- AMD to buy Xilinx for $35 billion in stock
- Wasn’t that long ago that Intel bought Altera
- Why such a merger?
- Sharing topography, technology, and manufacturing techniques
- Tariffs and the Election
- Engineers and Manufacturers have been decoupling from China for the last 4 years
- The big question is what happens next.
- If Biden wins, are the Trump tariffs gone? Or will we keep them? What happens if Trump wins?
- Tesla self driving
- Imagine learning there is no autonomous driving and it’s just a bunch of dudes mechanically turking your car
- This could be the end game with Starlink. Work from Home Uber to drive cars!
- Johnny Cab
- Revitalizing old products
- What and how to “modernize” electronics
- Design decisions
- Through Hole vs SMD
- Component Placement
- Design for machines
- Keeping the spirit of the original
- Beware of single sourcing
Special thanks to whixr over at Tymkrs for the intro and outro!
About The Hosts
Parker Dillmann is MacroFab's Co-Founder, and Lead ECE with backgrounds in Embedded System Design, and Digital Signal Processing. He got his start in 2005 by hacking Nintendo consoles into portable gaming units. He also runs the blog, longhornengineer.com, where he posts his personal projects, technical guides, and appnotes about board layout design and components. Parker graduated with a BS in Electrical and Computer Engineering from the University of Texas.
Stephen Kraig began his electronics career by building musical oriented circuits in 2003. Stephen is an avid guitar player and, in his down time, manufactures audio electronics including guitar amplifiers, pedals, and pro audio gear. Stephen graduated with a BS in Electrical Engineering from Texas A&M University.
Host 2 00:18
Hello, and welcome to the macro fab engineering podcast. We're your hosts,
Host 1 00:21
Parker, Dolman and Steven Craig.
Host 2 00:23
This is episode 248. So we have another acquisition in the electronic space. Dun dun dun,
Host 1 00:33
They just they just keep the the numbers that get thrown around in terms of the money. That just happens. It's just it's astronomical, in my mind.
Host 2 00:45
I want to know where you begin with a billion dollars, let alone $35 billion. And what we're talking about is AMD is going to buy Xilinx. And so Xilinx makes FPGAs and a bunch of other stuff. AMD builds, computer processors and graphics processors and a lot of other things, too. But yeah, so AMD is going to buy Xilinx. And it wasn't that long ago, that Intel bought Altera and Altera and Xilinx are kind of competitors and AMD and Intel competitors. And well, I really wanted to know was what has Intel done with Altera since I bought them. And what I can really find is there's some Intel chips. Like not even like workstation, but chips designed for like data centers that have basically an FPGA, like on the die. So you can write custom stuff. Like, I guess, if you want like a custom math processing unit, or something like that, you could write it there. Kind of like an old, like, you know, back way back in dating in personal computing, you'd like you had a separate ALU Oh, yeah. Or different math unit. So I guess that's kind of what they're going with there. And then Intel is also like, sharing gate technology with Altera so like they're starting to like align with like, what how their gate topographies are like I guess that's probably where AMD is going with Xilinx because you're not seeing like, Intel branded FPGAs. At least not yet.
Host 1 02:32
Oh, man. Well, I mean, I guess it's just it's a it's a Gosh, what kind of? That's not vertical integration. That's horizontal integration. Right? Yeah. Horse picking up a new sector.
Host 2 02:43
Yeah. But then, when you think about it, yeah. A FPGA has got a bunch of gates and the CPUs got a bunch of gates in the, you know, when you all boil it down. So I guess sharing, instead of doing the work twice, basically, these companies were separate. You do the work once on improving gate technology or manufacturing or something like that.
Host 3 03:04
I'm sure that they they go extensively through the entire catalog and consolidate everything that they can, right. You know, that would that mean? That only makes sense. Like, pick, pick the best technologies from both and then create something out of it. So have you ever used any Xilinx stuff before?
Host 2 03:22
Way back in college? I, one of my FPGA classes was actually all my FPGA classes were Xilinx. I tend to use Altera just from the Why do I use Altera? For my personal projects, I can't remember why I think it's actually one of my friends who I originally started using FPGAs. I've learned a lot from him. He used Altera. And so I use Altair a lot, mainly because the software, the IDs are really weird in FPGA land. And I never really liked Xilinx his package, like their free version of the IDE that students could have. It was always kind of crappy. The Altera one wasn't that much better, but at least it was better. I had a resource I could go talk to for it, which made all the difference.
Host 1 04:13
Every time I've looked at FPGA IDs, it's always just like, oh god, there's just so much in there.
Host 3 04:22
Like MP lab X. And and what what is the one for? Well, I guess it's still MP lag, because I was thinking back when AVR had all their stuff, but like AVR studio and stuff like that, but even those, those are super jam packed with stuff. And FPGA IDs, put those to shame when you start really digging into I remember I had to do some Xilinx stuff in in college, and I had a TA who was doing like a, an FPGA. I don't know what was like a lab. But they didn't explain anything. They didn't even explain that we were using an FPGA didn't explain what an FPGA was. They just said, this, here's this board, I'm gonna have you do some things on it, and it barely works. And, and it was it was something where, like, he showed us some code. And then if your code didn't compile, he was just like, try it five or six times, like, just keep pressing the compile button. And it might work on the fourth or fifth time, you know, like, my introduction to FPGAs was so unbelievably awful. That like, it sounds like it. I didn't even I just like, I just knew that there was a thing called Xilinx board. And it was unbelievably painful.
Host 2 05:49
What was the class for?
Host 1 05:51
It was digital logic, which was one of the biggest jokes known to man, that whole class. I squeaked by with a C in that class. Oh, it was it was so awful. It was unbelievably terrible. That that was one of those ones where it's like, Man, I'm paying $300 an hour for this. Like, it's one of those classes that makes you do that. Yeah, it was. It was just it was terrible. And I feel like I feel bad knowing what I know now about things which admittedly is limited, but like, even what I know now, I'm like, Good God, you think they might have like, even gave me an introduction to like, Oh, we're trying to get you to learn how to do this. So maybe you could do it in the future.
Host 2 06:33
Yeah, my, my digital logic class that we were learning FPGA we learned on Xilinx as well, like, our project at the end was like to make a snake game and hardware on the LPGA. That's because that was a lot of fun. Like EA had to make a display driver. I say drivers they're not drivers are their logic gates driving a, a, the signals for a you know, CRT monitor. Hey, that counts, right? Yeah, it was a lot of fun.
Host 1 07:02
Wow, that's awesome. I wish I got to do that.
Host 2 07:06
I wish I need to revisit that again is like see how much like, can I still remember how to do that?
Host 1 07:12
Yeah. So did you just were you just bit banging like a VGA cable?
Host 2 07:16
Basically, in the Yeah, you would build hardware that could generate the the stream? Basically, it's good way to put it. Cool. Yeah. But it's, the VGA isn't an analog signal. So you had like a, a, right? A digital to analog resistor ladder, right? To do that, and then so then you just built the pulse frequencies for, you know, horizontal and vertical sync and all that good stuff. And yeah, it was a lot of fun. And then you had you had to make a PS to interface for keyboard. So you could control the snake on the screen. Wow. It's a lot of fun. Yeah, I guess. It's kind of like, it was almost like a senior design project. But you got to do it as like a sophomore. Because you spent like, basically a whole year working on that. Were you in teams? Or was it individual? It was individual. That's cool. Yeah. I liked it a lot. I really, that's actually I'm actually surprised. I'm not more into doing like FPGA stuff.
Host 1 08:21
Yeah, I've only ever seen you do one FPGA thing. I think you'd be lbds driver. On what the pin heck board?
Host 2 08:30
Connect board way back in the day. Yeah, yeah, we're driving a 1366 by nine no 1366 by 768 768 LCD monitor for my laptop. Because you can buy those monitor or those LCD panels for like dirt cheap, like you can buy it for like 40 bucks. Anyways, so probably see that kind of stuff from the Intel you know, that from Intel. AMD Xilinx is like, consolidating technologies. Basically what it is, you're not going to get Zion AMD branded FPGAs or you're not going to get Xilinx branded graphics processors. That way you just get like what's under the hood will start to assimilate basically.
Host 3 09:15
Assimilate I like that okay, so get this. So I've got an interesting topic here that I kind of want to get your opinion on on some things. I'm actually working on a project right now. Well, I'm actually multiple projects that involve no
Host 2 09:36
Vitalizing old working on a project Stephen
Host 1 09:40
Oh, God you like yeah, no, I mean, I did. I just go to work and watch YouTube all day long. She don't tell anyone to get this. So I got I got a couple projects I'm working on right now that involve revitalizing some old not tech. analogy but but basically old products that we want to bring into modern electronics. And I'm using arrows there because you know it. It's ancient technology that we want to basically make manufacturable even though it is technically manufacturable, we want to make it more manufacturable. We want to make it small company 2020 manufacturable,
Host 2 10:25
You wanted to make it so it's more inexpensive to make.
Host 1 10:30
Well, and, and all that comes along with that, yes, but we don't want to end this is this is a word that we use at work, we don't want to lose the spirit of the original design. So you're like, This is great. Yeah, so what I
Host 2 10:53
Know, I know where you're getting that. I don't like putting it that way. Because it sounds very marketing risk. But I do know where you're coming from, I do know where you're coming from.
Host 3 11:05
Well, but but get this here's, here's, here's a decent parallel, let's, let's say that you were contracted to remake the Nintendo Entertainment System, let's say they were going to revitalize that. And you were the engineer to do it. And you were in charge of controller design. And you wanted to make it modern. But one of the things, one of the design criteria is that we want the spirit of the original controller to be there. So you want the feel and the buttons to be exactly the same. That's what I mean with the Spirit, like, because it would be very easy to just pick a new solution, and then it might feel crappy, and it might feel just not right. And it wouldn't feel like the original Nintendo controller. So kind of the same juju, I guess you could say what these projects. So one of the ones that that coming to mind right now uses multiple PCBs, that are through hole technology, they have cable assemblies that go in between them and things, very Allah,
Host 1 12:11
Multiple decades back kind of mentality, where, you know, I have a power supply board, and it's its own board, and it gets bolted to the back of a chassis over here, and it gets cable assembly to multiple different boards, or they get daisy chained through multiple connectors and things. So my mindset when I'm thinking about these kinds of things is Okay, I gotta make this work for our manufacturing process. And I usually cut tried to think of the guy who's trying to build what I make, when I look at this kind of thing, because, frankly, for the most part, I actually know the guy who's going to make what I design, and I and I'm thinking, Hey, Billy, I'm just using some, some random name here. I don't want to make Billy's life awful. And I have the direct control over being able to do that. So I don't want Billy to have to sit there and crimp wires and make cable assemblies, I'd rather just make everything more monolithic.
Host 2 13:14
You'd rather let Billy just watch YouTube videos at work.
Host 1 13:17
Hey, you know, that's the goal for everyone at work right now,
Host 3 13:20
Right? So here's some design decisions that I'm kind of approaching this. And I think this is a good idea. If you see something like say you go to GitHub, and you see a circuit and you're like, hey, I want to kind of revitalize this or make it here's some things to start thinking about when considering like, maybe you're going to make some and send this off to a manufacturer through hole versus SMT is really a big
Host 2 13:48
Oh, yeah, I know this battle
Host 1 13:51
Through hole is expensive, it's difficult, and most people don't want to do it. Especially if it's like through hole resistors and things like that, that a manufacturer is going to look at you kind of with a raised eyebrow, if you just have a board filled with through hole resistors when that's not required, right. So if you can push things to surface mount, go ahead and do that. That's a good idea. Actually, I had a conversation with a guy just the other day about doing some through hole and surface mount conversions. And this particular guy was very interested in through hole stuff because the spirit was in the through hole components and that's totally fine. But I let the guy know like, you know, maybe the spirit comes with an extra price tag. You just gotta you gotta watch out for that one.
Host 2 14:44
Spirit comes with Casper cost a little extra
Host 3 14:48
Just a little bit. Yeah, yeah. I mean. So the next the next design decision that's that's worthwhile. Keeping in mind is component placement. So say you You've got this design, like I mentioned earlier, you got five boards, and you found some fancy way to consolidate them down into one board. But if you start putting components all wacko on two sides of the board, and, and just go nuts with things, you can actually end up adding a ton of cost into it. So whenever it comes down to manufacturing, if you can shove everything onto one side of a board, that's one setup. That's one stencil. That's one manufacturing run, that's guaranteed to be your cheapest option.
Host 2 15:32
When one trip to the aeoi. One, yeah. Done, you basically have to work.
Host 3 15:38
Right, right. Not only do you have to work, okay, so I don't, I wouldn't say that you have to work you've done the amount of work that people expect. As soon as you put parts on both sides of the board, you've doubled the work. Okay. Yeah.
Host 2 15:53
That's a different way to think about it. But yeah, sure.
Host 1 15:58
And, you know, I get a lot of quote, requests at work where I see people, they've done like, 99% of a good job where they put all these components on one side of a board, but then you flip the board over, and there's like, two surface mount resistors. And, and it was just maybe at a convenience, or maybe they didn't understand the the impact of that. But I A lot of times, I have to explain that, that you've doubled the amount of setups and the amount of and you've significantly increased the manufacturing time on that. And so, you know, even if you have some trouble with routing traces, find a way to get those onto the other side of the board, if possible. So the next design decision that's important here is designing for machines. This kind of goes in line with the theme we've been talking about for a handful of months. Now. It's like if you know what your manufacturers machines capabilities are, or what the requirements are, start from the beginning on that. So kind of with the component placement, design decision. If you're if you're maintaining your clearances, if you're looking, looking out for things like selective solder ability, or wave soldering, and making sure all your components are, are obeying all of those rules, you're going to be setting yourself up for a lot more success. So if you don't know what those design criteria are asked, and ask way early on in the game, because it sucks to change things way later. The next design decision there is watch out for packages. And what I mean by that is what what component packages, a lot of times, it's convenient to put 1206 size, passive components on your board, but they eat up a ton of space. And they're not necessarily the cheapest option. So if you can use an O six, or three, or an O 402, were aware of 1206 might seem like it's more convenient, well, that's probably a better option, that that extends also to like op amps and, and other digital ICs. And things. Look out because you know, a T sub package might actually be, you know, X dollars, but an SLIC package might be cheaper in that case. And so if you have the ability to shuffle your bomb around and pick low cost components based off a package, a lot of times that's worthwhile also. So then keeping the spirit of the original so kind of what I what I go at with this whole concept is I'll take a design and I look at the schematic and I look at the things that are the absolute critical items for keeping the spirit. So in other words, like say, I'm doing a revitalized old guitar pedal or something like that, I'll look at the signal path. And I'll look at all the things that kind of abuse the signal and massage the signal. And I'll make sure that those components are identical. And make sure that everything works exactly the way that that should. But things that just do housekeeping things that just do like power rails or whatnot, or op amps that don't have critical functions to the tone like say buffers or things like that. Those are ones that I can massage cost out of and make turn them into surface mounted making them a lot easier. So kind of keeping that idea of like what are the the things that I absolutely have to keep, I don't know, vintage and special and magical. Those, I won't modify but everything else then I put through those design decisions that I just kind of worked through. And that that one product that I'm talking about that has multiple boards and cable assemblies, I'm already crushing it down into a single board that has one side SMT Uh, stuffing, and one side through hole stuffing, such that this one product that would have to go through our line five times, if not more for double sided. And all these cable assemblies and stuff now goes through our line one time. So just some things to keep in mind if you are going to start taking an old product or taking even even a new product and kind of revitalizing it for manufacturing.
Host 2 20:27
I do like that that term, though. I'm starting to warm up to it.
Host 1 20:31
Keep in the spirit spirit. Yeah, you know, we, we actually use that also, when it comes to our design. What is our design meetings, we usually use that word in relation to feature creep, because so we do at work, we mainly do modular synthesizers, and modular synthesizers, the way you play them is you physically take cables, and you connect signals to other signals, and you make sounds based off of how you patch things together. And that is the spirit of modular the spirit of modular is that you are in control by the patch things. It's really, really, really easy to start feature creeping and start connecting those wires inside of a module for the player. But But we always have to back off and say, Well, no, that's not our decision that needs to be the players decision. So we always have to keep the spirit of modular alive by saying, like, we could do this for you. But we're gonna instead give you the jacks, such that if you choose to do it, you do it.
Host 2 21:39
Yeah, that makes sense. So when you fry a component, and the smoke comes out, that's the spirit being released.
Host 1 21:48
Well, I mean, yeah, clearly, there ain't no spirit no more.
Host 2 21:51
That has no spirit anymore.
Host 1 21:54
So actually, I'm curious, what is the spirit of pinball?
Host 2 22:00
Oh, in hardware for me, so for me, it's just how well the play the game plays. Yeah, like the hardware that's underneath, like, doesn't matter at all. And the end, that doesn't matter. No one cares. So the whole reason of building a Pinball Controller is to make that part of the system cheaper. So you can make more margin on the pinball machine. The spirit is cache.
Host 1 22:27
The spirit is through whole MOSFETs. Right.
Host 2 22:29
Host 1 22:33
That's where all the magic is?
Host 2 22:35
Yeah. No, that's actually that's the biggest change. We went on penetrator from pintech dependent Tara is getting rid of through hole like across the board. The only thing left is connectors.
Host 1 22:47
Yeah, you guys, you guys went to single side SMT load, right? It's not
Host 2 22:50
Single sided. SMT load, it's actually pin heck with single side s&t. This is not mainly for because we shrunk the board down, it's cheaper to shrink the board down, in our case, then is to go to single sided load. Just because the board is just so big to begin with the way
Host 1 23:07
I've always handled double sided SMT load, which I do on occasion is if I'm willing to break the seal, and go to double sided load, then I go nuts, like then like, as soon as you do it, like, take not 50% of your components, but start throwing stuff on the back because it makes no sense to put one or two components on the opposite side. But as soon as you do put tons over there,
Host 2 23:32
You're 100% Correct. But on our backside, we just got some there's some resistor arrays, and then some diodes, there's probably about 40 components on the backside. But the big thing is to remember is if you're going to do two sided loaded SMT is all the heavy components need to be on one side. And then the other side needs to be in quote light components. So you don't they don't fall off and reflow and make your CM angry.
Host 3 24:06
Well, and I you know, it's been a while since I've been at macro fab but macro Fed did charge based off of double side load, right? No, no, they didn't. Okay, like, because, well in terms of if you're doing production they they might correct and production but yeah, but But yeah, for for prototypes, like because the way it works where you get to raid with other people. Yeah, every panel is double sided load.
Host 2 24:34
Yeah, basically. Yeah. Cool. So we have a topic here that we need to stay apolitical on. I don't know how we can, how we can but we're gonna try. So Misha, our CEO at macro lab actually sent me this a topic, and I thought it was enough at least we can just bring it up and chat about for a little bit
Host 1 24:56
Is because the apocalypse is happening happening next week. Right? As the moon is turning to blood and the oceans are drying up,
Host 2 25:03
And all the planets align and big laser beam fires out of the sun, right, right, right. Yes. So the election is in the United States, I should preface because we're a world, we're in a worldwide podcast. The election in the United States for president is next week, actually, like, six days after this podcast comes out. So one thing is, it's just very interesting is so the last four years as engineers and manufacturers here in the United States, we've been basically decoupling from China for four years. And I just kind of want, like, some thoughts and stuff on like, what happens if, like Biden, so Biden is one of the president candidates, and then Trump is the other candidate. And there's, there's a lot of other candidates, but those are the two that will probably go into when it's one of those two. So if Biden wins, are Trump's tariffs gone? Will he keep them? What happens if Trump wins, etc, etc? Or because those tariffs went in place in early 2019? Are we are coming up on close to two years with these Chinese tariffs? So yeah, good thing would be like, so how's it how's the current tariffs affected? You? I guess?
Host 3 26:30
Well, I mean, a lot that we purchase, at at work gets gets affected by the tariffs. So. And actually, I have a story just from the other week. In fact, I think I talked about it last week, podcast or week before I was talking about some sheet metal quotes that I got. And just for fun, I got a handful in the States and I got a handful in China. And the Chinese quotes, some of them were really attractive in terms of the cost, but then as soon as you start adding up shipping, and then 25% tariff on that, they start getting in the same ballpark as the American quotes that I got. And then you start weighing the question of like, well, okay, so I can talk in the same timezone to this person, there's not a language barrier, and I can just drive over to the place and pick it up, you know, at least for like, you know, the small quantity stuff that I was dealing with. So
Host 1 27:29
We deal with it quite a bit at work when doing quoting and things of that sort, where, like, now we kind of have to just add this arbitrary 25% to anything that goes for quoting which it's annoying, but you know, it's there.
Host 2 27:45
Yeah, for on my end, I'm not even going to dive into McWrap, just my personal projects, like pin, like Penetang, one of the big things I did with Penetang is none of it is sourced in China, because I didn't want to deal with that stuff. So it's all Taiwan, Japanese components, or Malaysia, and PCBs at Taiwan. So if you do that by looking at, you know, country of origin of components, it's interesting is that most most distributors don't tell you what that is, until we like you order the part. We always think it's kind of weird, but I guess they don't know if they have the order. But they have stock, because like Arrow arrow is the distributor that will tell you what their COO is Chief country of origin for a component is that they have on their warehouse floor. I just wish more more distributors would do that. Here that DigiKey Mouser.
Host 1 28:47
I'm sure I'm sure you can find it out. Right. Yeah.
Host 2 28:50
Well, a lot of times you can contact the manufacturer the part and they will tell you, but but they but component manufacturers have to put it on the components. So you get it when you get the parts. You can tell where it came from. But usually most manufacturer most distributors, say don't advertise what that part where that parts from? That's what I've been doing is just like reducing how much stuff I use that China in terms of component sourcing. Now, there's some stuff you can't, you know, avoid, but try to do as much as possible. But um, so yeah, I don't I don't know what will happen because I don't think I did some Googling and I couldn't find a source of basically, if Biden wins what happens to the tariffs? Because we know if Trump wins, it's just going to be continuation,
Host 1 29:43
As usual. Right? Yeah. Oh, here's my prediction. If Biden wins I think Biden would get rid of the tariffs. And only this might be a little too political here but but only for for the fact that they're Trump's tariffs, just because like Trump bad get rid of tariffs that That's, that's the concept. Trump did
Host 2 30:03
The same thing with all the stuff that Obama did if it was not a law, like it was an executive order or something like that, which is what these tariffs? I don't because the tariffs weren't even. They didn't go through Congress that day. I frankly, I don't know. I don't remember either. It doesn't matter anymore. But I do. I do think
Host 1 30:23
It's that black and white. Yeah, I would I would put money on that. If Biden wins tariffs go away. If Trump wins, then business as usual. Yep.
Host 2 30:33
Biden does have a thing on his website called Made in America plan. I didn't look at that yet. Check that out. I, I asked Misha for a topic, like 15 minutes for the podcast. And he sent me this. And so I did not get a chance to read that section of Biden's website. But again, I haven't seen a political candidate for presidency that actually said, actually did what they said they were going to do. It's been quite a long time. So I don't know what he actually what Biden promises in the Made in America plan? I don't know. Yeah,
Host 1 31:13
I'll have to check that out myself. Because I'm curious what what his thoughts on that are and what that looks like?
Host 2 31:19
Yeah. Because I really want to know is does our 25% tariffs go away? That's Oh, that's actually the other topic I want to talk about is my biggest beef with the current tariffs. Is it's interesting what, because we went we had a whole podcast talking about like, what they covered and stuff. It's because they cover components and like sub assemblies, but they don't cover. I think we actually did complain about this before. Probably talking about this, where they don't cover fully built things like toasters. But components for toasters, that's 25%. But
Host 1 32:01
Bagel pin ICs, that would have a 25%. Yeah, it
Host 2 32:04
Does if it came from China, right, right. Yeah. And so it's just stuff like that's like, if they if the tariffs were really meant to help, let's say everyone in America, in terms of like blue collar workers, or assemblers or manufacturers, it would be all like anything that was related to an overseas toaster, right? Because what the hat what happens is, it makes it so that the little guy that's only can afford to build sub assemblies in China now gets hit with that 25% tariff, but the big box toaster maker can just build them in China or, you know, fully built and then ship them here with no tariff. So I don't know if that's changed in terms of the full built stuff is tax to now or tariff now. All right, what are you saying? It's a tariff? I don't mean it's a tax. So
Host 1 32:59
I don't know. I think the what doesn't the Constitution handled us separately? Why are fintax?
Host 2 33:06
Yes, it does. So technically, the only thing that they that the federal government can get money from is tariffs, but whatever. We already crossed that bridge 100 years ago.
Host 1 33:23
Yeah, they actually I'm looking at I'm looking at Biden's page now about
Host 3 33:28
The Biden plan to ensure the future is made in all of America by all of America's workers. That's the title of the page. And it's considerable. It's very long. If you want to go check that out. Just I guess, Joe biden.com. And then you can you can search through and find that guy. Yeah. So a lot of a lot of stuff about Trump's tax cuts, and then how to
Host 1 33:57
It's all the same in the end, right? bring jobs to America. From both sides. Right. Both sides. Right. Great. Yeah.
Host 2 34:06
I haven't read it yet. So I'm going to read it probably after this podcast. So just see what's going to happen with tariffs or whatever, because I did search like Biden, Trump tariffs. And like, there's no official word on like, what he would do. Yeah. So
Host 1 34:21
Yeah, I guess you never know. Maybe you'll just hang on to it. And just like, Yeah, let's work. Yep. If it did, I don't know.
Host 2 34:28
I don't know either. All right. Let's get out of this. Go away election. Yeah. Bye. Bye. We'll
Host 3 34:34
See you next week. Yep. So I got a just a quick little news report that kind of goes back towards manufacturing, but it just kind of beware of single sourcing, because we got bit by this. Well, in terms of our manufacturing, not our own products, but
Host 1 34:53
The Japanese semiconductor fab of AKM As the name AKM, I'm not even going to pronounce or pretend to pronounce what that stands for. But
Host 3 35:09
The factory actually caught on fire. I'm looking at the picture that does not look good. No, no, it caught on fire on the on what is October 22 is when this news article came out the AKM factory caught on fire. And at this factory, they manufacture codecs, Dax, and ADCs for the audio industry. And they are a single source of that, and they are completely shut down now. And the world is just scrambling to purchase the codecs that they that they manufactured there, because they're some of the better ones or some of the best ones.
Host 2 35:43
So they're, you're talking about like an audio codecs. Chip. Yeah. Yeah,
Host 1 35:47
That's right. So and we have certainly been impacted by that, because we do We do manufacture with those parts. And what's interesting is, this, this website, we have a link that we can post that just, it's a Japanese website that translated to English, which, if you've ever done Japanese to English translation, it it's a really difficult language to translate word for word, but in the first paragraph, they say, let me see here, the fire broke out. Yeah, at the the fighting the fire was difficult due to firefighters complaining of skin abnormalities. Which, I don't know if you've ever if you've ever seen the stuff that's used at semiconductor fab, like there's some nasty chemicals there. So it's like I don't I don't know if that's a poor translation or something to be scared of there, you know. So we'll post this this article, if you want to check it out.
Host 2 36:45
I do like how it Yeah. Translating Japanese to English is it's just like Chinese English. It's just it just the language is don't are not comparable structure wise. And so you get really weird stuff like also is does not stand the prospect of extinguished in the 22 days midnight current older than time.
Host 1 37:09
Actually, there's a, there's a really great website, if you if you ever bored, called translation party. Check that out. It's basically you write an English sentence, it translates it to Japanese. And it's, then it takes that back translated back to English. And it goes forward and backwards until it reaches an equilibrium. And sometimes it takes hundreds of forwards and backwards. And it just butchers the sentence. It's check it out as fun. So the moral of the story here is be careful about single sourcing your critical components in your in your devices, because if you pick a codec that's manufactured in one location, even if it's the best in the world, like you're down right now, well, then
Host 2 37:53
How do you? Do you just sell a product with an inferior codec, but you'd have to do a whole new design most time? Yeah,
Host 1 38:00
Exactly. And that's kind of scary. So we'll just what this what this means is when you're doing your initial design work, contemplate using a codec that is fabbed in multiple locations.
Host 2 38:18
So the second paragraph here actually says what it's what the skin irritants skin admin now at aboriginality That's how, yes, yeah, Whoo, that's a word that comes from it's actually hydrogen chloride generated from burning vinyl chloride that was used on coating electrical cables in the factories.
Host 1 38:41
Well, they said that's what the offensive odor is, presumably.
Host 2 38:45
Well, hydrogen chloride would also cause like your skin to like, get eaten. So a small amount of flammable materials used in manufacturing high density, integrated circuits are placed inside the building. And although a small scale explosion may occur, there is no danger. At least no one was injured. That's a good thing. Yeah. Yeah. Wait. This complain about skin every now. And then no one was injured. Okay, whatever.
Host 1 39:15
Well, that's the firefighters. No one was injured in the fire itself.
Host 2 39:18
Ah, gotcha. Gotcha. Gotcha. Yeah.
Host 1 39:22
So, yeah, keep an eye out for for that it sucks. Designing stuff is not fun, because you have to contemplate so much more than just the design.
Host 2 39:31
Yeah. Well, especially if you're designing and you're maintaining the lifecycle the product. Yeah, because sometimes that's a different, that's a different job. Depending on the company and whatever. But yeah, if you're maintaining, if you're the designer, if your designer and the person who's maintaining the lifecycle is the same person or on the same team, stuff goes a lot better. Well, and
Host 3 39:55
Also whenever you see things like this happen, where you know a few days ago when this how happen and you see a pop up on the news. Start searching for stock and buying up stock as soon as you can, you know, yeah, stock of your components.
Host 2 40:09
Alright, last topic for today. It's on Tesla self driving. And so there was a kind of like a goofy video that was on the subreddit called Wall Street bets. Some people might know what that subreddit is, some people might not. It's basically a bunch of people posting about, like, they're really bad trades on Wall Street. It's about the best way I can put it, get some really good memes out of it. But it was right when Tesla announced their new self driving stuff that they're rolling out to some customers basically, like the drivers that the that they thought were safe enough to get it got this update. It's like a beta I guess. Okay. And so this was like the self driving portion, just like it was a was thinking some flags in the distance were like, street lights. Okay. Okay. So it wouldn't let the car move. Right. Anyways, like, there was a comment that was Imagine, imagine that self learning that there's just, there's no self learning or autonomous driving. It's just a bunch of dudes and like a call center that's actually driving your car around. And this actually got me thinking
Host 1 41:24
Is kind of kind of like the military with drones, right? Or they're just gonna blunder in Arizona or something.
Host 2 41:30
Yeah, that this is actually got me thinking is, what if this was the actual end game? For Elon Musk? Because he's building because Elon Musk owns Tesla. And He's launching all these satellites to make I think it's called Starlink. Which is that low orbit? Low Latency satellite internet. You could drive someone else's car anywhere in the world using Starlink.
Host 1 41:56
You mean, like you would have? Like the the technology to do that is what you say? Yeah,
Host 2 42:03
Yes. And so instead of putting all this effort into making AI, what have you just said, ginormous? Like, you could work from home and then do your Uber. Uber from home?
Host 1 42:15
Hmm, you know, I would totally buy a cabbie hat and just wear it at my at my desk as I drive someone's car with my wsa D on my keyboard.
Host 2 42:27
This has actually got me thinking is like, so when you're in a taxi cab, or an Uber or whatever, like, there's someone there driving you if that person wasn't there, but he was so or he or she was still controlling the car. Like, they can make some decisions because their life is not at risk here. Oh, yeah. Right. Right. Like they might try to slip into traffic more, like quicker or whatever.
Host 1 42:59
Yeah, I guarantee you they would. Yeah. Also a flat screen in front of you does not give you enough cues about
Host 2 43:08
Reality of it'd be a VR goggles set.
Host 1 43:12
You know what? Actually, you remember the car the Arnold Schwarzenegger movie on the on the Mars with a terminal not terminal. Gosh, which now we got to look it up now. Everyone's probably yelling at me but like arts this movie.
Host 2 43:32
Always the one where they go on? It's a total recall Total
Host 1 43:36
Recall Total Recall? I was thinking terminal velocity, but that's not no. Total Total Recall. Yeah. Where they have the the cab drivers where it's got like that Android in the in the front. But like somebody could put on VR goggles and then get back that androids vision? Yes. What are those called? Taxi. What is it called? Like Jerry cabs, Johnny cabs. That's what it was Johnny cabs.
Host 2 44:02
So. So we have this thing called Mechanical Turk. You know what Mechanical Turk is? I do not know. So Mechanical Turk is like a, it's a, you make a repetitive task. And then you farm out that task to set a price, a program called Mechanical Turk ing. Okay. So like, if you took a bunch of data, and you just needed someone to pull this one line out of the data, and you there wasn't a very good way to make an AI to do it or a script to do it. Like it took a little bit of context for someone to do it. You would set up this rule set and instructions for people to mechanically target and they get paid per action, basically. Okay. So yeah, what if we had self driving cars? But there were not self driving they were driven over Starling. by a bunch of mechanical Turks, it still
Host 1 45:01
Applies. They're driven by a self somewhere, right?
Host 2 45:05
That's true. Yeah.
Host 1 45:09
I don't know. That's okay. So in terms of the level of like uneasiness in terms of
Host 3 45:17
Self driving cars like, I want to drive my car, I feel uneasy if a robot driving my car, I would feel extremely uneasy if someone halfway around the world was driving my Uber.
Host 1 45:30
Because you're totally right, like, you know if something goes wrong, well, what do they care?
Host 2 45:34
They just log into the next car.
Host 1 45:38
And it drives past your wreck.
Host 2 45:42
No, it pulls up to your wreck expecting you to get out of the wreck. Yeah.
Host 1 45:45
Get into that. Yeah, that sounds like an awful idea. Parker.
Host 2 45:52
Yeah, it's pretty bad. I just thought that's like, my first thought of him like, Man that it would technically work. Just yeah, getting people on board with a computer driving you around is going to be the first hurdle.
Host 1 46:06
What if What if Microsoft Flight Simulator was actual airplane somewhere?
Host 2 46:12
You know, I had a friend who tried to, to make me believe that back in the day like back when I was like, eight years old. Really? Yeah. That's great. But then he was Yeah, exactly. Oh, times affinity a double dog dare you crashed that plane into the mountain?
Host 1 46:33
You kill a bunch of people.
Host 2 46:37
Oh, man. Alright, let's wrap up this pot. Yeah, I think so. So that was the Mac fat engineering podcast. We're your host, Parker, Dolman and Steven Gregg. Let everyone take it easy.
Transcribed by https://otter.ai