MEP EP#274: The Pinch Load

Podcast Transcript

Host 2     00:10

Welcome to the Mac fat engineering podcast. We are your hosts, Parker, Dolman. And Steven Craig. This is episode 274. My computer just like goes crazy when start streaming.

Host 1     00:24

I mean, there's a lot going on here, right?

Host 2     00:25

Yeah, yeah, CPU usage is pretty high. Um, anyways, continuation of the beating the heat topic from last week, which was thermal management. So I thought about this topic a bit over the weekend. And I kind of want to dive into because we kind of just touched like, like you were talking about, like, basically back calculating. What was the thermal junctions? Yeah, no thermal impedance of a component by like, figuring out how much current you're putting through and how much it was eaten up, and all that good stuff. But well, I want to talk about is kind of like interesting designs and devices that I've come across and seen. And you probably seen the same thing in the AR experience for thermal management. So I was actually testing some product here at macro fab. And they had some very interesting thermal management for like, to 220 devices that were like soldered, you know, flat, and then Deepak devices that were soldered onto the board. So they were soldered onto like a one inch by one inch square copper pad kind of standard for like thermal dissipation on the PCB. But they had a screw hole that was really close to those parts. That was connected to a pat as well. The The thing is with with those kinds of packages is usually like the output of the part of the component is like, the voltage output, I'm like, it actually doesn't make any sense, the voltage output to the component is the tab, or is connected to tab.

Host 1     02:11

A lot of times when not a lot of time to one to 220s are made that like basically the tab is stamped out effectively where the the die connects to the tab. But the die also that I'm sorry, the tab also connects to pin two. So whatever pin two on a 220 is, is the TAB TAB.

Host 2     02:29

Yeah. And so you can't just like connect that to your chassis, because then all your output rails are connected together. And that's not good. So what they what this design was, is it had copper fingers, so the ground plane that was connect or that grump on the the thermal plane, I want to call Yeah, that was connected to the pad, or the tab of the package was like a fingers like this. And I actually will draw this out for the podcast notes. But the live streamers can see this. But then then they had another pad that was like this. So like the thermal pad came over. And then they had this big, like chunky plain like this kind of like a zipper or Yeah, yeah, but so they weren't touching, and just a big zigzag between them. And then that went to a screw terminal. So they could sync it to the chassis. I don't know how well that works. But they did it. And it's an industrial design. That's odd. Yeah, I don't know how well it works. But it's got to work. So a bit, you know,

Host 1     03:47

You know, when it comes down to like D packs, and things like that a lot of manufacturers will put the information in the datasheet as to, you know, what recommended size pads and things of that sort. And then also, if if needed be via stitching around those pads to pads on the other side of the board for additional heatsink and capability. Most of the time that's in those data sheets, one to 220 packages, however, since they're through hole and not necessarily intended for lay down operation. I've seen it a bazillion times, though. You don't get that kind of information in the datasheet. And I read a really cool app note. I think it's by vishay. The other day about doing lay down to 220s on a copper plane on a PCB. What torque should you torque an M three screw to and does that have an impact on the thermal dissipation of the thing? And what's interesting is it totally does so like off the top of your head like it obviously makes sense. Like if it's really loose, it doesn't do anything. Okay, yeah, we can all agree to contact But as you keep stepping up the torque, its thermal dissipation gets better and better until a certain point where it actually kind of like tips over and rapidly drops off. I bet it's warping the board. It's not warping the board, it's warping the to to 20. Because the screw, the screw is not like straight where the dye is, it's on the tab, which is adjacent to the dye, and the body of the component will actually start to tip off of the thermal pad. And so and then of course, they took it to like complete destruction, they took it to a base shear off. So what's interesting, it's five Newton meters was acceptable for like long term stability, and it had like, you don't get any better thermal transfer effectively. Past that, and then I think for an M three screw, I can't remember what it was, like 20 Newton meters is the point at which the head will just shear off. So Oh, I'm sorry. Yeah, no, no, it's not my bad. Not Newton meters. That's that would be monstrous, right? inch pounds. So five inch pounds.

Host 2     06:09

Yeah, I'd say Newton meter was five Newton meters. I mean, it's actually not as high as you think either, though. No, but it's yeah, no,

Host 1     06:17

I had it wrong. It was inch pounds is what I'm looking for.

Host 2     06:21

Actually, I take the back, oh, it's 45 inch pounds, which is what that'd be four or five foot pounds. Not too bad.

Host 1     06:33

But for up to 220 with an M three screws. And actually, that would be counterproductive for bolting to to a heatsink on a PCB. Yeah.

Host 2     06:46

Well, yeah, on that, on that on this one product they had both sides were like that. So like the top where the component actually soldered down. And the bottom side was two plates. And they were all stitched together with vias and the fingers were on both sides as well. And then last week, I can't remember who posted this component from vishay. That is a thermal jumper, but it's a thermal isolated jumper. And I bet you this is like a component that like is able to thermally suck away or allow thermal to the transfer across a, a non conductive barrier. So you could do something like the finger thing, right. But this thing actually has like, what the specs since you know, thermal,

Host 1     07:37

I like the trademark name thermo wick thermal wick.

Host 2     07:41

Yeah. So I will share that into chat. And on the podcast notes,

Host 1     07:48

High thermal conductivity of 105, and D

Host 2     07:52

Is credence to that finger idea. Because this, this is basically the same thing. So if this is this probably works better. But it costs money. But it costs money. Whereas the copper finger thing doesn't.

Host 1     08:08

Hmm, interesting. I'm curious if anyone if anyone has used these and what is your application? I'd like to see, like, when did you need to specifically solder things down to channel heat?

Host 2     08:23

Oh, basically your so I'm looking at the substrate material. It's aluminum nitride. So your base, it's soluble. And so it's basically like a chip chip components like resistor or capacitor. But instead of a dielectric inside or, or a thick film resistive element. It's just a big chunk of aluminum nitride in there. Yeah, right.

Host 1     08:44

So it doesn't pass. It's electrically isolated. But thermally it's hungry.

Host 2     08:50

Yes. And then, but for my end, though, the most I've ever done is just having ground planes or thermal planes really. Like on on Patar. That's pretty much all we're using. I have done kind of like as a hack of had MOSFETs kind of close to like connectors that had big pins. And then you could like thermally sink some heat into those big pins. That seems to work okay, I probably wouldn't recommend that for like a production device but seem to work okay for the stuff because you know, the connector is not designed to dissipate heat. So, but it works if you need it. Yeah, I think there was a I gotta find the paper I read a long long time ago. Back when like you were first starting at macro fab. And it was like, when you like they did like the the thermal planes and they just kept making the thermal plane bigger. and like how big is enough basically? Or is there a follow up like, you know, clamping down that the N three screw? When do you get to like what's what's called diminishing returns when you get diminishing returns on how big your thermal plane is. And most of the time, it's for most component sizes. It's like one square inch, anything over one square inch, it just doesn't dissipate out far enough or fast enough to really matter for that component. Oh, sure.

Host 1     10:36

Like, I guess if you take a cross section and look at like a thermal gradient that comes that, yeah, it probably drops off fairly quick. Yeah, drops off fairly quickly. So anything over that,

Host 2     10:47

I'm pretty sure it was one square inch, I have to go find that article again. But anything over that it helps, it doesn't help enough or fast enough to where like you're keeping under your thermal limits of your components. Right. So basically, if you need more than one square inch of copper, you need to go to another means of dissipating like having thermal vias, a pat on the backside or thermal Wixey or a or thermal wick, therm wick is that what it was called?

Host 1     11:20

No, it says an A in it thermo wick, I believe thermo with

Host 2     11:24

Thick tea, Mm Hmm. Interesting. You don't anything crazy. Thermal was the SSPs. Right, we were we never really built that well,

Host 1     11:36

When we when we fired up the SSPs. And, and I will remind all of our listeners that does did work in pieces right now. But like it could generate a monstrous amount of heat, like just an absolute insane amount of heat. And we it's a very good age to place our PCB for a waterblock to go. So we never actually got around to making that portion work, like actually mounting all of our heat generating ICs to it. So in my most recent amp design, I have a handful of regulators and transistors that go in relation to those regulators. And I did a ton of like, heat load balancing in it. Basically, I wanted everything to kind of share all the heat as opposed to one getting super hot. And I've got some pretty cool thermal image images from it. In fact, I got a little video, I might even post that up where I first turned it on, you see the thing where like, all the hotspots start to appear. But that's where I use my to, to 20 footprint, which is basically it's like, it's basically one inch by one inch ish, with a with an M three screw through it, and then via stitching all around to an identical pad on the backside. And none of my none of none of the parts that I'm bolting to it get to ridiculously hot i, that was something that certainly really wasn't calculated. Some of it was so some of it I put enough calculation to just be like, well, this isn't going to explode. You know, like that level of all the rest. I was like, I'm gonna turn it on and see what gets really hot and then adjust from there. Yeah,

Host 2     13:28

Yeah, it's not gonna disorder itself.

Host 1     13:30

Right. Right. Which, which is funny because I actually I saw a god, what's it called? It's not an app note, like when a company sends out notes about changes to devices in the field,

Host 2     13:47

Or Rada, I guess, within a Rata

Host 1     13:51

Rata. They so it was it was an amp schematic that I found on a forum the other day, and it's fairly new. And at the very last page was like, Oh, we have, we have these AP notes basically, that these, these diodes that are in here are really not necessary. And we suggest to pull them out. And then they say, because there are some situations where they can get so hot that they d solder themselves from the board. And I'm like, Oh, my God brought a product change notification. Yeah, Fabio firmware? Yeah, I think I guess that that's probably a good. I think that's it. I don't remember what it's called.

Host 2     14:30

Well, let's go jump to this topic, because since we're on the thermal stuff, so last week, we started talking about that because you had some LM three, three eights that were getting a little toasty.

Host 1     14:42

Yeah, roasty toasty. So I may, I'm crossing my fingers here. We might know tomorrow, because well, okay, let me let me back up. Last last podcast, we talked about some of the thermal design that I've got going on in one of my new projects and you Using some lm 338 to crank out what is it 6.3 volts at 2.6 amps. And they're fed off of a supply that's like 10 to 11 volts somewhere in that range. So that, you know, they produce a fairly significant amount of heat. And I'm heat sinking them to my chassis. And I'm working through the heat issues, very similar to my amp design where it's just like, I'm turning it on seeing what's happening and then fixing the problem. But, so, I've got these LM three, three eights that are that are in my design, and I purchased these off of Amazon. And one of the reasons I purchased them off of Amazon is because Mouser and DigiKey are not particularly quick at the moment. And this was available a lot faster. So I just bought a like a little five pack of LM three, three eighths off of just whoever had them on Amazon. And so last week, I was I was discussing that these transistors were just starting the regulators were getting way too hot. So I had a heatsink solution where I was just going to slap a heatsink on the back, and and see what happens. Well, I did that. And it didn't solve my problem. In fact, it basically didn't do anything at all. And I even tried getting a load of ice and just sticking it basically on the back and just like force cooling them and even putting ice on the body of the the regulator to just see like, okay, is this mean not heatsinking things properly? And can I temporarily just like force cooling really hard? No, they're not functioning. And they seem to be dying around one and a half to 1.6 amps of current. Which if you look at the LM 317 It looks like that regulator. Its maximum output is about 1.51 1.5 amps ish, somewhere in that range. So I'm, I'm just questioning perhaps, did I get a some LM three three eighths or some LM three, one sevens that are actually marked as LM, three, three eighths. So I it's funny because I wanted to get things quicker. So I ordered it off of Amazon, and then I got potentially bit by this. So I went to Digi key and ordered some, and they show up tomorrow. So I'm gonna throw them in and see if they respond any different. And if so, the plan is to ship these off to Parker. So he can X ray them. I bought some extra LM three, three eights. So he can do a comparison comparison. I don't I don't remember which brand they are or which brand they're marked as because who knows, right? Yeah, so we'll see on that.

Host 2     17:46

In if we don't see differences under the X ray, I'm going to try to see how much it will cost to send it off to give them D caps. And just see because then we'll actually see what the difference is.

Host 1     18:00

Yeah, yeah. And I don't I don't like lm 317 has been around for a long time. So like I think we could probably find an image of the dye right somewhere and then compare if that's what we have there. So maybe this is a story not to buy off of Amazon just random parts

Host 2     18:19

This is a good story to tell your customers not to buy off Amazon well their production

Host 1     18:24

I mean yes for production that that that would never happen this is me and my basement right behind Yes yes. So yeah, I might I might have gotten bit

Host 2     18:37

Hi Ron found a China PCB copy.com But apparently they have an IC unlock service unless you send your parts and they decap them to d cap them and copy them for you

Host 1     18:53

Get a quote Oh,

Host 2     18:55

They actually will they actually can get the bin or hex file off a microcontroller to

Host 1     19:01

What they read all the bits. Yes. Oh god.

Host 2     19:06

I love this this this like footnote here. I see sample will be decamped. It will not be in work condition after IC unlock.

Host 1     19:18

100% refund if the samples do not pass the test.

Host 2     19:23

Interesting. I'm gonna have to look into that. But this is why services like this when we were talking about right to repair and your secret sauce being your schematic. This is why that's not a thing.

Host 1     19:40

Yeah, even your firmware can be extracted, right?

Host 2     19:43

Yeah. Your secret sauce has got to be like your service and your your brand, which actually leads to what we were talking about. Why do parents at what Apple wanting to protect is their brand and their service their prestige

Host 1     20:01

Pedigree.

Host 2     20:04

Speaking of that, actually, the right to repair stuff is there was a article that was sent to me about the whole ice cream, you know, when you go to like McDonald's, and you want to get an ice cream, but they always say the ice cream machines broken? Oh, is it a conspiracy? Yeah, that's actually a conspiracy infamous like meme now. But apparently, those machines are really finicky. And the service plant because like most McDonald's like franchises, and so you, you kind of have to go through McDonald's to get like your service for your equipment, and especially that machine. And it's like some special, crazy machine integrate machine. That is, the only way to get service is through the company who also makes it and they don't have a last service technician. So it takes a long time to get your stuff fixed. And there was a couple people out there that are trying to like hack the machines, and make them easier to repair and easier to figure out what's wrong with them. And McDonald's and I think companies called Thomson or Thomas, something like that are going after these people for like, making it so it's easier to fix their machines

Host 1     21:23

Right to soft serve.

Host 2     21:25

Yes, the right, this is Officer Taylor. Thank you. GraphLab.

Host 1     21:30

It's Taylor. That's fantastic. Yeah.

Host 2     21:34

Which is that's the thing that like, if you bought if you actually own the machine, so you're not like leasing it from Taylor that I don't know if that that difference. If you actually bought the machine from Taylor, you should be able to fix it however you want. If you're leasing it from Taylor, okay, you got some, you gotta wait for them.

Host 1     21:56

You know, back in college, I worked as a computer monitor repair technician, and the computer monitors. So like CRTs, CRTs, and LCDs. We had a flat team and we had a tube team. And the we Okay, so we had a we had a company that basically so like car mechanics and and cash one of my dealerships, car mechanics, and dealerships could sign a contract with us, where we would outfit their entire thing their entire place, we would give them all their computers, we would give them all their monitors, we would do everything. But if anything ever broke, they had to return it to us because they technically didn't own it. And we would repair it. So they didn't have a releasing everything from you effectively. Yeah. But not, but it didn't work out that like they leased the building worth basically. So they didn't have the ability to repair things. I wonder if right, we're right to repair would change that. Like if something broke if they didn't have to send it in if they just so honestly,

Host 2     23:05

No, because part of it is like the terms have to be explicit. So your terms would be explicit that hey, you don't actually that's the whole thing with leasing versus owning, owning, you have to pay for all the capital upfront leasing. You that like you're, you're leasing companies taking depreciation on the product, not the company who's leasing it from you. That's the difference.

Host 1     23:29

Right? But how does that apply with cell phones that you haven't fully purchased yet?

Host 2     23:35

If you haven't fully purchased it, then technically that company owns that phone

Host 1     23:39

Still, then you can't repair it right? Or you have

Host 2     23:43

To go through whatever the 18 T if it's a it's 18 t so I'm making the argument that is the same Sure, if you're talking about phones that you don't own, or you're leasing but that's actually the thing is a lot of people don't know that or they get their phones for free and they don't know it's it's a lease plan through their their bill or whatever. So those would be more explicit. So

Host 2     24:14

So chat is going crazy about belt, tailor McDonald's and ice cream. And why the McDonald's ones break all the time. Apparently it is like a McDonald's wants a custom that require a lot of finicky finicky adjustments to keep them running. Right.

Host 1     24:32

Which which is interesting because it doesn't it. I mean, it doesn't seem to be an issue at other fast food restaurants, right? No, it's kind of ubiquitous with McDonald's.

Host 2     24:44

I just got China notice. I don't know from where looking at like, oh, a bazillion monitors I have opened right now. Could have been slack.

Host 1     24:57

All right. Well See, I see a note up here about common DRC problems. I say let's Oh, yes.

Host 2     25:05

A little bit distracted. Yeah. Um, yeah. So common DRC problems. So I've been actually going through it kind of making like a DRC matrix file, not matrix file matrix grid, at macragge. For our all our teams, because the engineer engineers are like, we're generating DRC reports and all this stuff, and we go to like our customer success team, or at our sales department go, Hey, your customer has these things that are wrong with your with their design files. And then they take it and just hand it over to the customer. Well, this is kind of like training material. So that now like sales and customer success actually know like, what does it mean, when annular ring has a a drill to copper issue? Like, what does that mean, what's and what resolutions should they expect? And all that other stuff, basically, is training material, but I was going through all the all our old reports, basically building up the database for it. And I found two issues that almost always the ones that pop up. And so I want to discuss them and also be on because I know y'all do a lot of contract manufacturing is what issues y'all see a lot too. And I wonder if they're the same. So the first one is, is the annular ring being too small, around drills. Most of the time, it seems that most annular rings are the same they set them to or a designer will set them to the same as like a trace width. So if you have a five mil five mil trace with DRC, they set it to five mils. Well, usually, that's incorrect, though, usually your annual especially for really cheap PCBs, the annular ring would be like six mil in that case, because that what because what's different about it is your your drill hits happen after the copper gets etched away. And that machine has a different tolerance than your copper etching machine, like the wash. So that's the biggest. That's like the number one problem is like annular rings being too small. And then to copper to board edge issues. Most people for some reason, or designers like will put the copper right to the board edge every single even like internal layers, right to the board edge. And that causes issues of like when you start dealing, when you start deep analyzing boards, you start getting delamination, you can get water moisture wicking into those inner layers. If you have a trace, that's right next to it, you also get like if you deep analyze, you can that trace can get cut. It's like board houses for some reason, or magnet, the putting mouse baits right next to the trace that's the closest to the edge of the board. I don't know what it is. That always seems to be the case though. I don't know why how they just go oh trace next board must put tab there. Which causes delamination issues and stuff like that. And also, like components being too close to the edge of board for the same reason, basically, like when you start deep analyzing, especially if you like V score, and you're running your big ol pizza cutter down the board to be penalized the board and someone puts a little ceramic capacitor to close, that thing's gonna pop right off. So what DRC problems Do y'all see the most?

Host 1     28:41

I would say the top two is components on the edge. Like, not close to but like, like, Oh, right on the edge. Yeah, like the edge of a 402 is the edge of the board gotta think I've seen that before. And then silkscreen on pan. And actually, what's what's interesting is I get I get that probably more often than maybe you do, because in our industry, a lot of our customers tend to be very, they want artwork on the board. And so they create this really killer artwork and it looks awesome, but they've done it in, you know, CorelDraw and then they just import it and slap it on the board and they and they don't do a silkscreen removal. So that that happens more often than I wish.

Host 2     29:31

That's interesting. Yeah, cuz I actually don't see a lot of silkscreen issues but we do a lot more industrial stuff instead of artwork stuff. So everyone's trying to make the designators really clear. And so I guess it doesn't you know, we don't get a lot of art

Host 3     29:47

You know, one thing that I thought I would see more in general but I just don't see very often is somebody just like trying to get like a tooth out trace or something like like a one boundary. It's been like that it's good enough, you know? Yeah, I

Host 2     30:00

Don't I don't see that like to like traces being too small. You don't see that that often. Yeah. But yeah, the copper the board edge. Yeah. That's also that's another thing is having your part especially those ceramic capacitors you're talking about being too close to the board edge sir capacitors like to crack with during the penalisation. So that's like the worst place you can ever put a a ceramic capacitor? Nobody actually really fun experiment is how far away? Does a ceramic capacitor need to be from a board edge to be like safe?

Host 1     30:38

Ah, you know, actually, let me let me let me give you an example on something that actually is a little bit opposite of that, like too far away is also a problem. Because a lot of the boards that we do are skinny and long. And, and if you have a ceramic cap right in the middle of it, you end up creating this convex thing and it pulls the cap apart. And I actually got an RMA the other day because we d panelized incorrectly, and you could just press on the cap and the circuit would work. Because we so actually in that situation, you know, there's multiple solutions. First of all, the person D penalizing it, they just need to live some extra training, you don't just have the rails on the end and snap them right. There's a way to do it. But also, if the if that capacitor wasn't in line with the long direction, if it was 90 degrees, that could have been that could have prevented that or helped prevent it.

Host 2     31:35

Yeah. Now I wonder what the that distance would be. So it's you definitely don't want to be in the middle. Because then it gets both sides of the arc, so to speak. Yeah, aboard flags. But you don't want to close because that's where most of the pinch, pinch, I would say pinch load. There's probably some shear thing or something like that. I'm not a mechanical engineer, man. I don't know what that is. But definitely not right on the edge.

Host 1     32:06

I think you just named this episode, by the way. Oh. That's great.

Host 2     32:15

So and also, the other thing is, do y'all when you're one DRC, do run a tolerance on your checks. Because that's actually one thing I started looking into is because your run like let's say your traces, for example, or macroblock, the smallest is like three mil, three mil, three mil, trace width and spacing and annular ring. Well, when you get that small conversion issues happen, like conversion tolerances, when you convert from like millimeter to mils, or mill mills to millimeter happens, and you start getting like 2.98 mils, is that three mils? Right? But what's those an error? But it's like, that's not really an error? That's fine. Right? That's fine. So we started running a couple of months ago at macro fab 8.1 mil, tolerance on our DRC stuff. So you can get away with a 2.9 mil technically,

Host 1     33:13

You know, we're I think we're a little bit lucky we do, we do a lot less contract manufacturing, where it's just like somebody dumping files and being like build this. It's a lot more relationship building. So most of our customers actually ended up using dip trace. And we can work with them with their actual design files. So we don't get to get to we don't have to deal with a lot of tolerance issues, a lot of files, you're a generic contract manufacturer you like if someone's got money, and they say build this, you'll do it right. I mean, as long as they play by your rules and things, and we get to kind of like work with our customers a little bit more such that, like, we forced them to play by much more narrow rules than you do.

Host 2     34:01

Now, that makes sense. And you get to play with the native files almost all the time. So you don't have translation. Translation is not the right term, but good conversion conversion issues.

Host 1     34:15

We I think, I mean, units, maybe we have one client that where we just get Gerber's and a bomb everyone else we get their native design files. And that's a I mean, even if it's like eagle or KiCad or whatever. And that's, that's amazing. It helps a lot.

Host 2     34:36

Oh, yeah, it does. So, I've been deconstructing Oh, before we move on anything else, we're common DRC problems. I kind of want to talk more about this. Like once I think more about it. I think

Host 1     34:49

We should we should do a whole DRC episode where like we go a little bit deeper into all of it, you know? Yeah, we need to get a guest for that to run DRC

Host 2     35:01

Okay, move on. So I've been deconstructing a, a record a, like portable record player right there. And so because it's got one of those old school style record players in it where it's which is like the super heavy arm with no counterweights, it's got like a self loader, spindle and all that crazy stuff. Which you don't want to play a record on, because they tend to the needles basically tend to eat records. They're not very gentle on them.

Host 1     35:40

So you're just lighting it?

Host 2     35:42

Well, not really, because some people converted it, but you still have the issue of like, weight balance issues, because it's not a counterweighted. It's like a spring loaded arm and stuff. So I'm just going to, and the thing is, I'm never going to play record on it ever, right? I'm not going to, I don't need a portable record player ever. But I would like to have at my office, I'd like to have a record player at my office. So I can on Fridays, listen to records. And so I was thinking like, hey, I can get in like an amplifier and stuff like that. So I already have a record player that can bring up. But the problem is the I didn't want to really buy anything, I don't want to buy an amplifier or anything like that. So I started rummaging around what I had, I'm like, Oh, I have this old amplifier or own one, basically. And so I started taking it apart, and it's a tube amp in there. And so I took a pic actually has a schematic that was glued to the bottom of the box,

Host 1     36:45

Right to repair. Yes, right to repair.

Host 2     36:49

And so it is a I actually don't know what it is. But I put the picture in the show notes that would be easier for you to read instead of on the stream. And it looks like a basically has a preamp and then an amplifier that drives the speakers. And then a I don't know what that other tube does. The 35 W for that's a rectifier. So that's what's actually rectifying for the amplifier. That's the

Host 1     37:21

Power supply. Yeah. So you got 120 that comes in. That's, that's the rectifier. And then C seven is a 40 micro farad. That's your main filter capacitor there for. So that's going to be your highest voltage right at that point.

Host 2     37:36

So, and this is kind of like a self contained box. Let me go grab it.

Host 1     37:42

Oh, yeah, that's pretty neat. So how long have you had that for?

Host 2     37:47

I've had this for a couple of years now just sitting in like my attic. And it does work. I did plug it in. And I just tapped the needle on the record player. And it made Poppy noises some like okay, it works. Yeah. So I'm thinking about doing is, it's actually kind of like I like held the front looks. That's cool. Yeah. It's pretty cool looking. So I'm thinking about what I'm going to do is, and these are also not worth anything, like, you know, eBay, they've been hundreds and 1000s of these back in the day, and everyone kept them and they're not worth any money. So what I was going to do is, I'm going to like chop the box down. And show it's just the front part. I'm getting far away from a mic. Sorry, Josh, our audio editor. But I'm basically gonna chop the front off and make a new back enclosure for it. And so and then probably replace the speakers with something that's nicer. And then, and then basically use my own record player and plug into it. Nice. Yeah. Basically make it a standalone amplifier tube amp. It already runs directly off a pickup. It's already got a preamp in it.

Host 1     39:01

Yeah, yeah. I mean, you'd probably have to adjust a handful of things on it. But that's actually pretty cool. I like that.

Host 2     39:09

I don't think at the just anything though, because it's the pickup goes right into the amplifier.

Host 1     39:15

So you're just going to replace it with whatever your record player is, is what you're saying. Yeah, I

Host 2     39:19

Mean, my record player outputs the same voltage levels as this record player does.

Host 1     39:26

Well, but is it a pickup So isn't there two types there's moving magnet and moving coil types. I don't know which one this is. It's the same as what I have. Okay, okay. So I mean given I don't know how old it is, it's probably worth replacing the filter caps in it.

Host 2     39:43

Yeah, I was probably gonna replace recap it and that's it.

Host 1     39:47

Yeah. Oh, also. It looks like it has a death cap in it, which is so it's you got you got no ground effectively it goes through capacitor in there, which is C 10. And they call that the death cap where you just basically ground out the neutral. So I would get rid of that. That's usually one of the first things to do just replace the power cord with a three prong and properly ground one end and just get rid of that capacitor. I see what you're saying. Yeah, yeah, because it's got a, it's got a little two prong here.

Host 2     40:22

It's got a little two prong plug.

Host 1     40:24

I actually, I had a guy who I just did a repair on an app for him where he was like, Yeah, every time I'd get up to a microphone and start singing, it would shock my lips. I'm like, dude, just just stop. Stop playing that. Let me let me fix this real quick. Those deaf gaps are? Yeah, they're a bygone era kind of thing.

Host 2     40:44

Okay, yeah. So I will when I want to build the enclosure for and I'll revisit on the podcast and talk about what caps to get and what actually, like we'll draw on the schematic and, and figure out what we need to redo and stuff

Host 1     41:03

Like that. Yeah, it's kind of neat, because there's actually, I really, okay, so I love this, there's there is a potentiometer, that has a that's in series with a cap. And it's, it's tapped off of the cathode of one of the or the preamp stage. A lot of times in in amplifiers, you basically boost your gear gain, or you boost your signal coming in. And then you'll use a potentiometer and, and a voltage dividing situation to adjust the volume. But they're doing it completely different in this situation, they're actually adjusting physically the gain of the stage with a with a potentiometer in the cathode, which is you don't see that super often. On top of that it has feedback from the speaker. So the speaker actually sent a signal back to the preamp to linearize, the whole stage. So like this, this is kind of a unique little, little thing here. I mean, everything's just probably trying to make sure that it has a flat bandwidth as much as possible. But you don't see that very often where you just the gain of the stage with the cathode. So I'm curious to see, I'm just curious to see how it works out.

Host 2     42:15

Yeah, I mean, my this was my great uncle's before he passed away. Cool. And so I'm kind of interested in just like, revitalizing it. And it's something that I could actually use because I'm never going to use it as a portable record player. It's not worth anything to sell. And the front panel on it is actually in really nice shape the rest of it's kind of junky, like it's a little beat up, but I'm like, Hey, cut it, put a new back and, you know, put a new back and box on it and try to match I think try to match that Brown, vinyl. And then you know, vinyl of new wood I cut for it. And I'm gonna use the knobs are on the side of the unit. Which they're kind of dirty. Yep. So yeah, the front looks awesome.

Host 1     43:08

Yeah, that's I really liked that. We need to take a picture of it.

Host 2     43:11

Yeah. And it's got two speakers in it. So I bet you I'd just take the speakers out and then find something that's modern equivalent to replace them with what same Oh images and probably the same bar get really sensitive. I think it's only a What's the wattage on this thing? It doesn't look like

Host 1     43:32

It's not going to be much it's it's a single Pentode in class A operation. So I you know, five watts tops.

Host 2     43:46

Like I think it's like 30 on the box.

Host 1     43:48

I kind of don't think that but who knows? I don't I've don't know what this tube is the what is it? Five Oh, I can't read 50 C 550 C five let me look that up real quick. Yeah, including that thing Yeah,

Host 2     44:09

No, that's actually a good idea actually that's what I've done too. I funny enough I have another project here and we've actually talked about this one on the podcast before and and since the I've been going back to work and working at HQ again microwave HQ. I'm going to finally install my No

Host 1     44:32

Wait is that though the one that we pulled out of the ceiling? Yeah.

Host 2     44:36

No and on the back right it's got a bluetooth module

Host 1     44:40

The old school Bluetooth module when we gosh, we were playing with do music through speaker MC box face back.

Host 2     44:48

Yeah. So we took we took this as the old So back when we first moved to that location, what we call macro HQ. Now see Even I didn't have anywhere to play music in the warehouse. And so I pulled this out of the attic. That was that there. And this is an old paging system. And then I had a bluetooth module. I soldered it onto it. And then Stephen pulled an intercom speaker out of the attic and then put it on top of it took a cardboard box and made that the chamber for it actually sounded pretty, which am that thing for a long time? I mean, a couple of years.

Host 1     45:32

It was religion.

Host 2     45:34

I wonder, do you have a picture that? Gosh, be awesome to find a picture

Host 1     45:39

Navy? I wouldn't be surprised if the development team at macro fab. You set it on fire eventually. They were not a fan of it. No. Well, I

Host 2     45:50

Have the intercom system. And I put the speaker back up in the attic when we took that whole area apart. So the box of course is gone. But yeah, but like you pre cut a hole for the speaker to sit in. And then you cut and it sounded muffled and then you like, and then you ported it by cutting out a hole in the cardboard. It actually did sound pretty good.

Host 1     46:15

So So I pulled up the datasheet for a 50 C five and it's a it's a low voltage Pentode. So it's like 50 volts. Oh, wait, sorry. That's their heater voltage. It's about 110 volt is their application here. But at 110 volts in class A operation, maximum signal power output is 1.9 watts. So it's yeah, it's not going to it's not going to shake the walls. Let's put it that way. No. But again, I

Host 2     46:43

Don't I don't want this to know. And I like it. It's a fun little box at that point. Yeah, I want to put a I want to set it up on my desk, with my record player right next to it and then just jam out some tunes on Friday afternoons. Yeah. And not bother anyone at the Fed.

Host 1     46:59

Well was 1.9 Watts, you're not going to bother anyone? Yeah.

Host 2     47:03

And actually, I might have been just see how the speakers sound and just be like, yeah, that's acceptable.

Host 1     47:10

Old crusty tone. Yeah.

Host 2     47:13

It's called Lo Fi.

Host 1     47:15

Steven. The crest tone. That's what it is.

Host 2     47:19

Crushed stone technology.

Host 1     47:20

Yeah, but it's Lo Fi and then you go listen to Lo Fi beats on it. Right? Double Lo Fi. Double low. How low? Can you go? That's great. So I got I got a cool last little thing. Well, maybe not maybe not cool. It's a

Host 2     47:40

Suggestion. It's always cool.

Host 1     47:42

This is the coolest podcast and the best one. Yeah, we're rated number two by somebody on Reddit. The

Host 2     47:51

That's a, I think there's we should make that our tagline. voted number two best macro fab, or best electronics podcast by some dude on Reddit.

Host 1     48:02

That's a truthful statement. That is true. Yeah. So I in a design have been working on I, it's funny, I'm on. I can't remember, like the third or fourth revision of this design. And every every revision has worked fine. For one portion of the circuit. So I haven't even like considered adjusting that circuit. And then I get my new revision, and it's not functioning well. Now I did change the op amps that are in that circuit. But I changed nothing else about the circuit. I mean, I guess the layout changed a little bit but and it's in its operating a little bit strange. Well, I found out that these newer op amps that I put in there, they're absolutely phenomenal, but they just need a little bit more help. And they need the older op amps I had were fine without any feedback capacitors in them to help stabilize them. But this, these new ones need a little bit more love in that situation.

Host 2     49:03

So I'm gonna be battery is the new one. I'm just gonna guess the new ones can respond faster to transient changes.

Host 1     49:10

Well, that's the thing that's interesting is they have largely the same characteristics. It's not like they're, like 800 times better or anything like that. So the here's, here's the part that's really interesting. So this circuit is part of a window comparator, but not just like, upper and lower bounds, I have 10 boundaries on this thing. So it's like a window comparator stack. And basically what I'm doing is I'm putting a signal into it. And as it crosses boundaries, it does different things for every single one of those boundaries. So what I can do is basically take like a saw wave and I can hack and chop it at different points for all of those window comparators. Well, these op amps drive, the threshold levels of all of my window comparators in this circuit and the The thing that's interesting about it is they output static DC voltages. It's not these op amps are not the ones that are making are changing in voltage, the user can do things with the interface or change knobs and they will shift. But as soon as you stop they, they're fixed DC. Now, what's interesting is when when the incoming wave crosses the threshold on these comparators, that comparator switch. And that's what's causing instability on the op amps. It's, I still haven't figured out exactly what's happening because the output of a comparator, which my comparators are not feeding back to these inputs, are causing these op amps to have problems, it might be a little bit layout related and might be power supply, decoupling related. Regardless of simple fix to this, I've found as I'm just piggyback resistors, across my feedback resistor in my op amp, and that passengers on your feedback on the feedback resistors. And that that ends up creating, effectively a low pass circuit, yeah, low low power just a little bit. And I don't have to do anything crazy, I'm not putting it like really low, I'm putting just a handful of Pico farads across it. And that gives enough stability for the op amps to not ring on the output. Because what I was having is as my competitors would change, the leading edge of the competitor change would be fuzzy and nasty. It almost looked like bad hysteresis, which wasn't actually the problem in the in the comparator. All of this boils down to that I wanted to get here and this is a little bit of a reminder to myself is if you have the room on your board, leave space to populate those components. And it's really funny because I do that everywhere. I usually take a shotgun to my board and just pepper it with places to add capacitors for exactly this reason. But I didn't do it in this one situation. And this is the situation where I need it where you need. So that's just like a trip, not trip a tip to do all you out there. If you're messing with analog circuits, just leave yourself some pads for adding capacitance if needed.

Host 2     52:14

I wonder

Host 1     52:17

If you compared to data sheets, I wonder if the one that worked for you recently had slightly higher, like parasitic capacitance on inputs. If there's it was just enough to keep the ringing down. Something is okay. So something when the comparator, flips slate, something about its inputs are are changing in such a way that my op amp that's driving the inputs is not responding well to that. And then after some period of time they recover. So they ring for a little bit and then recover into whatever state they're happy with. Well, that's

Host 2     52:55

What I was saying is on your feedback loop, you just put a little just a tiny bit of capacitive delay. And that was enough to skew that change a bit. Yep. Where your new op amps are fine are fine now but on your old ones, something about the old ones were okay with it. And it could have been a little extra parasitics there or there were they actually aren't they response lower so they don't see that transient changes bunch, something like that.

Host 1     53:24

I believe it's I believe you're on the right path or track with it there. And the thing is, these newer op amps are new whiz bang, op amps and they probably are are just overall better than the old op amps I was using the old op amps might already have enough parasitic inside that they basically have something akin to that capacitor. Yeah. Whatever. The nice thing about surface Mountain is you can always piggyback in prototype land you know, so piggybacking capacitors, not a

Host 2     53:59

Manhattan style construction. That's right. Well, I think we should wrap up this part. Yeah, I think I think that's good. So that was the Mac fab engineering podcast. We're your hosts Parker Dolman and Steven Gregg live let everyone take it easy

Host 2     54:23

Thank you, yes, you our listeners for downloading our podcasts and to the people who are listening to our live stream Hello out there in internet land. If you have a cool idea, project or topic, let Stephen and I know Tweet us at Mac fab at Longhorn engineer or at analog E and G or emails that podcast at macro voices.com. Also check out our Slack channel. You can get to it by going to Mack fab.com/slack

Transcribed by https://otter.ai