MEP EP#108: Crazy Transistor Contraptions

Podcast Transcript

Host 3     00:10

Hello, and welcome to the macro fab engineering podcast. We are your host, Stephen Craig

Host 2     00:15

And Parker Dolman. This is episode 1080. Yeah, yeah. So Stephen. Yes, sir. You've been working on a new synthesizer design. That's right. So what's been going on with that?

Host 3     00:27

Well, so that's actually complete, and I'm using air quotes.

Host 2     00:31

That's what I mean, all your projects are complete air quotes,

Host 3     00:35

No complete as in like, the circuit does everything I want it to it just it the case comes in tomorrow. And I just have to screw it in. That's the complete part that needs to finish. So it's like nothing more electrical engineering has to happen to this it is done. And that you're right. That is like, very uncharacteristic. So yeah, we'll be we'll be actually talking about the synthesizer proper in a future episode. But today, I want to talk to you about an issue I had that was kind of interesting. Okay. So above and beyond the synth design, I've been making a MIDI to CV converter. Yep. Which we talked about a couple episodes ago. That particular design is still kind of in the digital realm right now in terms of like, My computer has the design, I haven't actually ordered it, I did a stopgap design in a way that was just enough to keep me in between having like the monster device. So what I made was it was a teensy device that just PWM doubt, whatever voltage I wanted, because I just blow past it effectively added a little bit of gain such that my teensy was able to spit out a zero to 10 volt signal. I used the Teensies like built in Arduino library to sniff USB and get MIDI codes. And now I can actually have my computer directly play my synth, similar to my old synth. But yep, this one is a bit more accurate and runs a lot faster. microcontroller. Yeah. Especially because you can overclock it to 240 megahertz. Yep. But yeah. So I can you do that? How

Host 1     02:09

Do you do that?

Host 3     02:10

It's a it's just a build option. Oh, yeah. You just go up to like the tools thing and just say go fast, go faster. I mean, like, it's already 180 megahertz like, yeah, you kind of have to need to overclock it in order for it to do that. But I've done it just because whatever. I've done it and let it run for like an hour and built the chip on the Teensy, and it doesn't really even get that hot. But I'm not really asking much of it. It's just like, reads USB and then does something. So above and beyond it having like a PWM voltage output. I also have it have like, turn up in high and low. It turns a pin high when it gets a note on command. And it turns a pin low when it gets the note off command. So I can I think

Host 2     02:52

You use that for an envelope. Yeah, that was a lot on your lesson. Yeah, that's right. Yeah.

Host 3     02:56

And that's, that's exactly what I'm doing on this. And my envelope, it's supposed to receive as like a zero to 10 volt signal 10 Being on, but it's just a transistor base at the input. So I can send it 3.3 volts. And that's more than enough to turn it on. So I don't even have to have special circuitry, I have the Teensy coordinate connected directly to that. So I get all this done. And it's like a great, let's start doing some stuff with it. Let's start playing some music. And of course, I go to Super Mario Brothers. And and I want to make it clear that like, I'm not, I haven't made this musical instrument just to play like video game music. But the thing that's great about Super Mario Brothers is everyone knows it, you know exactly what it should sound like. And it's a pretty decent test as like a kind of like a benchmark, because it's just a bunch of square waves. If you need to detect frequencies or whatever errors like it's just like a really it's like venchi on a 3d printer. But for Yeah, it's just a benchmark. It's a benchmark. So I started playing that. And it sounds kind of funky. It's it. All the frequencies were on I had it tuned right. And it played well. But like the notes were sort of not aligning up very well. So Super Mario Brothers had four tracks. The fourth track is just a percussion. It just has that kind of sound. So I didn't play that. The other three are actual notes. You have a high melody, you have a mid level melody, and then you have the baseline that goes through there. So I recorded three separate tracks all just my computer's spitting out MIDI codes and wasn't playing it and just recorded it back in. And it sounds like Mario Brothers. It sounds like what you expect, but it's a little bit off. And I couldn't figure out why it just the notes didn't seem to line up super well. So I started playing around with it and the entire time. I was blaming either my circuits or my code in my Teensy. I'm like okay, well, I've done something wrong here. And I end up spending like, seriously like six hours just at my desk banging My head on the desk, I rewrote my midi code in the Teensy four separate times, doing it four different ways for different ways. Yeah, no, I had I had like, straight up my main loop doing it. I had like case statements doing it, I had a different one where I mailbox it and had an ISR, looking at the mailbox. I'm doing all different ways. And every single time it responds poorly, and doesn't do exactly what I'm thinking. I've got my scope reading, like all these points on this stuff. And it's just like, driving me insane. So what I ended up pulling my my trigger signal off. So the one from the Teensy that goes from zero to 3.3. Yeah, on a note on note off, I pulled that off and just have it pipeline audio out. So it's not turning the notes on and off. It's just shifting pitch. And it's perfect. It sounds exactly like it should, it's just not turning the volume on and off. So I'm like, okay, great. So I'm getting the exact rate codes, the pitch is being sent properly across MIDI, but it's not telling it to turn the notes on and off in the right time. And so I pull up my scope, and I look at the signal. And it should be nice square waves getting dumped out. And they're not. They are square waves, but they're not the right length. They're all over the place. So we ended up creating another little MIDI track that just had four notes that just, you know, did at it up in scale, each one having the exact same length, and the exact same amount of pause in between each note. So it should have a very known square wave that comes out of Yeah, and my scope should be able to trigger on something like that. So I spit that out. And my scope starts freaking out like it can't see anything. So I'm getting like big pulses and medium pulses, and it's just completely random. So do you know of the Millis M? I? LL is command or function.

Host 2     06:53

Yeah, it's it's what you normally use for delay functions.

Host 3     06:57

Yeah, yeah. Well, it basically just takes a timestamp. Yeah. So the Arduino is like constantly counting milliseconds. And if you call that function and assign it to a variable, it will just say like, oh, the, the time in milliseconds is right now. Yep. And then later on, you can call that function again, take the difference, and you are able to time so so I redid my code such that when I got a note on command, I would call Millis. And when I got a note off command, I'd call it again, take the difference, and then spit out what it was getting. So every single one of my notes that I was spitting out in this little arpeggio was supposed to be point two, five seconds. And I was getting all over the place. Yeah. And you're spitting that out over like cereal, right? A USB. Okay, well, it's it's serial over USB. Okay, innocence. So, so So I'm getting like, my teensy has given me all kinds of garbage here. So I was like, Okay, I've got to try something different. Here. I went, and I downloaded just like a mini player, like, you know, just one where you could load whatever medium just spit out data. And I had it play the exact same thing. And it spits the data out perfectly. And my teensy spitting the data out perfectly, and it was showing point two, five seconds. Yeah, every single note. So all of that time I had rewritten my code I had looked over myself because I've done all that what it was was Reaper is terrible at middie

Host 1     08:24

And Reapers the software you're using.

Host 3     08:26

That's right. Reaper Reaper is a free DAW, which is like an audio recording software. Apparently, and and after, of course, I did some research and this is all over the internet. Reaper sucks at spitting out mity on time.

Host 1     08:42

And another thing is we were using Reaper for the previous synth.

Host 3     08:46

That's right, but we weren't doing it at a fast enough rate for you to actually hear it. Ah, okay, so So I actually have two tracks I have, well, I have my original Super Mario Brothers track where we'll put it up and you'll you'll be able to hear it. It has, it just sounds like Super Mario Brothers, but it's a little bit off. And then I took a particular segment of it. And and I slowed that down. And it's just like, it's literally like two seconds, but I extended it and slowed it down a lot. You can see that the notes that all should be hitting at the exact same time. Our shifts are shifted so early. Some of them hit late. It all depends on what Reaper decided to spit out at that one point in time. And the worst case that in this one two second segment was point 117 seconds off of what it should be. And that's way long enough. Yeah. Human to hear here. Yeah, even at the regular speed. The thing is like you don't hear it as like the notes being like played differently. It just the whole track sounds bad. Yep. So it was like, the whole time. I'm blaming myself, I'm blaming my design. I even got to the point where you know, I'm using someone else's libraries in the Teensy. I'm like I have no idea what their timing looks like. Maybe their timing is somewhat random, maybe, maybe a note on command takes longer than a note off command. And that's throwing things off. No, the entire time. It was my computer spitting it out incorrectly. So

Host 2     10:15

Now you have a reference, like setup for your software. Yeah, yeah, exactly.

Host 3     10:22

That was such a pain in the ass. But it's great. And like, as soon as I fired it up with the other program, it's awesome. It sounds great. Yeah, he's great. It's just like, Oh, my God, it's been so long doing this. But it's so hard to diagnose crap like that. Oh, yeah. Because like that, that was literally the last thing I was looking

Host 1     10:40

At you. You assumed? Oh, I do. Like you say,

Host 3     10:43

Would it be wrong? Yeah. Why would it be wrong? Yeah. And and the thing was, like, it's only partially wrong. It's, it's completely sending the right pitch data, the pitch data is was perfect. Yeah, it's just, it would send it sort of whenever it felt like it. And so it's just like, oh, my gosh, and I don't, from what I Googled, it didn't seem like there's like a mini buffer or anything that I can make it go faster, or make it think faster. It just seems to be like an inherent problem. And that kind of sucks. Because Reaper is awesome for being free. It's great for like, actually recording waveforms, but I guess it's middie engine is just need some love.

Host 2     11:21

Yeah. Or it's banked on. You know, it's it's running some kind of process. That's not a high priority for middie.

Host 3     11:29

Yeah, you know, I was that like, like I was saying with the buffer. I was I was seeing if there was some kind of setting. Because right now I don't need the middIe to be live. It could it could think about it for minutes and then play the the middIe code. I don't care how long it takes as long as it's perfect every time it comes out. But it's just not. Yeah. So. So that was that was Yeah. And that was on my birthday. I spent six hours on my birthday. I'm like, Oh my God. But yeah, we'll have to, um, I've said this a couple of times in the past, but now that I have it all, like put together I'm gonna bring this in then with a working software and have Parker play the synth on

Host 2     12:10

On the pond. So I guess at the end of this podcast, we'll just play the clip.

Host 3     12:14

Yeah, yeah, yeah, we'll play we'll play both clips, the the regular one, and then the one that's slowed down, so you can hear the notes being all jacked

Host 2     12:20

Up. Cool. So in a similar route realm, I guess I'm working on those digital analog converters. Yeah, using the PCM 5122. I think I think I've said that number enough. I'm pretty sure that's the correct part number

Host 3     12:34

I did, they're gonna change it. So that it's what you say now.

Host 2     12:39

Um, so we did a blind test today, with these DAX so I've been testing them, you if you want, like, follow me on Twitter, like I've been posting like waveforms I've been taking. So basically sending it a one kilohertz signal, recording the output on my laptop with the line input. And then doing like an FFT analysis on it and seeing what comes out with. And the biggest thing I've noticed with this DAC is how much you load its output is crazy, how much like changes what your your signal looks like? I guess. I come from a digital realm. So everything is basically you know, transistor input or output. And so it's like, you don't really worry too much about loading, like MOSFETs. Sure you have gate capacitance, and that affects how fast you can switch stuff. But the analog realm in terms of, of impedance matching is something that's completely different. That haven't really mess with you don't you

Host 3     13:42

Don't get the luxury of rules of thumb. When it comes to the analog? Well, it's not like, Well, 10k is a good value. So yeah, you know, like, you don't get that as much.

Host 2     13:50

So so because I first sampled the output of the DAC, with my scope, which has got like, you know, it's probably not this, but it's like a giga ohm input. It's really high.

Host 3     14:02

It's probably one to 10 Meg. Yeah, you probably put

Host 2     14:05

A, you know, off, off the shelf scope on it off some scope off shelf multimeter and apartment show open load. That's supposed to. So I pumped the DAC straight into that. Right. And I was getting a lot of harmonics. Yes, I'm sure No, no, no, no, no. A lot of harmonics. Yes. A lot of harmonics. So it's like a one kilohertz signal. And I was getting a 2k 3k 5k, blah, blah, blah, you know, you know, all your primary and secondary harmonics. And they weren't super, like, high compared to the primary one kilohertz signal. But they were there and I'm like, That's interesting. I wonder why that is and then person the Slack channel suggested I switch to plugging in and seeing, you know, maybe it's the scope can't like, because only like a 10 bit scope at best. Yeah. And so it's like, okay, it's outputting 24 bit audio. And I'm probably sampling. I'm sampling like, you know, what, 250 million samples a second, which is way over two, so I'm probably getting crazy Yeah, well, no, it's not a listen. It's like a when you want to ban bandwidth limit your sample, I can't remember what that term is called. bandwidth limiting. Well, no, there's a special term for it. Anyways, I should be bandwidth limiting. And so I did that, and it cleaned it up a bit. Yeah. And so I'm like, Okay, let's try to the, the input on my, my laptop. And so I got like a really fancy laptop that's got like, it's got microphone and line in on it. So at first it plugged into the microphone. And it was like it like killed everything outside of the one kilo hertz on the spectrum. Hmm. So it was only the one kilo hertz signal. And I like That's weird. It's completely different. And so not plugged into line in. And then some of those harmonics came back. And then I started doing some Googling and like, Oh, what a batch of that microphone input is, it's supposed to be really low impedance, because I microphone doesn't have a really big signal. And then the scope has a ginormous and putting peanuts, and then the line in is somewhere in the middle, which makes sense for that result. So I was using the line input on my on my laptop for most of the recording, because that's what it's supposed to be used for. It's a line level out DAC that goes into an amplifier. Okay, and so we did that. That was the how I recorded most of the tests. And what I found out was the most expensive deck, it's got the film caps, its harmonics were suppressed the most, but it had the most amount of low range fill on like a one kilohertz signal. Okay, so when you did an FFT, you had more. I mean, it was about five dB difference on the low end.

Host 3     17:17

I was like, extra

Host 1     17:18

Stuff, just extra fill.

Host 3     17:20

Okay. Yeah. They usually call that IMD. intermodulation. Distortion. Yeah. Which is basically, well, frequencies beneath your fundamental Yeah, that are being amplified. Yeah. Yeah. So there, for some reason, the film caps for adding some IMD.

Host 2     17:33

Yeah, I wish I printed off the the charts for you to see. Yeah. But yeah, so if you can imagine it was noticeable. Yeah. Yeah. So it was like this. And there's some harmonics on there drawn

Host 3     17:45

On it. Yeah. It was very accurate. So

Host 2     17:48

That section right there, right, yeah, I'll post the pictures on the podcast. That section right? There was larger. On on the so yeah, come caps, you're pointing to the low frequency and low frequency underneath the primary.

Host 3     18:03

One thing to be careful if you were doing one kilohertz. So this wouldn't matter. But like, if you're doing like 10 hertz and doing a FFT on Yeah, it can look really funky if you don't have if you're not doing the right FFT, because the the zero or the DC component in an FFT don't show up as infinity. Yep. So it can skew the graph, depending on how tight the graph is and how many points you're doing.

Host 2     18:29

So yeah, I wasn't trying to do a, like absolute measurements. I was just like, might set this all up and just change the DAC out, right? What am I seeing things go up? Things go, he's down? That's all I really cared about. Exactly. Right. And like by how much so I noticed that the film cap ones had more of this bottom fill, but they suppressed the primary and secondary harmonics of the the fundamental frequency better. Hmm, that's really interesting.

Host 3     18:57

I wonder if that has to do anything with them potentially picking up more broadband noise. And that's just like filling up. Maybe that might be, they might be better at just like having the proper harmonics go through. But it's also giving you a lot of other crap with

Host 2     19:13

Yeah, it's possible. And then I did a sweep. So I sent that 20 Hertz to 20 kilohertz signal. Yeah. And then I did like a 10 second segment of that and captured that 10 seconds. And then Brandon FFT. And the all the DAX no matter what the loadout was, had the same flat response, the same, you know, slightly downward slope you would normally see for a DAC. Yeah. So as the frequency goes up, its response goes down. It's just normal. But the film cap outputs were had a higher DB response, it was still the same flatness. So the so all the all the graphs were parallel, except that if you the film cap outputs in the

Host 3     20:01

The their audio tude was just slightly

Host 2     20:03

High. Yeah, yeah, absolutely empty was higher, but like the audio grade resistors and stuff, it was a slightly higher by like five dB over the entire range. Sure. And they both started to get distorted at like, you know, 12 kilohertz or whatever slightly and out to is actually interesting. Yeah, once that once that DAC hits like 17 or 18 kilohertz, it just like, plummets like rock.

Host 3     20:29

Oh, yeah, there's some kind of internal pole that's bringing it down. Yeah. Now, the one thing to be to consider in this case is you have one of each of those devices. That's true, the tolerance on those things. So it might account for the game shift. Yeah.

Host 2     20:45

And so going back while I was talking about earlier, so this morning, we did a blind test. And so I, I built these two boxes. And so

Host 3     20:55

We used we use priority mailboxes for a lot of experiments here on the map.

Host 1     21:01

So, because this is what we just had, and so

Host 3     21:03

Can I do a test on this? I haven't seen this. Yeah.

Host 2     21:07

Um, if you plug it in and plug your headphones in, it will play a dire straits song. Oh, nice. Isn't playing right now? Well, you got plugged it in.

Host 3     21:15

Oh, okay. I will let me let me fire up my laptop. I want to wait which dive straight song? I don't remember. Cuz that would be

Host 2     21:25

Amazing. Yeah. So yeah, you can I hook on hooked your headphones, you can bring it over. So I set up the boxes where they look mostly identical on the outside and taped them shut. And then I put I had the 3.5 millimeter audio jack on the outside so you can swap your headphones back and forth. And so is as close to a be blind test as I can possibly do. And so I had 15 People listen to it. This is first time, Steven, it takes a while for it to boot up. We got an operating system for the boot up and then it should start playing.

Host 3     22:03

Okay, well, I'm, I am now fully plugged in and just waiting for some dire straits to start playing yet to serenade me. Oh, there we go.

Host 3     22:19

Cheers, Parker. Hmm. And so actually, that's a that's a decent sounding deck. Yeah. Wait, so is the deck driving these headphones to correct that's, that's a fairly decent volume for that little chip to dump

Host 2     22:33

In. And that's actually only at so because of that offset. And the response, I actually dialed the good debt down the five DB by just empirically led just like moving it down and then re measuring it until the graphs lined up. Yeah. And so when you swap the headphone jack, one doesn't sound louder than the other. Because then you know, people the first test I did, I didn't do that. And people just said the louder DAC sounds better.

Host 3     23:03

God Yeah, of course. Yeah. And so

Host 2     23:05

Then I Stevens posing for a picture. So I dialed down the volume down so they both match. And we ran an A B test. So I basically I set him up at my desk, and just had people come over and they put the headphones on and whatever. And they told me if the left one or the right one was correct. But

Host 3     23:33

Is there a third rule?

Host 2     23:33

No. I only only had SD cards to run to pi. That's okay. So I picked the most expensive deck and the cheapest deck.

Host 3     23:42

Oh sweet. I can blind test to see if I can find which one is which I'm gonna get it wrong. Do you know which one is which

Host 2     23:48

A not anymore? I don't remember what that means.

Host 3     23:54

I'm listening to one right now. And it sounds great. I mean, zero complaints whatsoever.

Host 2     24:00

We can open up the box and see which ones which

Host 3     24:03

Wait wait wait this is this is this makes for excellent podcast material where I'm trying to be quiet here listening I'm sure our our listeners

Host 2     24:15

And we can't even record this because this is copyright. It's no music.

Host 3     24:18

Well, as long as we did less than three seconds and did not say which band it is which it is not dire straits. Alright, give me a second. Parker and entertain people.

Host 2     24:30

Okay, so the results of this test is people liked the cheaper deck. Oh, we didn't know why ours is. So I some of them told me like their notes. And the cheaper deck they liked because it sounded cleaner. And the other the people who picked the other deck liked it because it sounded more bacey

Host 3     24:59

I had have a preference. Okay, I this this one right here has a slightly better bass response. I'm not trying to get wonky on you. The kick drum is a little bit punchier and a little bit deeper on this. So open that one up. Yeah, let's open it up. And let's see. Yeah, cuz like it could be the shitty one.

Host 2     25:18

I think that one is actually, which actually has a, you're the only person that would have said that has a better bass response. Everyone else said, Yes. For bass bass. This one has a cleaner tone. And when you look at the frequency analysis for the one kilohertz signal, just pull the tape off the red board, and then read what the name says. It shouldn't say like rev one, A or C. C. So that's the cheap one.

Host 3     25:48

Hey, the cheap one. Sounds great. Yeah.

Host 2     25:51

And so what when you look in the thing is what people are saying about it. You're the only one that said that it was bass here. That one is it's it's punchier. Okay, so people were saying the, the most more expensive one with the film caps. had more bass interest. Okay. But and when you look at the frequency analysis of like a one kilohertz signal, it has more in the bass region filled. Hmm. Which, maybe that's what people are hearing who who knows, but the thing is out of 15 people 11 picked the cheaper deck for picked the more expensive one. And you're number 16. And so you pick

Host 3     26:29

The Okay, when you say the cheaper deck, it's the exact same deck, it's just cheaper components, the components

Host 2     26:33

On for the passives. Right. So the cheaper deck has a, you know, house parts for filt for bypass caps. And it has like, I think we talked about this podcast go it has a it has an n zero G caps in the audio stream NPO cog. Yes. Yeah. Yeah. And it has just run in the mill thick film caps. resistors?

Host 3     27:03

Well, they were both incredibly similar. And like, if you put either one of those in front of some person, at one at a time, they would probably just be like, Well, this sounds fine. Yes. But on a to be comparison. Like I said, the kick drum sounded like it, it hit a little bit like, I hate audio terms, but Fuller, it seemed like it just like punched faster, and was less subdued.

Host 2     27:29

So and the only other thing is that might be the harmonics that those ceramic caps let through that the film caps are suppressing,

Host 3     27:38

You know, you'd be surprised, like, a lot of times, like a laser perfect frequency response sounds like but, you know, a lot of times you want something that has a little bit of, you know, garbage in it, uh, to make it have some character to make it have something that makes it unique. Yeah. So that

Host 2     27:54

That was interesting. That's cool. I really like that. It I like it turned out that way, cuz that means I that this thing's cheaper.

Host 3     28:06

Probably actually significantly cheaper.

Host 2     28:08

In low volumes, it's a lot cheaper, like 20 bucks cheaper and high volume. It's about five bucks cheaper. Gotcha. But you know, five bucks over 500 units.

Host 3     28:17

You know, and the thing is, like, of course, this was not like an ultimately scientific No, no, no, yes. Like, they were inside us priority mailboxes. Right. And, and on top of this, like, we're playing it through headphones, normally, this these are going to be played through speakers, which those will have huge impedance shifts and all kinds of like, well, usually this press until the dough usually this

Host 2     28:40

Goes into a amplifier, and then to headphones. Right? And so that those headphones are probably you know, eight Ohm and so it's gonna be a higher impedance then the microphone input, but it's not as high as in line in. And so I bet you those headphones

Host 3     29:01

Are easily going to be the lowest low I mean, that the heaviest load on that for

Host 2     29:06

More than a microphone input. Oh, yeah. Yeah, yeah.

Host 3     29:09

I mean, yeah, so domes versus so actually, it's gonna be

Host 2     29:12

I wonder if I put the if I capture if I put those headphones in and then captured that on the the scope. So the scope doesn't have any influences as such high impedance. I wonder what that looks like. I wonder it cuts all the

Host 3     29:28

The low end crap, well, not all

Host 2     29:30

The loan crap, I wonder if it cuts all those harmonics out?

Host 3     29:35

Well, I mean, the thing is, if this if this deck is designed to be pumped into an amplifier, which is what it sounds like, Yep, I mean, its output is probably going to be seeing an input impedance of like a mega ohm, if not more, I mean some some preamps have like or amplifiers have like a 10k input but even a 10k input is orders of magnitude upon what a headphone is going to present. Yeah, so headphones are probably worst case scenario for this kind of thing. Yeah. In terms of it sounding bad, which it doesn't. So anything higher than this, it'll probably sound just great. Yeah.

Host 2     30:11

So, yeah, the ceramic caps. They're not the cheapest ceramic caps out there. But they're a couple orders of magnitude cheaper than film caps. Yeah. So,

Host 3     30:25

But But you did kind of, in a way, you know, if you neglect the idea that there's tolerance shifting, you did kind of show that film caps, do something. Yes, they do be good, but they use.

Host 2     30:36

And that's the thing is, I'm going to write up an article because I did an article on like, using the Mac platform to, like, price out your production crap with this. That's why I have three different versions is, you know, go read that articles, explains it way better than I can in like 30 seconds on the podcast. But I'm going to do a follow up article with all this test data. And then I'm going to do another article on like, a pseudo production run, like how would we test these in at in production and stuff? Ah, so

Host 3     31:09

We're gonna start adding film cap to the house parts?

Host 2     31:11

No, you probably not. Oh, yeah. So let's go on to the RF. Oh, yeah. Yeah, yeah. This is a couple weeks old, but we haven't been able to get to it. It's the discrete Pong project goes big adds a player.

Host 3     31:29

This is absolutely incredible. So this is

Host 2     31:31

A a, I don't know the person's name. So Person A.

Host 3     31:40

This was a Hackaday article that came out Yeah. So Person

Host 2     31:42

A is got a thread on the V blog where he's been. Or the guy's been like, building these crazy transistor contraptions? Yeah, that uh, that'd be the title this podcast,

Host 3     31:55

Trading crazy transistor contraptions. Yeah, that one's gonna say that real really

Host 1     31:59

Fast. So guy or player A or

Host 3     32:06

Computer fires up. I'll try to find his name just so we can give him the honor.

Host 1     32:12

If it's a he,

Host 3     32:13

Oh, that you write your I apologize. Yeah, yeah, it may not be a heat. Yeah. So although I read the article, I'm pretty sure it's that he.

Host 2     32:22

So the Hackaday article. Last year, this person had a nother project. It was a Pong like transistor project, but it was only like one player and you played against I think a wall.

Host 3     32:36

Yeah. And I think there was also some like, housekeeping stuff that wasn't in in his that that project. Like, I think the scorekeeping was a little funky way the way they did it. So he's kind of updated it now.

Host 2     32:49

Yeah. So the so yeah, the games he updated it, but it's how he built it. Because I've seen transistor Pong machines before. I've been to a couple of like, video game conferences. And usually someone's got one of the really old machines that's been based off the brown box by Ralph pants, his name,

Host 3     33:12

Growl, his name is just Ralph.

Host 2     33:16

Give me misremembering that though, but it's called the brown box. And it was a transistor based video game console. Built in the late 60s, if I recall, I actually had this schematic laying around somewhere. I've always wanted to build it. And that looked at the transistors and I'm like, Hmm, that is way out of my wheelhouse. Because those transistors don't exist anymore. And so you'd have to convert it to modern equivalents, right? And then change all the biasing and all that stuff to make it work. Right. But this guy's already done that stuff for us. Yeah. But again, that's how he built it. And it's a like, two foot by one foot solid clad of Fr for that IE Manhattan style built everything in it's all like, perfectly wired and who

Host 3     34:04

So if, if you haven't seen Manhattan style before, go to Google and type in Manhattan style circuit or Manhattan style transistor or what a Manhattan

Host 1     34:13

Manhattan style drink.

Host 3     34:16

What is a Manhattan drink? It is. It's whiskey. Okay. Yeah. Well, you go look up that and get one of those and then go look at this Manhattan style, because you need a drink to like understand? No, so in Manhattan style, it's actually like the majority of the players that use that are kind of like the RF and the high frequency dudes. Because what it is you have a copper clad base, that's all ground, that's all ground and what you do is you cut up basically little squares of copper and glue them wherever you need to on top of this base, such that if you ever need to attach to ground, the whole board is ground you just take a leg of what whatever component you have, and just solder it directly to the base. But all the interconnections between components that aren't ground are just floating in space. They're floating above this, like, ground level thing. So I guess it's Manhattan because it's like skyscraper. What are you building vertically? No vertical. Yeah. And a lot of times, you'll see, like, you know, the guys who do like crazy RF frequency, whatever stuff, you know, when they need to make like a 50 pole, you know, brick wall filter for whatever radio station they're dealing with. They'll just do it Manhattan style and shove it in a box. And they're not the most exciting circuits ever. Because it's just like, Okay, great. You made a filter. And that's, that's fantastic for what you do. But it's just like, not cool to look at. But this guy's is immaculate. It's art. Oh, absolutely. Not. You can hang this. Oh, for sure. And the fact that like, Okay, so what's what's so cool about it is like, when you look at a circuit, if you don't know much about electronics, if you look at a circuit, you can kind of get the idea that these little black bug looking things. There's something going on inside you don't get to see what that is. But like, Okay, those are chips who they do something, you know, something like that. His doesn't have chips. So it's even more like ooh, like, it's super

Host 1     36:16

Cool. I love it. Use a bunch of flip flops. And

Host 3     36:18

Yeah, yeah, but uses a bunch of flip flops. But it shows on a screen like intelligible data, you know,

Host 2     36:24

Oh, yeah. And that's just, there's a really cool video, a YouTube video that shows how the original brown box worked. And that's pretty similar how this works. So I did find that I had to find that I need to go watch that. Yeah, I wish I knew what

Host 3     36:39

I need to go get a six pack of beer and a pizza on the way home and go watch. That works.

Host 2     36:44

It's very interesting, because the first couple Pong news and like, the arcades. That's how they worked.

Host 3     36:52

Really weren't the first couple. I could be way wrong here. But weren't the first couple Pong units like you build it yourself? Did you just buy

Host 2     37:00

In? Or no? No, because Atari had the patents? Okay. Okay. I thought they were copyright, I would say. And

Host 3     37:07

I mean, when was the first like, build it yourself Pong, because those have been around for a long time, I don't know.

Host 2     37:12

But the interesting thing is on the tarean was Pong was big deal. And the first it was all discrete stuff. And then the big deal was when they actually came out with a Pong on chip, or puck. I don't know if anyone actually in the industry called it that by like a duck. I like that. And so it's like when you go from the the schematics between the old transistor based ones to the POC, it's like the POC has like a couple of resistances big rectangle. That's whatever the part number was written on it. And like a couple resistors and like the inputs and like, and it's like, it drives the display.

Host 3     37:51

Wow. And then like, so that one game was popular enough to drive that deep of integration into a single chip.

Host 2     37:59

Oh, yeah. That's, that's crazy. Yeah. What's even crazier is the some of the old video games, like the Odyssey, video game consoles, like the Aussie where your cartridges didn't have software on it, it changed the jumpers. And it was a jumper card. And so when you plugged it in, it just changed how the circuitry was wired up like an FPGA. And so it would play a different game.

Host 3     38:24

So if you knew what you were doing, you could just jump at yourself. Yeah, I mean, that was

Host 2     38:29

Pirating back then.

Host 3     38:32

That sort of feels like hacking in a way that that feels like a real is too high. Yeah. So like, the what would be awesome is like, what is the most integrated? You could possibly get Pong? Like could you get Pong to just be like a chip with just power into it? No passive? Like, how far can the rabbit hole go? You know,

Host 2     38:55

I mean, you can probably do that on like an 80. Tiny now, which needs which needs a bypass cap and power. Both things problem is getting the output out. And so how are you going to output you can probably there's probably a project altre to get at Tiny the output composite over one pin.

Host 3     39:13

Yeah, probably consumes a lot of its processing power.

Host 2     39:17

Yeah, but I mean, it's Pong. Yeah. I mean, you could just do it all on a look it up at Tiny composite output. Or

Host 3     39:25

At Tiny Pong. I'll look that up to scanning.

Host 2     39:30

That's got to exist. Pong. It's like, it's like that Doom website.

Host 3     39:36

Oh, yeah. Everything runs doom. Yeah. Or whatever. Doom on everything. Yeah. Yeah, I would not be surprised if someone has an Arduino that runs Doom would not even supply

Host 2     39:45

Maybe one of the higher end ones out of 32 bit. actually want to know us, man. This is very interesting for our listeners. So I'm going to go on to the stealth winners of the new iPhone x. Because you know, they had to skip the night iPhone nine right Yeah, because reasons Windows did same thing. I mean, it's even like, was it the x box? 1x? It's like, what the hell? Are they thinking naming these things?

Host 3     40:10

Well, okay, so Windows went to Windows eight. And then and they did, like, they did such a disservice to their people. They're like, we're gonna like upgrade twice to say I'm sorry for eight. Okay.

Host 2     40:26

So the stealth winners are these are kind of like, people have done tear downs with iPhone X already. But this is from electronic design comm where they kind of they when I don't know if they did a tear down, but they were going into more like the unsung heroes of like, what makes the iPhone X possible. And the craziest thing is, I think the craziest thing is how they are doing the stackup inside this PCB now inside the phone now, and so I printed off a picture. And I'll have a link to the picture in our description.

Host 3     41:03

I'm going to pause you for just one quick sec, kind of just take one step back and say Hackaday already has an article about a guy with an 80 Tiny 85, the small eight pin guy that spits out composite video and can display text on a screen so he can do Pong. Yeah, absolutely. Okay. Just maybe not at full speed. Okay, so forward onto apple.

Host 2     41:24

Yeah, so the PCB sandwich is so they have two main boards, and they're like most cell phones, okay. most cell phones that I've taken apart, they have a, like, a really high density connector on both ends, and then they click together. A board a board, where they call it mezzanine. Yep, mezzanine connector connector. Well, the problem with those is they still need thickness to work. Okay, cuz it's a connector and you have stuff there. Well, Apple decided to basically make a via pass through PCB. And so if you look at this slice through, he had this top board and this bottom board, and then on the edges, you see these vias that go all the way through?

Host 3     42:09

So this these are vias that extend from board to board?

Host 2     42:14

Well, no, because it goes from one board into another board into the other board. So there's three boards. Yeah, yeah. And so it's made out of like an a, like, thicker fr for, that's just vias all around.

Host 3     42:28

So they're making a PCB into a mezzanine connector. Yes, but it's thinner than a normal mezzanine connector. So how was it manufactured? Is

Host 2     42:35

It soldered together? Yeah, whole thing. Both two boards are soldered together.

Host 3     42:39

So okay, so they put the bottom board that already has components on both sides, like the 12 layer board, it's already got everything on it, right. And then they on top of that they put a small fr for connector boards with just vias. Yes. Then they stack another board on top of that that's

Host 2     42:54

A traveller board with a bunch of parts on it. Yeah. And then run it

Host 3     42:57

Through an oven. Yes. I guess they're filling the vias with with paste. Yeah. And then oh my

Host 2     43:03

God, they're probably in there probably vapor reflowing it is I can write. I'm saying Yeah.

Host 3     43:10

Because like, okay, so this this cross section image that that you have here, first of all, you can see the boards and see that there's like a bazillion and then these like big ol wads on the side, which I guess it's called via frames. Yep. Via frame via frame. Okay, so that's the that's the connector that so

Host 2     43:25

I didn't print off the top view. But they there's some guys that D soldered the top board. Yeah. And yeah, it's a frame that goes all the way around the perimeter about three vias wide, that is just filled with vias that does the pastor.

Host 3     43:41

You know, it's okay. So here's the thing that about Apple that's kind of interesting. They're big enough as a company and that one product, the iPhone n is big enough by itself to invent an entire new technology that other people may never use. So they're also pushing nology only for them. So

Host 2     43:59

They're pushing another technology in how the PCBs are manufactured now called M SAP or micro SAP, what's that? Well, it's not micro, I guess it's modified semi additive process for PCB assembly.

Host 3     44:12

Okay, Microsoft, so you can like change parts of a PCB.

Host 2     44:17

Well, it's semi additive, but a modified whatever that means, anyways, what M SAP is, so for normal PCB manufacturing, you have thick copper, you drill your holes, and then you played it. So you played the, the, the vias, and then you put a mask over it, and then etch the copper away. And the problem with doing that is you get your traces look like trapezoids where the top of the trace is thinner than the bottom of the trace the the

Host 3     44:47

Technical term for that is an isotropic. Sure, yeah.

Host 2     44:51

I don't think I can bounce That. That.

Host 3     44:53

That's that's the that's the word that they use for when you're when you're etching dyes inside have chips. Like if you want the gate of a transistor to be really straight, that's ISO tropic is it's straight up and down. But it's never isotropic it's always an isotropic it always has some kind of angle to it. Well, and it's the same thing with copper etching. Yeah. So

Host 2     45:15

What they're doing now is doing this semi additive process, which is a lot like building semiconductors, which is an additive process to you building on top of this growing crystals where you grow the crystal, but then you you build crystals and, and embed metal and stuff like that on top. So what they're doing now is you start out with a really thin copper, do your drills played it, whatever. And then you put a mat a mask down, and then you play that, and that builds the thickness of your traces up where there's no mask, and then you take the mask off. And so now your traces are more more parallel up and down.

Host 3     46:01

So okay, so instead of chemically etching away copper, you're chemically adding copper, to the, the OPPO electrical, electro lytic places. Yeah, plating, which is kind of, okay, I didn't know that. Yeah. Yeah. So so so they're just doing it in reverse, which works fine.

Host 2     46:20

Yeah, but it lets you get away with thinner traces. And sticking the traces closer because now the base is the same width as your top or really close. And so you can move them closer. And you can get down to point four mills. Trace with trace width and trace like separation.

Host 3     46:39

Oh my god. Yeah.

Host 2     46:40

Yeah. That's crazy, isn't it? Because you we play in like five mil five mil world?

Host 3     46:46

Yeah, no, the the the standard at microfiber is six, six. Isn't it? 5555. Should be Yeah, it probably Yeah. Yeah, but But I mean, in general, the five six world is like, normal. That's normal. Yeah. Anything less than that. You start to pay extra price. Yeah. But, and and to be honest, let's let's, let's lay it all out right here. That's still pretty damn small. Oh, yeah. That's all Yeah, that's small. But point four mil is like, you know, many 1000 pin BGA kind of Yep. Escaping.

Host 2     47:18

So that's the other unsung hero of the of the iPhone internals.

Host 3     47:23

Wow. Yeah. You know, you have to you have to assume that there's like some crazy stuff going on in under the hood in that kind of stuff. Oh, yeah.

Host 2     47:31

Like the the article talks about like why they switched the their men's manufacturer for the accelerometer and basically the IMU. It's because like, I think it was Bosch is the the new manufacturer forum. And they were able to like make theirs thinner by like, point three millimeters. I imagine that's enough. Yes. Not the portal slightly bigger battery in

Host 3     47:57

Oh, in in those phones, like every inch of space is consumed?

Host 1     48:02

Everything's customed to Oh, yeah. Oh, yeah.

Host 3     48:06

It actually kind of makes sense that PCBs eventually make their way to the same sort of technology that I sees do, yeah, where instead of like, effectively, the way you do ICS is sort of the way you do see and seeing you start with something bigger, and you chop away everything you don't want. But with ICS, you start with like, almost nothing and then make what you want from that. And that makes sense. But wow, cool. They have enough money to drive it. Yeah, that's for sure.

Host 2     48:36

Okay, so next topic. Last topic for today is this was on I think I'm going to start doing this more as pulling more topics, I find that questions on like, the eceee Reddit subreddit, and what we get in Slack channel and stuff, because this one's really good. Is someone asked is coding required for electrical engineering. And he asked, cuz he's a student, probably in high school, going to college. And he's not very good at Java, because that's what for some reason, they want you to learn in college, don't learn Java. That's what like when you learn, at least at UT, you learned like assembly and C, and the, quote, high level language you learned was Java, because for like, data systems or something,

Host 3     49:20

Shoot the highest level you should learn as Python. Like, that's super useful. That's very useful in industry.

Host 2     49:26

So I was thinking about making this as a general question is, is coding quired for engineering for electrical engineering, just like every electrical engineer should know. And like, I guess, like go into like, also, like how useful it is, even if like, you only do hardware, like, how useful is knowing how to code?

Host 3     49:49

You know, a great so before we go way into that, I think we should invite as many people to get onto our Slack channel, because I would love to see this conversation. Yeah, go on to the Slack channel. So for those of you who don't know, we have a Slack channel here at macro fab, where you can go and Parker and I are, are on it and other people here at Mac fab. And if you just want to come and shoot the shit and just talk about whatever, it's there.

Host 2     50:15

I'm usually on until I go to bed base. Oh,

Host 3     50:18

Yeah, yeah. Parker's on all the damn time. So yeah, come on there. And this is this is not a self promotion but because personally I think this this topic is actually really cool. And I think slack would be a great place to go to that. So what's, what's the method to get to this? Like,

Host 2     50:34

I don't know what else there's a link in the podcast, there's okay, there's a link, it's usually when we tweet about it, it's there too.

Host 3     50:40

So if you if you troll the website, then you'll find you'll find this. Okay, regardless. So here's my quick thoughts on this. And when I have done job searches in the past, one thing I have noticed is that things have changed significantly since I even since I left college, which was less than a decade ago, but almost a decade ago. When I left college, they didn't seem to require it as much. But now it's everywhere. Like, yes, it will be like, oh, you know, when you go see like a job posting it will say, like, you know, must have fundamentals of in this engineering must have analog circuit design, bah, bah, bah, but also, like, must have experienced with C slash C++ or Python or whatever. And that's just so much more relevant to today. So if you're just like an analog circuit guy, there's people who will want you, but they're really specify specific now and special.

Host 2     51:42

Well, even I was even thinking about like, what would be an electrical engineer that would never have to do programming? Like, oh, yeah, analog guy, blah, blah, like, no. What about simulation?

Host 3     51:53

I guess that it's really high level of programming?

Host 2     51:56

Oh, yeah. Well, when you have to build your own, like models for a device that doesn't have a model, you'd have to use models. Yeah, to build your own spice model, which is it's scripting, but it's still a that's still programming, right? And then when you go a little farther, it's like, okay, Python is kind of like a scripting language. It's a little it's people considered a proper programming language. But it's, you can view it as a scripting language, because it can be use it. And yeah, I would say python is like, like, the most useful thing, just from like, data management, because that's what an engineer is, is knowing how to look at data. Yeah, no matter what it is, is how to look at data and how to implement what you need to implement based on that data. And knowing a scripting language or something that can like MATLAB, MATLAB is amazing if you can afford license.

Host 3     52:51

And if you're in college, even maple, if Yeah, Maple Mathematica, or whatever they call it.

Host 2     52:55

Yeah. Because when you're in college, you usually get MATLAB for free. Right. And that is such a powerful tool to basically help you break down data and look for trends and in MATLAB

Host 3     53:07

Is totally see for dummies. Oh, yeah, that's, that's exactly what it is. In fact, MATLAB is like Arduino for, like, you know, that world, that math world because it just it already has, like, in fact, one of the functions I use all the time was linear space. It's li N, SP, whatever, I can't remember what it is. But like that one function, you just plug it in, and it would just create automatically an array of whatever you you, whatever, you know, parameters you gave us you gave it. And that one was like etched into my memory because like college was all about doing MATLAB and that

Host 2     53:45

Array mean, manipulation? Not array, though matrix. Yeah. Yeah, linear math and all that stuff. Do not miss that class.

Host 3     53:56

Actually, here's what I would say is maybe as a sort of like a minimum, maybe it's not the best, or I shouldn't say the best. I guess what I'm what I'm getting at here is you should be able to at least look at code and understand what it's doing. Maybe you are not capable of writing that code without like, extra thought. But like if somebody has code up on their page, and is like, Hey, can you help me like, look at this, you could at least follow it. And we're sort of like speaking language, like you can read it, but you can't speak it. It's the sort of the same thing. And

Host 2     54:33

I would go it looks like if you're a layout engineer, see, all you do is lay boards out, hey, you know, like, oh, I never had to, you know, no, no code. Well, most EDA tools support a scripting language, right? And if you have a function that you want to write like a something that you do all the time. Well, instead of spending eight hours to do it, spend two hours to write a script and then hit enter and Have it just chuck it out, I write a lot of Eagle scripts, right? That just like make my life easier like when I have is changing the how a schematic looks to how I want it to look. It's it's very similar on the development side they have like they have their own tab verse double space. Like that's gonna make people go crazy and slack. I'm a tad guy, by the way. So seems like I have no idea what you're talking about

Host 3     55:26

I get it, it just doesn't affect me.

Host 2     55:30

So I have a script, a couple scripts in Eagle that I run when I load up someone's board that changes the colors how I want to look and changes like all the fonts, so it looks exactly how I want to look, I can easily read it. And there's a bunch of scripts. Let's see what else is there?

Host 3     55:50

Oh, you wrote a script for the Mega Mondo ridges? Oh, yeah, I have like all 40,000 Yeah, I

Host 2     55:56

Had a script that wrote that just just procedurally went out and did 40,000 resistors. And it's connected all the nets and actually automatically routed the board.

Host 3     56:06

Right, right, right. Cuz that one was like a good candidate for that. Yes.

Host 2     56:09

Well, I only had to do the same thing. 40,000 times in a row.

Host 3     56:13

Right. So is coding required for electrical engineering? I think boil it down. Both of us are saying yes, yeah.

Host 2     56:20

And I can't. And I tried today like thinking of like, is there something that you could do that wouldn't require you to code? And sure, it's gonna your life will be easier if you do, though. Well, not to be a thing is there's difference between knowing how to code and being a developer. You don't have to be able to build apps or whatever. But it's actually going back to being able to read code. Because what if you get if you're a board layout guy, or draw out schematics, and someone hands you firmware, and says, We need this firmware at work? Well, if you don't know how to read that, how are you going to wire that microcontroller up to know what pins go to what pieces of hardware? Right? So like, oh, this port goes to our motor controller and this one's direction. This one's polls, and this one is fault.

Host 3     57:09

Yeah, right. Right. Okay. I'll give you I'll give you rock solid industry experience here. This actually happened to me at my very first job. I was a, basically a manufacturing engineer. In effect, it was like, if, if a product went down on the manufacturing floor, my phone would ring and they'd be like, come help us. And until like, one of the big products went down on the floor, and there was a bunch of shit going down. And they were like, hey, our testing software doesn't work anymore. It was written like, a decade ago in Visual Basic five. And I have no clue how to read Visual Basic, but I've generally been able to read see, and a handful of other things. And so like, sure, I maybe I wasn't jumping on it as quickly as possible. But I was able to kind of sift through the code and see that a there was a function that was kind of crapping its pants, when they were asking it to do a certain thing, even though I've never even seen Visual Basic before I got the gist. And I saw what it was doing. And I was able to fix it. And you know, it's nice to be the hero in that kind of situation when the when the big product goes down and you fix it, but like, I don't like programming, but I can generally read it. And that helps you know that like I didn't throw my hands up in the air. So like, that's why I'm saying like, just at a bare minimum read.

Host 2     58:28

Yep. Cool. Yeah. So that will be the back of engineering podcasts. We are your host Sparky going and Steven I'm so good at that. So yeah,

Host 3     58:38

By the way, we're done with this episode Parker. Decided that thank you for listening.

Host 2     58:46

Yes, see you next time guys take it easy thank you yes, you our listener for downloading our show, even though I might have cut it a little soon. Depends on what SEMA thinks. If you have a cool idea, project or topic that you want Steven and I discussed Tweet us at Mac fab or email us at podcast at Mac fab.com. Also check out our Slack channel. I hopefully we spurred or rustle some jimmies and we get some people talking about our two hardware people talking about software. So yeah, later

Transcribed by https://otter.ai