The Genesis of My Bill Of Materials

Welcome to the macro fab engineering podcast. We are your host, Stephen Craig and Parker, Dolman. This is episode 140.

So yeah. That RTL software defined radio, apparently I was calling it signal Defined Radio all last podcast.

People been correcting you. Oh, yeah.

I think I like 40 emails.

That's funny. And it's funny, because before last podcast, you you, we had talked for a while about software defined radio, like, and we were just calling it that you didn't even click in my mind.

Yes. So the software defined radio, I finally got something kind of working. Cool. It's called RTL underscore, FM. And it's kind of like a command that you can call it's part of like the RTL underscore SDR, like, package that you can install on Linux. And so you can like tune the frequencies and stuff like that. So it's pretty easy to use. You basically just like it basically outputs the sound through standard out. In Linux, it's like signed 16 bit or something like that. And then you just pipe that directly into like, player and plays music or whatever you're picking up. So C's work fine for FM stuff and anything like up to the ceiling of that software defined radio, which I can't remember how high it goes. I think it's like one gigahertz. But I can't seem to get am to work all too well,

really, like what's the symptom?

Well, so the minimal that this thing can technically tune to is 24 megahertz. Mm hmm. And am is like 500 kilo hertz to 1050 I think kilohertz, so like 1.05 megahertz. Anyways, so the make that work is you have to directly sample the signal. And there's some special modes that you can like, make it work. But it seems that the RTL underscore FM just doesn't really support that it's got like a direct mode option that you can use. But it doesn't seem to work. I can't get it to receive audio over am if I plug it into my Windows computer and run I think it's like SDR let me see SDR sharp, which is like a GUI based signal Defined Radio software defined radio. Software Defined Radio, pack, software suite, I can I can make the aim work, it has direct sampling, and you just click it and you can do it.

Well, and when you when you say direct sampling, it's not sampling. It's under sampling. Right? It's it because it's only trying to pick up the audio not the carrier, right? I mean, it's not it's not fully sampling the one megahertz wave, right? Yeah, that's what it's doing. Okay, so it's, it's, it's doing the full signal bandwidth and then rejecting the high frequency stuff.

You're you're reading the whole spectrum in Oh, that you're reading in your carrier frequency. And your your software is doing a D modulation.

Hmm. Okay. It's interesting that sampling pretty quick, I guess.

Yeah. And so the am works for some reason on SDR sharp but it does not work in RTL FM or RTL SDR. I've tried some like Linux based GUI applications I can't remember they escaped me which ones I've tried, but they don't seem to work either. So I don't know what the problem is there. And some some some people I've been asking around a lot of people say for me to use on upconverter. So basically it will up convert the am signal to into the band of the normal band that is a software defined radio reads which is an option. The thing with that is basically if I need to switch between just reading the antenna directly verse upconverter, is I had to switch like the audit the signal path on the on the PCB, I guess. And I don't know what the best way to do that would be I don't know if you can get away with relays or you should use like a special signal chip or

something. You could just use an analog switch I bet. Yeah, probably. Yeah. Like one of the Digi for whatever I can't remember the part numbers they I know they start with Digi but you could probably just Use one of those a bit. Yeah,

that will probably work. I've never done switching on, like analog stuff before. So

yeah, I guess you just have to look and see if it's gonna load it weird at that frequency, you know? Yeah,

make sure it passes the basically the entire spectrum that the software defined radio can.

It's gonna trip you up every single time now.

Oh, yeah. Because I'm overthinking it now, probably. So this, yeah. So may I make sure that the switch can can handle the frequencies?

mean, do you even listen to am?

Yeah, I've listened to some of the news and sports on am okay. Okay. And also, what's interesting is that the RTL FM doesn't support stereo sound, it's mono only. Which was interesting. There's some people who've hacked or I won't say hacked, they rewrote some of that software. And I set the compile like a different version of RTL FM. And I can get stereo then. So I haven't tried that yet. But give it a shot soon, maybe after the podcast. And then I started looking around for like other software defined radios. And I found this one, it's our 820 Ti p 7k. Five, if you just like Google that it'll pop up. But it does 100 kilohertz to 1.7 gigahertz. Well, it'll do the whole thing. It does the whole thing. But the 100 kilohertz now that's underneath the am band. So and this thing's almost about the same price as the one I bought. It's like $8 more. So I might try ordering this one, and see if it will do an basically out of the box and not don't have to worry about switching.

Yeah, that sounds like a good option. Just because switching sounds like a pain in the butt. You know, you'd have to design something around it, and then you'd have to work. I don't know. That's a little too much. It seems like especially for something that feels like it should be all in one.

Correct? Yeah. And so yeah, that's where the software defined radio project is that now? I'm hoping by next update. I'm probably gonna buy this. Our 820 t p 7k. Five radio thing, and probably just go that route. And yeah, and hopefully have some kind of like, interface making that work.

Cool. From last podcast, did you come up with any secret tricks on working with a Raspberry Pi and having it not be so messy?

No, it is a ginormous mess right here.

It's still messy.

I'm probably trying to figure that out. I started looking for like raspberry pi cases like desktop cases and stuff. Basically, like it needs like to contain the power supply and everything in the box.

Go get a USPS priority box and put everything inside of it.

Make sure it's a little greasy and wet and then run 240 volt through it at 50 amps.

Exactly right. Yeah. You got it. It seems to work, right?

Yeah. So for those that don't that haven't listened that episode that was like, man, it was a long time ago episode like 30 or something like that. Yeah. But Steven made a his electric brewery rig. And he needed to brew and so he slammed all the electronics inside of a USPS box.

Yeah, I've ran a few 1000 Watts through that box. And I got some delicious beer because

it did not burn down your your apartment.

No, it did not.

And then let's see what else I've been doing. The I've been working a bit with key CAD, but not like learning how to layout boards and KiCad

when you're going to get around to that. Never.

But I've been working with the CAD Python API. And basically, I wrote some scripts that can automatically generate the Gerber's, like your PCB Gerber's and the Mac Fred, xy RS formats. So because right now, on Mac crabbet, only when you upload a KiCad PCB, you only get the XY Rs, you still need to generate Gerber's locally and upload them. And so hopefully in the future, you just upload a PCB and it generates the Gerber's for you too for KY CAD like it does for Eagle.

Right, right. Right. Would you've had the eagle thing going for a long time?

Yeah, Eagle was pretty easy to get rolling on on the server side, but KY CAD it took a while for there. Thought API stuff to manifest, I guess is a good word for it. And it's been around for about two years now. And I finally got around to actually working with it. The problem with the key CAD API Python stuff is there is no documentation. So you just got to figure it all out. Yeah, the documentation is like the code.

Oh, on GitHub, right?

Yeah, well, they have like a dot, a Doxie. I think it's called Doxie. Doxie. Like documentation thing. Mm hmm. But it just, it gives you the methods and the classes and stuff. And you're like, Okay, if I call this this class, what attributes does it have? I don't know, I just have to start basically, ramming attributes, I think exists, and see if I get stuff.

So it's incomplete documentation is what you're saying.

There's zero documentation for the, for this thing. If someone knows where the documentation lives for the Python API, please let me know.

Honestly, if anyone would know it would be one of our listeners. Yes, yeah. So uh, what do you have working on that so far? I mean, do you have anything that actually does anything yet?

Oh, I finished it all up today. Oh, okay. Okay, so that's

going to go live sometime soon.

On the website, the factory. I don't know what the eta is because a developer in quotes. I saw I wrote some Python scripts that make it work. And so it's basically like, a proof of concept that if you implement it this way, it will work.

Okay. So they have to, they have to take that and they have to massage it and push it and make sure it does all the good jazz with all the rest of the plugins and everything.

Yeah, exactly. So they had to make sure you know that the build the infrastructure on it? I did. I did technically the easy part.

Yeah, for sure. Um,

it definitely was the most fun part. Because all the other stuffs not fun.

Well, that's why you don't do it. Exactly. Yeah. Okay. So you've proven that it works. In the next. I don't know what it might have been two months, it'll probably pop up. Cool.

And I'll we'll put an announcement on the on the podcast when it's up and running. Sure. So I think the next one I'm going to try to work on a lot is Altium. Now, yeah. Because that's, that's a big, a lot of our customers use Altium. And right now, we it's a ODB plus plus export, which is fine. It's just annoying. Yeah, I'd like to be more like, here's just my file, and then it all works.

I think you're going to spend a lot more time on that than you did on keqin.

Probably, you know, I only spend about a half a day, about a about one day, so eight hours on the cat stuff. And most of it was just like punching in stuff and seeing, okay, I need to get like the value out. But I don't know how to get that out of the class, because it doesn't tell you. It's like you click the class and it's just like, okay, that's the class that and there's like, description is nothing. I don't have the attributes, how do I get that stuff? You know,

that's actually one. One thing that I think I've found kind of annoying about the differences between EDA tools is not like putting schematic symbols down or you know, talking about footprints or patterns or anything like that. But figuring out, first of all, you as the designer, figuring out where to put your part numbers for the parts. And then if you're trying to read someone else's work, where they put their part numbers, because most EDA tools give you options on where you want to put them. And in some ways, I find that kind of annoying, actually, I wish there was just more, because maybe I'm wrong. In fact, I know I am wrong on this, but I like to have my manufacturing part numbers in my schematic of my, you know, electronic files. Yeah, I do too. But but but but not everyone does that a lot of people will just run a paper bomb that references the schematic. And so just references the references on there. And then so you always have to have, you know, an Excel sheet and a schematic up. And that's fine. And I've worked that way before, but I don't know, I guess it's just a personal preference. And it doesn't seem like there's like a really good universal way that EDA tools allow for just because I've worked, you know, on probably at least six or seven different EDA tools. And I shouldn't say extensively, but enough to know what's going on. And you have to do something. It feels very different in each one to just say this part is this.

Yeah. In Eagle what I do is I use the attribute function, and so I make a attribute NPN and then you just put the MP in there. Right? And then Matt Krav X We wrote the eagle parser to just pull the MPN out.

And you know, that's, that's, that's cool and all. And that's great because, like, you've created a really default way. But you created that eagle didn't create that correct. And that's kind of a little bit annoying. And you know, dip trace does the exact same thing, when you click on a part dip trace, you have a handful of attributes that you can apply to it. But manufacturing part number is actually not any of those values. There's, you know, there's reference designator, there's value, and then there's name. And none of those are manufacturing part numbers. Now, typically, I put the manufacturing port number in the name category, such that when I export a bomb, I can, you know, just dump the name. And then you know, if I want to I can change that column in Excel to manufacturing part number, but it doesn't have a default, you know, place to put a port number and I find that kind of annoying actually.

Yeah, I actually part of this parser stuff I wrote for KY CAD is I defined, how would you do MP NS and Ky, CAD McWrap. And it's interesting, because keep CAD. On the PCB file, you can't put custom fields for parts. Yeah, you can only do it in a schematic. And so you can add a MPN field there. And so what it does is, if you also upload your CAD schematic file, it will look into the fields and look for MPN. If MPN exists, pull that out and use that as the Manufacturer Part Number. Sure. So hopefully the next couple months, that's, that's live.

So it sounds like it's it's just a lot of housekeeping to get it there.

Yeah. It should be the three this should be three values for parts. Reference, next is Nadir value. And then MPN. Right, and then footprint or slash package. So for Well, yeah,

but but the footprint package thing can is sort of different in a way. In terms of like you're defining that, but you're not, you're defining it by like actually selecting a footprint, as opposed to just like typing it in a field, whereas I agree, the reference designator, which I shouldn't have to put that in myself, that should automatically increment when I place a new component, the value. That's that's up to me entirely, and the manufacturing part numbers up to me, but those three should always exist, in my opinion.

Yes, yeah. So EDA, tool designers, that listening do that,

oh, maybe we're missing something. Maybe there's some secret sauce that that, you know, maybe there's a good reason not to have manufacturing part numbers in your schematic file. I don't see why, but

I don't see why either. I do it all the time. Mm hmm. And that's our biggest, you know, thing from customers? Like, how can I get my NP ends? And easier, it's like, spreadsheets.

Right? Because there's not a good way. Yeah, I like having my schematic is kind of my master. My layout references, my schematic, my bill of materials. You know, the genesis of my bill of materials is my schematic. Like I pull my bill of materials out of it, I'm not creating my bill of materials as a separate file, I literally export it. And that, that way, I have a single point of all information, and that if I, if I modify my bill of materials, I typically start by modifying the schematic and then adjusting the bill of materials. It just, I don't know, one point of, of data. sounds correct. One source of truth. Yeah, there can only be one. Except most EDA tools have more than one. Yes.

And then the last thing I'm going to be doing is I'm going to the political spectrum conference in San Francisco this week, that's gonna be fun. So I actually I think when this podcast goes live, I will be there. And they the government have a bunch of workshops for IoT hardware and software, and I'm looking forward to that. It's been a lot of fun.

That's cool. Do you have any specifics of what you're going to be doing while you're there? Or are you just kind of like, wander around?

So they got some like, talks and stuff I'm going to go to? And then there is a couple workshops. Brandon satrom, I think, yes, that drum, who was on the podcast a couple episodes ago, like 20 or 30 episodes ago, uh, he is given a talk about mesh networking. And so I'm gonna go see that and I think it's like a hands on so like, we'll have devices and get the program that Oh, wow, that sounds like fun. Good. Dangerous. Yeah.

How many? How many days? You're going to be out there?

I'm going to be there from Tuesday to Thursday. Okay, okay. Yeah, so I fly back Thursday night.

Okay, so on next week's pile cast, you'll be able to give us all an update of all the cool little gizmos you played around with. Oh, yeah. Yeah, totally do that. All right, sweet. And I guess you'll be feeding this left channel with all kinds of cool stuff, right?

Oh, yeah. I can't wait. It's, it's one of the first conferences I'm going to be going to. I mean, last one I went to was like, like Maker Faire, like four years ago, three years ago.

You and I attended Maker Faire Houston? A few years ago. Yeah. I think that was probably three years ago,

too. Yeah. So about three years ago. Yeah. I can't wait.

He sounds so excited.

I'm just tired right now. Yeah, I feel Yeah, I'm like, oh, man, I gotta fly out there now.

And you got to wake up super early tomorrow. Oh, yeah. Yeah. So Steven, yeah, that's me working on more of the Vox in the box stuff that I've been kind of talking about that for the past couple of weeks. So I got the PCB layout done for the first iteration of the box in a box tester, which, for those who don't know, it's a, basically, I've made a power amp and a power supply in a pretty small box, it's about 3.6 inches by 2.6 inches, that'll all fit inside of an aluminum enclosure. And I've got a 12.6 volt power supply, a 350 volt power supply, an 18 volt power supply, and a 30 watt power amp all inside this little box. So I finished the layout. Yesterday, I think, well, I mean, technically, I did a few more things, I always do a few more things, when I'm doing a layout, you know, I always look at it like that one, a couple more little things. It's kind of like, maybe people won't like this. But it's kind of like artwork, in my opinion, where it's you never really finished, you just abandon it. You know,

this is a four layer,

it is a four layer board, I originally I was going to do a two layer board. But I opted to do a four layer because there was a couple of interesting little things I wanted to do with it that ground plane, multiple ground planes. This is sort of the first time I've done a whole bunch of split ground planes like a like, we'll post the pictures of the layers in the Slack channel and on the the show notes for this episode. But if you want to check out the layers that I did, I've got some sections that have ground planes on top, some that have internal ground planes I have routing in between. And the way I really don't see a star ground. Yeah, that's right. So the way I approach this layout is I have multiple subsections, or sub circuits that the only thing that they really share is ground and power. And I because most of the you know, the analog circuitry that will drive or communicate with the power supplies in the amp are going to be external to the box, I kind of designed three separate circuits, one being the heater circuit, which is the 12.6 volts at 600 milliamps. The other one is the 350 volt at 10 milliamp power supply, and then the power amp. Each one of them, I treated them separately in terms of how I grounded them, I followed, you know as many of the rules if not all of the rules from the datasheet as I could. And so some of them have ground planes on top, some have ground planes, internal. Some of them have multiple ground planes were there via stitched out the wazoo. I just tried a whole bunch of different things. Well, I shouldn't say tried, I treated them all differently and treated them all in the way that each one likes to be treated in a way. The one thing that I did differently that I'm hopeful to try with this is each one of those circuits has its own basic ground chunk, I guess you can say where it's its individual by itself. And I have connections on the outside of the box that allows me to access that specific ground plane. So like the high voltage 350 supply, I have a terminal block that allows me to connect directly to that ground. So if I want to return current to that ground, I just connect there. But on the bottom layer of the board, I have traces that all go to a center stood that's in the middle of the board. So I made a very specific star ground that everyone shares together. And I have terminal blocks that go directly to that star ground also. So when we're building test circuits, we can see what grounding works best, do we want to ground it directly to the power amp? Or would it be better to ground to the star ground and then eventually everything flows from the star ground out the DC jack? So I wanted to build this thing such that we could test you know other circuits and more than the analog stuff so we have all the power and all the the amplifier stuff. But I also wanted to be able to test different ground topologies as opposed to having to you know, re spin a board every time we want to try that I just gave options with it. So yeah, Cool little board after the podcast, I think I'm going to press go on it unless Parker finds any issues with it right now.

But do you have ratlines still there?

Yes, I do, there's actually two ratlines available. And that's because I'm playing outside the rules of dip trace. Dip trace is usually really, really explicit about its ratlines. And it'll be very upset. If you don't do the follow those ratlines. Dip trace does not have a good way of doing milled slots in pads. So when you define a component that needs a milled slot, that is a through hole pad. Dip trace doesn't like the top and the bottom being the same thing with a milled slot in the in the bottom. So what ends up happening is say if I have a component that has a specific node slot, and I want it to be grounded in a lot of EDA tools, you can you can select that, hey, this is a through hole pad with a milled slot. And it will consider both the top pad and the bottom pad as ground dip trace will let you only pick the top or the bottom as ground. And then it'll throw all these errors saying you're trying to mill in the center of a pet. Well, I know in reality that the top and the bottom are ground and it's going to be a plated hole. So I sometimes go outside the boundaries of the DRC because I know exactly what's happening. But it'll know better. I know better than in this situation. So it's telling me that my grounds aren't connected, but they really are. And at the same time I did a modification, I broke the rules pretty hard here, I made a modification that is not reflected in my schematic, I added a shave off board switch would. And the whole reason why I did I didn't go to the schematics because I didn't feel like drawing up another component to do what I wanted to do. So if you see off the DC jack, the 18 volt goes off to a pad that's like a through hole wire pad. And then there's another pad that comes back to the rest of the 18 volt circuit. I'm just gonna offboard wire to a toggle switch, switch.

I just got your power switch,

right. So it's very, it's very, very simple. And that's five minutes worth of work. But of course, now it doesn't match my schematics. So depresses also, you know, crapping it's balancing, hey, everything is wrong here. No, it's fine. It's okay. I'm willing to break the rules when it's simple stuff like that. Especially because it's not like gonna go into production with this. This is a, you know, a one off thing,

one off kind of thing. Yeah, I like that in your interlayer. One, you've got this like guard

trace. So yeah, over on the Okay, so over on the ZAP for the audio. Yes, that's that's the input audio there. So there's there's a positive and a negative, it's a differential signal. And what I tried to do is I tried to keep my two input signals, as far away from the two switching power supplies as possible. On top of that, it has a ground plane on the top layer, it is completely surrounded by via stitching in its layer, which is layer number two. And then on Layer number three, there's another ground layer underneath it. So the two input traces for the amplifier are surrounded, top, bottom left and right entirely with ground. So I basically tried to, you know, sort of make like pseudo coax out of the PCB. And I want to, I betcha that's going to work pretty well, it should yeah, it just because I want to make sure that all the switching noise is just completely gone. And I added a couple features like there's some resistor swapping, that if you want to, you can change the frequency of the class D amplifier by just you know, pulling legs higher low. So I just have a hand like an array of resistors where and there's like codes on the on the silkscreen that tell you, you know, populate this if you want one megahertz or populate this if you want 500k, or whatever. So there's a couple of things I want to try. And hopefully, I can find what frequency works best with the lowest noise, mainly because my traces right now we're running next to the output filters on the class D amplifier. So I spent a lot of time making sure that those traces were as protected as possible.

And they're out of the the traces or out of the current path to read. I'm looking at the return current paths of that class D and they're out of the way so should be pretty quiet.

That's that's the goal. And and in the middle of the board, there's a section that is for the high voltage switching power supply. And it actually has its own like ground stitching. And that's actually from the datasheet that Yeah, so there's like it's kind of neat to design the chip pretty well. Where the left side basically all the pins on the left side are the low noise section and all the pins on the right side or the high noise section or the switch and currents really. So you can treat the left side of the chip and the right side of the chip as separate grounds effectively, eventually, they all come back together. But you make this really nice ground Island, that's all surrounded by vias. And you know, basically, it's a ground trap in a way. And you, you treat that as super, super clean. And you connect that over to the main ground for that sub circuit at one location, which is the ground for the IC itself, the ground pin. And so hopefully, that'll turn out to be really nice. For that section, I referenced the datasheet, which they have some images of circuits that they did, and I followed something pretty similar to what they did in their example evaluation board. So I'm assuming that should work pretty well. This is also the chip that we were discussing last week where the app note for it. If you look at the circuit board that the guy designed for the app node, it's, it looks like garbage, which is Yeah, I mean, I love it, because it works super well. But what I designed here is like 10 fold better than what it looked like on that app note. So I'm assuming this will work really, really well.

And one more thing I found. So on on, Enter, Enter one and air copper one on those orange fat traces. Is there a reason why you got a little nubs stick in there?

That's an artifact of Detroit's that after I generated the Gerber's, I saw it and I was just like, Dad, I don't want to get rid of it. I don't care. It doesn't matter. DTrace put a little, a little extra stub trace coming out of one of the pads. And it doesn't interfere with anything. So I that's that's that stub trace is called Steven is lazy.

Ah, that would bug me.

There's a lot of other things that bug the bug the shit out of me and more like like the via stitching. I obviously went super OCD on that and got anal with it. Yeah.

And I would on I'm finding everything wrong with this with this low man. Inner Inner one. You got this like a ground like island. So you know what? You're talking about your switcher there. But just to the bottom of that you've got this little stick of ground. You probably should just get rid of that. Okay. I don't get it's just an antenna. Oh,

I see what you're getting at? Um, maybe I don't know. It's just, it's just a ground that is part of the ground plane. That.

I betcha that's gonna pick up something from that switcher.

Well, you know what, I'll leave it and we'll find out.

Okay. And it's on the inside, though. So you can't cut it.

Or you could drill it out.

Punch I want I've

done it before. Yeah, no, it's small enough that you could totally punch a hole. See that? Maybe just for you. I'll even if this works. Well, I'll record before punch a hole through that one little stem, and then record after it and see if it changed.

Yeah, see, we can pick up anything. We should. You should totally do that. See, that actually will make a difference.

Maybe. So yeah, that's that's what I've been working on. I'm gonna place an order tonight. Maybe I'll have it by next podcasts. Maybe not. I don't remember. It's like a four to five day turnaround. And there's a weekend between now and then. So I doubt I'll have it by then. But who knows?

Yeah. Are you gonna stencil it? Yep.

We're gonna buy. I think I'm gonna buy enough egg on this one. I usually don't get enough when I'm just building by hand. But myself and the other guy who's helping me at work, we're both going to build one each of us. So I think I'm just gonna get in a good a good stencil will do lead free paste and just tossed it in our oven network. And I because the ICS especially the class D amplifier has a huge power pad underneath it. I'd rather the oven take care of that.

And, and because of that big pad going to Enoch is going to be better for your coat. coplanar this of that that area. Yeah, right. It's gonna be a word for

that. I liked that. coplanar

well, because Yeah, cuz if you have the hassle, that part could just be you know, tip the bit. And not solder. Right.

Right, right. Yeah. So I actually already have a use for this. I mean, the whole idea of this box in a box is we're just developing a fun little thing that we'll play with. But a friend of mine at work actually just purchased a rotary lezlie cabinet, which if you haven't seen those, go check them out. They're super freaking cool. And if you ever have a chance to actually hear one in real life, do it. They're really really super cool. Allegedly cabinet usually connects to an organ. And you know, it produces that Rock Organ kind of sound or the blues organ kind of sound that you you've heard on a kazillion records, though they have, typically an onboard amplifier, that runs the whole cabinet. And they don't usually take a low level signal, they take a higher level signal that, you know, communicates with the original organ. Well, he's got this Leslie cabinet, but he wants to be able to plug something else into it. So effectively, all this boils down to is he needs a preamp, and I was like, well, we're designing a box in a box. Why don't we make you a Leslie cabinet preamp using this box in a box system? So we're going to give that a try with him and see how that works. This

thing is a I'm trying I'm looking at a GIF of it working. So it's spinning a horn at the top. And then it's spinning a I guess a sound letter around? Yep. Okay. Yep.

And so it has a motor that turns on both the horn and an a, like a reflector down at the bottom. And there's a speaker that faces downward basically. So you can you, there's a couple of ways to, you know, work with them. But there's like foot pedals and things like that, where you can change the speed of the rotation. So you can hit a note or a handful of notes on a keyboard and then change the speed of the rotation and you get a Doppler kind of effect. Gotcha. And it's kind of like throws the throws the sound, you know, as the horn spins around, it throws it in different places around the room, and you get a really cool like warbling effect. And it's, it's one of those effects that you can simulate it, you know, electronically, but contrives, even simulating it that the problem with simulating is, if you play it through a static speaker that's just pointed in one direction, you're just simulating it, the reality of a true lezlie. It's actually something spinning around. And it's just it's basically impossible to truly get it's basically impossible. It's completely false. And they sound absolutely incredible. So if you'd like I said, if you get a chance to go check one out.

Ah, that's why you said, Don't check it out on YouTube, because you can be playing through normal speakers. So you had you had to be there with one in the room. Yeah, gotcha.

Yeah, you'll you'll get the idea of what's going on. But the like the true like meat of it. It's just you have to be there. So, Nick, if you ever get a chance to go to visit Joshua studio, he's got one there. And I'm sure he'd be happy to give it a whirl for you. No, ma'am.

Oh, you should make headphones. That's too many ones. Because reasons. Because reasons. You totally need another project. Yeah, I

need another project. And the entire reason to have Lesley rotary headphones is so that you can accurately listen to a Leslie on YouTube. Right? Yes. That's the only reason I really liked it.

Yeah. Oh, but then it would have the rotation at the sink up as well.

Oh, geez. Yeah. Oh, man. It's Go figure. Yeah. Good luck on that.

Well, you forget about this idea. Yeah. Yeah,

no more. We're both not going to feature creep ever again. Ever, right?

Ever. Cross my heart. Hope to die.

All right onto the RFO

Arvo. So this is near and dear to my heart. Parallax, the creators of the parallax propeller BASIC Stamp educational stuff. The propeller two has been rumored to come out since before macro fad. And so the propeller two is basically a beefed up version of the propeller. The propeller is an eight core micro controller that runs spin and C. And it's a really neat micro controller architecture. And I always really liked it. The propeller too. I just was going through my Twitter feed a couple days ago. And they apparently got production samples and they work. Oh my god, like eight years after they said they were going to do like the prop two. And it's finally here now.

So what's different about the prop two from the prop one.

So the big the big difference is? Speed. It they went from 80 megahertz to 180 megahertz, which is pretty blazing fast for a microcontroller. The also the other performance boost. I think they were able to do this. I didn't see the documentation yet for it. But at the first prompt takes four cycles per instruction. So you're effectively only running you know, 20 megahertz or 20 MIPS. Apparently they have that down to one cycle per instruction now. So you're actually running 180 MIPS?

Wait, what? For every instruction for every instruction?

Wow. But I read that like four years ago. Okay, I don't know if that's actually true. But I need to go look more into if that's actually true.

Because if that if that is true, they just like, exploded, they let's like nine times the speed. Yeah.

Another big thing is they went from 32 to 64 io so that you have 232 bit banks. And then they started adding more hardware peripherals. So the original prop pretty much just had like a UART. And then it had a, like a what was it? It's like a TV? No, no, they didn't have that in there. They, there's some other kind of peripheral that's built into the IO on the prop one, I can't remember what it is. But now they added like a lot of stuff like serial, and they added a built in USB. They have DAX, and ADCs, and stuff like that built in. So that's pretty cool. They call them smart pins. And each pin is a smart pin. So you have 64 of these, like, super, you know, hybrid IO pins, which is kind of interesting.

Do they have it where any pin can connect to any function? Thanks.

So Oh, that's cool. Well, I think it's because every every pin is identical. So they can't do anything. It's not like a crossbar, like in a FM eight or a some of those picks, not picks, P socks. P socks have a crossbar to where you can, like, connect the UART to like a couple different pins, you know, right. But I think this DMS like that, too. I think this is because every single pin has all this hardware in it. So that's kind of cool. And now it has a debugger as well has an onboard debugger. And I think it probably works through the USB. So that'd be cool. So yeah, I'll put the link for the short form datasheet that they have for it, which has like the, you know, specs and stuff for

So did they just announced this,

uh, they've been working with the community a lot on it. The parallax propeller community, I just kind of saw that, oh, they have production samples. Let me go looking at it because it's like, I kind of, you know, I always thought like, basically since you know, three years ago, I thought that they would never make it basically. Like, it's been so long you even made a meme about Yeah, I didn't make a meme about that, like five years

ago. You need to post that mean? That's yeah, it's pretty great. So the the dye in the original prop the the actual silicon die looks super freakin cool. Because it was it was not necessarily designed step and repeat in a way like a guy actually like drew it all out which you know, that's that's kind of common CHIP.

CHIP Gracie design hand laid out that chip.

It looks incredible. The dye and I would love to see what the prop two looks like.

Yeah. How convenient is you're you're a chip designer and your name is Chip.

Just just a slightly could meet. So so does the is the sort of the original kind of like gimmick of the prop one is that it had one of the column cogs

cogs which are just eight cores. And so how that works is the each cog has its own memory. But it's really small. It's just enough to like, do scratch work, basically. And so whenever it needs to go to system memory, it has it's a hub. And so all the system memory is round robin, so each for every one out of eight cycles, the cog has access to it. And so use sometimes that's really fast, because then like if a cog goes, Hey, I need this address. And it's on its turn, it can get it right away. But if it's not, it can potentially have to wait seven cycles before it can get there. So it's kind of weird that way. And the thing is with original prop, there's no built in hardware, really. And so you basically build AV everything in hardware in software. So like an I squared C Driver, you don't have a surreal hardware. You know, your or spy bus in there. You actually build a bit bang you know spy driver. Yeah, but a lot of people like it because of the I do. I think it's really fast in terms of developing software for embedded stuff, just because you don't have to sit in the datasheet to figure out oh, what registered do I need to hit to make it do this? You're just like, Oh, I'm just gonna bit bang it. And then, you know, spit out eight bits. And it works. Yeah, but but

at that point, it's almost entirely based on how good you are at writing code.

True. True. It is very

user dependent. User dependent.

Yeah. And a lot of people write assembly for it still. Wow, that's because then assembly you can get one clock cycle per instruction, I think. Yeah, that sounds right. It's spin that takes four instructions, four cycles per instruction. Okay, and so

so so does the prop to have eight cogs or is it still

eight? Yeah, I always always wanted them to like basically double everything. And have, you know, 16 cogs, but they kept it today. Apparently, the cogs are really big. Like, they take a lot of real estate. So sure, I think the problem, that's expensive. Yeah, it's expensive. I think the prop two supports like native SD card booting, and stuff like that. It's got some really interesting features that have added into it. So I can't wait to finally get my hands on one.

Yeah. Well, speaking of that, is there any kind of idea on when, when you know, might be able to play with one?

I don't know yet? Uh, Roy Altham. She's one of my friends who does a lot of development work with them. He was like, Yeah, you won. So I'm like, okay, cool. So hopefully I get one yes. Sweet. And run it through a reflow. Do a do a first look, podcast, first look, podcasts. And speaking that we should invite Ken and Chip Gracie on podcast.

That would be really great. I'd love to, I'd love to hear about the development. And I'd actually love to hear about like, what their method is and how they go through it top to bottom, especially with the fact that this chip is really, really unique in a way no one else is doing it that way. So you know, what are the difficulties? And what do they go through with that? That would be a ton of fun. So if anyone knows them, and wants to suggest

I think I've met I've met Ken before, so I probably just asked him on Twitter.

Okay, well, that works to

cool, but I'd like to know more about man, we can make that like an Uber episode of like, talk about the Prop and stuff like that, like the development, the history, all that stuff. And then do another one of just like chip design and manufacturing. Oh, that would be great. Because I don't know a lot about I actually I design chips and you design chips in college as well. I never got my made, but I did lay out. But but very what VSI? I think it's what they call it?

Yeah, yeah. Well, very high density layout. VHDL. Yeah, that's

right. I think V LSI. Is something else? Nope, very large scale integration is V LSI. And that's that's what I did. I think it's different terminology. Scales. Yes. And scale. E.

They represent different skill sets. Yeah,

you're probably right. So, yeah, that'd be really cool. But

But I bet you these guys could tell us the difference between Yes, exactly.

And they were talking about if the prop two does really well, they're talking about going to a smaller manufacturing process, like 27 nanometre. And they said that simulations show that the chip can do like a gigahertz.

Wow, that'd be cool. A gigahertz with one. Execution per clock cycle. Yeah. Wow. That'd be crazy. That's pretty sweet. But a bit you a bit you the the price goes through the damn roof when you do that, probably

for the masks and just the manufacturing of it. Oh, yeah. Yeah, for sure. And I think it poles it was like, at 180 megahertz. Chip was testing it and it was running, full tilt to everything doing everything. It was like 980 milliamps at 1.8 volts. So it's pulling some juice.

Okay. Yeah, that's there's the army. I mean,

I think it's eight processors base, eight microcontroller processors. And they're both just ripping away. Yeah, yeah. So they said it gets a little warm.

I mean, that's what package doesn't come in. I think

it only is going to come in like a QA fn or QFP.

Okay, yeah, like an ad for 100 Something pin? Yeah, probably like 100.

pic. 32 is 140 It's probably like 100 ish. pin

right at 100. Maybe. Yeah. Well, I mean, if you think about that, that's okay. So it was nine 90 milliamps, or whatever you said at 1.8 volts, that's, that's, you know, between a watt and a half to two watts. Right. So that's in a little tiny package like that. Yeah, it's gonna get real high. It actually

has a pinout on that on this a dropbox link to take a look at that.

Yeah, I will join you on this dropbox link. I guess we'll have to supply that Dropbox.

Yeah, it's got it's 100 pins. So we were right. Yep. Okay, there we go. It's got a little diagram of like, the cogs and the hug RAM and stuff like that.

Yeah, okay. So all the if it's got 64 I out iOS, all the others are just power and ground and pretty much other. Okay. They call them smart pin. Yeah. Cool.

So we got DAX, and ADCs. But oh, I have all this stuff.

Oh, look at that. It has a huge as an exposed ground pad for for thermal. Oh, yeah, it needs it. Yeah, for sure.

But yeah, each pin has a eight bit DAC. Each pin has a delta sigma ADC. It's just crazy.

Wow, really? Yeah. Like, and you can have them all running at the same time.

I wonder if it's machst. internally? Or what?

I don't know. I bet you it's much, because that would probably be so expensive to have it on each

day. Wouldn't it be cool to do 64 ADCs

64 ADCs reads in parallel at the same time? Yeah, that would be cool. That would be cool. That would be rad

to see if we can get chipping in. And Ken on dope already set us straight.

That'd be like you guys got this. All right. Well, that is more reason to have them on. Yeah.

And then last on the RFO is kind of a discussion. Luke Roberson, from the Slack channel asks, How do you search for technical information? I am specifically thinking in regards to forums, I realize there's a niche forum for everything. But I was wondering if there were a few standouts people prefer, and presumably ones that are not laced with kids looking for answers on their homework. And he mentions ones that he uses which are Eevee, blog.com, and DIY. stompboxes. Calm?

Hmm. Well, in the what you just kind of read from his thing there. It sounds like you sort of answered your own question. What I mean by that is, you already know that there's a whole bunch of them with kids looking for answers to their homework, but the only reason why, you know, that is because you did search and like walk through things. And then you found Evie, blog, DIY stompboxes, which those are more specific. And so when it comes to searching for technical, technical information, I mean, exactly that way. You know, I've personally weeded through a bunch of forums where I'm like, Well, you know, maybe one out of every 1000 posts has something worthwhile. So I don't go to those ones as much, I suppose. I don't know. How about you, Parker.

Uh, so I used to do was it used to be a tab in Google Search called discussions. And you would search for something and click discussions, and it would just pull up forms and stuff like that. And that was amazing. You would search for a part number and click discussions. And then you would find everyone who's ever talked about that thing on the internet? And then Google got rid of it.

Wow, it was powerful. I? Well, I

think it was because people knew that it would just basically bypass ads, right? Because now if you search for a part number, the first link, like the first two pages, for part is generally how to buy it. Yep. And then maybe the PDF don't in there. And it takes a long time to get to where people were actually talking about it. I found searching like a part number and then like form in there will kick you know, it will actually look for a URL that has a form form in the URL, and that tends to help. I don't really have like forums I just go to every day. I do. I do read Hackaday every day and then I read all about circuits. I guess I read more websites and forms. So they're like electronics weekly, and E times and all those guys I basically look at those every day. And then I actually pay attention to Chris Gamble's the amp our Twitter feed, because that he usually has some pretty good links in there to read. I think I think some people at work use Hacker News, but I tend to think that the community there's kind of toxics or Stack Overflow, right? Yeah. That's not gonna overflow. It's not bad.

Yeah, Stack Overflow, you get a whole range of stuff. Yeah. You know, and I think actually come to think about it with this question a little bit deeper into it. I think it actually, one of the big things to consider is how you actually search for something in terms of like what you actually write, because if you ask a question in, say, Google, that's like, show me a constant current source, or something like that, like, I'm just using that as an example, then you're going to get a lot of like, really generic ideas about what a constant current source is. But if you say, show me a constant current force for this particular type of circuit, you know, if you get more specific, then you're going to start finding more of those forums where people are talking more intelligently about the thing that you are specifically looking for, as opposed to just a Wikipedia page that says, this is

this. Yeah, this is a current source. Right?

Exactly. And and it gets even better if you say something like, constant current source and put in a part number that you know, as part of that constant current source, or a very, like widely known or used part, well, then you start to get really specific, you start to get people who have actually built that thing. And you can see pictures of it. So I think the rabbit hole, you start with the like, cast the net wide with technical information, like constant current source, then you might find something that allows you to dig a little bit deeper with Okay, now I'm looking for a conference current choice in this situation with this part number, then you'll get exactly what you want.

The only problem with that is eventually narrowed it down to that one forum post. And the person's response has the same question. You know, this is our good friend Viper. You never know. The the best investment I think you said this last time too, though, but the best ones are like they respond. Oh, I fixed it. And that's it's not how, because then you know, it's possible. Right? But you still don't know. That's the worst. That's the worst. Yeah, that's the worst.

Don't even tell me you fixed it. If that's the case.

So actually, I had a moment like that. Okay. Well, I was looking on Reddit, and someone posted I fixed it, and they didn't post the fix. And this was like a Reddit post, like five years ago. Yeah, I actually, his his account was like, yeah, like posted like three years. So I just shot up, you know, a pm over him over to him. And he responded in a day. No, I answer and I'm like, Yes, it was like a Jeep question, but like, I was just amazed no Viper. Yeah, I found my Viper.

I think we need to make T shirts that say something like that. I found my Viper. Actually wait no, I just realized that's a really awful thing.

Don't be a viper. Post your answers online. Yeah, there we go. Yeah, something like that. Maybe we can make that new T shirt sometimes it's

got to be like Viper one to seven or something like that. Yeah.

Oh, cool. Yeah.

That was the macro five engineering podcast. We were your hosts Stephen Craig

and Parker Dolman. Take it easy. Later everyone say I didn't forget that time.

Yes

thank you. Yes, you our listener for downloading our show. If you have a cool idea, project topic, or Viper like story that you want Steven and I discussed Tweet us at Mac fab or email us at podcasts at macro calm. Also check out our Slack channel. Because if you post answers there, you won't become Viper 142 If you're not subscribed to the podcast yet, click that subscribe button that way you get the latest episode right way releases and please review us wherever you listen, as it helps the show stay visible and helps new listeners not find Viper