: Yeah. So, and at that point I decided, you know what this, all this Latin and Greek, those are dead languages anyway. Who wants to bother with that? Electricity is the future. Let's, let's switch direction here and go and do electricity. And so that's how I moved back from, you know, a general science, you know, people that study languages or arts or, or general, you know, Latin, Greek mathematics, which then eventually flow into become, you know, engineering. I switched to, to basically what it's called. I don't know if you call it the vocational school or, or what you want to call it, but it's something that was much more practically oriented. So, and that's, that's how I rolled back into electronics. And, you know, I still go watch fishes while scuba diving, but don't ask me what the Latin names are. Yeah. Right.

Chris Gammell: It doesn't matter. You don't want to cut them open either. Right.

Free Electron: Exactly. If I catch a really good one, it'll end up on the barbecue. Oh, there you go. Yeah.

Dave Jones: And at 18, you took your first maintenance. You got a tech job at a semiconductor fab.

Free Electron: Well, there's a step in between. Actually, what happened is. Oh, there's a step in between. Do tell. There's a step in, yeah, there's a step in between. So what happened at 17, actually at 18 years old, you know, you're supposed to do like a little graduation work where you have to, you know, you have to build something, assemble something. And most of, most of the students, they go for one of these Veloman kits and, you know, they put them together. They write a three page report about it. And they go in the lab and they take a few measurements and they hand that in. And I don't know exactly how I got a hold of it, but I had gotten a hold of a satellite tuner. So just the metal, the metal can and the low noise amplifier that goes on the dish. And I found a parabolic dish that actually was, was for one of those space heater thingies. But hey, it's parabolic. Right. I mean, it is a parabolic dish dish. It does work. So I found the, found the, the, the, the, the focal point and put that in LNA in there and hooked it up to the tuner. And well, the video quality was so, so, but I could actually pick up a signal. And I figured out, well, I had been thinking with, um, Electro magazine had had published around that time, a project where they had an 8051 with a built-in basic interpreter in it. And I had, I actually had built that board. So I figured, well, if I hook up a D to A converter with a little, with a little amplifier behind it to drive the zero to 30 volt tuning voltage. And I also drive a stepper motor that can position that dish, you know, from, from, from satellite to satellite because there's multiple satellites above the equator. And that became my final project. And it happened to be that the Rotary Club at that time, um, they did a kind of a, um, a contest between students that were in that last year. You could hand it in and you had a chance of, you know, being picked, um, and winning a prize. And the jackpot was a two month during the summer holiday, a two month trainee, uh, application in a company of your choice. Oh, and.

Chris Gammell: Did they pay your salary or what?

Free Electron: Uh, yes, they did. So you, you, you were basically, you would be for two months an intern in a company of your choice. And luck had it that I would, I managed to pull off number one there. And because.

Chris Gammell: I, I choose the, the, uh, lingerie store. Are we talking to now? Okay. So yeah. Okay. We're talking to now. Okay. Okay. Okay. Okay.

Free Electron: Yeah. And since, since that metal can said Salkomp Finland, and I looked, I had looked up Salkomp, which is a subsidiary of, of, uh, Nokia. I just blurted out. I want to go and visit these guys that made that satellite tuner because Hey, this is RF electronics. This looks kind of cool to be space, space communication and the rotary club. I mean, these guys have connections all over the world. So they got in touch with, with, with Nokia and Finland. Nokia said, hell yeah. Yeah. He can come over. Wow. We, we have, I mean, this is problematic getting people who are interested in this stuff and that when I come, you know, we're 80, we're 80 miles above the polar circle here. Um, okay. It's summertime. You got 24 hours of daylight. I wouldn't want to be there in winter, but so I ended up in Finland, did my two months of internship. And, and one of the things that this is actually made like a, it was an automatic test jig. So they use stepper motor to adjust, to tweak the inductors in that, that thingy. So I wrote some, some software in that time it was turbo Pascal and I had, they had a GPIB card and I was reading, you know, getting data from a spectrum analyzer. So basically what we had to do is, you know, tune two coils to have the low end and the high end. So I wrote some software to do it. And they had a mechanical department that built the jig with the stepper motors and the adjustment, uh, uh, little, little metal blades that would, you know, slide into the, to the inductors. And at the end of my, my, my two months stint over there, these guys said, you want to come back next year? And I'm like, hell yeah. So I did that actually three times in a row. So I went three times for summer holidays. I went to Finland and put it around. And the second year they were, they were introducing the first HDTV, the first 69 aspect ratio. And they used a, um, it was still an analog signal, but it was multiplexed in time. I'm trying to remember what it was called, uh, DMAC or D2MAC. It's digitally multiplexed analog component, something, something. So they basically, they transmit luminosity followed in time by color. So you don't have those, those bleeding artifacts anymore. Um, and you still, you still need a frame buffer to decode it, but it's still an analog transmission, but it is high definition. I mean, it was 1920 by, uh, not 1080, but 1025, I believe the 1000 and no, 1250 lines. So it was the old PAL standard, which was 625 doubled, which was 12, 1250 lines of which you only have about 1100 visible ones. But so that's, that's, so, and I were talking 1989 now. So this was as an 18 year or 19 year old kid. This was like, wow, look at that. I'm sitting in front of a $5 million spectrum analyzer. So these were big honking 19 inch rack spectrum analyzers that did 30 gigahertz. And I mean, they could feed the, whatever they picked up from a satellite straight into that machine. Um, so yeah, that, that was pretty cool. And then I went on and I thought, okay, I'm going to do a engineering course that was, you know, to become a real, uh, BSE or an MSC. I looked at that and it's like, wow, this is way too much theory. Um, but there wasn't, there wasn't a different kind, uh, of course in Europe, which is called a graduate course. Um, I don't know if there is an equivalent in, in the, in the English system. They, they now scaled it up. So it would be the equivalent of a BSE now, but it's, it's not quite, it's not exactly the same thing. So I thought, okay, well, let's, let's go for that one. That's a little bit more practically oriented, I hope. And then in the first year, because you have this, this influx from people that, uh, my, my ex friends that started, you know, when I was interested in biology and studying Latin and, and Greek, all of a sudden these guys pop up again and they have never seen anything about electronics. And, and me coming from a, uh, you know, a more vocational oriented school where we, we did all, you know, we, we, we, we did learn how to repair, uh, radios and televisions that we, we, we, we could design an amplifier, you know, it was circuit with an op amp and integrated and we could, we, we were thought how to do all these things. So now you, you start this, this bachelor's equivalent course. And because they have the influx of the people that came from a non-technical background, here's lesson number one, Ohm's law. Start from the beginning. Yep. And by the time you hit the middle of the year, they're still messing around with a diode and I'm like, geez, this is boring. So the end result was that I basically flunked for almost there. And then the problem was you had also the obligatory chemistry lessons and physics and geography and history and the math and all this other stuff that I, I have no interest in. Um, so I, I knew I was going to flunk this thing anyway, not because of electronics, but because of all the other courses, because of boredom, um, I decided to lack of enthusiasm.

Dave Jones: Yeah.

Free Electron: Um, lack of, lack of proper stimulation. Yeah. So I decided to spend the last, you know, four, four weeks of that nine or 10 month, uh, year, uh, to help the seniors finish their projects, which was, which was far more rewarding for both me and them.

Chris Gammell: Right.

Dave Jones: So I finished a many, a senior project and I, and I look back and I got, went, geez, I worked cheap. I did that for a free lunch.

Free Electron: You know, basically. Yeah. You, you keep, you keep me happy and supply ample amounts of Coke and chips and yeah, I'll decide your, your whatever.

Dave Jones: And word soon gets around that. You're the guy to do that. You know, you're the only one in the class that has any clue, you know, we, uh, we, uh,

Free Electron: we got it to the point that I was so well known that we had a, I mean, that's good. That school had a big, a big, uh, covered, uh, uh, a break room and I would have consulting sessions every lunchtime. And people would come up to me and say, when's the next available slot? I need about two hours of your time.

Chris Gammell: I'm imagining like the Godfather music playing like Don Vincent. I need your help with, with his op amp.

Free Electron: That was brilliant. That wasn't exactly, we didn't exactly go that far, but yeah, it was, I mean, it got to the point where, and especially as crunch time, you know, I mean, as the year was progressing and, and some of these guys were in, in, in getting into real trouble, uh, uh, yeah, then it was really, okay, who gets priority here? And there was many a fight over those things and once in a while I would pull a practical joke as well. If there was one of these guys that I couldn't stand because I, I mean, I could detect that this is a guy that really, you know, he's got no clue what he's doing. He just wants my help because he wants to get that piece of paper at the end of the run. I would just give him a deliberately wrong answer. And the most funny, the most funny one was one, one guy had a problem with a 7805, you know, five volt regulator that kept overheating. And I told him, I said, well, you just need a bigger heat sink. And the guy goes, well, there's no room in my box for a bigger heat sink. So I started this elaborate explanation and said, well, you know, that the, the ability to release heat to, to your ambient environment is, is equal to the surface of the heat sink. So if you start drilling holes in it, you create all these interior surfaces. You can get an infinite, infinite surface area. Swiss cheese heat sink. It was, it was more than Swiss cheese. It was to the point if you would hold it by one corner, it would unravel downward. It's an air core heat sink, right? He must have drilled a hundred holes in that thing. Well done. I couldn't stand people like this. Like you, you want to come to, you know, I have no problem. I mean, there were students that genuinely had problems understanding basic things like, you know, the classical two transistor flip flop, you know, where you have two, you make two LEDs blank, right? I could sit there and explain that a hundred times in a row. And that wouldn't bother me. And at least if they show genuine interest, wanting to understand how it works, I can understand that a person doesn't catch on immediately. And I have, I am patient with that. And I will, if I will explain it 150 times or a thousand times if I need to be. But if you're just doing it because you're doing it so you can get it, leave it behind you and move on and you don't have any genuine interest. Yeah. You got to watch out for me because I may be giving you a wrong answer. Beware forum members. Beware.

Dave Jones: A 20 page reply and it's all bullshit. And everyone's laughing in the background.

Chris Gammell: Did I see Laura Mipsum in there?

Free Electron: Not really. Love it. Oh boy. So, and then there was the mandatory military service I had to do. So, in Belgium, I was one of the last people that had to do, you know, it was mandatory. It was one, you know, one shift behind me and then it became, you know, voluntary. And I ended up in a depot of the military where we basically had to exchange the radio transmitters on their, you know, their Jeeps and their, you know, light vehicles and stuff like that. But the problem with any of that military work is that you've got to be certified. I mean, you can't even put a solar joint in unless you're certified to mil spec, yada, yada, yada, yada.

Chris Gammell: Yeah.

Free Electron: So, it turned out that was pretty boring because all we were basically allowed to do is unscrew the plug in the back of the unit and do two bolts, throw the unit in the bin and put new one in. I mean, we wouldn't, we weren't even allowed to change a cable going through the antenna. Oh, that's going to be done by a certified. Right. So, that job turned out that I was more on the road and those machines were, I forgot if it was Siemens or Philips that made that equipment. And because I was from the part of Belgium that that factory was in, I was always the Chinese volunteer because the surgeons always told me, you know where that place is, right? Where they build these things. Yeah, sure. I grew up very close to them. Well, we've got three broken antennas here. Go take them and drive them over there and wait until these guys sold a new cable on them and then come back. Okay, fine. Whatever you say. So, that was pretty much military service. I mean, it was only a few months, but, you know, it was kind of fun, but it was also kind of boring. All right.

Dave Jones: Well, let's get on to the interesting stuff.

Free Electron: And then the interesting bit is, so now I'm, you know, all the mandatory stuff, education and military service and all that stuff, it's all done. So, what do you do? Well, you go and, you know, and apply for a job at a couple of companies that, you know,

Dave Jones: in Belgium. Get a real job. Whereas nowadays, you would go and put something on Kickstarter and do a startup. Back when we were boys, right, Vincent? Yeah. You would go and get a real job. I got a real job, Dave.

Free Electron: Right. Yeah, I know. When I came out of school, there was no internet. That didn't exist. I mean, it only existed like a few years later. See, Chris, I told you.

Chris Gammell: It's true. Oh, Dave hasn't been lying.

Free Electron: No, no, no, no, no.

Chris Gammell: I've heard that more than anything. Next up, it's two uphill both ways and in the snow. Yes.

Dave Jones: And exactly.

Free Electron: I mean, I didn't pay horseback.

Chris Gammell: That's right. Yeah.

Free Electron: The early internet was limited to a 9,600-boult modem. And if you had to walk two hours through the snow to get to the internet, you did that. You know, now they just…

Chris Gammell: Anything to get on the BBS. Yeah, exactly.

Free Electron: Well, that was before there was internet. There were bulletin board systems. That's right. There were a couple of big ones in the US, but those were expensive to dial into, especially from Europe. But there was one bulletin board that was actually operated by Philips out of Holland, and they had a mirror site in Belgium. So that was kind of interesting because they had all the data sheets for their AD51 clone microcontrollers in there. And there was all the stuff from the Cignetix people. So that was kind of… Anyway, so Belgium has had quite a history in electronics. I mean, we have had probably five or six companies that were making television sets in the 70s. But by the time it was, you know, 1991 and I was, you know, entering the labor market, most of them were gone. There was only one big player left, which was Barco, which is still around today. They make these big kick-ass projection systems for big events and shows and control rooms and stuff like that. I mean, those guys have been around forever. So you'd walk in there hoping to get a job and, you know, you get turned down because, you know, you don't have that gold ribbon-plated paper that they want, you know? And so, well, what do you do? You go to these interim, well, not websites because there were no websites at the time, but they had actual… A tab agency? Yeah, agencies, you had to walk in the door and fill out a bunch of papers and, you know, and I don't know, a few months later we get a phone call and, is this Mr. Hempy? Yeah, it is. Well, we have a job opening in a company in Aldenada, which is called Metech, to do, you know, it's only an interim job. We'll be paying you weekly, but it'll be involved. You're going to be working in a clean environment and it's precision work and it's electronics. Are you interested? And she says, sure, yeah, I'll do that. Hey, I'm an employer. I'll sweep floors. Anything, you know, I'll paint the lines in the parking lot. I don't care. So, I show up over there and they go, oh, your name's Vincent Hempy. We were actually after a different guy, which turned out to be my dad.

Speaker ?: What?

Free Electron: What? Yes. He was unemployed at the time and he was in the same agency. No way. That was actually the guy that they were shooting for. But anyway, I got an interview. That was, yeah, that was like, oh, well, that's my dad. Hey, I'm here now. At least do the interview, guys. And one of their maintenance people came out and, you know, it was a very simple interview. You know, okay, can you make me a circuit? You know, you have a transformer, rectifier capacitor. We want you to make 12 volts. So, you know, draw the schematic with four diodes, two caps and put a 7812 in there. And then there was some other questions. They said, okay, well, this is basically, it's an interim job with a potential of becoming a permanent hire. It turned out that what that company did is they got their manpower on a six-month trial period through an interim agency. So, you would work on a weekly basis for your first six months. And if after six months they liked you, they would offer you a permanent position in the company. And this was maintenance. So, I was actually assigned to the group of people that did the maintenance in the clean room of the etching and iron implantation equipment. So, that's actually how I rolled into semi-combustible.

Dave Jones: Chris's eyes just opened wide up.

Chris Gammell: Yeah. I started an etch as well, except that was probably like open etch chambers and like crazy stuff like that back then, right?

Free Electron: We still had wet etching at the time, but most of the stuff had switched over to plasma etching. So, basically, you put your wafer in the reactor, you pump it down to near a vacuum, and you inject reacting gases, and you bombard them with an RF field so that the molecules fall apart and they react with whatever you want to eat. That's it.

Chris Gammell: See, one of the guys I used to work with, though, he said that, maybe it was earlier on, but maybe it was like 80s, because this is probably past the 80s. But he said they used to have open etching chambers.

Free Electron: Yeah.

Chris Gammell: Like, not wet etched, though. Like, there's a dry etch.

Free Electron: Yeah, but those, yes, that is a possibility. What they do is they are open in the sense that you can actually see through them, and you could actually stick your hand in there where they're running. But the way that those things work is they use two high-pressure currents. So, what they do is there is actually a zone which is at near a vacuum and filled with the reacting gases. But they use, it's not really an airlock, but what they have is they have a nozzle that projects a gas, which is typically nitrogen. And it spits out that gas at such a high velocity that it actually, for all means and purposes, behaves like a solid. So, you basically, you decrease the pressure of the gas flow, you slide the wafer through the gas flow, and you increase the pressure. So, now you have the wafer sits in between these two curtains, left and right, of high-speed, high-pressure flowing nitrogen. And top and bottom, there's a metal plate. So, that is indeed, that's what they call an open reactor, because the reactor is actually closed, left and right, because of those two curtains of high-pressure nitrogen. So, but yeah, those things do exist. Although, for etching, they weren't used as, I mean, there are instances where they use them for etching, but they were typically used for vapor deposition, where they were, you know, growing new layers on top. So, that's typically, what are they called? CVD, chemical vapor deposition machines. That's what those things are called. Yeah. So, here I was hopping around in my bunny suit. I mean, mine was blue because I was maintenance, as opposed to white, which is typically for operators. Blue and white color workers. Yeah. Well, not really. No. No, no. Operators are not white colors. Yeah. No. And then there's orange, which is like for the plant and facilities and intervention teams. There's a chemical spill or whatever. Yeah, emergencies. The guys with the gas mugs and the yellow rubber boots.

Chris Gammell: Did you get the training? So, this is what I always tell people about with the training and stuff, how they talk about like HF, like hydrofluoric acid. Oh, yeah.

Free Electron: And while HF is, I'm okay with HF because it's a liquid. You can see it. But it's the things you can't see that are nasty. I mean, like if you're going to do any kind of the doping materials used in semiconductors, you got phosphorus, you got boron, you got arsenic, which is extremely poisonous. And you have antimony, which is a heavy metal. So, the problems are arsenic and phosphorus. So, the way that they get these, antimony is a solid metal. Okay. Okay. Boron is a gas which gives off white smoke. You immediately see it. If it comes in contact with air, you see, oh, there's a boron leak. We got white fumes coming out. Get out of here. Run, run, run. The problem is that the other two, arsenic and phosphorus, they use hydrogen as a carrier gas. So, it's hydrogen phosphide and hydrogen arsenide. Now, I am told, I don't know, I have no personal experience, but I am told that one of them smells like garlic and the other one smells like rotting fish. The problem is that the level…

Chris Gammell: If you smell it, you're dead. Bingo.

Free Electron: The levels at which a human will detect the smells are the lethal dosage. Great. So, these things are… Yeah. These are really annoying. So, the problem is that in an ion implantation machine, they use these little steel cylinders. They're about the size of a one liter bottle. They have about 800 psi of pressure in them. So, that's about 50 bar, give or take. 50 times 15, yeah, 750. So, they got a volume of that gas in it, but if that gas were to escape from that bottle, everybody in that clean room is dead.

Chris Gammell: Yeah.

Free Electron: So, you have these big elaborate gas detection systems, and whenever that cylinder has to be changed, which is about once every two weeks, that little tank is empty. You got to go to security, get a work permit issued, and then they put the alarm systems on extra alert and whatever. And this is one of the key training sessions that you go through when you're being initiated in a clean room. If there is a moment in time where all the lights, all the normal white lights, we're not talking about lithography. Those are the yellow light guys. Those are… Yeah, that suck. That's a separate room. They're the yellow light people. We don't associate with them. Yeah. That's right. No. Anyway, that… They're weirdos. That's a whole different clean room. There's no interaction. So, the area where the ovens and the implantation and the etching is done, that's a separate clean room. So, that's where the poisonous gases are. Lithography, they don't have any poisonous gases. They're breathing alcohol and that kind of stuff all day long and all the solvents. But there's no poisonous gases.

Chris Gammell: Right.

Free Electron: So, but the main clean room, basically what they do is they shut off immediately when there's a gas leak, a dangerous gas leak. The normal white light is turned off. The light goes to red and all the doors pop open. So, they have these electromagnetic locks on the doors that actually shoot a deadbolt and it shoots the doors open. When that happens, you run for your life. And I've seen instances where… And you're… I mean, as a technician that has to do these gas changes, you're being trained. I mean, you're wearing self-contained breathing apparatus when you're doing that stuff. And everybody is aware, okay, they actually have this announcement system where they tell the operators, okay, the guys are going to start a gas change on implanter number three. Be on the alert. If the light goes red, go. You know, don't drop what you're doing. Even if it's…

Chris Gammell: They make it green instead of red.

Free Electron: Even if it's a box of wafers, you know, don't bother putting them down. Drop them where they are and run because it's not worth, you know, killing people over it. But the problem is here you're faced… I mean, you've got your head stuck in this machine that has a high-voltage cabinet. And it's a cramped space. You're trying to get that bottle in there. You've got it hooked up. You do your pressure testing and everything's fine. You turn the valve and the gas escapes between the seal around the stem of the handle. So it's not escaping into the system. It's actually the seat. It's not the orifice where the gas is supposed to come out to. But it is leaking between the stem of the handle and the bottle itself. The alarms go off. What do you do? Oh, my God. There's no way to shut the damn thing off because if you close that valve and it keeps leaking because you broke the seal, you're faced with a dilemma. This bottle, no matter what you do, this bottle is going… I am fine. I'm wearing a gas mask. I'm on self-contained apparatus. I can waltz out of here if I have to. I could spend dilly-dally for half an hour. I'm okay. Everybody else is already out the door. But what do I do? Do I let this gas bottle empty itself in the clean room knowing that probably in the next couple of days nobody's going to be allowed in here until the air filtration system has done its work? So what do you do? So in an almost instinct reaction, you yell to your colleague who's… You're always working with two or three people. There's two people watching over you basically holding onto your belt ready to yank you out of harm's way. So you yell to your colleague, okay, open the outer door. And you, as quickly as you can, you undo the buckle, the clamp that holds the cylinder in place. You undo the gas line. And you just throw it as hard as you can through the open door into the… Oh my God.

Chris Gammell: Where everybody's now standing? Into the farmer's potato field next door.

Free Electron: And then you basically have to go to the farm and tell the guy, dude, whatever you're doing, we just lob a gas canister in your potato field.

Chris Gammell: Don't eat those. Don't eat those, buddy.

Free Electron: Don't go there. We will tell you when it's safe and we will pay for your destroyed crops basically.

Chris Gammell: Those are going to make for some bad potato chips. Yeah.

Free Electron: Well, that's the other thing because you're mentioning that potato chips when that factory was actually… This was before I rolled into there. But they had an advert. They had a, you know, this was all high-tech and especially in a small country like Belgium. Wow, they're going to open a chips factory. And that's what… People didn't know what an integrated circuit was. They knew. They thought it was french fries. Microchips. You know, Joe in the street knows what is a microchip. He doesn't know what is an integrated circuit. But microchip, that's something that he sort of heard of. So it turned out, and this is hearsay, okay? I have not witnessed this myself. But what happened is when the factory was opened, they had the big inauguration day. And there was a big article in the newspaper that says, you know, research center, Belgian research center in collaboration with some investment partner, opens chips factory in Audenarde. And it turned out that actually a few farmers actually showed up asking if they could have potatoes.

Dave Jones: Sorry, we only take saying.

Free Electron: It was a chips factory. So these guys go, oh, well, they're going to probably need potatoes to do that. Oh, that's great.

Chris Gammell: What's the name of the town? Or maybe can you spell it?

Free Electron: Audenarde. It's O-U-D-E-N-A-A-R-D-E. The factory actually today still exists, and it belongs to On Semiconductor. Oh, there you go. So they make a whole bunch of mainly automotive parts for On Semiconductor. So that way for fun, this still exists today. So, yeah, that was 1993. And what happened is, so I was hopping around in that clean room doing the maintenance stuff. And this is the kind of maintenance, you know, I wrote it on the forum, actually. If the manufacturer says every two weeks the bearings in the drive system have to be changed, you change the bearings. You don't go like, well, they still look good to me. No, because if, you know, one week from now that machine comes to a halt, there's way too much money involved. You do your preventative maintenance even though it looks, you know, like a waste of money.

Chris Gammell: Trust me. Well, then you go take it out and you refactor all the consumables, right?

Free Electron: Exactly. So you don't cut any corners in there. But what happened is that they had a three-shift system. So one week I would have the morning shift, which is 6 a.m. in the morning to 2 p.m. in the afternoon, I believe. And then the late shift, there was one week late shift, 2 p.m. to 10 o'clock in the evening. And I had actually one week of night shift. So the night shift is typically the shift where very little is going on. That's where they're running the long-term processes. So they have been started in the afternoon shift and they will run until somewhere in the morning shift. So basically the people who are running the night shift, the operators, are only there to keep an eye on the machinery. They don't really do any, you know, hands, let's call it hands-on work. They're just babysitting. And if the machine goes wonky or something is off, they have to be there to keep an eye out for what's going on. But it's a much lower pressure than during the daytime.

Dave Jones: Are you able to give us an example of what things would typically run overnight? What are these long-term?

Free Electron: Long-term things could be stuff that would run in an oven and sit there for 8 to 12 hours baking. They would have, for example, a whole bunch of wafers that had to be exposed. And so, okay, well, we're going to pre-coat. We know that next week there's going to be a peak in demand. So let's pre-coat wafers with photoresist, for example. That would be typical jobs for the night shift. Got it. Was this a research place? No, no. This was very hard manufacturing. We were a sole supplier for Bosch and for Wapco and Alcatel. We'd made a lot of chips that ended up in telephone exchanges and telephones and stuff like that. So no, no. This was real, real production.

Chris Gammell: So I used to work night shift, but it was a couple years ago. It was only like six or seven years ago now. And so it was all just automated. So it was like a lot of SPC type of stuff and like keeping an eye on things. And like you said, it's like babysitting. The way I always used to say it, it's like being an ER doctor where like a wafers, a patient, you kind of, when they crash, you go figure out what's going on.

Free Electron: That's exactly what it is. Most of, and that doesn't mean that these people aren't working because yes, they are filling out the paperwork and they are doing the statistical process control and all that work in between time. But it is basically, you have to be there and you're responsible for one or two machines and you have to babysit that thing. And at a certain point in time, somebody comes with a box of wafers and then you got to do your thing. And then you got to sit there and wait until the machine's done. And then you got to do a little bit of another thing with some paperwork and hand it off to the next guy. But yeah, I mean, it is still, you have to be really focused and know what you're doing because if you screw up your, I mean, they call these things thirds. So every single step in the process of making a chip is a deliberate action that is being done on a machine. Now, between a bare silicon wafer coming in and an actual, you know, a wafer leaving the fab to where it is handed off to testing, you're looking at 100 to 150 process steps. So if you screw up one really early and you only catch that yet, you just lost seven days of work. So you have to be really focused what you're doing and don't, you know, don't make any mistakes, double, triple check everything you're doing. And they've, they've got, I mean, those processes are so well thought out that it is really hard to screw them up. I mean, it only, really, it only screws up when a machine screws up. But operator error is almost, there's so many checks in the line that you, it's almost impossible that there is an operator error. It's either.

Chris Gammell: Well, they're removing operators too. That's the other thing. I mean, operators is like actual decision makers in the process. It's more like they're just kind of shuffling stuff around. Yes.

Free Electron: Yeah. Yeah. Because the process flow is worked out and you just follow it. You follow the dots. That's it. But still, you have to be focused on that little dot that you are working on because it's got to be done right. So the human factor is always there. I mean, the message that I'm trying to give is this, don't get the idea that, oh, well, these guys that are, you know, hopping around in the buddy suits, all they do is, you know, put the box down, hit the start button. When the machine goes, ding, they take it off and hand it off. No, no. There's a lot more involved than that. There's a reason that these, I mean, if they were replaceable by a start button and a bell, they would have been replaced long time ago. There is a reason these people are in there. I mean, they cannot be replaced. Their job is valuable. But so night shift, you know, it's quiet time. And night shift is also typically where you do a lot of that preventative maintenance because you've got the machines that are heavily occupied during the daytime. But sometimes it's boring. And the tendency, in a way, you know, as a maintenance guy, it's boring because you're sitting there looking at this big terminal hooked up to a big mainframe computer. And you're hoping that, please let there be an equipment crash. Please let that thing say. Yeah, exactly. Nothing makes the night go faster. You know, oh, look at that. We got a flask matcher that just went down. Yay, I got something to do.

Chris Gammell: So I found out. So how did you transition?

Free Electron: I mean, maintenance. That's what I was going to explain. So what happened is during a lot of these shifts, I found out that we actually had a room where there was one of these, they call them tested measurement benches. So there was a scope and a power supply and a variable AC transformer and a whole bunch of other things. Now, when you're repairing, when a machine actually goes down and a part breaks, you don't really troubleshoot component level. All this stuff is modular basis. It's not even circuit board. I mean, they're really metal boxes with a connector. And the machine would say, okay, there's a problem in the optical endpoint module where you would yank the module out, go and get to the warehouse, get a spare one, pop it in, and production would resume. You know, it's all about minimizing the downtime. And then you would send that unit back to the manufacturer and it would come back with a $5,000 price tag attached to it. So some of these modules were driving little chart recorders and they would fail very, very often because, you know, that cable would come out. Somebody would trip over it or yank on it and it would short circuit and then the output wouldn't work. I was bored one night and I decided, you know what, we got two or three of these modules laying around here. I was actually filling out the paperwork to have them sent off. And curiosity got the better hand of me and I took a scooter. I took the damn thing apart. And here you have this big 19-inch cartridge that has big honking connectors in the backside. There's a big circuit board in there that holds two decoupling capacitors, two resistors, and an LM7401 op-amp. Well, there can't exactly be a lot wrong with these things, you know. I don't see any other active components apart from that op-amp.

Chris Gammell: I might be able to figure this one out.

Free Electron: So I went to the lady in the night shift that did the stock and I said, by any chance, are we producing any LM741 op-amps? And she goes, yeah, that's one of the products we actually manufacture. Can I have a tube? So I pulled the tube of, you know, 50 LM741s and DIL package of the product, fresh off the production line, desolded the op-amp, put in a socket, put in the op-amp. And the next time, there was a few days after we had maintenance that had to be done. And while I was doing maintenance on the machine, I thought, I'm going to see if actually my module works. So I swapped out the working one with my repaired one and my repaired one worked. So I went to the boss and said, hey, I know how to fix these endpoint. You know those endpoint modules that fail every few months we have? I know how to fix them. We actually make that part. You know, it doesn't cost us anything to fix these things. And then the boss goes like, wow, we paid $2,000 to have these things fixed because we've got to ship them all the way back to Petaluma, somewhere in California. And then they've got to come back and you'll be recertified and whatever. So the boss quickly figured out that I could do a little bit more than just, you know, swap modules around. And I evolved to, you know, actually start to troubleshoot things on a component level. And I built them. One of the etchers had a problem that you could only store the recipes. You could only store four or six recipes. And those were stored in like a 2816 style eSquare Pro. So I made a little... Jeez, really? Really. And so what I did is I made a circuit board with four sockets. And I had a 74138 decode. And I had a BCD, one of those thumbwheel switches. So I just would switch the chip select for one of the four on my board around. And hey, I just quadrupled the memory storage of the etcher. And they thought, wow, this is really cool. We don't have to reprogram these things all the time. So I started doing stuff like that. And then I automated the wet etcher. I made a board with an 8051 microcontroller so that they could, instead of having, you know, the operator sit there with a chronometer, you know, a stopwatch basically counting down while it was etching. We had a little motor that could lift the wafers out of the acid. And they just had the key and the lot number. And I had a, this was, it wasn't an RS-232. It was an RS-422 current loop. So this, my little board would basically ping the big, those were digital VAX computers at the time. Big honking mainframe would ping and said, have lot number 793, whatever here. Am I supposed to be at this process step? And the mainframe would answer, yes, you're supposed to be here. Because that thing was tracking, you know, all the, everything that was going on over it.

Chris Gammell: Operators are keying in anyways, right? Yes. It's just you automated that.

Free Electron: Exactly. And then, so that was step number one. And then it would cue the mainframe and say, well, how long is this supposed, is it supposed to be at station 3, by the way, reporting? Is it supposed to be at station 3? Which has the action for aluminum in it. And again, the VAX, big VAX would reply, yes, you're supposed to be here. Okay, well, how many seconds are we supposed to at station? The VAX would reply, 34 seconds. So the little LCD display would get, the number 34 would appear. And it would say, okay, question mark. And the operator had to push the okay button. And it would lower the wafers in the asset for 34 seconds. And then when they're done, just like the fries are done, the wafers would pop up and there would be a bell that goes ding. And that does go ding. The popper is go when it goes ding. Yep. So the Pavlovian conditioning of the operator would kick in and he would take his wafers off and take his little check mark on the run travel, which is a sheet of clean room, dust-free paper that accompanies the wafers. He would put his little check mark on there. And he says, it's checked off. And then he would press the check mark button that would tell the mainframe computer that, okay, this process has been completed. And you know, it's going to move on to the next guy. And so at that time, they were looking for somebody in the R&D division to basically be the lab technician. You know, what you call a gopher. Exactly. You know, make, I mean, they were wire wrapping boards at the time.

Chris Gammell: Oh.

Free Electron: Oh, yeah. I mean, they were primus. Or those big, you know, what do they call it? Vero board or Vero board. The big blue boards with all. Yep. And somebody, you know, who could design simple circuits and maybe write simple programs on a PC, collecting some data. And so there was a job application posted on a billboard. So I noticed that and I applied for that. And my boss over there in the Wafers Fab was actually supportive of me. He says, yeah, I mean, what the hell are you doing in the Wafers Fab over here? I mean, you've done this automation stuff when you were at Nokia in Finland. I mean, you've talked GPIB and then I knew how to do all that stuff. He says, go for it. I will support you. So I get an interview by one of these, you know, and actually it was a gray beard. You know, I'm not joking. The guy had a really gray beard. And we started chatting and he was like, yeah, well, we have a problem. We're looking for somebody who can write some code. Maybe we're starting to develop a chip that has an I2C interface. And I'm like, oh, yeah, I know all about those. And by that time we had the internet had started. So I was one of the first. I believe my. It's here. Yeah, it's here. Yay, it's here. My internet user number was 751. And you can actually look it up. If you, you will find, you will still today find references.

Chris Gammell: There's like a registry or what?

Free Electron: I don't know. I don't know exactly how it worked, but the Belgian internet provider, which was called Ping, which was personal internet gateway. They handed out a username, which was just Ping and then numbers. So my username was Ping 751. So I was the 751 person that signed up to their service. And you will today, if you go to Google and you Google my name, Vincent and I2C, you will still find references to Ping 00751. Because I wrote, I was one of the very first people that wrote the I2C. I wrote the I2C FAQ, which is still floating around on Usenet and on the boards. And it's still out there. So this guy goes like, oh, so you're the guy that wrote that document. Yeah, that's what we found on the internet. Our designers are using that. And that was basically, I was hard on the spot.

Dave Jones: That's what I said. You publish stuff. And that's why you have hobbies, folks. Yeah, that's why you publish stuff. It gives you an instant reputation. Yeah, I've walked into interviews and people have gone, oh, you're that guy who wrote that article. Yeah, that's me. You know, like you're hired. You know, that's, yeah, that's how it works.

Free Electron: And it's really weird because you, I mean, that I2C FAQ, if you just Google for I2C FAQ, there's going to be 2 million hits with my name on it.

Chris Gammell: Actually, this is a Russian site.

Free Electron: That was the whole point. I mean, this was posted on news groups and it was, you know, free for all to copy. And as long as you, you know, leave my name in there, I was perfectly fine with that. Please copy, share and enjoy. You know, that was the motto back at the time. And I don't know exactly what's the title of the book, but there is a book about developing Linux kernel mode drivers. And there is a paragraph in there that points to an I2C FAQ written by Vincent Hippie. And you can find it under ping 00. That book is still being published today. Brilliant. So, yeah, so I was hired on spot. So I moved to a different location, which was in a different town. And my, you know, you have to understand that, you know, coming from, I mean, my parents were just, you know, day laborers or, you know, not white collar or whatever you want to call that. I mean, they had just had a regular paid by the hour job, not a monthly salary. So you kind of look like, you kind of look at that engineering community as like, ooh, these are going to be like a bunch of uber nerds, almost godlike sitting on a throne. I got to dress up when I walk in there. So I got a, you know, got a nice suit, got a tie, a white shirt. Oh, jeez. I walk in there and talk about a cold shower. I mean, you, you, you, you pass that reception area, you walk past that computer room, which is a big glass aquarium full of these VACs and some mainframe machines. You look, your eyes are popping out of your sockets. I mean, you've never seen some, this is the stuff that you only see in movies like War Games. You know, it's like, wow, look at big honking mainframe running the spice simulators and layout tools. And then you get to meet your first colleagues and there's a guy sitting in a sleeveless shirt. We're wearing shorts with his leg in a, in a plastic cast on his leg, chewing on a ballpoint pen with his keyboard in his lap. And you're like, oh crap, I overdressed for this one.

Chris Gammell: You only do it one day. It's fine. Everybody does that the first day.

Free Electron: So, yeah, that was my introduction to the, to the high tech world. It's like, I mean, it, it, it, it really is like that. Okay. Probably not in all companies and, and probably maybe in the seventies and the eighties, you always see these pictures from national semiconductor where everybody is, is, you know, those are the black, but these are the official pictures when the camera is not around that. It wasn't the way people were dressing, you know? I mean, it was, it, it is a much more relaxed environment. I mean, today they call it business casual. Yeah. You don't, so you don't show up in shorts if you have a customer meeting, but the tie is not really mandatory. I mean, it's, it's, you're there to do a job and, and, and it's, you know, unless you are a hard, hardline German, a sauerkraut swilling, whatever, maybe. But even those are, are very rare these days because I have a lot of German colleagues and they don't do that either. So, yeah, that, that, that, that framework, um, that, you know, that, that, that idea that people have about that stiff environment. Um, it's not really there and, and, and it's not only Silicon Valley, it's everywhere in the world. It's, it's, it's much more relaxed than what people believe it is.

Chris Gammell: So anyway, I imagine it back then was kind of like how it is these days. Cause I mean, like, like, like a lot of software people are these days where you have like these rockstar software people. Oh yeah. I imagine that they treated it like that because where are you going to get people otherwise?

Free Electron: We had, I will throw in a couple, a little bit later, a couple of anecdotes about rockstar designers. And I got a couple of real stinkers. I mean, stuff that you, you cannot, I mean, this is Bob P's quality stuff. Okay. All right. Excellent. That's what we wanted to do. So anyway, yeah. So I ended up in the lab and I'm introduced to one of the designers that is working on a chip for a vacuum cleaner, Hoover. And Hoover had the, Hoover had the idea that they're going to make a vacuum cleaner. And they're going to put a little microphone in the incoming and the air intake and basically listen to the, the cluttering noise that dust makes as it's traveling in a plastic pipe. And they're going to, they're going to do a peak, a peak detector, integrate that. And they're going to drive a silicon controlled rectifier and control the motor speed relative to the amount of dust coming in.

Dave Jones: Why not just let the motor go 100% all the time?

Chris Gammell: So they're making a physical oscillator.

Free Electron: Well, they wanted to have a product that stood out from the competition. It's a wank in other words. And they thought, bingo. But they thought housewives are going to go for that. Techno geek husbands are going to go for that because they'll be vacuuming. And it'll be, oh, it's clean here. Let's move on somewhere else. So 1993 arbitrary waveform generators don't exist. Well, they probably did exist, but we didn't have one. So we had a bucket of calibrated dust and a piece of carpet. Bucket of calibrated dust. How does one calibrate dust?

Chris Gammell: Is it like a grain size?

Free Electron: What they did is they figured out that dust particles, they, they did, they run a spectrum analysis, probably one of those dynamic signal analysis. And they kind of figured out where the peaks are in the spectrum for your typical household dust, like, you know, breadcrumbs and whatever being sand and dirt coming in. And they made a comb filter that was, you know, peaking at those frequencies. And then they would integrate that. And that would be a measure for the amount of crap being sucked into the vacuum cleaner. So they, they did what they did is they, they had filtered some sand, I believe. And they, they had very fine sand and then had coarse sand. And this was a blend of different kinds of sand that sort of exhibited that noise spectrum. So it was equivalent to what you would find in a household environment. So we had a little bit of the lab that was, had a few square meters of carpet. And every morning I had the duty to take my little scoop and scatter some sand in the, in the calibrated box, put the vacuum cleaner ready for the lead designer because he was going to come in and do some measurement. And he'd sit there looking at his oscilloscope, you know, with probe needles on his die, one hand on the handle of the vacuum cleaner, vacuuming his carpet, looking at his oscilloscope at me. I mean, that, that is my very first project I ever worked on. And he's going, shut up guys, I'm measuring dirt.

Chris Gammell: Did you guys, serious business here. I was going to say, did you ever make jokes? You're like, Hey Phil, come over and measure my apartment later. It's really dirty.

Free Electron: Bring your DSA and your, and your, and your, and your, and your, your, your, your bucket of vacuum, bucket of calibrated dust. Oh, that's great. And then we had, we did some, we did some car electronics, central locking system, speed control for waper motors, speed control for air conditioning systems, sensor pickup heads for anti-lock braking. And then come 1995, our company, which was, that company was called Metic at the time, was absorbed. We were kind of a subsidiary of Alcatel, but in 1995, we were completely absorbed into Alcatel. And another division of Alcatel, some really clever people, we call them the team of seven. I mean, there were seven people in that team. They invented DSL. So what is today, everybody is using ADSL or VDSL or one derivative of that. So the basic signal theory had been known for a long time since the fifties or the sixties, but nobody could do it because you just didn't have the compute power to actually make the quadrature modulation and modulate 64 bits and do all the sine cosine transforms and create that signal with ESP. It wasn't heard of. So these guys figured it out and they had a, I believe it was an I-960 CPU from Intel, which is a big, massively.

Dave Jones: It was big in its day. It was fast.

Free Electron: One of the big beasts, you know. Yeah, it is. It was a 32-bit risk machine that had so much horsepower that, I mean, the actual, the Tomahawk missile flies real-time under control over under a signal. I-960 is guiding that thing to its target, reading topographical maps and reading the infrared sensor. And it's flying a supersonic missile. I mean, a single CPU clock that 16 megahertz could do that. I mean, that thing is, I mean, it has horsepower. So they had this massive board with an I-960 and a bunch of ASICs and whatever, and it was a big 19-inch rack. And we could actually send across a twisted pair. We could, you know, do about 6 megabit a second, which was like, holy moly, we got 56k modems, and these guys do 6 megabit a second. Wow, this is amazing.

Chris Gammell: I can download a web file.

Free Electron: Yeah, and most of the websites those days where you would hand-code HTML. Of course, yeah. Animated GIFs were about as far as you could go.

Chris Gammell: Yeah, dancing babies.

Free Electron: There wasn't any HTML5 or Flash or Shockwave or PDF didn't exist. Hell, that came two years later.

Dave Jones: And no online video, for goodness sake.

Free Electron: Oh, no. So they had that crappy Intel coder that could do 320 by 200 whatever in 60-bit version back then. Oh, yeah. Like it was 160 by 16. No. Windows 95. No. This was the one that shipped Windows 3.8. I think it was called INDEO. I-N-D-E-O.

Free Electron: Yeah. INDEO video. Intel video. Yep, that's right. That was 320-240. Wow. Oh, yeah. That preceded QuickTime and anything else out there. So, anyway. So we decided 1995, we're going to start developing this stuff. So I was rolled into that team. I mean, at that time, I had a rebuild. We had moved to a different building. I was appointed the architect. So I did all the, figured out where all the plumbing and the vacuum and the compressed gases and the power out let them to be. And then I designed the lab benches because everybody was always bitching about how lab benches are never deep enough. So I had, I went to a company that could do, you know, aluminum or metal chassis welding. And I had them make custom depth lab benches. You could put a full depth spectrum analyzer on your bench and still, you know, have a work area where you could put another analyzer in front of you. Nice. So it was double, double the depth. And I bought some of that wacky fiberglass material that they use in hospitals to make the bed boards out of, which is virtually, the name eludes me now. It's virtually indestructible. So we had custom benches built and power put in and I put a GPS receiver in. So we had a stable 10 megahertz clock because we were going to do, I mean, that whole DSL thing is timing critical. So you want to have all your signal generators. The inspector monalizes not only be timed, but also phase aligned. So we had a distribution app with matched length coax cables going all the way throughout the lab. So everything was synced in time and set that up, rolled into DSL. And then the thing is that, well, we're going to, first thing we're going to do is we're going to develop an A to D converter and a D to A converter. That's going to be clocked at 17 point something megahertz. And it's going to do 16 bit. Holy moly. Never, nobody ever attempted that. And then, well, how the hell are we going to test that? I mean, where are we going to get the data from to feed the damn thing? And then not only the data, but that communication pipe was a time interleaved four bit wide bus. So you had to time interleave the data. So you had to clock it in not at 17 point whatever. But at 63 point whatever, it had to roll in and nibbles. And there was not only data, but there was a four bit opcode that would tell you if it was a data packet or if it was like a change the output gain. Because we're seeing signal noise degradation here. Change the output gain of the transmitter a little bit. So you had some other stuff in there. So we didn't really know how to tackle that one. It was 1995. And I figured, you know what? I'm going to make myself an ISA board, stick it into one of these PCs here, hook up some really fast memory, and I'm going to make a programmable pattern generator. Excellent. And I started drawing.

Chris Gammell: Just for like test codes and stuff to throw out the A to D basically?

Free Electron: Basically the same thing. So it could do both things. It could generate data and spit it out in that format to the D to A. Or it could sample the output from the A to D, store it in a – we had a one megabyte. Wow, one megabyte. Or one mega sample buffer memory. And that could be dumped to the PC and you could visualize that in MATLAB. I could dump it to a file in the MATLAB or MATCAD. I forgot which one it was. You could visualize that. And then you could do a Fourier transform and look at the spectrum and see if there's any spurious signals in your capture. All that kind of wizardry. The problem was I started designing that with, you know, 74 TTL chips and that thing just ballooned. And I'm like, holy, well, this ain't going anywhere. And I had heard about a company called Xilinx that had like this little configurable chip, which is called, you know, an FPGA. And I thought, wow, it would be cool if I could do that. And that's how I learned Verilog. Nice. So 1995, learned Verilog from one of our digital designers taught me that because they were – those guys were already using it in silicon design. But at the time for FPGAs, Verilog was new because everybody at that time was still either doing schematic entry, which was done in some wacky tool called – oh, what was the name of that thing?

Dave Jones: But even today you can still do schematic entry for FPGAs if you really want to.

Free Electron: You can – if you really want to, you can – and it is actually still – even on ASIC design, it's still done in – oh, we need an additional driver in here. We'll just plunk in a salad that's done in schematic level. But, yeah, they would – you know, schematic entry was your one way or they had these proprietary languages like AHDL and Couple, which Couple was from logic devices and AHDL was a data.io came up with that one. But those weren't real synthesis languages. These were just a Boolean notation that you would do. There was some somewhat synthesizable, but you really – the big bulk of the work had to be done by the designer. You know, you had to understand what your primitives were and it could optimize and do the logic expansion and minimization, but that's as far as they would go. So, since our digital designers had been using Verilog and that Silenx tool, which wasn't called ISE at the time. I forgot what it was called at the time. Exemplar logic or something like that. So, I learned, you know, basic good old-fashioned Verilog and I wrote them because that sampler was basically a big loop with a start vector and a stop vector and a jump vector. And it would run and then the logic to the PC, it's an ISA interface. You know, it's a simple memory bus with a read and a write strobe and an address latch signal. So, it was very easy to describe it in Verilog. And so, that's how I rolled into FPGAs in 1995. And then I decided that the Silenx tools, I didn't really like that. I went to some convention the year after where Altera gave a demonstration of what was then called Max Plus. Yes. Which is now Quartus. And that was way… It has been ever since. It has been ever since.

Chris Gammell: Was that Quartus 1? Because I've never heard of Quartus 1.

Free Electron: Well, Quartus… It was… Actually, yes. You could call… Yeah, because they released Quartus always as Quartus 2. Yes.

Chris Gammell: It's been Quartus 2 for as long as I've ever heard of it. Exactly.

Free Electron: There was no Quartus 1. Quartus 1 was called Max Plus. And that was much nicer because that was a tool that, first of all, didn't require SunWorkstation to run it on because the Silenx tools were… If you didn't have a Sun at that time, you were stuck. PCs were unheard of. But Altera had their tool running on a PC under Windows 3.0. And this was actually very comfortable. Look, I can use a simple, cheap PC and borrow my schematic and do my little Verilog thingy. And I have my little programmer straight hooked up and I can sit here on a bench. I don't have to go to a workstation, an office space where there was one of these expensive SunWorkstations with those big honking 21-inch monitors. I can sit here in the lab with my dinky little PC and breadboard the damn thing. I mean, if I would walk into the design center with my soldering iron, I would get kicked out. You know, that was like, you don't do that because the fumes are going to get into the fans of the SunWorkstations.

Chris Gammell: We paid a lot for those.

Free Electron: Oh, no, those were pricey. That was… But it was that or nothing at the time. It was a SunWorkstation hooked up to a VAX, a digital VAX VMS system as the backbone. That was it. Everybody was on that. So, anyway, we did that. And, you know, ADSL started rolling. And now come the really… The Bob Pease anecdotes are starting to come. We're getting into that territory. Sorry. So, we had hired a person. I'm not going to disclose his name. But he's well known in the industry. Who is… I wonder if he can. He's a real wizard. He is a… We had an argument or a discussion once over lunchtime. And there were some people, you know, joking about, oh, well, these engineers, it's always the same time. And the same thing with engineers, you know, how nerdy they are. And the guy goes like… He had a very dry sense of humor. He goes like… I'm not an engineer. I am a scientist. And we go like, wait a minute. What do you mean that… He says, no, I don't have an engineering degree. I am a scientist. And we go like, well, I am a astrophysicist.

Speaker ?: What?

Free Electron: Excuse me. And he goes, yeah, I have… My specialties are string theory and collapsing black holes.

Chris Gammell: Oh, this isn't going to end good.

Free Electron: You mean you were dilly-dallying around with quarks and quasars? What the hell are you doing on an ADSL system design? Well, I was hired because of my mathematics background. And I did some work for ESA, which is the European Space Agency. He made an infrared spectrometer or whatever to do some deep field, something, something. And that thing has flown on the Mir space station. And later on, he did another project for NASA, which is a project that is actually up there. It is still there on the International Space Station today. And he said, well, I did all… I had these postulated, these theories during my doctorate work. And I was hired. He said, well, this is an interesting theory, but we would like to see it proven. Because, you know, these guys come up with wacky theories, but then it's up to the engineers to figure out how to prove it. But the engineers were basically scratching their hair going like, holy moly, we've got no clue what this guy is on about. And he said, that's how I rolled into electronics. I had to learn what a transistor was and, you know, do all this design work. And come up with a piece of electronics that could actually sense this radiation or, you know, prove that it was there. And that's how the guy rolled into it. But this guy had a totally different grasp of electronics. I mean, electronics to him was, you know, he didn't want to hear about op-amps and stuff like that. For him, this was…

Chris Gammell: Oh, he wouldn't go to first principles and stuff?

Free Electron: He would solve…

Chris Gammell: Yeah.

Free Electron: Everything was blackboards full of Maxwell equations. Wow. That guy was…

Chris Gammell: That makes for a long weekend.

Dave Jones: And here's how you calculate a lead dropper, folks. Two hours later.

Chris Gammell: Yeah, right?

Free Electron: Yeah. Let's start with a principle of how many photons are going to be emitted by the LED. Oh, God.

Chris Gammell: I had that happen in an interview once where the co-op started diving into that. We were like, you know, we're trying to get balancing between, like, the charge and two capacitors. And he starts diving into, like, the low-level stuff. And I'm like, dude, dude, chill out. Just, like, tell us the answer.

Free Electron: So, here's… He's still a good friend of mine. He's retired now. He's still… He's from France. He's still a good friend of mine. And so, he comes to me one day. This is anecdote number one. He comes to me at one point. And he says, you've got to explain me something. We have this ARM microcontroller in our new chip. I mean, we had, you know, the analog friend by then was designed. We were starting to integrate it. We had… I made a first circuit board with an ARM CPU and a memory and an eSquare problem. I wrote the bootloader so they could, through a serial port, upload their code and debug it. And he comes to me and he said, yeah, I really want to start exploring this microcontroller to do some of the testing on the filters that he was supposed… He was in charge for all the signal processing. And he was specializing in all the digital filters. And what were those things called? Convolvers or convolutional encoders or something like that. Something wacky. Convolvers or something like that. And he says, can you explain to me in a few hours how I go about programming this microcontroller? And I asked Bernard. Okay, well, there goes his first name. I asked Bernard, have you ever done any kind of program before? And he looks at me and says, no. Okay, this is going to be hard.

Chris Gammell: Let's start with Hello World.

Free Electron: And so I start, you know, we had a C compiler for him. He says, no, no, no, no, no, no, no. I don't want to hear about this C stuff. I've been reading this data sheet here. And I've kind of memorized all these mnemonics for, you know, shift left and whatever.

Chris Gammell: Oh, for like assembly? Yes.

Free Electron: So he wanted to do a Dahlia and assembly because he said, I don't trust, I don't know what that compiler is going to make if I write high-level code. This is going to be time-critical, high-speed optimized code doing signal processing. I need to be in control of the CPU.

Chris Gammell: That sounds like a physicist.

Free Electron: That's what, I mean, that's what those guys do. They really need to know at a fundamental level what is going on. So he had sort of worked out the algorithm, how he was going to encode it using the available instructions. So I showed him how the debugger worked and how he could open a hexadecimal viewer and an assembly window. And he would sit there and start typing assembly mnemonics and, you know, run it through the assembler, spit it on a hex file, upload it, run it, hit the debugger, hit a memory dump. And he'd sit there reading hexadecimal code. And I'm like, Bernard, we have a disassembler, you know. He says, no, no, it's fine. I know what F673D2, I know what that means. I can read the Matrix. I can read the shift. Yeah, exactly. That's a shift left through the caddy with the flag set. And so that was my first interaction with one of the gray beards. You know, it's like, holy moly.

Chris Gammell: Yeah, Sheldon Cooper right there.

Free Electron: And at number two, a couple of months later, I'm walking into the lab and Bernard sits there, comes over to me, says, do we have any of these whiteboards? I mean, he had, every bench was set up so they had a big whiteboard on a wall. And he says, I need a couple more whiteboards because it ain't quite large enough. I always end up wiping half of it out when I'm in the middle of a calculation. So I wanted a couple of whiteboards. You know, whiteboards on wheels so he could move them around from setup to setup. He'd sit there all morning scribbling one side of his whiteboard, flipping it over, filling the other end, filling the one on the end of the wall, grabbing another mobile one, scribbling that full width. I mean, these are matrix operations because these are digital filters. So you have all these seed coefficients to start the filter. And then as the filter trains itself, all these coefficients change all the time. So this is an iterative kind of, I mean, a number of generations later, you have no idea what these things are going to do. Or at least the normal human has no clue. Bernard did. We didn't quite grasp it, but he did. So he sat there, you know, scribbled it full, looked at it and he goes like, yeah, I think I've almost got it right. So he starts, fires up. It wasn't MATLAB. It was MATCAD on a PC. Starts punching in all these equations, filter coefficients, fires up the computer and he says, eh, it's about two o'clock in the afternoon. I think it's going to be running about for, you know, probably a day or two. Give me a phone call when it's done. Keep an eye on it and see. The screen is going to change. There's going to be a table with numbers. Give me a call when it's done. So he went back to the floor above, which was, he didn't want to come down all the time to keep an eye on that computer. So he went to the floor above and went off and did something else. I call him when it was done. He comes down. He looks at this big honking. I don't know, it's a table with like probably like a 40 by 40 table of numbers. He looks at it. He looks at his hand drawn, you know, his papers and he looks at his whiteboard and he goes like, hmm, something is off here. And in the middle of that equation, he just, with the palm of his hands, he wipes out a two, which is, you know, a power two. And he puts a three in the place. And I think that that should do it.

Chris Gammell: So he's like adding a pinch of salt to a recipe or something, right?

Free Electron: He changes that one number in MATLAB. The thing runs for another two days and now the numbers line up. And I'm like, holy shit.

Chris Gammell: Good lord.

Free Electron: So that was situation number two. Situation number three is that up till today, there is in the DSL algorithmic code, so the signal processing code, there is today in the C source code, a block of inline assembly written by Bernard with big exclamation marks around it. And it says, do not touch, we have no clue how this works, but leave it in.

Chris Gammell: Every good system has that. Come on. Nobody. It's like using boot code.

Free Electron: No, this is the algorithmic code. Actually, so the algorithm is very well understood. Okay. So the equations behind them, plenty of people know exactly how it works, but nobody up till today has been able to code it in a higher level language with the outcome being so compact and efficient and time aligned as the handcrafted assembly from Bernard. It's like big exclamations. Don't touch this stuff. Do not. If there is a mop in line, do not remove that. No operand. It needs to be there. And it's been that way for 18 years, right? And it's been.

Chris Gammell: Right. And make sure your clock's accurate too.

Free Electron: They call that the golden. Don't touch anything. It is the golden library. Do not touch this thing. Yes. Yeah. So this is the same guy that got away with situations like the big boss coming in, you know, arms waving, yelling at people. What is this going to be done? How come this isn't done? Why is this so much delay? He would just stare at his computer screen doing whatever it is he was doing. And when the boss is done ranting and raving, he wouldn't even look up. He would simply ask the boss, are you done now? Go take a leak and leave me alone. I got work to do. It's like big boss would walk out of the lab without saying a word. It's like we try that and blink and you're at the front door escorted by security. Yeah. Yeah. You're at your box. Yeah. Yeah. He's the same guy that got away with shoving a computer from the third floor through the window. Nice. That is a real Bob P story. So what happened is that, I mean, 1995, we had 286 computers with a math coprocessor. He had been begging IT to get a new computer. I want a new computer. This is not fair. The boss's secretary and the admins, they have gotten 386s and 486s with mathematical coprocessors. And they have to wait, you know, two hours for a calculation that could have been sped up. No, Bernard, you can't do that. Your computer is only two years old. It's not, you know, it's not written off yet. Computers have, at that time, five year write-off time. Yes, yes. You can only get a new one. You can only get a new one if it's broken. Ah, okay. So he opened the window. He just shoved it through the window. Well, he calls IT with a very dry comment and says, my computer crashed. And the IT guy goes, where is it? On the pavement out front, you can go and pick it up.

Dave Jones: And with that, the entire DSL modem market advances two years.

Free Electron: Exactly. That was the week we all got brand new machines.

Dave Jones: That's great. So that code still exists in?

Free Electron: That, probably, I would, honestly, probably not. Okay? Because stuff has evolved. I mean, we are, I'm talking 1995. We're 2000. We're 18 years down the road. They've probably gotten, okay, you have to keep in mind, at that time, everything needed to be written so tightly to the hardware. Because we only had 16 megahertz CPUs. Now we have dual-core ARM CPUs doing the video. Now you can probably write this in a high-level language. And yeah, we've got plenty of cycles to do this. Yeah. But at the time, it couldn't be done. So we're moving on to the year 2000 now. ADSL is, you know, we've got the big telecoms boom of the 2000s starting. And they're like, well, we're going to set up a design center in Raleigh, North Carolina. Yeah. Because we...

Chris Gammell: So wait, you were in Alcatel?

Free Electron: I was in Alcatel.

Chris Gammell: At 2000. Wow. In 2000.

Free Electron: In Belgium.

Chris Gammell: How were the benefits then? Oh, yeah. Were they pretty good? Because I imagined they'd be really good then.

Free Electron: That was the golden time. Right.

Chris Gammell: Yeah. Yeah. I remember my dad told me to buy stock in them in 2000. They had like... Yeah, that sounds great. No, not great. Two years later.

Free Electron: They have 216,000 employees in the peak. Wow. Yeah. Oh, Alcatel was huge. Wow. They did everything in-house from etching their own boards to writing their own board layout software. They really did that. Alcatel was a powerhouse. They had subsidiaries that made capacitors that made resistors. They had their own semiconductor files. That was a... We got vertical industries that have vertical industries. We were vertical before the term vertical was invented.

Chris Gammell: That's right.

Free Electron: So, the year 2000, here comes the demand from the likes of IBM and Compaq that are Windows 95, 98, Windows ME. So, we're in the timeframe now where everybody has a computer with Windows 98 and Windows ME on it. And everybody wants to have, you know, jump on the high-speed bandwagon. And Compaq, this was before the merger with HP or even Digital. It was way before that. They had the idea to start selling computers with a built-in ADSL modem. So, this was going to be a PCI card. And they had set up shop in Raleigh, North Carolina, which was called Research Triangle Park. So, Alcatel said, well, we're going to open up a design center over there. We have already an Alcatel site that makes the business end of ADSL, meaning the central office, the big honking machines that have the line cards serving 50,000 homes. Right. But now we can, you know, put a secondary search center that's going to do the customer side, the little modem, whether it's an Ethernet, USB, or PCI plug-in. And since I, you know, I had done all this, you know, I had introduced the circuit board layout tools in 1993 when I joined that design center. Like I said, we were wire wrapping boards. The first thing I introduced was, guys, do you guys know what ammonium persulfate is? That's an action. I can make boards with that shit. So, I was, you know, on a little plastic tub. I was hand-hatching boards. And the designers loved it. Oh, wow, this is so much better than these stupid wired boards that always were full of bad contacts. And then I had gone and begged the boss to buy me a piece of software. So, this was 93, which at that point was distributed on a single 712K floppy and ran under DOSA. It was called Autotracks 1.21, made by a little-known Australian company called Protel PTY. Yes, in Tasmania.

Chris Gammell: I've heard of that. Yeah.

Free Electron: So, that is how long I've been using that tool. So, for the people that know me on the forum, yes, I am an Altium aficionado because I've got history dating back to 93 with that thing.

Dave Jones: 93? I was using it in 89, I think, in the late 80s. Oh, wow.

Free Electron: That seems before me.

Dave Jones: It extends before that, yep.

Chris Gammell: Yeah, it didn't escape Australia until later. Yeah, I think. Yep.

Free Electron: But, I mean, it was great. You had a schematic tool. You had a PCB tool. You could send it in that list. The library was all through whole at the time. But everything I wanted was there. And, you know, it could run. It could actually exploit EGA mode in 1024 by 768. 768.

Dave Jones: Yes.

Free Electron: Which was unheard of, you know.

Dave Jones: Oh, I think I was using Hercules mode, you know. Oh, 720 by. Yeah, 720 by 384, is it? Yeah. Was it? Or something like that? I don't know. That was the duck's guts at one stage. 720 by 384 or something, yeah. And then it had Visa extensions to it. So you could get up to 1024 by, you know, 768.

Free Electron: Whoa. And then you had, what was that? They had separate plotting tool and the printing tool. Yes, yes. And that was Traxplot, yes. Traxplot, yeah. Traxplot and Traxprint. So I would set up because we had a big plotter to do, you know, they were plotting the chip layouts on that machine. We had this big, you know, 42-inch wide roll-fed huge ink machine. And I could spit HPGL to that thing and I would plot it on vellum, which was that really thick, glossy paper. And that machine was actually plotting with Chinese ink, the really black, you know, art, yeah. Calligraphy ink or something. That's what was being used. Wow. And I would expose that. I had figured out what the doctors were using to do to make x-rays. You could go to a photographer and buy that. The BASF was selling, it was called light x-film, which is what they do to make chest x-rays. And I bought a box of those, you know, the company bought a box of those films and I would transfer my vellum plot onto a transparent plot with the same ultraviolet exposure unit. I mean, that film was light sensitive enough. I only had exposed it for a few seconds and develop it and you had a perfect, crisp, crystal clear foil with perfect black traces and I would expose my boards and my boards always worked, you know. So I had done all that stuff and they said, you know what, we're going to open that design center over there. We need somebody who has in his life built a lab. You've done that. You've moved the building. Who can do the board layout and get the thing rolling. And you want to go to the US? I'm like, oh yeah. So 2000, October 2000, I moved to Raleigh, North Carolina and I get introduced to a bunch of really nutty American people that mean that are, there are no limits to what can be done. You know, one of the first thing my boss says, you know, this is really cool. So we can, you know, by the time, at that time we already, we were at ProTel 99IC was already out. So that's where we were. And we would have boards made somewhere, but we would always have to go to some subcontractor to have them built. And the boss said, you know, it would be really cool if we would have our own assembly line. So I investigated, found on the very young eBay at the time, somebody who was selling a Juki, I forgot what the model number was. Some used a pick and place machine, including 80 feed lines for tape and reel and a couple of stick feeders and a tray feeder. So we spent $40,000 buying our own pick and place robot. We spent oodles of cash at mouse and digi-key getting every possible resistor and capacitor value, full reels, having them in stock. And we hired a person to run the machine. And we made so many boards on that machine in our lab that after the first four months, the machine was paid for. I mean, we would have spent so much money on subcontractors and labor alone. And we had the damn thing. We had a little pizza reflow oven and a manual screen printer. And we could do the typical flow as to the point that we would tape out. Well, tape out is a terminology that comes from the chip world. The 70s?

Dave Jones: Okay. Well, they still call it tape out because when your design is done, you write it to a tape.

Free Electron: You send it to the mask maker. So even when we were doing a board, we would call it tape out for the board. We would send the board Gerber data out on a Friday evening. We would have the board spun on a three-day turnaround, which would mean that we would get them back. There's a lot of board companies in the U.S. that work over the weekend. So by the Tuesday morning, they would come in. Okay? So tape out Gerber data Friday evening. Monday morning, place the order with digi-key or mouse or whatever is missing or have that shipped overnight. Tuesday afternoon, the bare boards come in. The parts are here because they came overnight with digi-key. At the same time, we had the stencil made somewhere. That came in and that was only a shop around the corner who would do it in a few hours. So Tuesday morning was spent me converting the pick-and-place file that came out of Protel, loading it into the machine. Rich, which was the technician, would load up the reels on the machine and we would cross-correlate the positions on the machine with the pick-and-place data. And by the time the bare boards came in, the stencil was bolted in the frame for the screen print that the pick-and-place machine was ready to run. And by Tuesday evening, we would have five or ten boards populated and nobody could beat us to that. I mean, you'd tape out on a Friday and Tuesday evening, you'd have ten boards ready for design people to test that thing. So that was an extremely interesting adventure I embarked on. And then the bubble burst. So the big telecom scroll. So everybody became flunky basically overnight. And my Altium shares plummeted. Oh, disaster. Companies like Broadcom, Qualcomm, EMC, Virata, Globespan, Coppercom, they virtually all went basically bankrupt overnight.

Chris Gammell: Yep. Anyone with fiber in their 10K report anywhere, mentioned anywhere, they're gone.

Free Electron: So that was the same deal. The same thing was going on for Alcatel. And Alcatel, at that point, like I said, there were like 216,000 people. And they shed over the next few years so many people that they ended up being down to about 40,000.

Dave Jones: Wow.

Free Electron: So that is the amount of – that is the amount of – there was – of course, there was a bunch of, you know, bare layoffs. But there was also people retiring. And then there was – they spun off a whole bunch of the division. They sold some of the stuff to Lucent. They sold some of the stuff to – like their circuit board division became an independent company. So they spun off a whole bunch of divisions. Yeah. That's true. That is actually what I was looking – I was trying to find the terminology for it. But that's what you would call it.

Chris Gammell: That's good. It means you'll never be in management. That's all that matters. I don't do complicated words.

Free Electron: So here is this, you know, Alcatel microelectronics department. That is still worth some money because these guys made ADSL chips that are still active. They're selling still pretty well in the market. But we, you know, as Alcatel, we want to focus and become a systems seller. And we don't want to really be, you know, messing around with the sand, the low end part of it or the base part of it. And they figured they could get some money for us. And they basically put us up a lot, stock and barrel, the two wafer flags in Belgium, the design teams all around the world. And Siemens was going to buy us out because Siemens is a big telecom giant that, you know, had ISDN. Well, they're an everything giant.

Chris Gammell: They're a German giant. Yeah.

Free Electron: And so what happened is that these guys put in a bid and they were going to – I forgot what it was. It was $600 million or something like that. So Alcatel was like, oh, yay, that covers the loss we had in the last few quarters. That would be – and then what happened is that somebody dropped the bomb. There was this Franco-Italian company called Esteem Microelectronics that stepped in and said, excuse us, guys, but there's a little thingy here. If you're going to be selling that – so they went to the Alcatel management and said, if you're going to be selling that, I mean, we were okay with you guys doing that. But we seem to vaguely remember that when you guys set up shop and build your own wave fabs, you came knocking on our doors for the chemistry and the process to run in the fab. We're sort of okay with you selling the buildings and the people, but you ain't going to be selling chemistry because that's ours.

Chris Gammell: Oh, yeah.

Free Electron: So the $600 million became poof, vaporized overnight because if you ain't got chemistry, all you have is some scrap equipment basically.

Chris Gammell: If you don't get Walter White, you just get the RV, right?

Free Electron: Yeah. So – and then what happened is Este said, well, look, we know that Siemens bid whatever. We're going to play it fair. We can always use an additional design team, and we're kind of interested in this ADSL technology, and we are already supplying 50% of your – Alcatel, like many other companies, everything needed to be dual source. So Este was supplying 50% of the chips, and so they said, you know what? We're going to take on that group, and we'll give you – I don't know what the exact number was, but – Yeah, $600 million? No, it wasn't half. It was between half and the full price, and they said, you know, we're going to take it on, and we're going to keep on being – you know, because we are a supplier to you, and we want to keep up a – you know, keep a business relationship. And we're going to take that on, and they signed the paperwork, and everything was hunky-dory. And three days later, Steve turned around and said, by the way, that wafer fab over there in Belgium, that's only running six-inch wafers. That is like all old technology. We don't really have an interest in keeping that wafer fab, but there's this little American company called AMI, which is called American Microsystems, not to be confused with Austrian Microsystems.

Chris Gammell: Austrian Microsystems, that's right, yeah.

Free Electron: Different company. Both are called AMI, but they're totally different companies. Is the American one?

Chris Gammell: Is that the one in Mississippi? Is that the one I'm thinking of?

Free Electron: That's the one in Idaho.

Chris Gammell: Idaho, okay.

Free Electron: Boise, Idaho, I believe. So the American company, at a certain point, they added an S to their name. They became AMIS to stop that endless confusion with the Austrian Microsystems, whatever. Anyway, so they said, well, these guys basically do a lot of work for the military, and these guys are still, you know, messing around with four-inch wafers. So for them, ooh, six-inch wafers is an upgrade. You know? 0.2 micron? Sign me up. What? Oh, you're jumping the gun, not 0.2.5 micron. Oh, I'm sorry.

Chris Gammell: Super modern.

Free Electron: That was, I mean, those guys were still dealing.

Chris Gammell: Hey, man, an analog, that still flies for me.

Free Electron: Oh, yeah. Oh, yeah. So they sold off the wafer fab to AMIS and kept the design team. So that's how me and a bunch of my design people and my design colleagues, we ended up with where I am today and have been there ever since. And then a whole bunch of, you know, all the people that used to be in the wafer fab and all those people ended up with AMIS. Yes, that a couple of years ago, the lady that was the big CEO cashed in on her. That was, so it was not a public company, but she cashed in by selling all her stock in the company to on semiconductor. And now that company is, you know, absorbed and part of on semiconductor. And they're still doing good. I mean, those way, the old wafer, the fab one is closed, but fab two, they actually, fab two was all, when I was there, was only half full with equipment. And there was what they called a level two, whenever, you know, business picks up a bit of bio-more equipment.

Chris Gammell: Surplus capacity.

Free Electron: Yeah. But under Alcatel, it never happened. That room was always vacant. I mean, the air conditioning was on it. It was clean room. There was just nothing in it. So now, actually, being on semiconductor, that room has been completely filled up. And they're doing very well. I mean, they're doing a lot of automotive parts. And they have little embedded microcontrollers that do things like airbag control and sensors, you know, signal conditioning on the sensor, basically. Lin bus, can bus kind of stuff. Yeah. Lin transceivers, can transceivers, smart sensor signal conditioning with a microcontroller on board. The stupid ones they put in your tires that never work.

Chris Gammell: Yeah. Stuff like that.

Free Electron: You know, that's the kind of stuff that they do. So they're doing very well.

Chris Gammell: Well, at least most people don't realize that a lot of six-inch fabs, I mean, like, a lot of analog wafers are doing that. I mean, it's very few fabs. Very few analog companies are in 12-inch fabs. I think TI is the only one that I know about. But most of them are eight or six or four-inch wafers.

Free Electron: For pure analog, yeah, the big wafers only make sense if you've got large dye surfaces in your given technology. And the way to – I mean, we are, you know, we're at 18-nanometer technology. The only way to get a large dye surface is to make a multi-billion transistor device. And then you will end up with a chip that is still, you know, five-millimeter by five-millimeter square. And then it does make sense to run these big honking wafers. If you're into analog, you can't do analog in 60-nanometer. The leakage of the darn technology is so high that all you can do is discriminate between the one and the zero. Yeah, exactly. Because there's so much leakage in between that. Schmidt triggers are about the best circuit you can build in that technology.

Chris Gammell: Well, they're figuring it out now. But, yeah, it's going to take a while. I mean, it's still, yeah. It's like 180-nanometers standard analog these days, I think, or something. Something like that.

Free Electron: Oh. Maybe higher, 250. Yeah, but that is 250. I would say 250. 180-nanometer is if you got like a – Like a mixed-mode BCD kind of process. Yeah, BCD kind of process. And then if you're going to look at it, yeah, technically it's going to be, you know, 180-nanometer. But you'll find five transistors. Yeah, right. Or five that are actually 180 and everything else is like 750.

Chris Gammell: Right, right. Yeah, yeah, yeah. You're doing trench rats and huge shit. Yeah. It's like –

Free Electron: Hey, I mean, this is another one of these things that a lot of people are confused about. I mean, people hear about, oh, they're making transistors in 18-nanometer. Yeah, sure, those are the real transistors deep buried inside that are flicking some bits around. But once you hit that IOP and you have to start driving that 10-nanometer, it ain't going to be 18-nanometer anymore. You know, we're in the micrometers now. Right, yeah. And especially like what I'm doing, I mean, I'm in the hardest business. If you look at those, the output that the H-bridge driving that three-phase motor, that thing can bump. That's three-amp current. Hell, our chip is, you know, three-millimeter by three-millimeter and two-thirds of the surface is just the six-power monster that's driving the motor.

Chris Gammell: Right.

Free Electron: Yeah, yeah, yeah, it is BCD and it is made in, you know, 120 nanometer, but not the output transistors. So that is – so yeah, for analog, it doesn't make sense to do it. On the other hand, a lot of companies that are, you know, traditional analog makers like Maximum Linear Technologies, at one point they used to have their own fabs, but they have gotten rid of their fabs because it is cheaper for them to just charter wafers with a big – TSMC, all that crap. TSMC, UMC. There's a whole bunch of – there's Alliance Semiconductor, which is a big chartering house, they call them. So because the problem is it doesn't – there is no economical point anymore to be running four-inch wafers. It's not viable. Unless you are doing wacky stuff like analog devices that is making a 24-bit gold-plated ATV converter that's going to sell for $25 a pop. Fine, make those on four-inch. You'll still make money. But if it's, you know, your average chip that's going to cost between half a dollar and $3, even six-inch is not economically feasible.

Chris Gammell: See, I always complain about this stuff though. So I always complain because I feel like a lot of these companies are going to lose their edge with, you know, process technology type stuff where they're – you know, there's kind of squeaking out benefits in the process technology. But then again, I also benefit from these ridiculously inexpensive parts. So I shouldn't complain.

Free Electron: Well, that's actually – so it's a double-cutting sword. It's a double-edged sword. So what you're looking at is if you're going to go and you're going to buy your op-amp that costs, you know, $1.79 a digi-key. And even though it says analog devices, yeah, it is an analog devices design. Was it running an analog devices fan? No. Probably not. Okay? If you're going to be buying that AD644 in a ceramic side-braised dill package that costs $33 a pop, that's going to be – that is going to be running their wafer fabric.

Chris Gammell: Run through by, you know, pristine versions, right, on the wafer line. Yeah, striking their beads. Yeah. Yeah, so that's – I think we're talking about different versions, Dave, but that's pretty funny. Okay.

Free Electron: Yeah, but that's the nature of this whole business. And so that's where we end up in situations. I mean, if I'm following the forum on almost anything that kind of piques my interest and you see these things fly by like, oh, how come the XR 2206 signal – well, dude, sorry, but XR is a really small company that had to let go of its wafer fabric years ago.

Dave Jones: Yeah, that chip.

Free Electron: Yeah, that chip. Yeah, that chip. Yeah, that chip was designed in the era when we had modems that were using, you know, frequency shift keying that would do 75 bowels. That's what that chip originally was designed for. You had a 2206.

Chris Gammell: Wait, so what is the background on this? Because I didn't have all this thread on the forum.

Free Electron: Okay, so there was some guy from university was – there was some professor teaching a course and they are using a 2206, which is a – Function gen chip. Function generator chip. Yeah, it's been around for like 25 years.

Chris Gammell: Oh, okay.

Free Electron: Well, the thing there is that people have always used this as a function generator to make a signal generator on the bench with it, but that's not what it was designed for.

Chris Gammell: No, right, of course not.

Free Electron: It was designed to be used together with a 2211, which was the FSK filter chip, to actually become a modem and you could pre-distort your sine wave and there was some trickery you could have. Yeah, okay. The details are lost in the gray zone of time.

Dave Jones: I don't know. The chip ended up becoming the 555 timer of the function generator world. Right, exactly. Everybody references it.

Chris Gammell: Everybody builds their projects with it. I get it. Yep. Okay.

Free Electron: And now this guy is, you know, he's in a pickle because, well, they're still teaching that course, which kind of makes me wonder if that professor is still dilly-dallying with a 40-year-old chip, how good the quality of education is going to be. Because, people, we've moved on. Have you ever heard about a DDS synthesizer? That's right. Exactly. We don't make pierce oscillators anymore or, you know, the thing with the transistor and the three capacitors where you make the 90-degree phase. We don't make sine wave oscillators like that anymore. You know, because they're very tricky to get to run.

Chris Gammell: Right.

Free Electron: So, and he was asking, you know, if there was anybody that knew where he could get these chips and the answers, you know. China.

Chris Gammell: Maybe.

Free Electron: If somebody has some old stock, yeah, probably that's your best. So, I blasted back. I was probably one of the first to reply. I said, dude, what are you doing with a chip that was marked obsolete five years ago? There was already a notice five years ago. This is a last time buy. It's game over.

Chris Gammell: Yep.

Free Electron: You know, you've got to follow up with, oh, well, we, we, we, and that is, that is something that these older, these older wafer fabs are disappearing because nobody has an interest in these older technologies anymore. A lot of times you cannot, you can't just say, okay, well, we have this technology that is now. What would prevent us to actually take the mask from the old chip and just run it as it is, send it, even though this, this technology is capable of so much more. So, it doesn't quite work that way. I mean, you got so much drift and shift in the technologies. You can't really, you, you would really have to do a substantial amount of maybe partially re-engineered schematic and the values and then do a new layout in the new lithography. And these guys look at it and go like, hell, we sell, we sell, you know, two sticks of 22.06s a year. Right. 20 cents a pop or whatever, right? Yeah, yeah. Yeah, exactly. Right.

Chris Gammell: It's a business reality of chip making.

Free Electron: The fab's gone. Just, if anybody comes along that wants to buy these chips, tell them that we dropped the masks for them and that's it. We broke the masks and we can't read the tape anymore.

Chris Gammell: Like, actually drop them, like they're shattered off more.

Free Electron: Yeah, they shattered and we went to a company to read the old, you know, the old 9mm 4-track tape and nobody says, we don't have tape drives for those 9,600 BPI tapes anymore.

Chris Gammell: That's right. That's right. Can't get the data back out. Go away.

Free Electron: Actually, that brings me back to another anecdote. This was when I was still at Alcatel. So, this is about 1998. Yeah, just before I moved to the US. We get a call and says, okay, well, we have finally conduct, finalized all the testing on your prototype chip and your chip has been accepted as, this was a pin diode driver. So, a pin diode driver is a diode that is used to switch higher frequency signals. If you polarize it the correct way, it basically becomes a switch. So, we had made a, in a very wacky 2.4 micrometer process, a pin diode driver that was radiation hardened and whatever, yada, yada, yada, yada, yada. This thing was designed in 1988. And in 1990, whatever, we finally, I think it was 98 or 99, it was shortly before I moved to the US. They finally came through and they said, your chip has passed all the testing that we want to do. We are going to buy a few million of these things.

Chris Gammell: Oh.

Free Electron: Oh.

Chris Gammell: Yeah.

Free Electron: Let's see. Does anybody remember who worked on that? Let's see.

Chris Gammell: Let's go to the file cabinet.

Free Electron: Yeah. This dude retired. That guy left another company. This guy's dead. This guy.

Chris Gammell: Yeah. Dead, dead, dead. Yeah.

Free Electron: Basically, stuff like that. This guy's got Alzheimer's. Yeah. No, it's not going to happen. Yeah. Worse. Does anybody know what a tapes for the damn thing are? I think they're in my basement somewhere at home.

Dave Jones: Yeah, right.

Free Electron: Exactly. So we start digging through this old. We have a room where they stored all these 9,600 bout per inch tapes. We start reading through these tapes. They can't find the tapes. Damn. What happened? I mean, we must have created a tape to hand off to the master. So they call DuPont, which was the DuPont chemicals, which is also doing paint and all stuff. It's one of the largest mask makers in the world. And they were in France somewhere. So we call DuPont to see if they by any chance would have still a backup copy. No, can't be found. Whatever. So somebody realizes, oh, but wait a minute. That thing was designed before we were on digital, on Vox equipment. We were using some dedicated Kalma workstation. Those are probably stored on disk packs. Wow. So these are these big cookie jars with nine platters in them. You put it in a washing machine. And then, you know, the heads come out. Kind of like that big disk drive that you disassembled. Yeah, yeah. But the platters are about five times that size.

Chris Gammell: Then what happens next is Nick Cage has to steal the Declaration of Independence. And there's written in back, on back, an invisible link, right? Exactly.

Free Electron: So we do find that disk pack. So we finally find the disk pack with the project limits written on the handle. It says, oh, jeez, we found a disk pack. Luckily, we will be able – and there is that – so like in the PCB industry, we have Gerber files. In the chip industry, we have a similar format, which is called GDS. I forgot what it stands for, Graphical Design System or something like that, which was created by a husband and wife team that were here in Silicon Valley. And it was called Kalma Systems. And it was because he was called – his first name was Calvin and his wife's name was Maria or something like that. So that was Kalma Design Systems. And they had a custom-built machine that ran on a bit-slice CPU with a very two-dual-screen display. So no operating system, no whatever. Hooked up to this big washing machine inside disk array pack. So here we're sitting, you know, a bunch of four or five designers sitting around the table looking at this big disk pack in the midst of them and going, holy shit, where are we going to find somebody that still has one of these machines? Because ours was gone, you know. Quick to halt it. You know. Holt looked at that and they go, like, that's way too bothered. We don't have that shit.

Chris Gammell: Yeah, right.

Free Electron: We don't store that.

Chris Gammell: Get out of here, whippersnapper.

Free Electron: So they eventually finally tracked down a company that – there was a company specializing in data restoration. And they still had one of these drivers that they managed to, you know, get the data stream off. And then when push came to shove, eventually they did – they actually – they ran away for a few months making these chips because it turned out that what that – we didn't know what that chip was because we were, you know, contacted by some company to design a chip according to their specification, which was a pin diode driver. It turned out that this was for the European – like the European Air Force. So this was a partnership between all the European countries' Air Force departments. And these were pin diode drivers for a tuned array radar dish. So they were going to make these big, massive radar dishes that were basically consisted of 100,000 individual receivers. And each one had a pin diode and a pin diode by a chip. Yeah.

Chris Gammell: Wow. That's a hell of an array. Yeah.

Free Electron: We designed it in 1988 and it took those guys 10 years to qualify the bloody data.

Chris Gammell: Right. And that was science, right? That's a quick turnaround. Yeah. And so – European military contractors? Yeah. That's as speedy to me.

Free Electron: Yeah. So this is to give you an idea how quick that stuff is evolving. By the time these guys approved and signed off, not only had the chip technology gone out of mode, but also the design equipment used to design a chip in the first place didn't exist anymore. Unbelievable. So this is like walking into a lap and going like, you need what, an oscilloscope? We haven't used that in 20 years. We don't do that. Oh, that's great. So that is – and that's why you end up in, you know, oh, I need an XR 2206. Ah, it ain't going to happen, dude. It ain't going to happen. Sorry. And Maxim was in the same – I don't know if you remember that one, but there was another one. Maxim had the Max 038, which was like a 2206. Yeah, the Max 038. I published a project with that.

Dave Jones: Yeah. Yeah.

Free Electron: That was the same thing. So Maxim only has – oh, yeah. I mean, it was way better than the 2206. Yeah, sine square, triangle to 20 megahertz. Oh. Man, DC to daylight. But, yes, and the harmonic distortion was better because the 2206 always had a little dimple on its side wave. There was always like a little – Yes, it did, didn't it?

Dave Jones: And there was a way to take that out. People came up with clever circuit configurations to take it out. I remember. Two diodes or something. There was an outmode about it. There was a couple of diodes, yes. I remember seeing that on my analog scope. I went, what's this little dimple on this 2206 that I breadboarded? Yeah. Yeah, sure enough, it was inherent in the –

Free Electron: Yeah, something saturated and something couldn't flip around. And with the two diodes, you were limiting the delta on an amplifier stage. Yeah, something like that. Anyway, Max 038 was really good until four or five years ago they pulled the plug and that was a big sting. Oh.

Dave Jones: Well, the backstory – I still get email over that. Where can I buy a Max 038? Because I've got my – it's still on my website somewhere. The project I had published in Electronics Australia 15 years ago, you know. Still get emails. Where can I buy one?

Free Electron: Well, the backstory there – yeah. No, because the wave fat burned down. So what happened?

Chris Gammell: Burned down? You said it's gone? Yes. Gone. Like someone tried to throw a canister out of the potato bowl and they had a door and said –

Dave Jones: This sounds like Milton from Office Space, you know. Yeah, exactly.

Chris Gammell: I did not get my paycheck.

Dave Jones: I want my stapler. This is the stapler guy. No, no.

Free Electron: What happened is that for that particular process – and that was not only the Max 038, but there was a point in time, I believe it was 2004 or 2003, where a bunch of Maxim chips all of a sudden became obsolete. And what had happened is this was – these were chips that were run in a wafer fab that was chartered by Maxim, and they had an accident. Something exploded. And the company that owned that wafer fab said, well, this is an old wafer fab doing four-inch wafers. We're not going to reinvest. We're not going to turn these lines back on. Now, if you want to continue making these chips, you're welcome to use our six-inch different technology, but you have to re-engineer your design. Maxim said, whoa, holy crap. We're looking at over 100 different chips here, and we're only selling so many. You know what? Let's kill them off and that's it.

Chris Gammell: See, I don't get why they don't just say, you know what? Go to Rochester, right? That seems like that would be the default thing because then Rochester Electronics is going to just charge you out the ass anyways for all these things.

Free Electron: Yes, you've got companies like Rochester. There's another one. The name eludes me now, but there's another one. But Rochester wasn't interested in that because Rochester is only interested in chips that they can sell to the military. Because the problem is the… Exactly. So, exactly. Because you have to understand the way that a company like Rochester… So, Rochester actually has a wafer fab. The way that those guys work is they buy when a chip is going obsolete. They approach the original designer of the chip and say, hey, can we buy? We don't need the schematics of the chip because we know that there's IP involved or whatever.

Chris Gammell: Yeah. They just want the mask, right?

Free Electron: They want the last known good mask or the tape that was used to create the final GDS file, so to speak, so that they can reproduce the masks. And they have an old-style wafer fab using probably not even second-hand equipment, but probably third-hand or fourth-hand equipment. Very advertised. It's been paid for for a long time. Oh, it's been paid for over and over. And they have, you know, if it turns out that this thing has to be etched on a really old etcher that can run some wacky process, they still keep one of those machines around just in case somebody needs that chip. So, they have this big, huge clean room full of old, obsolete equipment that they just keep around and stand by just in case somebody comes around and says, I'm really desperate. I need 50 of these because my tank doesn't work anymore. And they go, okay, show us the money. We'll make it. $100,000 a chip.

Chris Gammell: Makes you wonder about, like, the kits they have to use. Like, you know, I think about the etchers and stuff. You know, they have, like, quartz consumables inside and all that stuff. But, like, do they go and machine, like, customs for all these old machines that they have?

Free Electron: Probably. Either what happens is Rogers probably approaches companies like Teagle, which are etcher makers, and say, by the way, you're an old Model 802. You guys don't make those anymore, right? Nope, we don't make them. You still have spare parts laying around? Yeah? What if we, okay, you guys want to make some room in your warehouse? We'll write you a check for a million dollars. Wow, whatever you still have. The military just gave us a lot of money. Yeah. We had a lot of dough to play with. We'll give you a million dollars to just raffle through your warehouse and take whatever we like.

Chris Gammell: Yeah, right.

Free Electron: And that's what they do. So that is the problem. That's the reason that these things from Rogers are so expensive, because these guys not only have the old equipment, they're also hoarding all the old parts for the old equipment. Yeah, they're the gatekeepers. And so if you are a wafer fab and you still have one of these old machines, and that thing goes tits up.

Dave Jones: Which is a proper technical turn, folks. It's game over.

Free Electron: Yeah. It's game over, because you can't even get spare parts anymore, because Rochester got all of them. And that's what is happening.

Chris Gammell: Cue the Godfather music again.

Free Electron: Yeah. And oh, yeah, but Rochester won't let go, because that's our business. That's our bread and butter. We ain't going to sell you these parts. They're ours. Right. So, and that's why you don't, yeah, I mean, I have Rochester Electronics, that is their business model, and they're very successful. I mean, there's only two or three companies that do that. Yeah. And it is an issue. I mean, I know a guy, oh, this was maybe two years ago. I bought some power supplies off of eBay, some Fluke power supplies, PM2811s. And they're pretty nice to have a nice display and push button, but they started to annoy me, because it turned out that these are not, you know, classic transform linear regulator. They're switch mode internally, with a sort of a quasi-linear post regulator. Yeah, a lot of filtering and stuff. Yeah. So they're kind of noisy. They are kind of noisy for what I'm doing. And they don't have a rotary dial to, you know, very quickly to be able to play, you know, you want to sometimes go up the voltage a little bit and see if something fluctuates. You always have to key in numbers. And I figured, you know what, I bought these off of eBay. They were broken. I restored them. I'm going to toss them back on eBay. So I get contacted by this old gentleman that says, oh, you're selling those? Yeah. Can you tell me what is the software revision in them? Yada, yada, yada, yada, yada. Because he was looking for some really old firmware in those supplies. And it turned out that they had six of them. And out of the six, three had the firmware that he wanted. The other one were too new. They had two updated firmware.

Chris Gammell: Yeah.

Free Electron: And I'm like, okay. And he said, yeah, I'll drive. I'm in Fremont. I'll drive out there. I'll buy them off of you. He paid cash for them. And I asked him, I said, why don't you want the other three I have? And he goes, no, no, no. You have to understand. I am still servicing old Intel computers. And they're using the old multibus.

Chris Gammell: Ah, wow.

Free Electron: And these are machines that are being used to test the flight systems for some jet fighter that is used by defense, probably maybe an F-15 Tomcat or whatever. And the military still once in a while buys it.

Dave Jones: Sorry, I had to correct you there.

Free Electron: Yeah. The old. Or an F-14 Tomcat.

Dave Jones: Sorry, I do know my planes.

Free Electron: It's the stuff they fly in Tomcat. That's an F-14 Tomcat. Yep. Okay. So these are the really old guys that today don't fly anymore. At least I don't think they fly anymore. But he says, yeah, military. So their test system is like a 19-inch rack that they can roll out on an airstrip and hook up to the plane. And it has an old Intellec multibus-based computer. And they use these PM-2811 power supplies to feed the onboard avionics when it's in test mode. And the problem is that the test software in that thing is written in ADA, which is the

Chris Gammell: military. Wow.

Free Electron: It is certified not only to mil-spec, but also to aircraft compliance.

Chris Gammell: Yeah, like FAA type stuff.

Free Electron: Yeah. R-Rink or something they call that. And they said the problem is that the newer software in those power supplies, they changed the command set. And there's one command that was changed. One quick. That it looks like we lost Dave.

Chris Gammell: Oh, yeah. He'll probably come back on.

Free Electron: Okay. So it turned out that nobody was willing to touch that software. So they said, no, no, no, no, no, no, no. We got to find the original power supplies with the old firmware so that we don't have to touch the ADA software because we'd have to resend it through certification. It's going to take three years.

Chris Gammell: Exactly. Yeah. So this... It's amazing. It's amazing, too. I hear about this stuff with like Copy Exactly for a lot of these fabs. You know, like obviously you've been in fabs. And like people paying like, you know, 10 grand for a 286 computer because it's the exact same one that was in Etcher back in the day. Yep. You know, like they... And like people make businesses on this because they know they warehouse this crap and then they sell them for gobs of money for something that hardly works anymore. Exactly.

Free Electron: And that's... So that's Rochester's business model. And that's what this dude was doing. He said he was lucky enough to buy the old stock of all the existing boards that Intel had ever made. He had a big warehouse packed with brand new, you know, stuff that nobody wouldn't touch with a 10-yard pole. And he says, I still sell boards every month. And these boards, they sell for 10 grand a board.

Chris Gammell: You know, the only downside to this is that it really enables hoarders in their mindset, right? They're like, oh, I'll need it someday. And it's like, no, no, this is very rarely the case. You know, like you have to sift through a lot of crap and warehouse a lot of crap in order to get something actually worthwhile.

Free Electron: Well, that was... You have to keep in mind, if you are a small hoarder, like I'm hoarding some... I have a tendency to hoard some... I don't think you're alone on that one.

Chris Gammell: Since Dave's not back.

Free Electron: There's people, you know, hoarding old vacuum tubes in the garage. Dude, you don't stand a chance ever selling one to the military because, first of all, they don't know you. And second, you're not a proof channel. Wow. So, you can... I was at the De Anzac Foothill College, the last swap meet of the season, and I overheard two old geezers, you know, they were staring, you know, mesmerized by some old vacuum tube. And one goes like, I tell you, this is what won the Second World War. And the other guy goes, no, no, no, you're wrong. Because this is a Penta-what base, and the World War I was a different base, and that is even more rare. It's valuable, I tell you. And I thought, yeah, it's valuable. I'll add scrap metal, probably. Yeah, exactly.

Dave Jones: Oh, goodness. Oh, man.

Free Electron: Fantastic. Well, as we... So, yeah, so that's, in short, how I rolled into all this stuff and how I am where I am today.

Dave Jones: And that only took two hours and 11 minutes, folks. He did promise a long show. What are we doing in the remaining six hours? Let's move on to what you've been doing for quite some time, which is hard drives. Because you work at a company, which shall remain nameless, designing chips, right, for hard drives. Yeah. So, well... So, you're the man to answer any hard drive-related questions. Amaze us with the technology in current hard drives.

Free Electron: For as far as there's stuff I'm allowed to disclose. I mean, this is an extremely competitive market. We have to be very careful. I mean, there's only three hard disk makers left. There's Western Digital, there's Seagate, and there's Toshiba. Everybody else is either gone the way of the wind, was borged by somebody else, was destroyed, was resurrected from the dead, destroyed again, went away again, came back. Like, they're all gone. There is no way to even... For a new player to enter that market, you don't stand a chance. The Chinese did that with their... I forgot what it's called... Excel store or something like that a couple of years ago. And they bought out some division from Hitachi, and that went nowhere. Because they could... Yeah, they could make the existing platforms, but they weren't capable of designing the new... You know, the next generation, they died. But, yeah, so what happened is after the big telecoms balloon in 2003, DSL was basically... You know, had been flogged to death, and everybody could now make an ADSL modem. If you were Uncle Wong Hen Lo in China that had a deep fryer, could make circuit boards with an ADSL modem. You didn't need...

Chris Gammell: The cotton baller.

Free Electron: Yeah, you could... All you would... All you need is a little shack and a reflow oven. You could make DSL modems. The way everything was so integrated, know-how not required. Just slap the chips on the board, and off you go. So, I had to do something else. I dabbled a little bit in Bluetooth, because that kind of looked interesting. For the first about two weeks, I looked at it, and then I realized that, oh, wow. This is another one of these ISDN disasters. I don't know how many people are familiar with ISDN, but ISDN is such a disaster in the sense that it was designed in the 60s, and it didn't roll out in mass quantity until the mid-80s, and by then it was obsolete. ISDN is Integrated Services Digital Network. Yeah, something like that.

Dave Jones: It was basically a competition with ADSL at the time, wouldn't it?

Free Electron: Or the very... We may have to snip this out. There's a phone call. I thought I had turned, because I don't know. Do you guys hear the phone? Yeah, it'll go away. I thought I'd turn. I'm going to go put that thing on mute. I thought I'd put that thing on mute. Hang on. Give me a second here, because I think who I know it isn't calling. I'm going to notify that I will call them back. It's probably my mom trying to pull up FaceTime. And because the Mac is occupied, it switches to my cell phone. It tries to pick it up over there.

Chris Gammell: Okay.

Free Electron: Yeah, it's her. She's trying to connect. It's her. It's her. It's her. It's her. It's her. It's her. It's her. It's her. Okay. So that was Dutch.

Chris Gammell: Nice. Excellent. It sounded a little bouncy. Usually Dutch is bouncy to my ear.

Free Electron: So I told her I'll call her back in a few hours. Nice. That I was busy. She said, yeah, I didn't tell her that I was doing this. And yeah, we started at what, five o'clock at 746. Oh yeah, we're making good progress. Anyway, lost my train of thought here. What was I saying? Okay. So I decided to dabble a little bit in Bluetooth, which was, like I said, ISDN. Been flogged to death. Designed in the 60s. Rolled out in the mid 80s. Obsolete by the time it came available for the large public. And Bluetooth is exactly the same thing. Because Bluetooth was originally envisioned as a system that was going to be, you know, it's going to replace wires. And we're going to be, we're going to do things like, you'll be able to take a picture with your camera and send it wirelessly to your printer. Well, see, that may have been true when they first envisioned Bluetooth. Because we had 0.8 megapixel cameras. And we had Inget printers that could do 300 DPI with three colors. But it took them so darn long to design Bluetooth. By the time the first Bluetooth links were established and the first chips were available, we had four megapixel cameras. And hitting that print button on the camera turned out at 700 kilobit a second. Then it took two and a half minutes to transmit that picture to your printer. Guys, USB is doing 12 megabit. I think we're going to go for the wire, okay? So that is the problem with Bluetooth.

Dave Jones: Well, it found other markets, though.

Free Electron: It found other users.

Dave Jones: Because it's huge. Bluetooth is enormous. Not really.

Free Electron: It is enormous. But see, the thing is that when Bluetooth was originally designed, they designed it keyboard, mouse, wireless speaker, wireless speaker, microphone, wireless headphone, you know, what we do today with the earpiece, all those profiles, that was envisioned from the get-go. You know, they had these ideas, but they also had wireless printing, wireless networking. They were going to stream internet connections across Bluetooth links. They were going to exchange business card profiles. They were going to do all kinds of stuff. And the only thing that ever got anywhere was the wireless headset. Everything, every other single protocol died. Printing has gone away. Nobody ever even supports it anymore. The audio protocols have been dumbed down because originally it was possible to transmit high-quality stereo sound across Bluetooth, and it still is today. But here comes that little flute company from Cupertino that says, oh, you're going to be streaming stereo files that you brought through our store. We ain't going to allow you to do that because you're going to be ripping him. So we're going to deliberately cripple our smartphones so they won't be broadcasting stereo audio through Bluetooth. It'll be mono. They did that on purpose. And then the Finnish company that made cell phones and everybody else said, oh, look, the fruity company gets away with that. We're going to do that too. So boom, now you just blew stereo high-quality, high-definition audio off of the protocol. And then that data exchange didn't go anywhere because, well, 780 kilobits a second, you know, that only gets you so far. Plus the problem is that Bluetooth is not a point-to-point network. It's a mesh network. And if five people around you also have their Bluetooth link between their gizmo and their other gizmo, that's heating into the total available bulk bandwidth because it is shared. And it basically got nowhere very quickly. And some, you know, there were some companies that said, oh, well, we can still, you know, if we put Bluetooth base stations in laptops, we will be selling Bluetooth, mice and Bluetooth keyboard. But they overlooked a little problem is that since Bluetooth is considered to be an always-on link, you can't just shut down your transmitter on and off as you want. You have to, you are a node in a sort of a mesh, in a mesh network essentially. So you can't just, you know, wander in and out of that relationship you have with everybody else in the network because somebody else may be using you as a repeater. So the problem is you had an existing keyboard mouse that was transmitting in those ISM bands, 430 megahertz or even 27 megahertz, and they would run for two years on a set of, you know, double A's. You try to do the same thing on Bluetooth when that transmitter has to end at that baseband, which is the digital chip, has to be running all the time. And after two days, the batteries are gone. Who's going to buy that? Exactly. You know, you're in the middle of a sentence and your keyboard conks out. Sorry, dude, batteries are empty. So that went nowhere fast. And that is the problem is that Bluetooth is one of these ideas that got, you know, leapfrogged by the jumps in technology to the point that 99% of the original ideas that were envisioned went nowhere. And the only application that we still see today is wireless headsets. And yes, it is big in that market and it's great in that market, but that is its only, you know, reason of still being around.

Free Electron: And it may not be around much longer. Because there's other ultra low power networks like Zigbee and other things that are coming online. And that is why now for Bluetooth, they have these special power profiles, which are almost starting to become ultra low power. It's almost like near field communication they're trying to do just to preserve, you know, power consumption. That's what they're starting to do. Because otherwise they know that they're going to be dead and in a few years nobody's going to be using it anymore because it is too power hungry and too elaborate for what it is being used for. It was over designed to be, you know, the end all of connecting peripherals to a computer. And it's, you know, only being used to do a microphone and a speaker for a wireless headset. That's where we are.

Dave Jones: And you saw the writing on the wall, so you didn't get into that market. You moved into hard drive.

Free Electron: And two weeks after two weeks, bingo, after two weeks studying the real Bluetooth spec and seeing what had been fallen along the wayside, I was like, well, this is dead end. You know, let's start looking for something else.

Dave Jones: So you actually sat down and thought, right, I want to get into a new industry. And you went and studied the industries, right? Yes. Okay.

Free Electron: Yeah. Yeah. Yeah. And so something came up, you know, they wanted help again with basically the same thing like I had done before in my ADSL time and even in the vacuum cleaner time. You know, we need somebody here. We have a lab where we are doing, you know, we already are an established player in the hardest market. But that stuff is evolving. When I entered the hardest market, I mean, the largest drives you could get were the 100 gigabytes. And that was it. And we were still doing, you know, the vertical recording. There was everybody was all classical, big heads, coils as not only writers, but also as pickups. And they were just starting to roll into what they call the magneto-resistive heads. So the writing is still done inductively with an inductrit. And that was still for the 100 gig drives. Oh, yeah. Oh, yeah. Okay. And then once they moved beyond that, the pickups, they became basically quantum physical elements. I mean, a GMR is basically, I always, you know, I explain it to people. It's similar to a Hall effect sensor. You send the current through it one way, you measure voltage across from it. And under the influence of a magnetic field, you will get a positive or a negative polarity out of it. But it has nothing to do with a Hall effect. The underlying physics are totally different. But for all means and purposes, if you are an outside, it behaves, it's like a Hall. You send the current through it, you get a voltage out of it. But this thing, it is a quantum effect. Hall effect is not a quantum effect. Anyway, so they were looking for somebody to, there was a person retiring, and they wanted a replacement for him, but they wanted also at the same time, you know, pull somebody in who could do maybe some automation on the test setups and do some circuit board and whatever. And that kind of meshed very well with what I had been doing, you know, in my ADSL days and the days before. And that's how I rolled into that one. And I've been doing that for, I don't know exactly how long, but somewhere since mid-2005 I've been doing that. You've been designing chips as well, right?

Dave Jones: Actually drive?

Free Electron: Well, yeah. So my job is, my official title is I'm a staff engineer. Staff engineer is a kind of a honorary title that you get bestowed. You can go in any kind of industry, you will find that. It means that you are a person that is an all-around technical expert. Meaning, we have a problem, we need manpower, go get one of the staff engineers and put them on it. You know, they should be capable of doing this stuff. Because they've done, you have done, I mean, it is a title that you typically only get at once you have like 10 or 15 year experience. I think Bob needs to do,

Chris Gammell: you know, staff over here, staff over there, it's just staff.

Dave Jones: I think Bob was famously a staff scientist at National Semiconductor.

Free Electron: He was, he was a staff scientist. And so that's what I'm saying. This is something that the semiconductor industry, it's, it is like an honorary title that they bestow on somebody who has a number of years experience, not only doing chip design, but who has also done system design, board design, who has done work in the way for fab, who has been, who has touched every single aspect of what is involved in semiconductor industry, and may be able to see connections where if you're only, if you're only a pure guy that is doing nothing else but spy simulations every day long, you just can't connect to what's happening on the board because you don't know you're way out of your water, way out of your territorial waters. I don't know how to do stuff scientists. I don't know how to do stuff. I don't know how to do stuff.

Dave Jones: People they keep around just in case, you know, when, when, when shit hits a fan, right, okay, hey, we knew we kept this guy on for something, here we go. Yeah,

Free Electron: and I am, I am, I am probably the worst silicone designer of them all out there because I have, I have done little, you know, I have done a few of a few IO cells. I've designed a few IO cells in CMOS technology. And then I know how to do that. You know, I have done the layout of these cells. I have seen them through fabrication. I have characterized them on bench, in the oven, under, under temperature, under over voltage stress, whatever. And they actually ended up in a pager sold by some Scandinavian. I forgot what that kind of Scandinavian. I mean, they were the driver cells driving the LCD. I have done that. Does that make me a chip designer? Hell no. I mean, if somebody comes to me and says, can you design me an op-amp that is as good as an AD? I don't know. You know how to design an op-amp. Okay. You, you will have to hold my hand here, but if somebody, but if somebody, give me a couple months though. Yeah. But if somebody comes to me and says, we've got a problem in a system here, and we're seeing this wacky, wacky behavior, can you take a look at it? I haven't, I have been in touch with enough different, various aspects that I can very quickly rule out. Okay. This ain't the board problem. This is not a layout problem. It's not a passive components around. Oh, but wait a second. Oh, you're using what you're using a ceramic capacitor. Look, if you're tapping on the board, ceramic capacitors are susceptible to microphone. Yeah. Oh, put a film cap in there. See if it. Oh yeah. Oh yeah. Look at that. The filter does work now. That's what they keep me on for because I have done all these different things that, that, and I make the, I look at these things. I know just enough about everything that I can very quickly figure it out. And this is unlikely, unlikely, unlikely. This is, these are potential root causes. And then I have to investigate. And it takes me as much time as anybody else. And, and that is the, that is the, that is the kind of job where you employ a staff engineer because these guys are gonna, I mean, I always, I never assume anything. And if people, I come to a problem to say, we, we, we try this, this, this, and that. I say, okay, thank you. I will try it again because I don't know what you have done. I am not, you guys are super engineers. You guys are super designers. I am nowhere. And I admit that I am nowhere near the level of the silicone designers, but I will do my own test and I will look with my own eyes because you guys have a tendency to be too focused on your little bitty thingy. And it may actually have nothing to do with your circuit in the first place. It is something, some fringe element that is having an influence, but to find it, you, you have to know. And sometimes, you know, ADSL, here's an example from the ADSL world. We have a problem. Um, common mode rejection is pretty bad on a certain circuit board. And we're like, this is weird because this is a filter that we qualified. You know, it's, it's a, an LC, it's a multi-stage LC filter. We've built hundreds of these filters. These filters are okay. Okay. Customer, please send us your board. We look at these inductors. And we, so we look at that board and everything looks fine. Yeah. The inductor, we measured the values. We sweep the filter. Filter doesn't behave properly. This is weird. All the values are, all the values are intact. And you know, it is, that's not it. So according to the SPICE simulation, everything should work. We measured the inductors. We measured the Q factor of the inductor. We desolded these parts. We measured them for drift and tolerance. Everything is fine. Put them on the board. Doesn't work. It turns out that if you buy these little air core inductors from TDK, the way that the, the core, the, the, the winding is oriented inside the package is actually horizontal. If you buy them from Murata, they are actually, so how to explain this? If you leave, until you have the surface of the PCB and you're going to start with the piece of wire, you're going to make, you're going to make your turns for your inductor. So you're going to solder the wire on the pad of the PCB. You're going to go up and you're going to make, you're going to make your first loop. So your, your loop is standing vertically up off the board. And then you place the next loop to the right hand side of that one. And the next one to, so you make, you make a cylinder that lays flat on the PCB in the Murata inductors, the cylinder sits vertical. So what happened is that, and this was something that we, we only realized afterwards because the proximity that these inductors were on the board. So this is a differential system. Yeah. The way that they're positioned, if the cylinders are horizontal, there is enough coupling. So you get very good common mode rejection. This thing behaves like, even though they are two, two independent inductors, you have a common mode transformer. It's their core. If they're vertical, it doesn't work anymore. That took us three days to figure that out. I've seen similar things. It's a bastard of a thing. And this, and this, and these are the kind of things, and it's the same thing, you know, there's this discussion going on now again on the forum about capacitors. And there's a couple of really knowledgeable people pitching in about it. There's one of these guys that he's from, from the, I don't know if he's Norwegian or whatever. His name escapes me right now, but he pitches it. He says, by the way, do you know this, this, this, and that? And people go, really? Yeah. I mean, if, if I tell you, if this is a critical analog design, and I told you to use a TDK 0805, 25 volt X7R, don't come back and bitch at me. Yeah, but you're only putting three volt in it. Have you seen the curves? Have you seen what happens if you put three volts on a, on a 25 volt capacitor? That's supposed to be one microfarad. If you're really lucky and it's a really good one, you'll have 0.90 microfarad. If it was a Ching Chong Wi-Fi V from the lowest quality, if you have a few nanofarad left, you're really lucky. Because the, the dirty secret of the capacitor industry is capacitor values are specified at zero volt DC working voltage. Nobody has a circuit that runs at zero volts. Yes. They do. It's not very useful. It doesn't consume anything. Yeah. So, and that is the problem. And that is, that is the kind of practical, the practical stuff. I, I happen to know these things because when I did ADSL, yeah, we sat there tweaking capacity. I mean, we, something stupid like power distribution network. Oh, you just scattered, you know, 0.1 microfarads for every digital chip. Guys, 0.1 microfarads worked great when the clock speeds were 10. megahertz at a hundred megahertz. They are an inductor. They don't do anything. You know, you might as well throw them off. And still, there is this convoluted notion that whenever you see an internet schematic, it's the sound of the, there's a hundred, you look at the last day, it should enough. There is a hundred nanofarad decoupling. Are you guys saying, what are you doing? And, and this, and, and for some of these critical, like in what we're doing in the hard disks, where every penny counts, we have switching regulators. We optimize the switching regulators. We tell them, look, these things are optimized to run on a specific frequency. Here's the amount of bulk capacitance you need to get rid of the, you know, the, the, the rectification ripple under your load, because you know, you're going to be drawing one amp with this week switching frequency, this duty cycle, you need 22 microfarad to get that ripple below 10 millivolt. But then the problem starts, then you get harmonics. So your 22 microfarad is not dampening the harmonics. And now you get the EMC problems that drive is going through. EMC, and it fails miserably. Put it in the anechoic chamber. Oh, spectrum looks like a hairball. So what do we do? Are we going to scatter a hundred nanofarad? No, because they cost money. So we're going to tweak the PCB layout and we're going to, we're going to actually, what we're going to do is we're going to take that board. We're going to take that switcher off of it. We're going to take the switcher off, feed it from a clean analog supply, power up the device. You know, the thing is going to be running and we're going to hook up a network analyzer with a biasing T filter in it. And we are going to make a spectrum plot of the impedance of the power distribution. Great. And we're going to, and we're going to see that, Oh, look here in the EMC spectrum, we are a four DB over, uh, what is allowed in the 100 to 120 megahertz band. Yeah. So what are we going to do? We open, we go to TDK, we download their capacitor software, select the tool that we say, we need the capacitor that is, that has the, the, that is best tuned for a frequency of 110 mega megahertz. And we need it. Uh, we're going to be running at this DC voltage. We're going to be running at this temperature range. And the tool will think for three seconds and it'll suggest you five capacitors. And I will call TDK and have a bunch of samples shipped overnight to me and I'll stick them on the board and I will read on my, I will read on my network analyzer. But if you explain that on the forum, there's people that are, exactly. Their eyeballs are starting to go, Oh, Oh, fuzzy. And they're, you know, they're going to fall off their chest. Like what the, what are you smoking?

Speaker ?: Dude,

Free Electron: this, this is how stuff is done in the industry. If you have to opt, you optimize the snot out of it because there is a cost factor. And sometimes you have to do crazy stuff like, yeah, fine. Let's put the impedance of the power grid. What? Nobody does that. Yeah, I know. Yeah. And there's, there's other situations. Here's another simple thing that people don't understand. Um, the difference between a series and a shunt regulator, a problem, all your classic regulators, 7805, 712, switching regulators, whatever, you know, they do, they do one of these exercises you always do in schools. Oh, we're going to calculate the output impedance of a regulator. And you are learned that for, to have a good power supply, the output impedance should be very low because, low. Yeah. Okay. So here you have, here you have, I'm a good robot. Okay. Now here's, here's the $10 million question. You have a 7805 and you, you are a very lucky person. You have a 7805 that is making exactly five volt. You can measure it with an eight and a half digit, you know, and obtain your multimeter. It's 5.0 at infinitum. Okay. What happens if you have hooked up to the output of that system? You have a couple of muscles, which sit there and bounce some charge around and they inject some charge back into the rail.

Chris Gammell: Can you do that? That's what ships do. That's what the Latroger does. No, but you had that perfect reference though. You ruined it. You're switching charge around. Now you need a perfect buffer. Oh,

Free Electron: ah, good. You get, you get, you get five points. You get five out of 10 points. You're on a good track, but here's the problem. You calculate the output impedance of your regulator to be very low and very good, but that is only true. As long as your supply is delivering current, because the moment you inject charge in the rail and you manage to lift that rail just above that five volt, that past transistor.

Chris Gammell: You have a broken DSA, right?

Free Electron: Yeah. Well, roughly. So, but what happens is the conducting, the regulating element, the regulation loops is, oh, we've got more than five volts. I'm going to stop conducting my transistor. I need to correct. So what I'm going to do, I'm going to close the valve. Well, there goes, there goes your rail impedance. Your rail impedance for injected charge is not low impedance. It's extremely high impedance because you're, you're all the capacitors that are in front of, before the regulator are all cut off because you cannot send electrons in reverse through the regulator. You switched off your pass transistor. You switched off the transistor. So you're perfectly calculated. Very, very low ohm.

Chris Gammell: What if you had like an AB stage like regulator or something? That's right.

Free Electron: That's right. That is why I gave you five points because now you do have a push pull. You have a push to feed five volts. If somebody back feeds, you shunt it to ground, which leads me back to these.

Dave Jones: Multicodural power supply, folks.

Free Electron: Let's get into that. Four. Exactly. And that's, if you remember on the forum, there was a discussion about the HP 6624 power supply a while ago. And I brought up, I said, guys, these things have a, um, oh shoot. Uh, uh, uh, not a name. The name escapes me. A down programmer. What the hell is a down programmer? Well, that is that power supply actually has not only a transistor coming from your transformer and you rectify it in your capacitor.

Dave Jones: Pass transistor.

Free Electron: That your pass transistor. It also has a transistor going from the output to ground. So if that supply detects that you are trying to force, if it tries to make five volt and you externally trying to make 5.1, it's going to activate that transistor to ground and it's going to sink that current to pull your rail back to five volts.

Dave Jones: Actively. So here. And that is the difference between a single quadrant and a multi-quadrant power supply. For those who don't.

Free Electron: So here's the thing. You get a junior, you get a junior designer in the lab who's trying to troubleshoot something. And he unwittingly, he unknowingly is right now using one of the super duper clever two quadrant power supplies. He only the 6624 was available. So he grabbed it out of the shell, hooks, hooks in the stuff. Oh, look, it works perfectly fine. Sends it to a customer, customer plugs in. Have you looked at the hair on this power supply? My God, there's five volt overshoot on this rail. Well, it's not on my bench. Yeah. You've got a power supply with a down programmer. Huh? What? You get that. What is it? Professor Farnsworth from a, from a Futurama going, huh? What? What is that? Programmer? And that is the kind of, that is the kind of situations where you need to have somebody in the lab that can, you know, casually walk and say, Oh, by the way, I see you using this machine. You do know that. Yeah. What? I thought it was a power supply. Let me tell you something. Let me show you good. Grab a chair because grandpa has a story to tell. Oh, I love it. And that is, that is, that is, that is, that's, that's where it's tough. And that's where people like Bob Pease, they had gone through all this. They had firsthand experienced all this kind of misery, figured out why they got this misery, figured out how to solve it. And that's where people like that are invaluable. I mean, I'm not going to compare myself to Bob Pease or, or Jim Williams. Hell, I couldn't even, I don't even reach to their ankles, but, but those people have,

Chris Gammell: you know, because in the lab, they are,

Free Electron: they are generalists. They have done customer support. They have seen all the wacky stuff that can and will happen. And they do not trust anything you feed these guys, because you can pull these guys in a meeting and go, yeah, but well, you don't have to argue, but we already tried that. These guys are going to go, yeah, sure. But you have tried it under what conditions? And are you certain that you understand your test? Do you understand your test equipment? Have you guys heard about burden voltage of an ammeter, by the way? Oh, what's that?

Chris Gammell: Yeah. Yeah. You know, it's, it's usually you find that engineers who have been out in the field and, and have worked directly with customers and seen a lot of that stuff, you know, like that they've been there, they've been yelled at, right? Yeah, that's it. You know, they've had to deal with a lot of banal, like customer complaints, and then some very serious customer complaints and they know what they're looking for.

Free Electron: Here's another, here's another, one of these stinkers. you're going to make a little, you're going to sweep an A to D converter, right? Or no, you're going to sweep a D to D. So you make a little program, you have a D to A and you're going to sweep it between zero and its maximum code. And you're going to measure the output with the multimeter. And you're going to plot that. And you're going to try to do INL, DNL. Yeah. And you're going to, you know, you're going to have some noise and you're going to be doing, you're going to take a hundred readings per code. And then you're going to average that. And you're going to do some statistical, I call it nonsense because you don't understand what you're doing. And you're going to plot it. And you're going to see where the hell do these previous signals come from? Dude, when you were sweeping that thing, did you have your multimeter in auto ranging? Yeah. You do realize when that thing changes ranges, it's changes accuracy. Oh, that's where those steps come from. I see. So they started out, you know, that multimeter had a 10 millivolt range. So for the low codes, it is measuring with the accuracy it has in its 10 millivolt range. And then it switches to its hundred millivolt and it's 300 mill. And every time you have this step and these guys go, Oh, we've got a problem in the D2A. No, you don't have a problem in the D2A. Turn your auto ranger off. You're not using your instrument, right? Oh yeah. And then, okay. They turned the auto ranger off and they run it again. And now they have a different kind of weird behavior. The INL, DNL films, the step to step error, is all over the place. You can't draw a straight line in that. Where the hell does that come from? Oh, it must be the D2A. No, you do understand that, that 34401, which is, you know, a really good multimeter. It's using a multi-slope converter. Oh, blank stare. Okay. Okay. Come here, squirt. Sit down. Grandpa got to tell you a story. Multislope converter. You, you take a known, a known reference. You charge a capacitor for a known amount of time.

Chris Gammell: Really, really good. Yeah. Yeah. Yeah. Exactly. No leakage.

Free Electron: Well, you can live with leakage if you, if it has been, if it's been characterized. Yeah. If you can characterize it and calculate it away during, that's why a multislope is so good, because it's a charge balance. Yeah. And you don't, yeah, any, any, everything can be totally off, as long as it's all, relatively, it's all right.

Dave Jones: You don't know how much. Yeah. And it doesn't change.

Free Electron: And that's,

Dave Jones: yeah,

Free Electron: exactly. So here's a problem. When you're running this, this multislope algorithm for every sample that is being taken, they are zeroing the system. So you don't, you do not have a way to find out for, before, just before you took the reading, how far off the true zero was, because you zeroed it, the sound, the conversion system, zeroed itself. And that is the random noise you were seeing, because that is the random noise in the digitizer. It has nothing to do with your DAC. So the way that you have to support, the way you are doing these measurements, please, and Agilent actually has a function for that. You send in a command that says, I'm going to start my, I'm going to start my measurement. I'm going to start my sweep. Please do an auto, go to a fixed range, do an auto zero, and give me your residual. And it'll give you that number. And then you do your sweep, and you deliberately do not auto zero, the converter. Because all that's going to happen, you do the nature of the beast, it is going, it is going to be an incremental charge accumulation, with either a positive or a negative sign. So when you hit your maximum code, and you're done with your sweep, you tell the machine, now take a reading, give me that reading, auto zero yourself, and give me your residual. And that is your, if you, you know, that line is basically AX plus B, that residual number you just have got, is your offset, to level the line. Exactly. Down, and the other, the original number, is the gain correction factor. And now you're going to look at your output, and go, look, all that hairy noise is gone. Magic. Yeah. Magic. Because, the noise was injected, because of the converter. You, that, the way, the nature of that converter is, it, it gets rid of its random crap, but because it, it eliminates the random stuff, you see it in the output, and there is no way to retrieve, what it actually nulled out. So you take a reading at the beginning, you take a reading at the end, you apply it again, and an offset, and look at that, you get a perfectly straight line. Oh, love it. Oh, but that, but that involves reading a 300 page manual. Exactly. Well, isn't that what we're supposed to do? I mean, if you're thinking with electronics, isn't that what we're supposed to do? You have to understand, your, the limits of your test equipment, and, oh, it's a bad, scope, it was under sampling. Yeah. No. Yeah, well, did you, did you, did you at least have an idea of what kind of a signal you were, no, we had no clue. We just hooked up a probe, and looked on the screen.

Speaker ?: Yeah,

Free Electron: exactly. When I have, when I have an unknown signal, I have learned, okay, the modern scopes are typically very good at not, because they have a frequency meter, and they will just show you, you know, a band telling you, dude, you're under sampling. That's it. But I have learned, if I, if I have an unknown scope, that I've never used before, I put that thing in its fastest time base setting, apply my unknown signal, and then I start dialing it downwards until I see something. That's right. because then I will never be under sampling. But, and, and these are little, tiny, little, stupid things,

Dave Jones: which, if you have never been out in the field, you just don't know about. Well, often, even if you have been out in the field, often you may never come across them. It just depends what sort of stuff you happen to encounter. You know, it's, you can be, you know, you can be a 40 year field veteran, and never see any of these issues. It purely depends on, you know, what random stuff comes up.

Chris Gammell: There was this interesting article about, about like artificial intelligence the other day, and it was talking about like, you know, computers can be the best in the world, but like actual intelligence is based on recognizing a situation that you've been in, and then like reapplying it basically. And like, this is a perfect example of it where it's like, yeah, seeing it, recognizing it, and then reapplying that data in a troubleshooting context.

Dave Jones: And even if you haven't seen an exact.

Chris Gammell: Nothing beats that.

Dave Jones: And even if you haven't seen the exact, you can, you know, get little subtle clues in your mind, puts, you know, oh, like, I think it might be that, that's worth investigating. Yeah. And that is,

Free Electron: that is the thing. If you, if you have been doing, if you've been in electronics long enough, and you've done all kinds of different things, you've done analog, digital, RF, all these kinds of different things, you're going to end up in a situation. I remember years ago, I have seen some, it may have nothing to do with it, but I remember there was some, can we check that one? Exactly. Oh, that turned out to be that one.

Chris Gammell: Yeah. Yeah.

Free Electron: And, and that is something that sadly, I mean, you can read all the books in the world. There, there is no substitute for experience. You have to, and this is a joke that is often made, you know, the experience of an electronics guy is proportional with the number of blown parts. That's it.

Chris Gammell: Yeah.

Free Electron: And that's,

Chris Gammell: that's it. The number of nights spent in the lab. Yeah. Yeah.

Dave Jones: And it goes back to what we said at the start of the show, that you learn something new every day. It doesn't matter how experienced, you can be Bob Pease, and you're still learning something new every day.

Free Electron: Every single day, there's something, something new popping up. Yeah. And you're like, Oh, wow.

Dave Jones: Didn't I know? Okay. I'll go investigate that.

Free Electron: Geez.

Dave Jones: And that leads you onto something else.

Free Electron: Oh boy. And now, now we are fortunate enough. You have the internet, but you have to be very careful because it's extremely addictive. I like it. Oh yeah. You read about some obscure thing and you go and look it up and then that, Oh,

Chris Gammell: the Wikipedia hall. Yeah. Yeah.

Free Electron: That obscure, one obscure thing leads to an even more obscure thing. Yeah. Yeah. And then you, and then you have the other danger that you're, you're faced with a problem. And the first thing you're thinking about is all the obscure crap. Guys. Right.

Chris Gammell: Isn't something simple.

Free Electron: Is it, is it plugged in?

Chris Gammell: Is the power on? Is the power on? Is it plugged in? Is a reference there? Is it the right value? Yeah. Yeah.

Dave Jones: And people want a troubleshooting guy. Yeah. People think, Oh, give us a troubleshooting guide from, you know, that we can apply. It's like, no, it's like, it's not that easy.

Free Electron: You know, that, that is a problem. There is no recipe for, I mean, Hewlett Packard tried to do that at two certain points. They had those signature analyzers. So what they would do is they would take a note, known good. They had a special device that they would run a special piece of test software and you would hook up a logic probe and it'll give you like a four letter number key. And that was, and so these guys actually said,

Chris Gammell: And then you went to the huge manual or something? Yeah.

Free Electron: And they tell you, okay, if, if this, if this error appears, okay, flick this jumper, power on the device, hook up. And, and I have, I have one of those, those machines. They, they combine, they're called signature multimeters. They fly by on eBay once in a while for like 50 bucks. If you ever can grab one, grab one because they're useful. Yeah. Especially for the, the older equipment and in the test manual, in the manuals, service manuals, they will give you the signature. So what you do is you poke that probe and that machine will tell you F3ZX. And you look at the manual. Yeah. Okay. If F3ZX, that means that this chip is okay. Now probe that point. You should see that number. Okay. I see that number. Now probe this point. Oh, that number is different. Okay. The manual will tell you if that number is different, it happened between the last step and this step. And the only, they will tell you it is exactly that chip is the chip that is in the way. That's the guy that's broken. They're magic. Those are signature analyzers. They are magic. Yes. Actually, that actually the principle is very simple.

Free Electron: it's incredibly simple. They generate. It's, it's a pseudo random stream that they send through the, through the whole machine and they record it on a known good one. You know, what they do is a known good test. You find it is a known good test. Golden board. They call it. You start. Yeah. You start at a certain point. And as long as, as the, the, the bit pattern is undisturbed, the codes will correlate the point where this, the disturbing, the disturbing in the pattern is, or the disruption in the pattern is happening. The code will be different. That means the error is between the last point you measured and the point you measured. Exactly. And the manuals are, are formatted in such a way that they always, only go one chip. They only go, they go from one chip to another. And that's how you, you, you, you nail those, those things. So yeah, if you ever get a hold of a, if you ever see those things fly by, get one. They're neat. Yeah. I've never had one.

Dave Jones: I've used them, but, uh, never had one in my own life.

Free Electron: They are, yep. Extremely useful.

Dave Jones: Well, if you just, if you don't have the info, like if you don't have a golden board with the info, then they're useless, right? They're, they're a, you know, a doorway.

Free Electron: Okay. You, you don't have, you don't have to have the golden board because the golden board is described in the manual.

Dave Jones: But if you don't have the manual that's been tested with that specific instrument, then they're just dead weight. They're a boat anchor, you know?

Free Electron: Correct. But the thing is that any, any piece of test equipment from HP from the mid seventies to I'd say the mid nineties has those codes in it. So they have been using that for a very long, it may actually be, if you are able to get a hold of the clip for your signal analyzers, I'm suspecting that you will find signature codes in there. Most likely.

Dave Jones: Yes. I'm working on that now.

Free Electron: So to get, so if, if you get a hold of the clip and you see signature, analyzer required, get on eBay and get a model 5005. Because that's what you want. I'm not that desperate. Don't bother with the, don't bother with the 5004 and 5003. Those are too old.

Dave Jones: Well, we have entered. I have, I have. I was just going to say, we have entered our fourth hour, Vincent. Okay. And I don't think we're going to make it to the end of four hours. So we probably should get onto our Q&A.

Chris Gammell: I can't, I cannot make it. Okay. I don't even know if I can make it through a Q&A.

Dave Jones: Okay. Do we want to do the quick Q&A? A quick, quick Q&A. Yep. Run them in five minutes and. Let's run some.

Chris Gammell: Oh, I don't, I don't believe you, but yeah, we can try it.

Dave Jones: Okay.

Chris Gammell: Shoot.

Dave Jones: Well, I've got one straight up. Oh, let's see. You've written a couple of books, of course. We hear it from, we ask probably everyone who writes a book, how easy is it to write a technical book?

Free Electron: Uh, it's an awful amount of time you put into it for very little revenue. You, I mean, technical, unless you're a guy like Stephen King, you won't be able to make a living off of it. Uh, pulp fiction and, and romance novels. There you go. That's, that's where the money is. Do not bother writing a technical book. I'm, you're doing it because you. You love it. Or you, or you, you like doing it and you like, you want some fun and you think that you got something. You have a story to tell. You have some know-how that you want to see condensed to the book for somebody else to pick it up. And that's the only reason you do it. Otherwise, don't start. Exactly. There you go.

Dave Jones: Words of wisdom.

Chris Gammell: Oh, that was, that was good. All right. So, uh, another question from, uh, Reddit was open design tools. Uh, do you ever touch any of those like magic, IR SIM, anything like that? Yes. Alliance tool chain.

Free Electron: Yes and no. Okay. The, the, the, the dirt, the dirty and ugly truth is that even though I'm not an advocate for open source tools, yes, whenever there's a new version of the tools, I do dilly dally around with them. I, I will download key cut and, and whatever one, you know, once every six months and see what they've been doing.

Chris Gammell: Um, the, the, are you saying you are not an advocate for open source tools or you just not, you don't use them?

Speaker ?: Um,

Free Electron: or, or, or you, or you've,

Dave Jones: or you've tried them and found that they're just not good enough.

Free Electron: the problem is due to what I do professionally, I am used to a certain quality.

Dave Jones: Yeah. Yeah.

Free Electron: Of tools that I just, I just can't find it. And I'm sorry for the open source guys, but it's junk. And it's,

Speaker ?: I'm,

Free Electron: and it's, I'm, it's really sad because there is a lot of talented programmers out there, but the problem is with open store. And I honestly believe that this is the root cause. Open source is nothing, but let's, I don't like you. I'm going to fork the design and go my own right. Yeah. And you end up with 25 versions of essentially the same thing, each having their own unique problems.

Dave Jones: And then they dilute the talent, the programming talent across. And that's the problem. Yeah.

Free Electron: And then, and then some, some poor sod says, okay, I'm going to try and collect all these little patches and, you know, arm wrestle it back into the mainstream. And it turns out that, well, but this guy changed the architecture. This is not compatible with that. And it turns out to be a tremendous amount of effort, basically spinning around in circles. And it all, I am, I mean, I only have, there's only 24 hours a day. I'm used to tools that just work. I cannot, and I will not, even if it's for hobby work, I get frustrated if it doesn't work in the first five minutes. Yeah, me too. Okay. Because I am used to, I am used to tools that I grab off the shelf, I hook them up and I am debugging. I'm debugging what I need to be. I am not debugging my tool. I'm debugging my, the damn thing I'm trying to get to work. That's it. And open, open source. Yeah. It is, it is good if you have a shoestring budget. And if you're willing to, if you're willing to live and invest the time in it. Personally, me, because I'm used, I can't do it. I mean, I get frustrated because I'm used to, to work with a different standard and different category of tools. And, and I agree that there's going to be a lot of people say, Oh yeah, but you know, Altium is $7,000. And if you want a Kyle compiled for arm, it's three, it's three grand. Okay. Well, you don't have to spend three grams for the Kyle compiler because face it guys, there's a 32 K compiled code free version. Yep. How much, how many microcontroller programs have you written that are more than 32 K compiled output? Oh yeah, but we want to do it. We don't want to do that because if we use this closed source tools, then it can't be compatible. That's code is sources incompatible with GCC and everybody else is bitching at me. So what? It's their problem.

Chris Gammell: So you're going to take the lone engineer stance on that one.

Free Electron: I am. I am. My, my point of view is if I need to put in, if I have a need to put in a screw that mandates me to use a Phillips screwdriver, I will get a Phillips screwdriver. If it mandates me to use a flathead screwdriver, I will use a flathead screwdriver. I am not going to sit there with a bunch of files and an angle grinder, trying to convert my Phillips head into a flathead to put the damn screw in. You know, around the corner, five minutes, I go get the tool. And there are a lot of these companies give, you for free base versions of tools that are perfect. I mean, Eagle. Yeah. I bitch about Eagle because I'm used to, I am used to higher standard, but, but Eagle is a per Eagle is a perfect. If you're going to design a board, you know, Eagle has free versions. Dip trace has free versions. They are perfect. Why don't look around with, with a key cut that is full of, even the Gerber output is wrong of that tool. Why would you, why would you want to spend time troubleshooting them? And on the forums, and I know it's very valuable giving these designers the feedback that they need, because otherwise the tool will never improve. But I have, my first task is I want to make a board because I have some doohickey. I want to fabricate here. I don't want to deal with your bugs that you put. I mean, you know, I expect you to, to work. If dip trees gives me a free version that works, I'm going to be using that one. If I can get my hands on something that's better than dip, I'm going to be using that one. I, I was lucky enough to be able to buy an Altium license when it was, you know, $2,500. They had a special promotion and I bit the bullet. I said, you know what? I'm using this thing enough. I'm going to buy my own license and that's it. Would I, if, would I buy it if it was now seven grand? Hell no. Of course. That's too much for, I mean, that's, that's way too much. I, I can't justify that, but I can sort of justify the $2,500 for Altium because I've done a couple of books. All the boards that were designed for the books were made with Altium. I have my prototypes. I have revenue from the books more than covers.

Speaker ?: Yep.

Free Electron: Exactly. You know, the, the, the, the tools that I, I can afford myself a couple of nice teas and luxuries because I have some income that was, you know, from, from, so I guess the lesson is if you go back to the, why would you want to write a book so you can afford a better, really, I'll do it. You can buy a better scope.

Chris Gammell: But that's about as far as you're going to get with that. Yeah. So on the, on the, on the topic of cost then too. So the last question we had here was the, a lot of, so this is on the EV blog forum. Peter asked, a lot of old school micros, sisters 8051 and 65 or two are still in a lot of touchscreen controllers, sensor modules and stuff like that. And is it still worthwhile to use them? And do you think they'll be around in 10 years time?

Dave Jones: Is it worthwhile learning them? I guess he was getting at it.

Free Electron: Yes and no. Yes and no.

Chris Gammell: Oh, right. Sorry.

Dave Jones: Okay.

Free Electron: It's, it's the, the answer is, is twofold. Yes. They are interesting to learn because these, these older CPUs have a couple of features that can give you some insight in certain things you can do with a CPU that you will never get to understand if you're only dilly dallying and writing some Python scripts and whatever, you're never going to figure that out. That's number one. Are they going to be around in 10 years from now? Yes. They're going to be around 10 years from now. Are they going to be very widely used to the point that hobbyists will get into contact with them? Probably no. Of course. The reason, the only reason that they're going to be still going to be used 10 years from now. And the only reason that these cores are still very widely used today, like this 6502, the 8051 and so on, is that these cores are free. Meaning they, you do not pay royalties as a chip maker to embed these cores in your chip. If you, as a chip maker, make a touchscreen control that has an arm in it, you are going to pay $1, dollar cent per sold chip to arm. Right. As it, and it's not, it is. Well, not that much probably. I'm giving it. It depends on your volume. It depends. And, and for some really large volumes, it is a, it's a fixed fee up front. It doesn't depend on the volume. It is a fixed fee up front or it is a fee per sold device that you are paying. Okay. 6502, the patents expired, the patents. And now we're getting into that whole hairball of patents again. Okay. You can't, you cannot patent the, the, the core itself, but you can patent the mnemonic. You can patent the instruction set. Okay. Right. These patents have, have expired long ago. So everybody is free to make a, an implementation of a 6502. You can, you fire up your very long, your very long tool, your VHDL tool. And you're going to define from the ground up, something that implements the instruction set of a 6502. And you're going to release that. You're going to put it in a, piece of silicon. And there is nobody that can play it. You're infringing on an IP here. No, you're not because it's written off. Intel has released the 8051 years ago as public domain. It is officially classified as public domain. Okay. You don't even, you don't even have to put copyright Intel on it anymore. It is gone. And that's why you're still seeing today, you know, out of the far East, all these Chinese companies making little, you know, display controllers, touchscreen controls, and there are 8051 or whatever. Yeah. That only 85. Why? Because it is what a very well-known court has a large history. It is a really nice court, a little micro control.

Dave Jones: It's got very mature compilers and tools.

Free Electron: It has extremely mature tools, debuggers, and circuit emulators, JTAG, whatever. Which is a big deal. Which is a big deal. The bloody damn thing now knows how to do 16-bit arithmetic, even though it's only an 8-bit machine. It knows how to multiply and divide in hardware. And it has the capability of very efficiently making multitasking. Because the bloody damn thing has four register banks. And if you want to switch context, you don't. In a traditional CPU, you got to shove everything you're doing on the stack, switch context, do what you think, go back and pop it off. On an 8051, you don't do that. Because you have a sign that this routine is using bank one. This routine is using bank two. And these are chunks of RAM memory. So when you have to switch context, you just tell the CPU, you're no longer running in bank one. You're running in bank two right now. So everything is left into place. And nobody is going to be able to modify anybody else's data. So it is a very efficient machine to make it a little multitask. I have to have a routine scanning a keyboard. I have a routine scanning output to a display. And I have two background tasks. One which is sampling an A to D and doing temperature logging. And the third one is controlling the relays. I can write this in 50 lines of code. I can write a thermostat that's multitasking. Beautiful. Try that on any other one of these. Yeah, do it on an arm. AVR architecture. You need oodles of code to do it. And that is, so the answer is, does it make sense to learn? Yes. From a historical perspective. Learn it for you kind of thing. Learn it to discover some of the clever techniques that they employ. Because some of these techniques are still valid today. If you're not at the level where you take an AD51, throw Linux on it and run Python as an interpreter. If that is your level of working, by all means, stay away from. Exactly. Go learn AD51. Okay? Learn Python. You don't need to know even what a microcontroller is. Because to you, it's a black box that runs the operating system and runs a Python interpreter. Learn Python. Have fun. Memory? What's memory? You know, memory is something that comes in a stick and you just plug it in. Exactly.

Chris Gammell: Yeah. It comes four gigabytes at a time.

Free Electron: Learn Python and become a Python wizard. If you're going to be delving around with small projects using small microcontrollers, take a look at the good old guys like the 6502, the RCA 1802, the AD51, the Motorola 6802, 6809, 6805. They have interesting instructions that do interesting things leading to interesting techniques that may help you become a better programmer that can exploit the capabilities the machine has as opposed to a code monkey that just writes 500 kilobytes of code to blink an LED. Exactly. So from that perspective, I say yes. From the other perspective, is it worth investing time to really learn it? Probably no, because you're only going to find these cores deeply embedded in some specialty chip. You're not going to be able... I mean, yeah, you can still go on DigiKey and buy an 8751. The question is how much longer.

Dave Jones: It's one of those things where if you have to ask the question, the answer is no. If you have to ask the question, you don't know why you need it. The answer is no.

Free Electron: Yeah. The answer is it only makes sense either from a purely historical perspective or from a commercial perspective. If you're being a designer that wants to, you know, have a custom chip made and you go to one of these outfits that, you know, roll your own chip, basically...

Dave Jones: In which case you wouldn't be asking the question.

Free Electron: Yeah. And it's like, oh, well, we need a programmable core in there. Well, look, you can go pay royalties to ARM or if you're happy with, you know, a few 8-biters, here's a few 8-bit cores. Well, why don't you have picks available? Well, because, you see, all the picks are still under copyright by a microchip. Well, what about the Atmels? Well, those are all under copyright by Atmel. I cannot, as a custom chip maker, unless I am Atmel myself, I cannot make you a chip with an AVR core. They're going to...

Chris Gammell: Well, you could with a really big legal contract, but in a bucket of money.

Free Electron: Yeah, you can license it, maybe. No, no. Even then, no, because that's not their business. No, you're going to go to Atmel and Atmel's going to say, you're going to do what? You're going to make a knockoff of my CPU? Weird selling guy. Yeah. You're going to... What's that chip for? Oh, who's your customer? Let me talk to your customer here. And you are going to go out of business. These guys, these guys do not... They don't want to have any other players, you know, stirring their soup. I mean, they are living off that pot of soup, running on their number of cores, and they don't... They don't... Anybody making knockoffs. Yep. Intel has always... And this is the difference between... I mean, that is the reason why you don't... You do not find a knockoff 6802 or a knockoff 6809. Only Motorola made that. You don't find a knockoff pick. You don't find a knockoff AVR. You don't find a knockoff whatever. But you do find knockoffs of every single bloody chip Intel has ever made. Whether it's... Whether it's... I mean, the 8000... The 8080 became the Z80. The whole PC line has been cloned by AMD, Zyrex, IBM, and all these other companies in between. The list is endless. The 8051 has been cloned to death by over 300 companies. The 8048 has been cloned to death by hundreds of companies. Every single micro... Why? Because Intel says, we don't care if you... We make these things as standard products, and these things are now... And for us, end of life. 8051, Intel doesn't make them anymore. We don't care what you want to do with them. We have a ball, guys. Implement them. We don't give a... For us, this is... You know, in the line. Yep. If the guys that are making AVR and pick are not gonna... They're not gonna give away the baby. Because if they give away the baby, pretty soon, Uncle Wan Hung Lo is gonna start making... And he's... Are you damn sure? He's not gonna be paying or asking. Exactly. So they... So they play... They bluntly, they tell you, sorry, Bob. You ain't gonna get out of court. And that's what they do. And yeah, there's... There's people on open cores that, you know, try to... They snuck out of here. Those are... They are like half implementations. Some of them... Some of them... Actually, some of them are... Actually, the 8050... There is a Austrian company that released an 8051 core. And I believe you can find it on open... They released it on open cores, but they also have it on their website. I actually tried it once. So it's a whole... They give you the Verilog source code. You're working 8051, including emulated RAM and ROM. You can just download the Verilog, load it in the Xilinx tool or in the Altera tool, dump it in an FPGA, and it runs straight out of the box. So I thought, hmm, this is cool. I'm gonna, you know, take that thing, synthesize it. But I'm gonna change the memories. Instead of using external asynchronous memory, I'm gonna use the memories who are available in the FPGA. So I tweaked it. I tweaked it a little bit. So it could... It didn't require external RAM and EEPROM. It would run out of emulated ROM in the FPGA and the synchronous RAMs in the FPGA. And then I said, I have the source code to that 8052 AHBASIC interpreter. Let me change the serial port routine. And let's see how far I can drive that FPGA. And I actually had the 8051 AHBASIC running at over 600 MHz. And it was pretty fast. It was incredibly fast. And that was, you know, that was an FPGA that you could do old-fashioned, basically with line numbers. And it would execute with an 8051 core. And that was a fun experiment. But so, yeah, I mean, for stuff like that, yeah, hell, if you want to have... I mean, there's some bozo on the internet that took an 8051, made a PC emulator, run on an 8051. And he's booting Windows 3.1 on an 8051. It takes three and a half hours to boot. If you move the mouse, it takes two minutes for the cursor to move. But it does work. Brilliant.

Chris Gammell: Oh, boy. I love that. And he proved he can do it. Well, and we proved that we can have a three-hour-plus show and still be going strong. But I think we do need to cut it off here. I think either, you know, our audience has... Either... It's doing great. We'll see how they... Either... Yeah, we'll see how they're doing.

Free Electron: There's two scenarios. They either fell asleep an hour ago or two hours ago, like, oh, the old guy is harping on about all the crap he does. Or there's guys like, damn, we need more test equipment. Anymore.

Chris Gammell: Well, we will have, I'm sure, a vigorous debate on all the things discussed here.

Free Electron: Oh, I expect a shitstorm over this one.

Chris Gammell: Hopefully from everyone except employers.

Dave Jones: I'm sure we missed a bunch of specific stuff, which if we want to get you back for a specific show.

Free Electron: We can do that. But, you know, that's what the forum is for. I'm eyeballing the forum enough that, you know, if there's questions, just fire away. If you've got a question that looks interesting to me, there's a pretty good chance I will post a few lines. A few lines. A few hundred lines. If it's a boring topic. Let me put it this way. If you put a question on the forum and I haven't picked it up after the first nine posts, it's probably because I have no interest in it. Exactly. Me too.

Chris Gammell: Right, yeah.

Dave Jones: That's how it works. All right. Well, thank you very much, Vincent. It's been awesome. All right, guys. Officially our longest episode ever. Oh, God.

Free Electron: By far.

Chris Gammell: I don't think we're going to beat this one because, yeah, I don't know.

Free Electron: Well, who knows? Maybe in a year from now, we'll have to do version 2.0. There we go. Because, you know, stuff evolves so quickly that maybe a year from now, who knows with electronics. But I mean, now they're talking about quantum electronics and using graphene transistors and, you know, going the photonic way and very tins.

Dave Jones: We have printed circuit boards before you know it. As in, you know, chip printers before you know it. Printed chips before you know it.

Free Electron: I think that one of these days, somebody is going to take one of these MakerBot machines and figure out how to squirt liquid metal with those things or silicon for that matter.

Chris Gammell: Whoa. Whoa. Do we have someone on my side here? No, we don't.

Dave Jones: Because he knows all the practicalities behind it. Trust me. Trust me.

Speaker ?: You know what?

Chris Gammell: That'll be show number two. Show number two will be there. All right. Me and Vincent will gang up on Dave.

Dave Jones: And you'll both lose. Severely.

Chris Gammell: Probably.

Dave Jones: All right.

Chris Gammell: Thanks again, Vincent.

Speaker ?: All right.

Chris Gammell: Have awesome. Thanks, mate.

Free Electron: Have a good night, guys. All right. Bye.

Chris Gammell: See ya. Bye. Bye.

Chris Gammell: Bye.

Speaker ?: Bye. Bye. Bye. Bye. Bye. Bye.