Chonky Motherboards

Welcome to the macro fab engineering podcast. We're your host, Parker Dolan and Steven Gregg. This is episode 232. So, a couple of weeks ago, Steven and I did a brew day where it was on a Saturday and we basically go brewed over slack. social distancing brewing. Yes.

2000 some odd miles apart. The best thing was like you were almost done growing by time I got started. But semantics.

We wake up at different times on Saturdays. Well, it wasn't just that I had to go to the local beer store or brew store. Oh, yeah, you had to go get all your stuff.

Yeah, because the one that's actually close to my house shut down due to COVID. So I had to go halfway across town to to the other one. And then so I got there. And they only did on, they only did pickup orders. And so I had to go back into my car, and order everything through my phone. And then wait 30 minutes for them to fill it and then call me saying it was ready. And then I went in and got it. That's that sounds like a that's indicative of modern technology. Right? Yeah, we're like you're 10 feet away from all this stuff you need to buy but you have to sit out in your car and run your car run burn gas, and like, order everything online and beam it up to space and back. So you can get a 10 feet away.

Yes. It wasn't it wasn't that bad. It would have been faster or better if I just had a laptop because brought trying to browse through like I need to order point eight pounds of like flight barley. It's kind of annoying to have to input that into a phone. Laptop. And I actually went to the website when I got back home. And it's completely fine. Navigating on on browser.

It's just not made for web mobile. It's okay on mobile. Yeah. But you know, it's local Bruce through store so I'm like, I'm surprised that it even worked in the beginning with you know, and that's because that's because basically you were boiling I think your second batch when I finally was like, okay, like I was like milling migraine.

And like getting that my hot liquor tank up the temp. But yeah, my brew day went great.

hit all the numbers correctly. I was surprising. Like, just like,

hey, for a first run on a brand new system. Sounds like it went really well.

Yeah, I was. I was point one five gallons off my my Deadspace calculation. Point.

So off. Do you mean like you had less volume than you thought you had more volume? Oh, okay. Got it. Yeah, so I was point 5.15 gallons too much. Oh, so yeah, you just put that into your calculations and offset? Yep. I already did that. Actually. I updated beer Smith, which is always a Smith. Beer Smith. Beer Smith. Yeah. Which I don't like that program at all, but I didn't think you would. But it's what you and Roz use. So I'm going to use it. It's

I you know, I've used it for so long. It's basically a program that you can plug in all of the information of all of your equipment, and the recipe and everything. And it will tell you all your temperatures and your volumes in blah, blah, blah, and spit out like when and how you need to do things. And the user interface. I don't think it ever started out well, and they haven't fixed it. They've just adjusted it, you know, like things just moved, but it didn't get better.

Yeah, it's just like how things are nested is weird. And yeah, it doesn't flow. That's the problem with that program. It doesn't like it doesn't seem like you have to walk forward and then backwards and forward a couple more times to find what you're looking for. But I mean, the thing about it is like, everything is there and it's super powerful. It's just not linear in the way that you think. But I was able to get it

working, you know, a couple hours in the afternoon. Yeah.

Get my recipe all set up. So pretty happy. How did your day go?

You know, mine mine actually went really well. So I had a couple of unknowns also. So when I was back in Houston, I was brewing a handful of times just before I moved up here one thing that was always happening was my my boil off calculations always seem to be a little bit inaccurate. I usually I put in my calculator that I would boil off one full gallon of water in 60 minutes of a rolling boil. And that never seemed to work super well in Houston. It seemed variable. It was odd, but it worked out super well here and I actually Thought I would, because it's so dry, I thought it would boil off a lot more. Not really, it would ended up being perfect. So I've got that set, which is great. It's always nice when you get hit the number because you like, click Save. That's what I will always use. Now from here on out. The other thing is, I usually didn't have my efficiency sent super well back when I was brewing in Houston. And I was still like figuring things out with my rig. And your efficiency is basically how well you are at extracting sugar from the grain, or how good you are at it. And

this time around, I actually ended up under shooting my efficiency, which I've never done before, I've always overshot my vision and got weaker beers. This time I pretty heavily undershot and got much stronger beer this time, which whatever, I don't care, that's fine with me. I did try one thing, though. So this time was the first time I've actually ever milled my own grains. And I tried a new technique, which I'm going to do from here on out, which is awesome. It's called conditioning your grains I've ever heard of this burger? No. Okay, so check this out, what you do is before you mill your grains, you weighed all the grains out. So you know 1512, whatever how many pounds of grains you have, and then take 2% of that and get that much water in a spray bottle. So like 15 pounds of grains ends up being something like four ounces of water, it's not much water. And then you just mix the grains up and you spritz them down with the water. And what it does is it just hydrates the whole of the grain and it like so it doesn't get anything wet on the inside of the grain, just the whole. And so you can actually pulverize the living snot out of the grain and it doesn't turn into flour, it doesn't become dust, because it's a little bit more leathery as opposed as opposed to like brie. And I did that. And I think my efficiency just shot through the roof. Because I ended up grinding my grain on a 20 5000s feeler gauge in my grinder, which they recommend like 39,000. So I went really, really tight and I didn't have any dust. So I the thing. Also, with this brew day, I've never once been able to run my pump wide open in my mash done because it always it always gets clumpy and it gets it gets stuck. It's not a stuck sparge it's just, it's I can't run the liquid through fast enough. This time conditioning my grain grinding super fine. I ran my pump wide open, so I could go way faster with it, which means I got better temperature control through my arms coil. So all of it was better. And all I had to do was just spray down What 2% Water. So it's like, it's a no brainer for me. I'm gonna have to try that. Yeah, give it a shot. Yeah, your efficiency will go through the roof, which hasn't really ever been a problem for you. You've said but like, it'll get even better than what you already have.

Yeah, this was actually the lowest efficiency ever got I got 70% efficiency, which a lot of people online say that's normal. Yeah, I'm used to getting like 80% which

defaults to 70. Yeah. So I hit 70. And I was like, man

I think I think I know why I need to. I think it's my research. My research is I have like a fan jet that comes out. I really need to go I need to have like a, a. a, I guess a rod that's got holes drilled in it a manifold. Yeah, that just sticks in the middle of the mash. For the research. What I've heard is like if you can distribute the fresh hot liquid equally across the top of the grain bed such that it has to flow through the grain bed. That's the best situation. It's hard in practice, you know?

Yeah. Well, the problem is with me what you can do that because your entire mash tun is what the grain fill. Oh, right, because you have a steel. Then I have a steel mesh, a jacket mesh. Yeah, mesh basket. And so it's kind of hard to do that. Because if you if you hit the top, it just want to run off the sides. Yeah. And that'll give you garbage efficiency. Yeah. And so I'm thinking about trying because the ones I've seen that have this kind of setup, they have basically a a tube that's got a bunch of holes in it, and you put that just right in the middle and it's that way it spreads outward. It's

radially outward through the grade and that's really interesting, huh? I'm gonna try that. That's kind of cool. Actually, what I like about that is you get like kind of this hot ring that spreads outward. Because with mine I've I've certainly measured a gradient where it's hot on top and it's cooler on the bottom. So I have like a Even though I'm recirculating, you would think the whole mash tun would be at the same temperature. I know it's not. Yeah, in fact, I stir my mash pretty regularly once every 10 minutes or so, just to kind of like, equalize the temperature throughout. And that's the one thing I'm gonna actually I was thinking about trying to shut the mash. And actually those chugger pumps, you just don't have the the oomph to do that. Yeah.

Good. I was hoping like the, because I have a fan outlet off my pump for it. And it just it I was like, like, my random, I'm like, oh, yeah, we're like agitated now. it agitates the top half inch. Yeah, about that. I'm like, it needs like agitate the whole basket. Oh my god. So you need a jet. So that's actually what I was thinking is like, maybe for season two, is like figure out a way to actually have like a movie jet that's in there, that just stirs it up constantly. But I'm gonna try that next time. So I'm gonna start I'm gonna start more. And because I didn't start at all, well, besides the very beginning the mash and I stirred. And then when I did my bat, my spark out, I started as well.

You know what I like about these home systems, the systems that recirculate the wart, you can take the top of the lid off your mash and do something to it, and you're not going to mess things up. You know,

I remember when we would brew in our I know, we've had like, so many brew episodes, and we keep

we had we had we would brew with our measurements would be igloo coolers, igloo coolers or schoolers. And I just remember like, opening the lid off of one and just watching the thermometer just drop, like five degrees. Yeah. Can't lift that lid. No can't. Okay, nothing about burn. Yeah. Hopefully, hopefully people who do not want to listen to us talking about brewing, come back now. Well, I mean, and we give some updates in the Slack channel, come and check us out.

So since the breweries done, I said, two weeks ago that I'm gonna start working on other projects, faster projects, that time just like I like I don't want a long term project anymore. For the podcast for a while. The brewery was like two and a half years, it was a long one. So that walk the brewery will come up again next year for season two of the brewery. But till then I'm going to work on some smaller electronic projects. And so the one I talked about was the cat.

Like, monitor feeder thing. So so give a description for those who don't know what it is.

And so the the problem with owning a cat, when you have a lot of people in the house is the cat will always meow want to be fed. And this is how cats get fat or become tankers.

Oh, Lord, He's coming. Yeah, as the internet knows fat cats to be. And so in an effort to make sure my cat does not become a chunker.

I want to make a I don't want an automatic feeder, because you can just get those I'm like, Yeah, that sounds like a lot of maintenance. And they always break because I had an automatic fish feeder for a while back when I had fish. And that was gummed up. And it's like, then you forget about it. And you're like, Huh, I wonder if the automatic fish feeder has food in it. And it's always know when you remember. So I'm like, Okay, I just need something that's so that people know, hey, the cat hasn't been fed. So we partially feed the cat. And this way, not multiple people feed the cat. The cat doesn't have any problem with that. Yeah, exactly, exactly. And so I'm like, okay, it just needs to have an LED that's like, if it's green, that means the cat can be fed. And then when you feed the cat, you press the button or or it knows, like by opening the lid on the on the food bin or whatever, that the cat's been fed. And then it has a timeout for like 18 hours. So you can't feed the cat again until another 18 hours comes on or whatever, 24 hours. And so the main thing I want with this project is kind of working with maybe doing a circuit design in a different way I've done before, because an easy way to do this, because oh yeah, I mean backup of it. The main thing I want to do is also low power. Because I want to, I want to basically be able to put a triple A battery or coin cell in this circuit, and it works for like a year don't have to worry about it. And, and that's actually the bad thing to think about is if I have oh man, I just thought of something that were in my idea of because because the idea was to have a red LED In a green LED, red LED is don't feed cat green LED is feed cat. But I was like, Oh, I don't need one to tell me why don't need to feed the cat. Just green means good, right? Well, if the coin cell dies, and the LED will never light up as you're never feed the cat. Well, okay, you could also for lower power, you could also have it where you like, go and check the thing. We have to press a button to like, check. Ah, but I mean, that would require you want to be able to just look at it. No, yeah,

just as you're walking by and the cat's meow and like crazy. Yeah. Feed me feed me, Seymour.

I mean, if you had it, where you would get out of sleep mode, because you pressed a button and then illuminated an LED. You can make it last forever. Yeah. And that's I was the thing I was thinking about was an easy way to do this would be like if a mate sleepy B, it would pull like nano amps. Yeah. And it would probably last until like, my cat passed away. Yeah, do do a CR 2032 coin sale on the back of it. Yeah, you'd last forever. Yeah. And because I've already done like, the macro watch project. And so like, I could actually just like lift half that code, because it has an RTC setup and have a timer. And I'm like, you know, that's not really learning anything new, though. And so I kind of want to do it a different way, if it's possible. And so I started looking into maybe doing like a 555 timer with like a decade counter. And then it would roll up. And then when the ticket counter overflowed, it would light up an LED, basically, the overflow bit would be the LED lighting up, and then you press the button and it would reset it. Only problem is I couldn't find a way to make that super low power, like five for five timers or they're hungry. And yeah, in comparison are hungry, because they have you know, they have the one of the reasons is they had that they had that resistor chain in there. And that just holds power because it's a resistor chain across a resistor ladder across power and ground. Go figure. It's 5k 5k 5k. Yeah. And there are some, quote, low power ones that use different resistors set up. So column five for five timers, but they're different, I guess. Yeah. So the, the, the 7555 is the CMOS version. And you can get those I'm looking at it right now in 60. Micro amp is the power draw for that. Okay, so that's not too bad. So yeah, the traditional five by five will, just to your your power budget. I tend towards the I've used the 555 a lot in in circuits, because they're super useful for the synth stuff we do. But I lean towards the 7555, mainly because they have a way more predictable output in terms of their amplitude. So if you're using them for like purely digital switching, and you have the power budget to get away with a five by five, that's fine. But the 7555 is a lot nicer if you need things to be constrained. Yeah. So we could probably use a certain 5.5 then to do that. And then you would feed that into a counter. And it takes probably a decade counter. So what kind of you because you seem to know more about these kinds of ICS than I do. Because I didn't know if 7515 was a thing. Yeah, the CMOS 555 is they're more expensive, but they they get the job done a lot better. So hey, maybe maybe we breathe life into your project again doing it this Yeah. Well, no, I want to do it a different way besides a microcontroller, right. The 74 HC Gosh, which which is the decade counter 4017 Yeah, yeah. 4017 That's it. What is the what is the like quiescent draw on that guy? Well, you might not go with the HC version for that. No, for the low power stuff. This This one probably won't work because it looks like the PDF has been copied a bazillion times. This the TI version of it is a fresh one you can you can look at that. It's probably all the exact same information. I you know, you gotta love that when when you get those old. Actually, I the ones that are really fun is when you're going to look up a transistor and you don't know what the transistor is ends up being an old transistor. And it just ends up being a photocopied page of a not only like a datasheet but it's just a photocopied page of a list of transistors that were available at one point and It's all in there somewhere. And it just says like the name and the package. It's like, Wow, great. Thanks. Okay, cool. Yeah, actually, the seedless 7484 017 has a quiescent draw of one micro amp. No point one micro amp. Wait, that doesn't make any sense. That's really low. Maybe that's it. I don't know, I gotta have to dig into this a little bit more. But ya know, it's pretty low, you could probably get away it looks around there. Yeah, you can get away with it. Yeah. And then then we will need a, a way to modulate the LED output. Because running at, like one milliamp is probably too much. So we definitely need a pulse with the, the LED. I don't know what the best way to do that would be. Oh, you could use another 77555. And then whenever it's active, it could like trigger a, like a triangle generator and do a pole PWM circuit. If you want to do this, like, if you want to do this zero microcontroller, do it all analog, you can have that that whole circuit be like dead until another 7555 turns on. And it activates it pulls a ton of juice at that time. Yeah, because I think we do need to have a it needs to have two LEDs, it needs to have a red one and a green one. So you know what if when the battery does eventually die, you don't end up not feeding the cat.

So you need to know when not to feed and when to feed?

Right. So here's the thing. Here's a good question. Let's do the exact opposite problem. If you had an LED lit when you're not supposed to feed and then what goes off you're supposed to feed. You're just continually feeding the cat. Right? Well, once again, your cats not going to mind. Well, what mine the other way, not this way. Well, okay, so check this out. Here's here's an interesting problem, I guess. So let's say let's say you have it such that green light turns on at five o'clock PM. And that's cat feed time. And let's say on Monday, you You're perfect. At five o'clock, you see it turned green, and you immediately feed the cat and you press the button. And so it's like, okay, great. It cycles over until five o'clock on Tuesday. But let's say on Tuesday, you don't catch it until, I don't know nine o'clock at night. And you're like, oh shit, I gotta feed kitty. And then you feed cat and you press the button. Well, now it's going to go off at nine the next day. How do you make it such that it goes off at five? The next day guaranteed? Or whatever? If you're the one who's wrong,

given the fact that the cat will not make sure you don't make that mistake?

Yeah, but that's the whole point of your box. It's not the the whole point of your box is to not pay attention to the

cat. No, no, no, no, no. The boxes. There's though that the cat can't request more than one food feeding.

Okay, so the cat is always going to say feed me feed me, feed me feed me. And then you look at this box and say can I feed you?

Right, but what I'm saying is if you're doing this in an analog format, and you and it has to complete 18 hours or however many hours you said in between, then like if you get it off, then it's off. You know, like, Yeah, that's fine. Okay. I don't think that's a big problem. Got it. So So you need to so Is it is it set up where if you press that button, it resets it no matter what. Yeah, it just resets the timer. Okay, so it's just really simple. Like, you could, you could reset it at 5pm and then wake up the next morning and reset it at 7am. And yeah, it will just start again. Yeah, it's just waiting for probably make it so it waits like, you know, 18 to 20 hours before it says okay, there is enough time pass where cat can request food again. You know, I've never had a 7555 go for 18 hours. Like the problem. The problem with that is you'll you'll calculate it but with tolerances and with temperature drift.

That's why it doesn't really matter. Yeah, that's why you don't have that's why you don't need precision in terms of like waiting 24 hours. I'm like, you don't have to like all that matters is your the whole idea is to not feed the cat twice a day.

Got it. Thank you. Yeah, okay, I think you can make this thing you could probably make this thing accurate to within say two hours. Yeah, that's that's plenty. Like that's what you're gonna get. Yeah, the

whole goal is to make sure cat does not get fed more than once a day.

You know, I think I've actually seen someone do a 7555 that lasted a week, but it was A big resistor and a big cap. Yeah.

Well, no is that you're thinking about that's a single shot. This is it's hitting a decade counter. So Decca counters counting up. Oh, you've

right my bed I was thinking about it as like as like a full on analog solution. Now with the decade counter, you're right, because like if it just has like a set frequency, then it'll be way more accurate. My bet yeah, it has a set frequency that's hitting the decay counter and the counter when it hits a certain number, right? We will. So you set the frequency and then you choose whichever bit on the decade counter or even multiple decade counters is equivalent to what you're looking for. Yeah. Okay, nevermind, that's way more accurate than the way I was thinking about. I was like, well, there's a cap, that's when you're thinking about doing a single shot. Five timer setup, which technically would work, it would it would it is way not the right way to do this. Now, your implementation is far better. I hope. Yeah. So the only thing is to figure out make the LED draw less power than what it would like if you just put like, I guess you can just currently, yeah, you could just currently Yeah, I mean, that's, that's the easiest way to control. Just what Okay, so here's the thing that I've learned at work. Most LEDs, you know, they're up to 20 milliamps of current draw, like like your your Joe Schmo, Jelly Bean 20 milliamp red LED, they don't need anywhere near 20 milliamps for you to be able to see them turn on. In fact, 20 milliamps is like annoyingly bright, in most cases. My rule of thumb now is to give an LED between one and five milliamps as a first kind of guess. Like if I'm doing a prototype, set it for one to five milliamps, and then like that might even be still too bright, then fine tune from there. So you don't need much juice.

Whatever that one milliamp, though you're talking what a coin cells got, what 300 milliamps?

Well, but it Okay, right? Yeah, so but that's just the thing. It depends on what you how you want to accomplish this. Do you just want it to be like a dull glow? Or do you want it to be like visible all the time, LEDs will kill a battery? That's the thing. That's the thing is so how do we how do we have an indicator that yes, it's working? And basically, there's gonna be an LED lit all the time.

Or it could blink? Yeah, I guess you can make it blink. Yeah, blinking is probably the best bet.

Blinking, you know, do a 50% duty cycle and you got less power draw. Yep. Well out the blanket. Yeah. Okay. Sounds like a good good start. The designs done, print it go. Ship it. It's started to Kickstarter now. Yeah, give $5 million. Right now. We'll figure it out later. Yeah. It's only sad because it's true. You know, I have not looked at Oh, yeah. And we have a name for the project, the cat feeder and reminder. The CPU, the CPU. Are you going to are you going to 3d print a little box for this and stick it to the

wall? Yeah, I have a little box. And then, you know, sticky tape it to the the captain.

I like you know, one of the things I like about this project is that you've added all of these constraints that don't need to be know that like, make it hard. You're like making this hard on yourself. Yeah. I'm gonna try and learn. I'm one project I can learn something from. Sure. Yeah, no, I totally respect that. That's, that's super cool. So one of the things you should do is put a spec for how long you want it to last? Yeah, I think last podcast we talked about this. We mentioned something like one year. Yeah, one year. I'm thinking probably two years is what I want. Yeah, we'll stick with one. Let's do one, you know on on your prototype, you should you should put pads in like a zero ohm resistor so you can pull it off and then actually read the current draw and see if you're gonna hit one year or not. Like that's your goal. Do it do it do it like NASA shoot for one year and get 10 Yes. Cool. So there was a there was a synth that was at work. Go figure I work at a synth place. But this was this was a like a full keyboard synth, not like a modulation that was actually in our front room. I don't even remember why fully why it was there. I know somebody wanted to repair it at one point. dump. I don't even know whose it is blah, blah, blah. But it's been sitting there for years. So it's like whoever it belongs to, obviously didn't want it anymore. And so I asked the guys, I was like, Hey, come and take this home and look at it, and they're like, fine, whatever, it's broken doesn't work anyway. And then everyone at the at the shop was like, I've tried to fix it. It's not, it's not working. So I was like, ah, that means I gotta fix it. So I brought it home. I started tearing through it, and it would power up, but it wouldn't do anything like none of the LEDs would light up, but I noticed it was pulling juice. And it was pulling like 800 milliamps at nine volts, something like that. So this, this particular model was the Korg, Ms. 2000, which is it's not an analog synth, but it's sold kind of as an analog synth. But it's like, when you open it up, there's, it's a DSP chip, and a bunch of other stuff like It's like what, okay? Regardless, like, it's got two boards in it. One board is like, the pots and the switches and all that jazz. And it's basically all it is, is a giant like switchboard with IO board IO board effectively, like Yeah, buttons and LEDs and screen and things like that. And it's got a nice little ribbon cable that runs off to like, the motherboard. You know, I was, I always find that funny when people like, whenever they see like densely populated circuits, they're like, oh, that's the motherboard, you know. But from here on out, we'll call that the motherboard. I think the service manual cuz it's something else, but whatever. So So I started probing through that. And, and there's there was actually two issues with the board that I found immediately. So one of the, so there's, there's a line filter that comes in, that's basically just like a, like a common mode choke effectively. And that had some problems. So it was just like, Screw it, I pop that off and shorted it out. Because like whatever I like I don't I don't need that for my own personal use here in the shop. And then further down the line, I noticed that there was a surface mount fuse with the number 20 written on it, and I looked it up and that actually ended up being like a one amp fuse. And I just took tweezers and shorted it out and and fired it up and it pulled like 800 milliamps and and nothing was happening. And I was like okay, well that's not good. Like where's that 800 milliamps go like it's going somewhere and if you let go it was praying entropy that's that's the funnest part about electronics because like something's happening, but you have no clue what it is like, you can't see it. You shouldn't smell it. You don't want to taste it like all of these things. But like something's happening right? My power supply the fan is starting to wear up because it's it's dumping a lot of juice into this thing you know, so I ended up just doing the Parker trick where I held the tweezers onto it for a while and then touch the DSP was like oh damn, that's hot. Like okay, this DSP is done ski. So I ended up you know, yanking the DSP off the board which it's it's one of these like 120 pin DSP. So it took like, a ton of hot air to pull this board and I ran the part number on it, which is an X CB 56362 P V 100. For anyone who cares about that whatsoever, which is like a 24 bit DSP that apparently is like ancient Chinese magic. Because it no longer exists. You can't find it anywhere other than like, some skeezy places on eBay that are selling like one offs for like 10 bucks. So I was like hey, you know screw it let's give it a shot. So I bought one and and it came in and I realized it was like wait, I have no clue if this DSP needs to be programmed. I have no clue if it has an internal or external rom right? I have no clue. So I you know the thing is, I did have the service manual, but this is one of those service manuals that like the schematic is not really intuitive. And it has things like it does have ROM chips and it has RAM chips on it. But it doesn't say that like oh the program for the DSP is stored on this ROM. In fact it has, if I remember right, it has one ROM chip that is connected directly to the main processor. So this whole board has a master processor, it has a DSP chip, the DSP chip has its own RAM chip, and then the processor has its own ROM chip. So how does that work out in terms of like, does the DSP get its program through the processor on startup or like how does that work? Or does the processor just handle all the controls to the DSP and the DSP just functions? Also, good luck on finding a datasheet for this DSP, I have no idea. So I just don't want you found this datasheet Yeah, is it in Chinese? No, really? Okay, well, I didn't have much luck finding one Regardless, it was the first link of I searched a certain part number of Google. Maybe, maybe things. Maybe Google's different in Houston than it is in color. Or just I don't know why it's on data sheets, three sixty.com. Hmm. Okay. I don't know is that it's got an IEEE logo on the website. Let's be legit. I don't remember seeing a datasheet for this. Well, it's one of those were like, you would immediately skip it because it doesn't look like it looks like one of those, like, datasheet aggregator sites that you can't get anything from. Yeah, that's what it looks like. But you can actually the gray market of data sheets. Yes. Well, maybe yeah, maybe maybe that's what I did. Regardless, like, I don't even know if a datasheet for this would do me much good anyway, because like, Okay, I had a bad Chip, I bought a new good chip, like, what's, what's the datasheet really gonna do for me? Other than tell me like I'm boned or not? Or what do you see? Well, how it handled its program? Yeah. Well, as I keep, as I keep going, maybe maybe you can tell me if, if I'm right or wrong here. But, but regardless, I ended up just taking the motherboard to the to work. And I soldered on the new chip, which, surprisingly, for this thing, because it was made in the year 2000 Is the vintage of this thing. And, and 90s and 2000 specifically, audio gear have that really crusty PCB technology where you know, we're like, the solder mask, like, begins to warp and foil and do that with this one didn't do that. Like it stood up to me putting some heat to it, which was really nice. So super easy to solder back on. And then I just, I set up a bath of denatured alcohol and I just dumped it in there. And before I left for work, I just pulled it out. So I think it spent seven or eight hours under alcohol, which Hey, I you know, it works. So I ended up getting back home firing it up, you know, connecting everything. And everything functioned fine. Everything fired up. You know, I was shorting up that that surface mount fuse, which I have since pulled off the board and just put a resistor leg across the turbine. So there's a play music. It totally does. Yeah, everything. Everything functions great now, like I remember, everyone was looking at this at the shop and I don't know, there was like 1000 different reasons why it wasn't working. But no one tried Parker's trick of just let it run until something gets. Hey, sometimes that works, you know? And, you know, it always works. Well. It always works. It just might not be repairable. I was gonna say it's not always recoverable. Always works and finding what the problem was. Yeah. Yeah. So that's it. That's, that's a fun one that that was, you know, I'm also trying to do so a handful of smaller projects. And that was one that was it was 10 bucks. I just had to buy something off of Ebay and re solder a chip. I'm surprised it was a legit part you got you know, that's just if you if you put it on there and you powered it up, and it didn't work you'd be like well I guess it needs a program that's what they would have been like you know, at that point I would have just tossed the whole thing. Yeah, but cuz I did I was trying to find like Could I just buy the whole new board because whenever I was running my repair shop that was a lot easier just like buy the brand new motherboard slap it in charge the customer go and man people like this since goes for like 600 bucks online something like that. And a new motherboards like 350 a you know, so it's like I don't Yeah, 350 would make it work but I don't want to spend that to just have it all spend 10 bucks and now I got a cool keyboard synth that I can play with so

yeah, without without a PGCE fuse on it. Yeah, I mean I wonder how that chip was that did originally

I still have the old chip maybe I should I wonder if someone I wonder if if someone plugged the power in wrong maybe but I would hope you know I haven't looked at the schematic too in depth in terms of the power supply but we're hoping to have reverse polarity on it now oh you know what might have happened hmm so this this whole synth runs on a DC barrel Jack I wonder if someone put AC like a high AC into it because it just runs on nine volt DC power jack could be it i that might but but it was just the DSP that went bad Everything else is fine. So well it could have it could have fraud you got when you do something like that. Usually the weakest thing guys the weakest thing go die. And then the weakest thing died, which was the DSP and it pulled a lot of current and then blew the fuse. Yeah, that makes sense. Yeah, that's likely. And the DSP is probably the weakest thing in there. Probably because everything else that's connected to 3.3 is the processor, which is probably a little bit beefier and some other logic chips, which are probably the most beefy in those string of things. And I wonder if something with the DSP is, with the output or something. Georgia got coupled with that AC signal, maybe? Who knows what happened? Because I mean, the when I first inherited, you said the line choke was bad to the lot. Yeah, the line choke was was crappy. And that could maybe be something there, the DC power jack had been dishonored from the board. And like, somebody just ran wires to it. So that's probably someone trying to figure out what was wrong. Probably, it could have been someone accidentally put power down the line input or output or whatever it's called. And it's just plugged mains into it. I mean, who knows, but like I'm saying is power came down there. And that could have fed into the DSP somehow, which, you know, the output of the DSP probably goes onto that circuit, somehow, or through a filter or whatever. But you know, yeah, it's a mystery. It works now. And it sounds great. After spending seven hours under alcohol, it's got a very clean sound to a little drunk, though, just a little bit. So So on a on a second note, I've got a cool chip that I wanted to share with everyone, it's actually an LED matrix driver that I've been looking at at work. So get this it's a 39 by nine LED matrix driver, which means you can drive 351 individual le LEDs, or you can set it to run 117 RGB s, and do like full color craziness with it, it's made by I SSI is he or her we say that, and it's the is 31 FL 3741, for any one of those who wants to play along at home. So this, this, this chip is in a small three millimeter by three millimeter QFN 24 package, which is kind of crazy, if you think about driving that many LEDs out of this thing. So it's a matrix driver. So you get like rows and columns and stuff. And it handles all the all the, you know, everything internally, so you can individually address each LED in the matrix, and you get 256 levels of PTO PWM control, but you also get an additional 256 levels of DC current control, which is pretty awesome. So you get a ton of brightness control on this, which I suppose if you know, for the individual colors, if you're, you know, you're just doing like red LEDs like, that doesn't matter too much. But if you do in RGB, then you get a ton of color control with this thing. And just it being so small, and it makes it kind of awesome. But the fact that it's actually not that expensive, either in quantities, it gets, you know, sub dollar for this, this part, and it singles. It's like $2 250, something like that. So, yeah, if you have a need for a matrix LED driver, this thing pretty much fits the bill pretty well. And oh, I see how they're handling that. So it's 39 current sinks by nine sources. That's right. Yeah. So you're actually running at a 1/39 duty cycle? Well, yeah, I mean, it's gonna scan your matrix. Yeah, all I'm saying is, you're, you're that's how they get that power density. Yeah, is it's not a 1/39 power cycle, or I believe in the datasheet. I think I read earlier today that it can handle up to 38 milliamps in that one. I see. So you can get a pretty bright, you know, you can you can hammer it. So one of the things is I've been kind of working out with one of my firmware guys, though, like, questions of layout VS code, because one of the one of the applications that we might use this for at work, we don't need 351 LEDs, we only need about 120, which is still you know, a respectable amount of LEDs on a board but we don't need all 351 So, but, you know, if I only need 120 I get to choose as the designer where they go, but if I choose where they go, am I making my firmware coders life harder, based off of where I put them in what pins they are? Or does it even matter? You know, if they're all individually addressable, and every write requires you to write the same number of bits doesn't matter. And so like I don't know, I like I kind of like this part of design where you have to consider like, Yes, I could do everything on my own such that My life is the easiest. But I might not be thinking about the next guy in line who's got to design around my garbage.

Yeah, usually what I do with that is I'll go into datasheet. And you're probably just looking at the hardware. So if you got look at the software stuff, oh, no, I've been doing that all day, trust me. And you got, so this is running off I squared C. That's right. And so with i squared, C, it's a, you set up a the master address, you set the address out, and then usually you can send a lot of data bytes. After that, you can almost say, I think there's probably a limit. But I think it's basically you can send as many data bytes as you want until you send that stop it. And so if you, let's say you're writing a giant matrix out, is, from a firmware standpoint, as long as they set up the that data packet, correct? It doesn't really matter what the order is, right of the LEDs on the output. Now, is it easier to write them all sequentially versus having to jump around depends on how what your display your display is? Like? I'm saying like, let's say you're doing like a VU meter? I don't know if that's what you're gonna do or not. But like, it might make more sense where like, the VU meter, is everything that's in a single bar is part of a is yeah, they're next to each other in the address space.

It just makes code cleaner. Yeah, that's what I'm like, because I could just connect LEDs. I don't want to say fully willy nilly, but I could just connect them in a way that it's laid out willy nilly? Well, yeah. Especially Oh, that's just a thing, especially with a with a package, I was kind of the three by three millimeter package I was I was praising it a little bit earlier by it being really small. But that also adds difficulty, right? Because now you gotta route signals in a small package to wherever they need to go. And it's nice if your LEDs are all the actually matrix on your board. But on our board, they're not, we're using a matrix controller, but they're not physically arranged in a matrix, you know. So that adds a little bit of complexity there, too. Yeah. I don't know, it's a pretty cool little chip, especially for the cost on it. And the ability to control each LED individually, with with, you know, some simple rights on it. The one thing that I'm investigating right now is yeah, it runs on ITC, which ITC is not the fastest thing on earth. You know, it's default 400 kilohertz. And this chip supports the faster one megahertz ITC. But that's the maximum speed that it offers. And in order to write to an LED, you got to I two C's got a little bit funky with this. Well, and specifically this one, because it handles everything in terms of pages, because you don't have enough bits to actually write for the all the for the whole thing. So you have to like write your control register, then you have to write your page register, then you have to write your address register, then you have to write your PWM register. But before that, you need to write the slave address register. So like in order to tell this chip, hey, I want you to look at this LED and make it this brightness, you have to write five words to it at 400 kilohertz, if you need to do that 100 times, how fast is your total write speed?

Yeah, that's why a lot of times actors look at this and does support it. They you can do a burst, right? Where you can say, you can set all that stuff up, and then you tell it hey, I want to send you like 400 Bytes worth of information. Yeah, do it. Yeah. And then yeah, it's going to go in that way. You don't have to keep sending address information.

Right. Just why don't we just keep the address open? And I'll just keep sending you bytes. Yeah, I don't know. That's, that's a lot of stuff that like, if anyone's ever worked on a project where they're not the firmware where the just the hardware guy. I think those are things that are really important to kind of keep in mind. You have to you have to work with your firmware guy, because last thing you want to do is piss them off by just being like, well, here's my hardware. It's really great for me, it's super shitty for you. You know, like all your traces are beautiful coming out, fan out. Yeah. You know, and actually so we use STM at work SD Micro for a lot of our microcontrollers. What I've started to do, and I really liked this personally, my firmware guy does, I don't know he doesn't super care, but I like this. I like this a lot. But I will start the The program or I will start the the STM cube MX project where I've named all of the pins, and I make sure that all the naming of the pins matches exactly my schematic and that everything is all set up such that I can just hand him the cube MX project. And whenever he starts to write code, it automatically sucks in all of my pin names, which matches the schematic perfectly. So it's like super clean transition between me and him. So if he's like, Oh, hey, what does net XYZ do? He can just tell me the name and I can be like, Oh, I've already set that up as a input on my process or in its XYZ. You know, like that kind of thing. I I don't know from a hardware designer, like it just doesn't. You can't just think about PCB land or schematically and you got to think about all the other stuff in there too. You got to think about every land every layer. So that was the Mac fab engineering podcast. We're your host, Parker, Dolman and Steven Craig. Later everyone take it easy.