Stephen and Parker rant (again) on bad datasheets, ARM being bought out by SoftBank, and general project updates.
Podcast Transcript
Host 3
Hello and welcome to the macro fab engineering podcast. We are your hosts, Steven Craig
Unknown Speaker
and Parker Dohmen. And this is episode 25 or 2525. So Steven, what are you been working on?
Host 3
This week has been kind of a hodgepodge. The macro ramp PCB finally came in. I kind of cut a deal with our manufacturing floor where I said, You guys put on all the surface mount stuff. I want to do all the through hole just because I don't know if through hole soldering is kind of therapeutic for me. So So I've got all the parts in the holes, nothing soldered yet.
Unknown Speaker
Yeah, it's all wireless. You know, bent over.
Host 3
Yeah, yeah. And I want to take it back to my shop and I'm going to do a whole bunch of pictures and that should all pop up on on Twitter coming up here soon. So Mac ramp isn't
Unknown Speaker
also you take into your shop just so you can use leaded solder
Host 3
that is that yeah, that isn't a thing I love soldering with lead solder is so much easier.
Unknown Speaker
Host 3
you don't have to put flux on it. And you can just kind of anything will melt it, I love it. It's fun. So got also got the SSPs analog board in this week. So kind of fun week for getting getting boards in. So this is the this is the all the analog side that interfaces with the digital and does all that nice jazz. So now we're finally put it mocking up a bit of the energy on cube that we've talked about so many times and will be thrown something together here soon with it. The way it works with our power supplies, we've really, it's hard to test because we need so many inputs and at high voltage that really the best way to test it is to just kind of hook everything up. And you know, bring it up to voltage with a with a current limiter and a variac. So we'll be hitting that soon and see what smokes first. Yeah, right. I did already test the high voltage rails by just not powering anything else just putting high voltage and nothing smoke. And it wasn't pulling. It pulled a little bit of juice. Nothing. Nothing bad. So that's good.
Unknown Speaker
Yeah, um, I guess I need to get the part numbers for the DAX that are on that board. Oh, yeah, I can run my prop driver. Yeah,
Host 3
Unknown Speaker
to you. I can probably hammer that out in like, two hours max. It there's nothing special. Yeah, yeah. It's what SPI right? No, I
Host 3
think our I squared C was IC. I think so. Even easier. Yeah, no, I don't think I chose spy. So above and beyond that, I've actually been putting a significant amount of time into some DFM work for a customer of ours. DFM meaning designed for manufacturing. So occasionally, customers will ask us to review their board. And that that entails looking at every element on the board, checking every footprint, looking at every datasheet. So it's a lot of work. And it takes a lot of time. So I've been spending a good chunk of my time on
Unknown Speaker
instead of it's the first time you've used Altium as well.
Host 3
It is yeah. So good. This customer gives us the board. And they're like, Hey, can you do DFM work? And we're like, what package Altium? And I'm sitting here like, yeah, I can do that. So I download the two week free version. I'm like, Okay, I've got to do this in two weeks. So I had to effectively learn Altium and then check this guy's board. But the funny thing was, maybe it's just I have enough experience with other EDA tools. I was I was flying in Altium within maybe 20 minutes. It's really, it's generally intuitive. I mean, I don't think I'd be fantastic at like designing a board from scratch in a short period of time. But in terms of navigating and finding all the things I need in order to check his board. It's
Unknown Speaker
easy. That's cool. Yeah. So do you like the software package?
Host 3
If I had 15k to drop on Altium? I would do it. It's awesome. It is it. I don't know, I don't want to push too much on it. But everything I'm using on it. It's like this is so killer. The only thing that I'm really annoyed with right now and just from navigation is that I don't I can't find a really simple way to change layers. I have to change whatever active layer I'm looking at. And I find that kind of annoying. I did trace all the number of buttons are your layers. And so if I want to swap to another layer, I just press a button bam there with Altium you have to activate them. You have to hide them. You have to choose Layer sets and all this crap. It's like that in Eagle. Yeah. So choose what maybe it's just that I I'm complaining about something that I don't need to complain about. If you tried the function keys, the function keys do something
Unknown Speaker
else. Okay. Switching layers.
Host 3
I'm sure there's an easier way and there's probably one of our listener who's just like, all you do is press this one button. Yeah. Although actually the letter Q on the keyboard swaps between mils and millimeters. And like all of this is amazing.
Unknown Speaker
That could end up really badly though. It could. It could Yeah, if
Host 3
you're not paying attention, you're not paying attention. But I think it's it's awesome because a cut some of the customers parts that he put down, some of them are in mils, which is odd. That doesn't really happen. I mean, where they only show mils on the dimensional drawing, so it's just like, oh, okay, great. Q. I'm already there.
Unknown Speaker
No need for converting? Nope. So what you've been up to the, the macro watch, which was the thing we gave away at last year's Houston Maker Faire? Yep. I've been working on a rev to basically make it even, you know, making it more cheaper, I guess, reducing the bill of materials. It was already cheap. It was already cheap, but I'm making it even cheaper. Yeah. So I'm in the old one uses a pic. 16. Yeah, I switched over to Silicon Labs. EFM. Eight slip up. Mainly because that is the cheapest microcontroller. Pretty much you can buy. Yeah, it was something dollar 60 cents 60 cents in signals. It gets up to like 38 cents in quantity or something like yeah, it's ridiculous. It's got a what's built into this thing is ridiculous as well. It's got a built in real time clock. It's got built in. It's crazy timers, it's got like four or five different timers that are all like different. All configurable from like 812 16 bit. It's got a built in ADC. It's got a built in
Unknown Speaker
I can't remember what the last thing. Oh, power mount management. Okay. It actually can figure out how much power it is consuming by itself. Wow. Yeah, it's good, um, power manager module in it. They just have to enable
Host 3
that's kind of crazy. So it's a kitchen. It's sort of a kitchen sink. Yeah,
Unknown Speaker
it uses a Ed 51 core, which is kind of old school.
Host 3
It's really not kind of old school. It is way old school. Cool. Um,
Unknown Speaker
but it works great. The I mean, the thing only pulls how I have it configured. It only pulls 2.9. Mac micro amps. Yeah, that's nothing. It's nothing. The old macro watch pulled 15.8 My grandpa? Yeah. Which is still not a lot. But we have a you know, a eight time saving on power. Yeah. So I think the old macro watch was like eight was about a year worth of battery life. This gives you eight years battery life?
Host 3
Depending on how often you check.
Unknown Speaker
Yeah, how often you check, because when you do turn it on the LEDs that draws up to 100 milliamps, right? Yeah, up to 100 milliamps, which is quite a bit that's, that's a ton before only a short period of time. Yeah, about five seconds. And
Host 3
I remember I was watching a YouTube video a few months back was talking about this chip. And they they chose the at 51 core. They didn't have to go that route. But their lead designer chose that because there was so much support out there. And there was already tons and tons of code already written for it. So it's kind of just like a hodgepodge in a way. Yeah, but it's, but it's all been like tried and true and proven.
Unknown Speaker
Yeah, the one room like about their chips is you can pretty much connect any of the peripherals that are inside the chip through any IO pin. They have this thing called the crosslink is what they call it. And you just connect to what you want through what pins and then hit go. And it does Oh,
Host 3
it's sort of like a stripped down prop that has peripherals internal.
Unknown Speaker
Yeah, it's a lot like a cypresses P rock system is a lot like this too. But the P rocks cost quite a bit more. So so far, I've been pretty happy. It was some weirdness with their they have a IDE called Simplicity Studio do or something like that. And it's based off Eclipse. So it's like it's a fairly standard IDE except it's got some weird, it kind of looks like MP lab a little bit. I think MP lab is built on Eclipse as well. Yeah, everything's pretty much.
Host 3
So you are having you're having some a little bit of trouble with it. Right.
Unknown Speaker
Yeah. So the documentation for this stuff is not the best least so far. First of all, took me a long time to find the actual datasheet for this chip. You keep finding like the family stuff and all that stuff. Like no, I don't want that stuff I want what registers do I need to hit to make this stuff tick? Right. nitty gritty, yeah, the nitty gritty stuff. So I finally found that and that fix some of the issues. But then I came across the issue why I couldn't drive the IO pins. Like I was trying to just basically blink an LED at this point, why can I blink this LED? And I looked at their code for Blinky. They have a blinky code, and it works. And so I copied the code directly into my project and it didn't work. And I'm looking, it has a really good debugger. It's on like, it's running the code and the variables are correct. And it's hitting the output register, but the pins not going up. But the pin is not, it's either going high or it's not going going low. It's at 1.5 volts. So it's, it's high Z, well, high Z, you just be it high, z would be zero still. Yeah. Cuz the moment you put a load on it, and get dragged down. And so I'm like, why is it 1.5 volts. And so I'm just digging through, and eventually, I compared the hardware config file. And I found that I had one bit different. I had not enabled the crosslink. And so none of my pins were actually connected to the core.
Host 3
So the core was doing everything it's posted. Yes, yeah. Well, so
Unknown Speaker
what's interesting is when the core when the cross links not connected, you have 1.5 volts, which is probably a residual voltage from probably the ESD protection protection. Yeah, I'm gonna say, that's probably that's what that is.
Host 3
So it's kind of funny. On the on a typical day, at macro fab, Parker and I were goofing around where we're swinging between each other's desk, Hey, check this out, you know, blah, blah, blah. We've been labeled as loud before. And And it's funny, because this week has been very different. Because both of us have had our own individual projects that really don't involve the other one. And we just, it's been kind of silent. But the other day when Parker was was working on this, all I heard was through the silence was just grumbling. Like, all of these little issues just piling up and making Parker and then
Unknown Speaker
I finished I got I got it all working at like, 530 Yeah, and I'm like, I have 30 minutes left, I am done for the day.
Host 3
And it's funny cuz it your day wasn't wasn't about writing code. Your day was just finding that one bit that was not set, and it screwed everything over.
Unknown Speaker
Yep. And then this morning, I basically ported my old pick code over and it was almost drag drop. It was pretty simple. Yeah, I actually copied pasted. And then because my old one was just a state machine. Yeah. It's like you just copy my state machine over. And all I had to do was change how the state machine interacted with the Yeah, what plugs in and out. Yeah, I was on the pic. I was running in a straight loop all the time. Right. The PIC was never actually going to sleep.
Host 3
It couldn't because it had something about not recovering. Right? Yeah. So when
Unknown Speaker
the pick 16 hours using went to sleep, it turned off the timer. Right. That's it. The timers were used to keep the time in the watch. So you can't actually go to sleep that way.
Host 3
While the sleepy beat doesn't have that problem,
Unknown Speaker
while asleep yet sleepy doesn't because you can make the timer's run independent of the core. And the real time clock runs regardless. And so And actually what the cool thing about is the real time clock has a direct hardware interrupt into the ad 51 core. And so you can actually wake up based off the RTC clock.
Host 3
Oh, that's cool. Yeah. So doing real time interrupts
Unknown Speaker
you can do real real time interrupts
Host 3
that should that should be on the, you know, the front of the datasheet that features real real time data in
Unknown Speaker
RTC interrupts. So that's pretty cool. I guess you could do that with an external RTC that had a trigger, and then pipe that into a pin edge. hardware interrupt, but you know, whatever that would add to the bill. material cost? Yeah. Oh, yeah. Another really cool thing with RTC. So you have to you can actually use an internal oscillator for it. But it's not that accurate. But the cool thing is you can add an external. But you can do the loading caps built into the chip.
Host 3
Oh, yeah. You were showing me that. So you don't even have to put your loading caps on there. Correct.
Unknown Speaker
You just get a drop down of like, I think it's about like 18 in the most popular loading caps. And Bam, done.
Unknown Speaker
That was pretty loud. I don't know if Joshua, we'll do anything with that one.
Host 3
So 18 loading caps 18. The most popular loading caps? Yes.
Unknown Speaker
So you get like two pico for adds up to like 20 pico frets. Yeah, that you can just select which is pretty cool.
Host 3
Well, and it has an auto scan mode too, right? Yeah.
Unknown Speaker
Well, actually, we'll step up through and find where the gain is the best.
Host 3
Yeah, the optimal cap. That is awesome. Yeah. I mean, it's kind of funny because they actually have that feature, which, I mean, it's not really a problem to put on loading caps, and you don't really save a whole lot. So I wonder why they chose that feature?
Unknown Speaker
I think it's because it was probably pretty easy because they already have that cross link. So it's probably built into the cross link function.
Host 3
Ah, you right, they probably just turn it on using the crosslink.
Unknown Speaker
Yeah. And putting caps actually on die. Will how small value those caps are preferred arrange is actually pretty easy. It's metal. Dielectric metal.
Host 3
Right, right. So it's kind of easy, and then they just put they just put a switching transistor that turns on whichever cap is necessary. Probably. Awesome. Yeah. Sounds like a pretty cool little chip.
Unknown Speaker
I like that chip a lot. I'm gonna probably, I'll probably do a article on our, our blog about starting to use these chips. Like what do you need? How do you set it up? All that good stuff.
Host 3
Oh, sounds like we should. We should knock out a dev board for it.
Unknown Speaker
Yeah, they have a really good dev board already, though. Oh, did they? Yeah, there. Oh, that's right. Yeah, we have one. We have a couple of them. No, like 30. This has turned into an advertisement. Yeah, it totally is. But um, they have like these crazy dev boards that have like a screen, a joystick, RGB lights, and they come pre installed with Space Invaders. Now they have Space Invaders built in that you can control with the joystick? No, like 30 bucks. Yeah,
Host 3
it's way more than 30 bucks on that board.
Unknown Speaker
Yeah. I'm like, looking at the screen alone is like 1215 bucks.
Host 3
Yeah. So but I was thinking more of a, like a simplified one more like your prop def stick? Yeah, just like just a breakout basically. Yeah. drops onto a breadboard. Yep. Yep. Yeah, that's good. Yeah. Didn't seem like it would take that long now.
Unknown Speaker
Yeah. So let's go into the RF section. Sounds great. I saw today that KY CAD is adding SPICE simulation built in to their schematic software. So I think is a really cool idea. Yeah. That why it just limits basically, you are going to be removing an entire package from your process, basically. So instead of having to draw your schematic once and you know, multi sim or spice, you draw it once in KY CAD, then you can simulate it right there. Then go straight over to the PCB layout. You know,
Host 3
I haven't spent a lot of time with KY CAD, and they keep adding stuff that makes me say, you know, I just need to spend some time with it because it keeps getting cooler. And people are starting to jump onto it, especially with the whole free aspect. You know?
Unknown Speaker
Yeah. So I think I actually talked to one of the devs today, and they said, it's probably going to roll into the 5.0 release. Okay. I don't know if you can get I know it's not in the latest stables. It might be in the in the 90s that you can, you know, compile yourself if you're, you know, a masochist. I don't want to know what kind of dependencies you have to install to get that thing to compile,
Host 3
Unknown Speaker
But yeah, I'm gonna be cool. Probably wait till 5.0 comes out before I try and keep that then
Host 3
I still am looking for. I don't even know if it exists, but I'm still looking for a PCB level simulator.
Unknown Speaker
I'm like, I will kind of piece to be simulated.
Host 3
Think of simulation, you have circuit simulation, but it takes into account your PCB. So all the capacitance all the inductance of your tracing systems of your traces, yeah. Everything that's involved with that, and I know that's like, obviously, that's ridiculous. But I bet your cadence can do that. Yeah, yeah, you're right. But it's but it's like a $20,000 upgrade package.
Unknown Speaker
That and you have to use cadence. Yeah.
Host 3
Are you okay? So I used cadence for four years. And those might have been the worst four years of my life. I hate cadence. It was so bad
Unknown Speaker
that we use cadence in, in college for designing chip level, like the actual transistors, no, for that. It's good. And so we had to use a our department only had like, 10 seats. Yeah. And their server seats. And so you had to log in and actually tunnel over x 11. Mm hmm. So you were the the lag in a design? Usually you don't think of like lag on a mouse? No, the fact you move the mouse click, and they're taking about five seconds for that move to happen. And then the click and then whatever action happened. So it just took everything just took like 10 to 15 times longer.
Host 3
clunky, just clunky. Yeah. It was because we were having to tunnel over a really crappy connection. That sounds terrible. But even if you use it in the lab, and of course you know, of course they the electrical engineering department is the one that has that problem. Yes, of course. Now guarantee the business school did not have that problem.
Unknown Speaker
Well, I don't think the Business School has any like expensive software packages like that. Unless they have one called Stark stock market simulator. That's probably a game you need on Steam.
Host 3
I you know, I bet you and it's one of those 99 cent games that you can get
Unknown Speaker
nobody really awesome the simulator, you're actually playing someone that's on the stock floor. And so you get the wave tickets and stuff like that.
Host 3
How funny would it be if they just had the game where it was stock market simulator where it wasn't a simulator it was just the stock? Oh, actually pulled. They just pulled real data and bad kids were like, oh, man, this is awesome. into your credit card.
Unknown Speaker
And their parents like, Oh, that's a purchase.
Host 3
That'd be an awesome game you
Unknown Speaker
bought you bought 20,000 stocks and Toys R Us son toys
Host 3
I think a track a track tapes are coming back. I'm investing
Unknown Speaker
a track day. Ah, the big news of this week, arm purchasing arm was purchased by Softbank for $32 billion. So there's been a lot of speculation and a lot of talk about what this means for arm. His arm is fabulous. Right? They just own the IP for all the arm microcontrollers. And Softbank is a telecommunications company in Japan, who doesn't own any hardware.
Host 3
But they they're fairly diversified. Yes, they they make really strategic moves. If you look at what they've been doing in the past, so they have something up their sleeve with this.
Unknown Speaker
Yeah. So it's it's why did they buy arm? It's probably their telecommunications company. And so they look at arm as Oh, vertically integrating kind of thing. Sure. Lisa, I think of Yeah, yeah,
Host 3
that would that would make sense. And, and I actually heard or not heard, I saw an editorial piece. I can't remember the website, I think it was electronic news weekly, or something like that, where they were talking about the telecom industry is really kind of being yanked by IoT devices, because that'll all end up integrating and working together. So potentially Softbank is doing something a strategic move towards where they own the actual chips, the IP, in IoT, they own IP that
Unknown Speaker
runs IoT, right. Sneaky. Yeah, exactly. It was just interesting. Some of the things that people have been saying, like a soft bank says that. That the, that was not attention because, you know, Brexit happen, what, two weeks ago now, longer than that, I think so long enough, that it doesn't really matter for America. So they were like, Oh, is it because of Brexit? Because arm is in Cambridge? Yeah. In the UK. And with the pound being so devalued towards the the yen. You know, was that but apparently this was in the move before? Way before that. I mean, you got to think is a $32 billion deals. Not something you do in a week?
Host 3
No, no, no. They have been they've had their eyes on this for a long time.
Unknown Speaker
Yeah. So it just probably made the deal sweeter. Sure them, it may
Host 3
Unknown Speaker
Yeah. Probably. But the the founder of arm. I think his name is Herman hawser. He says this is a it's kind of interesting. He's the founder. And he's founder of the ARM Holdings, which I guess is like what they call the the top level of arm. He told the BBC that he thinks is a bad move. And a bad day for UK. technology, the technology segments because they you know, they're not owned in the UK anymore. They're owned in Japan. Sure. Which would be true if, let's say Softbank said, Yep, y'all have to move on with Japan. But Softbank says they're going to increase you know, fivefold over the next couple years in the UK. But it could be you know, one of those Softbank says they do that and then they don't do that and they just start shrinking tank everything. Just tank it. Arms got three, three divisions all over the world. So really yeah I think ones in the UK ones in Italy and one somewhere else I
Host 3
wouldn't know. Now I'm just curious to see if they're going to allow arm to continue with the progress they've been making or if Softbank is going to try to steer them heavily towards telecom basically his arm going to become a IP
Unknown Speaker
yeah the comments but Broadcom pretty much makes telecommunication chips right
Host 3
so is now yeah Softbank buying arm so they can compete with Broadcom,
Unknown Speaker
possible, you know? But yeah, um I guess I'm kind of indifferent on this as long as, as long as the arm IP stays how it is yeah where anyone can license it because I could see where Softbank starts locking down like the licenses harder, where it's harder for like, let's say ti who sells their chips to a different telecommunications company like at&t. And so, like, oh, you can only use these chips if you're using them for our telecommunication product products. Gotcha that can see that happening. Yeah. But the moment that happens someone one of these other architectures that's like, you know, just you know, sitting out there and some research lab will just kick up and you know, dominate the market like arm did, right? Right.
Host 3
Oh, some something new because I mean
Unknown Speaker
arm overtook what? at 51 and 68k. And, yeah, all those other guys. And arms in everything now. Yeah, but it doesn't mean that another architecture can just come in and sweep the floor, basically. Yeah. Yeah. If soft bank locks down so hard. Speaking of acquisitions, we've talked about this before. There's the microchip. AVR Atmel thing. Yep. You found this article.
Host 3
Yeah. And I'm actually pretty excited about it. It so apparently an article came out where? Microchip it's funny to say this microchip has come out like Rommel. Yeah, my microchip has come out saying that AVR is not only are they not like hacking and slashing all of their product availability, they're increasing it. So this summer, there's a handful of new offerings that are not microchip or pick, I should say, offerings. They are AVR offerings. So I'm
Unknown Speaker
the AVR eight offerings, right?
Host 3
eight bit microcontrollers? Yes. So it's not like they're adding just another 32 bit. One of the arm. Yeah, or another arm? It's it's another eight bit. So I'm I'm really excited about this because we might start to see the meld of both worlds. Getting better. The fine finally like the AVR pic, you know, coming were great. Take the best of both worlds and smash them into one eight bit architecture.
Unknown Speaker
Yeah, I can see that. Yeah. This just reminds me what microchip tends to do is okay, we made this really awesome chip. Now let's make 300 SKUs for it.
Host 3
Oh, yeah. If you go to the if your noodle electronics and you like I need to pick a microcontroller you go, you go to pick and you're like, they might have something like 10 or thing or so options to choose from? No, there's like 3000 things to choose from. It's like, how do you pick the right one?
Unknown Speaker
They have a telephone book of microcontroller. Part numbers. Yeah.
Host 3
Yeah, it's it's absolutely ridiculous. And in fact, whenever I first started working with pics, I just, I literally just picked one at random because it had 20 pins. And I was like, that's good. And I learned that one and then and then just started making projects on that one. I think it was a 16 F 690. Was was the one and and it was just like fine. It works.
Unknown Speaker
Yeah. Um, actually, the thing about this is what micro microcontroller company has the least amount of skews.
Host 3
Silicon Labs, right? Well, Silicon Labs
Unknown Speaker
still has no, they have the C 8051, which came before the FMA. And there's a ton of those. There's probably at least 50 of those. Parallax might be the small they have the basic stamp, Prop, javelin, which doesn't exist anymore.
Host 3
Are you are you considering ones that don't exist anymore?
Unknown Speaker
No. So then Java, so it's two. And I want to know if they make the basic stamp anymore either. So there might be done. The one Arduino doesn't count. Do they know that? Well, have you seen how many different Arduinos there are out there now? Xe right yeah. Now all use different Atmel chips. Yeah, you're right.
Host 3
So that that doesn't necessarily work. Yeah, it might be. It might be that parallax
Unknown Speaker
Host 3
Technically two, right because the basic stamps still available.
Unknown Speaker
Yeah, so it's actually one of the biggest products last time I checked. Just education. Just eat those
Host 3
things up. Yeah, cuz they're easy.
Unknown Speaker
Host 3
Unknown Speaker
Hey, man, I staved off making an arm and a leg joke.
Host 3
Oh, geez. This might have to be the end of this week. Yeah, I think that's, I think that's gonna be the puns are coming out now. Well, this was episode number 25 macro fab engineering Podcast, episode number 25. We were your host, Stephen Craig
Unknown Speaker
Host 3
Take it easy guys. Later guys. Don't
Unknown Speaker
lose an arm and leg. Oh geez.
