Parker learns Python and OpenCV and Stephen gets silly with transformers.
Podcast Transcript
Host 2
Hello, and welcome to the macro fab engineering podcast where your hosts Parker Dohmen, and Steven Craig, this is episode number 74.
Host 3
Hey, listeners, if you enjoy the Mac fab engineering podcast, please let others know about us. Tell your co workers, your friends, your family loved ones, and share it on social media at macro fab, or follow us on Facebook.
Host 2
And I still need to add it to this list. But the Instagram is Mac fab Inc.
Host 3
That's right. At some point during the show, we're going to announce a secret code word. If you email us the code word and your address, we'll send some cool macro fab swag your way, the email address is podcast at macro fab.com. And I
Host 2
would like to thank you for downloading our show. And if you have not, please subscribe to the podcast. And make sure that you basically if you subscribe to the RSS feed or to iTunes or to the Google or wherever you can download this thing. It will let you know when a new episode comes out. That's right. And if you could leave a short review on our iTunes Page five star reviews only. They'll help other listeners find us. Yeah, thanks a lot. Are you thanking them before they they give a review?
Host 3
I'll think again at the end. What's been up Parker?
Host 2
Um, so this past couple weeks I've been learning Python, right and open CV which is open computer vision. I think I talked about this a little bit two weeks ago.
Host 3
And probably in the past. Also I know we've mentioned it, but yeah,
Host 2
basically I just like, like about two Monday's ago, I sat down from my computer and I was like, I am going to figure out how to do like, basically web development and open CV and stuff like that. And so I set up a Ubuntu virtual machine and then got pi charm, which is a Python IDE installed. And I basically just started hitting our API endpoints on McAfee comm. Right? And pulling in order information and stuff. And I'm just like, I felt this way web developers have it's so much easier than hardware. This stuff just works.
Host 3
Yeah, yeah, you write it you write whatever command it is, and you just get whatever you need. Yeah, it couldn't be awesome. Not that easy. I know. It's not that easy. It could
Host 2
be though, that our our web developers made a really good API that's easy to use, possibly. It's probably both
Host 3
probably, yeah. Because they've probably been through the crap before,
Host 2
probably. And then the, for the Open CV side, I enrolled in like this, like free 10 day. Like, class, the, the blog is pi in image search. com, I'll put the link down below. Okay. But basically, you sign up and the guy tries to sell you his ebook, like almost every other turn. But he's got like a 10, free day clear of free thing, like, like a lecture series kind of thing. Kind of it's, it's very example, based.
Host 3
Oh, okay, download this and just look through it kind of thing.
Host 2
Well, and then he has a blog post that goes with it that, you know, he'll have, he'll break the code down line by line of what's going on. Know, what does this function do? And then what I'll do is actually take the function and go look in the open CV documentation and see exactly what it does.
Host 3
Okay, they break down the syntax and everything. Yeah, cool,
Host 2
but and they'll, but they only tell what that function does in relationship with what we're using as in the project. Whereas I want to know, like, everything. Yeah. So I will go and read that documentation. And so yeah, I got a basically I got a webcam working. So I can pull images in. I can crop I can. I actually just today got Lens Correction working.
Host 3
Right. So you can take all fisheye Yeah, basically,
Host 2
yeah, take the round this off. And that works really well. So then, I am basically taking pictures of PCBs and cropping out all the parts and stuff, right, using
Host 3
the XY RS data, or that just the XY locations, xy locations
Host 2
that I am pulling through our API. Well, plus,
Host 3
do you also use the XY size of the part? Yes. Okay. So you so you crop the image based off of the XY location and size,
Host 2
put a little margin just so you know, right? It's a little more area 10% Or something like that. Yeah. And then so pull that image. And so the idea is to be able to do quality control on a prototype level, visual QC, without having to like sit there and Remember, if you've looked at the entire board,
Host 3
right? Or, or instead of having to just, you know, go all day long with your eyes under a microscope, you can take all these cropped images and do QC on the screen. Yes. Right.
Host 2
Yeah. Cool. Don't be the guy who's even doing QC doesn't even need to be at that bench anymore. either. He can be remote. Yeah. Yeah, watch sends it to your phone and you just like, it's like,
Host 3
you're in the bathroom and doing keys. What's that
Host 2
app? Is it Tinder? We swipe right or swipe left? Yeah, so that's what's except, right. So you swipe right for good parts and swipe left if there's bad,
Host 3
and it shows the heart or the heartbreaking based off of a sleeping writer PCB Tinder, I like that. I think that might be the code word. PCB Tinder, that PCB tender. Alright, that's the code word. Send that in with your address to podcast at macro fab.com.
Host 2
So that's, that's that project I've been working on for the past couple of weeks. The Raspberry Pi three compute module board landed on my desk today. Yep. And this is the board that Steven and I have a basically a six pack of beer riding on whether or not it'll work correctly the first time or not? Right? Yeah. So the
Host 3
so you've already powered it powered it up and didn't blow up. It pulled like 20 milliamps. Yeah. Which is fine.
Host 2
That's basically like, just quiet. Of the regulators and right and leakage through capacitors. So
Host 3
you don't all right so far. Yeah, it's there still might be something wrong with it. Yeah. Cuz I saw
Host 2
out the plug the, basically the Raspberry Pi into it. Right. And I don't get one of those until my friend gets into town this weekend. And so we'll know next week. Yeah. Cool. Yeah, listen to us. We'll know next week whether or not I messed up or not. But I didn't mess up on the
Host 3
Host 2
I counted. The compressor IoT board that's got the Particle Photon on it, right. And that's the project where I'm gonna stick this board, basically glue it to the side of the compressor, and it will measure how long the compressor runs for it can turn off the compressor. If there's a, you know, an error. Whenever it detects that the compressor is running for a long time it goes okay, something's probably wrong. Shut it off. It can it can read the temperature of the room in case it gets too hot. You know, bunch of environmental stuff I can do. Yeah. So I plugged it in, and it pulled an amp and a half at five volts, which is not supposed to right. And then I bet you know, the first thing you do, of course is like you start touching the board and forget what's hot. It's I put my thumb on the op amp, which was a 45554558558. Yeah. Generic. I think it's a diode in one. Okay, brand.
Host 3
Yeah, it's a jelly bean, Jelly
Host 2
op amp audio op amp. And it like it's singed, a, you know, so I see a square into my thumb. And then I'm like, oh, yeah, that's hot. And then I took my thumb off and immediately let the smoke out. In the power supply dropped down to like 20 milliamps,
Host 3
which is what you would expect. That's what I wanted. So that op amp was taking everything. Yeah.
Host 2
And so I'm like, okay, that's weird, because I've used this part before. On other projects, I'm like this
Host 3
differently. You used it on the, the jeep. Okay. That's part of the audio switching. Sorry. Yeah.
Host 2
The Jeep Bluetooth switcher has these has like, two of them on
Host 3
it. Right. And they worked. I mean, as soon as the board came in, you powered it up. And that worked fine. Yeah, that was
Host 2
another board. I made that work. First, just flawlessly. Yeah. Yeah. So
Host 3
what happened on this one? So?
Host 2
Because and we pulled up the board layout? Yeah. And we're like, and you were like, you knew what it was supposed to be? Pin a on a standard dual op amp package is positive voltage pin and pin four is negative for ground and so I right clicked on that and I had an backwards Yep. So I'm like, that's weird because you know, the outputs pin one. And so we went to pin one and it was output. Yeah. And the the negative and positive and second or third? were correct. Yep. And so we went back to the schematic and we're looking at it and I right clicked on it and hit properties and I had mirrored the parts vertically vertically to switch the the positive and negative rails the flip them around so the schematic would make more sense when you do it. But I did not label the pins on the op amps symbol for power. They were just legs coming off. Yeah, why they didn't have like VCC and VDD right you
Host 3
didn't you didn't realize it but you flipped ground and power on the schematic symbol. Yes. And then just connected them up that way. Yeah. Opposite so yeah, I had Oh wow. Wow. So that brings up a pet peeve of mine. Okay. So when you go and look at pull up any op amp example on like go to Google and type in like op amp circuits or whatnot, the almost 99% of the time you see an op amp, the non inverting terminal is above the inverting terminal, so the minus sign is up and the plus sign is down. I don't know how Eagle does it by default. But in dip trace, all op amps are drawn. The other way around such that plus is up and minus is down. So pretty much no matter what if you plop down a default op amp from dip trace into a schematic, you have to do a vertical flip. I know it's just like a right click, but it's also like everyone always has the non inverting terminal. I'm sorry, the inverting terminal facing upward. So just drop that way. Even if you're doing a non inverting amplifier, you still have the plus, typically it's traditional to have it down. So I don't know, it's just a pet peeve of mine.
Host 2
I always draw mine. That way. I just flip them around sometimes when it's the makes the schematic look cleaner.
Host 3
Yeah, yeah, well, and another thing too, is that the exact same thing has happened to me, in terms of flipping them vertically. On in dip trace, they have the positive terminal, the positive input, and the positive voltage on the same side of the op amp. So if you flip it upside down the way it is drawn in like 99% of the world, then you have to connect power on the bottom side of the op amp for the schematic side. So it's just annoying. I have to go in and modify the package. I hate doing that. Like, just I don't know. It's a pet peeve? Yeah. Maybe Eagle does it the same?
Host 2
I don't Well, Eagle you. I actually have never looked at the default libraries of Eagle. You've always done it all yourself or has done it myself. Eagle. So Steven, yeah. Any since stuff?
Host 3
Yes. And no. And yes, some more. So I think two weeks ago, I talked about building some more modules. Yeah. So I'm still I'm still kind of chugging through those right now. So I'm building two envelopes, and one filter. And one of the reasons why I've kind of just haven't gotten them out is because I'm still waiting on a part. So
Unknown Speaker
why are you building three envelopes now?
Host 3
Well, I'm canning the one that's on the on my current PCB. So I'm building to it, the intent was always to have two envelopes, such that you could control the amplifier, the voltage control amplifier and the voltage controlled filter separately, are they designed the same? They're identical, okay. So across the faceplate, in fact, in one of our episodes, I can put another picture of the faceplate, I have switches on the front such that any module in the synth can access either envelope one or two, you select it from the front. Gotcha. So you could have the amp on both of the amp and filter on one envelope or opposites or whatever. So So yeah, I'm building those. And those are pretty simple. I already got the strip boards and I and I cut all my traces and things like that I already have. I've got all the parts sitting on my coffee table. I just need to kind of put them together. Kind of another reason why I haven't finished building them yet is because my soldering iron broke the tip of it. And not not the actual tip itself. But the the the plastic sheath that goes around it Radio Shack firestick. It was it knows a mark Ethan, from Amazon. It was like 6070 bucks. It's a great iron, except for the fact that this plastic part breaks
Unknown Speaker
front fell off the front. Yeah, the front fell off. Oh, well,
Host 3
yeah. So I'm like that. That's a YouTube video. Right? Yeah, so that happened. So the I've got a new part coming in tonight. So I'll be getting on that. But the the part that I'm kind of waiting on for finishing the filter is the ca 3146, which is just a five up array of NPN transistors in a DIP package. And that's kind of a traditional IC that's used in these filters. And I didn't have to use it. I could have gone a bunch of different routes. But I wanted to build like an old school filter. So this is like a traditional design. So I bought that IC from a place that claimed to have it in stock. They even had like, we have 94 of these in stock. Yeah, exactly. Yeah. And yeah, exact number and then I contact them a week later because I haven't gotten the shipping notice. And they're like, Oh, well we've had to buy it from another place. They have to get it in then they're gonna ship it to me. So it's like, Oh, great. I should have just bought it from eBay. You know, because then like you do actually know what you're getting. Yeah, this this. This chip has been obsolete for years and years. So why
Host 2
are you using an obsolete part for traditional? Okay,
Host 3
that's it's purely just, I think it's fun. I think it's Want to use like old stuff like that, you know, like, I would love to eventually get into si d chips from like a Commodore 64. You know, those are obsolete, but you can still get them on eBay. It's it's more of like a, I want that specific filter from a specific year. And that chip was used on that filter. So they were the place a bottom from they were like 80 cents. And I thought they had old stock, but I guess they didn't, whatever, I got five of the chips coming so I can build some a couple filters. So I'm also working on some some new power supply circuitry for the tube mics that we've talked about here. Yeah.
Host 2
How long ago was that? We gave eight episodes
Host 3
ago. The tube mics? Yeah. All maybe more than that. We gave Josh?
Host 2
Oh, no, it was the absolute with Josh. Yeah, that's absolute, like 60
Host 3
something, it's something like that. Yeah, we gave him one of these mics for a Christmas present. We built him one of these mics. And his power supply works fine for it. But I want to do some mods too. And I basically what I want to do is just kind of adjust some voltages and stiffen things up because the power supply that comes with the mic, or the kit that we bought, it's all right, at best, but it's not great, I can do a whole hell of a lot better. So one of the things that this this power supply needs it since I've got 120 volts available to me, it needs a 6.3 or a 12.6 volt rail, DC. But it also needs something in the range of 200 to 300 volts. And I don't want to spend a bunch of money getting like a custom transformer wound. So I have a whole box full of these little flat pack trans transformers at my guitar shop that they're like, they're meant to be a you know, dual primary. So you can either put 220 or 120 into it, and you get out 10 volts, or 20 volts. So it's kind of ridiculous. But what I'm going to do is I'm going to take 120, put it into one of these transformers, and I'm going to drop that down to 10 volts, such that I can get the heaters for the tubes off of that. But then I'm going to take a second one of these transformers, flip it around, put the 10 volts back into its 10 volt line and flip that around to 220 and rectify that, ah so that I can get both the heater line and the high voltage line. It takes two of these transformers. But I literally have 120 of these transformers in a box that I've had for years. So it's just like story behind
Unknown Speaker
this box for transformers. You know, I
Host 3
was building guitar pedals a while back. And there's a there's an electronic store here in Houston that had been given
Host 2
like it was like on the side of the road. And like you have like a box of free kittens. It's free.
Host 3
Oh my gosh, that would have been amazing. The box is like you know, when greasy looking and Oh, well. So I got these transformers to build guitar pedals. And I used only like a handful of them. But I couldn't deal with an electronics during Houston and got that whole box for nothing. And of course, you know, my style, I kind of hoard all the parts that I get my chops so full of friggin parts. And I knew one day I was going to use these transformers, and now I am using them. And it's gonna be Yeah, I'm gonna use two of them. The 120 So yeah, that'll be fun. It's it's kind of ridiculous and way overkill, but it's free. I don't have to go out and buy, you know, 20 $30 worth of transformers? Yep. So that's what I've been up
Host 2
to. I guess we'll move on to the pow. The pick of the week of the week. So SparkFun always like SparkFun. Yeah, they came out with 100. But this is what the closure of Radio Shack
Host 3
right? Because it's a little bit of an homage to radio show. Yeah.
Host 2
SparkFun came out with their 130 in one electronic playground. Starter Kit thing. Yeah. Which is basically you get an electronics book a bunch of parts a piece of, you know, really fancy plastic slash cardboard with springs on it. And you make circuits.
Host 3
It's kind of, in a lot of ways. It's it's people's first electronic kit.
Host 2
Yeah. And what with the closure of Radio Shack? Where would you buy one of these? I guess maybe fries, but fries is everywhere, right? Same thing with like, Microcenter or whatever. And so now you can just buy a kit on SparkFun for only 50 bucks. Yeah, it's actually a really good deal.
Host 3
It's like half the price of radio. Yeah. Yeah, I remember I had a couple of these when I was a kid. And the thing is, like, they're really cool at first and then you realize that you can't do a whole lot with them. I mean, most of the circuits in there, like, learn how to turn on a light bulb or something like that. There was always like radios 20 130 Yeah, it really is. Like yeah, only five or six are like really interesting or something. The thing is, they're they're made as cheap as possible. They got these two plastic side plates. But the actual board that you work on is made of cardboard. And if it gets even slightly moist, it just gets completely gross. And I had one or two that was just like real floppy and nasty. And
Host 2
what Hey, I built a bunch of circles spill your milk on it. That's my chocolate milk. So on to the RFO until our foe. So we have three this week, like usual. The first one is fail the weak museum buttons found on Hackaday. I picked that one because you are doing a museum project. That's right. So that was going to be an interesting one would Lichtenberg that we as a project, slash tutorial on instructable. Instructables. Yeah. And then Qualcomm Qualcomm fingerprint sensor enabled virtual home button,
Host 3
and that was electronic news weekly.
Host 2
Yes. So the first one, fail the week museum buttons. This is like I think actually one of the editors or writers on Hackaday had an article about a museum piece that they are designing. And they were first using industrial control buttons. And those, you would think that'd be really good because it's got industrial in the name, right. But they actually are really bad for this kind of environment. Because this has got kids you know, how he says it picking prodding and licking switches. And they're not designed for that they're designed for a human with a glove the press it, right, that's it, a person in a, you know, industrial plant isn't going to, like, pick at a button or like lick it or whatever, kids will destroy a button. So industrial switch is more of it can it has long, legit blin longevity, right? In just actuations.
Host 3
Right. And in some ways, it's kind of guaranteed for that number of activations
Host 2
and it's, it might have a IP writing, it might have a div writing,
Host 3
yeah, some kind of chemical rating or Yeah, explosion proof, whatever. Whereas
Unknown Speaker
for kid proof, you need vandal proof switch,
Host 3
again, Bulletproof Bulletproof.
Host 2
And so he goes into some things with like, so that didn't work. And so they switched out to vandal proof switches, but those are pretty pricey. They're like 30 bucks a switch. Those are the kind of switches you see in like elevators and stuff, where they are really flat. And when you press them, there's not a lot tolerance, so you can't really get a pic or a pry bar in there or anything. Or, you know, spit. And so they tried to do capacitive touch. Yeah. Which worked in their lab, but they installed it at the museum, and it didn't work at all. Oh, yeah. And so they had some weird issues with like, grounds or something. They actually, from the time of the article, they haven't figured out that solution yet. And so I was going to ask you is What are y'all doing for the drill?
Host 3
So yeah, real projects for them for our drill projects. So we ours is continual motion. So basically, when the museum opens up, they effectively plug it in, or turn on the light switch, and it activates and it just runs nonstop. So yeah, we didn't we have, our drill is purely for visual, there's no interaction with it, there's going to be like a little kiosk around it that has like, you know, here's a drill head and you can watch it kind of thing. But but no actual buttons. Now, if I were to use a button in that, in that situation, I what I would likely do is actually purchase a handful of buttons, and just beat the living shit out of a bunch of them and find out which one
Host 2
works. I would use an arcade style button. Oh, yeah, yeah, deep body solid plastic and then with a micro switch on the back. Yeah, cuz there's, if they can survive an arcade, they can survive a museum piece.
Host 3
Well, and then the thing is, like, it's guaranteed to have a boatload of kids coming by slamming it as hard as they possibly can. Or hitting it like an elevator button just you know, just like pounding on it. It is it is guaranteed to see its actuations very quickly, or it's it's limited limited numbers. Yeah, yeah, it's a brutal environment.
Host 2
Should have kid ratings and OKRs KR
Host 3
KR six man, that's awesome.
Host 2
Well, no cuz you have a physical kid rating and then a, like, moisture kid rating for like spit and juice and food and stuff like that. So it'd be just like an IP rating except KR, so KR 67.
Host 3
Well care 67 has like all of them put together. Yeah. Yeah.
Host 2
So 60k or 67 is like, a kid can never destroy that button.
Host 3
Yeah, right. It can, it can withstand, you know, gallons of apple juice being spilled on it, and hours of button pressing. So maybe our new testing facilities, we just, you know, give our products in an elementary school come back in a week, and they give us a report on it and how destroyed it's a good idea. Yeah.
Unknown Speaker
So second RFO. Yeah, would
Host 3
Lichtenberg, this is sort of an internal one in a way.
Host 2
Justin, our Director of Operations at macro frapp, VP of Operations, VP operations. So you got promoted?
Host 3
No, no, it's always been, you know, he's always been VP. Okay. Just not to you.
Host 2
Um, he came to us a couple of weeks ago, with this, this side project, I guess, yes, this link, and it's really cool. It's basically a piece of wood, a lot of high voltage, which is always fun to play with. And it makes like lightning bolts seared into the wood.
Host 3
Yeah, yeah. So you actually burn a pattern of lightning ish things into the wood.
Host 2
And so what, what they do is you take baking soda, and then you mix it in the water. And that's your electrolyte and current carrier. Yeah. And you soak the wood in it. And then you apply, you know, like six to 12 kilovolts to the either ends of the wood, and that arc through, right. But it's, it's kind of slow, it burns its way through, which is really cool. There's, there's a bunch of videos on YouTube that show this. And so most people use like a neon sign transformer. And so we picked one of these things up for like, 60 bucks on them. Yeah, they're really not that bad. And this one does. I can't remember the part number, but it does 12 kilo volts at 35 milliamps, right. And so I was like, Well, if the power supply used all its power to burn the wood, then the wood would have to be around what 342,000 ish fold.
Host 3
You just did that calculation in your head didn't chew on,
Unknown Speaker
Host 3
What I like is what Justin keeps saying about this project. He he'll keep saying it won't kill you. It'll just severely mess you up if you get shocked. But yeah, we're gonna we're gonna attempt to be as safe as possible when doing this. Yep. It really it really isn't that bad. You just kind of you marinate the wood a little bit in baking soda water. Yeah, you drive nails into either end. Clip them on with, you know, basically jumper cables that are far apart from each other. And then just roasted. Roasted. Yeah, it's really cool. So yeah, well, maybe
Unknown Speaker
I do drink. Oh, that'd be cool. Yeah, we'll do a Facebook Live Stream. Yeah. Well,
Host 3
we'll have to figure out exactly when we're going to pull it off. Yeah, yeah.
Host 2
third topic. The Qualcomm fingerprint sensor enables virtual home button.
Host 3
This is really neat. Because of the implications of where things are going, in my opinion. So basically, what this is, is Qualcomm has now been created a fingerprint sensor or an ultrasonic sensor that can exist underneath the glass in a touchscreen. Oh, yeah. Mainly phones. Yeah. And it can act as a home button without actually having a physical tactile Home button. Yeah. It's
Host 2
interesting that it's ultra. Ultrasonic. Yeah. So the resolution is high enough for that sensor to read the ridges in your finger.
Host 3
That's right. Yeah. And so the article that we have here doesn't, doesn't go into a lot of detail on the how it works more of just like this now exists. Yeah, this now exists. But one thing that they were talking about is they can actually detect your heart heartbeat, which is not that crazy, but they also say blood flow. So that's curious. And I want to know, what they're actually doing to detect the actual flow of your blood. You know, how far are we away from being able to you know, do some diagnostics, just by putting your fingers on your phone
Host 2
and some other stats they had there? It goes through the sensor can read through 800 micro meters. Yep, that sounds like a lot but the glass on your phone is very thin. Yep. So yeah, that's the glass and they can go up to 650 micro meters of aluminum signal through solid aluminum. That's
Host 3
impressive. Yeah. The cool thing here's here's what I'm I'm kind of liking and they they said that the I Phone eight. I don't know when that's coming out. But that's supposedly supposed to incorporate this. So they're getting rid of their one button, you know, altogether on the front. What what I see us moving towards Apple already got rid of the headphone port, you know they do Bluetooth, also the green that they're getting rid of buttons entirely. So you could you potentially have wireless charging wireless audio, and no buttons on front. Could you have a hermetically sealed phone? Probably so that they get the audio out audio in and out? Well. Okay, that's the last one that I see the speaker. But can you not come up with some way to vibrate the case?
Host 2
Yeah, you could just have a transducer on the case. That's right. Yeah.
Host 3
So can we have fully sealed phones? I think that's where we're headed towards Proko.
Host 2
I was actually thinking about the fact that it can read your heartbeat and blood flow. Yeah. And so Apple can read data on like, what apps make you excited? And like what videos you're watching at the time and stuff like that?
Host 3
Oh, yeah. This is where we put our tinfoil hats on. I guarantee you they're doing crap like that. Anyway, they can find to make more money off of
Host 2
them. They will or like, oh, this ad board them. So you know, because their heart rate went down or something like that. Yeah, probably.
Unknown Speaker
Yeah. Cool. Neat. Yeah. Awesome.
Host 2
Do we have I don't think we have any other news. We had the. We had the meetup. Yesterday. Yep. That went pretty well. So we'll have another one in July. end of July. So go is a sign up up on Eventbrite yet. Not yet. Okay. So when that comes up with Harvey by next week. Yeah, keep your eyes open for it. Yep. So that was the macro fab engineering podcast. We are your hosts, Parker Dohmen. And Steven Craig. Later Everyone, take it easy.
