When an Alien Craps

Welcome to the Mac fab engineering podcast a weekly show about all things engineering, DIY projects, manufacturing, industry news and DRC problems. Were your hosts electrical engineers, Parker, Dolman and Steven Gregg. This is episode 317. So before we start, macro fab is now a member of the ERAI. The ER Ay ay ay is a global information service organization that monitors investigates and ports issues affecting the global electronic supply chain. In a nutshell, becoming a member of the E. Rai, and the EMS industry is sort of a badge of quality and prestige. So, we're going to have Cody Ed lick on the podcast next week to discuss this and the ongoing supply chain electronics problem we've been having for like, since 2018.

Yeah, I'm pretty excited because frankly, I honestly, I didn't know much about this agency. And so I'm really curious to kind of sink my teeth into as well.

Yeah, the before the whole supply chain stuff. They were kind of like the supreme authority on like, quality and counterfeit and mitigation products and services for the CM world. So I offer this way. So Cody is a supply chain strategic sourcing person here at macro fab. And I didn't even know E AR AI existed until he said we remember of it last week.

I will be on one comment about it is I think they they might, they could stand to modify their name to make it easier to pronounce or say like E RA or something like maybe there is some like easier way of saying it. But e AR AI is like, that's actually kind of difficult to say, you know,

especially for me, because like I was like pausing making sure I was pronouncing each of those four individual letters correctly. Yeah. Because I'm terrible at pronouncing things

too bad. It's not easy. Like, you know, I Tripoli or something like that. Oh, yeah.

Or ISO or something like that.

Yep. Yeah, exactly.

Yeah, it's gonna be interesting next week, because I have no idea what that I basically, Cody said that we were a member. And I'm like, you want me on podcast? He's like, sure.

I'm like, Yeah, I

don't know. We're gonna talk about but yeah, let's do it.

Now, it sounds it sounds great. I'm really what I'm curious about is how do they enforce what they're what they're going for here? Yeah. So it because

that's what I want to know is what does this really mean? Besides, you know what we're like, you go to CMS website. And they have like, also like, all those little like, icons at the bottom of the screen.

Yeah. That says like, the woman does the icons.

Yeah. Well, firstly, I got somebody like this website's Adobe nine compliance. You haven't seen that a long time, though. No. I always wanted at backfat when we had like a feature. We weren't done yet. But it was like out. I really wanted to do like those old like, late aughts, early 90s, like our late 90s, construction gifts. Were like this site is under construction. Yeah. Like I shut down fast, though.

You know, I was I was just joking around the other day with a buddy of mine, because we're building a website together. And to make things easier. We started by actually making a sitemap. And do you remember back in the day when Sitemaps were like a thing that you could go to? And it was like a tree that you could like?

You could see. It was like the index of the website. Right,

right. Yeah. Yeah. Whatever happened to those? I mean, like, obviously, they're just not really necessary. But

a lot of websites still haven't. They just don't. It used to be like the way to go find something on a website. Right, right. Yeah. But now everyone, basically, Google, you can like integrate Google into your website, and so that you can easily search your website faster that way. People like to do that instead of searching like an index.

Yeah, you can borrow right to the information you're looking for.

It's like, it's like a, like a modern PDF. Like you have a table of contents. And you might even have an index, but people just go Ctrl F and then type in what they're looking for. And then hit enter a bunch to like, get to it.

Enter a bunch. Yeah. And whenever you do Ctrl F and it shows zero, which Oh,

no, oh, no. Oh, no. I got the wrong PDF. Or it's not text searchable.

Yeah, actually Maybe a hint for something we'll talk about later here. But if you are providing a schematic for people, make sure that your, the schematic you're providing is not just an image of your schematic, make sure that it's something that you can search. So you can pull up like ref Dez. Or you can pull up your, like part numbers and things like that. I've got a story about that.

Let's go right into it then. Okay. It's given the topic three out of three today.

Sure. Okay. Yeah, well, well, yeah. Why not? Because this I was going to do this as part of another one. But you know, it's somewhat separate. So okay, I'm actually repairing a synthesizer for a gentleman. It's a Moog Voyager. It's actually called the Moog, Voyager old school. And it has some some issues on the inside that are preventing one of its features from working fully. It partially works. As in one mode, it does something and then another mode, it should do the same ish kind of thing. And then it just flat out doesn't do that. So I've actually tracked it down to being a CD 4053, which is an analog multiplexer. And it's kind of funny, because in this story, there's like layers of things not working well. Sometimes you get those repairs where like, you can go find the schematic easy, and then you can search the schematic and then you easily find the part that's wrong. And then you fix the part. This is one of those ones where it's like, okay, cool. Search for a while. Do I have the right schematic? I'm not sure it's, it's odd. Can I search the schematic? No. is the schematic easy to read? Absolutely not. And so like, everything has been against me, and everything is like you climb up the mountain. Yeah, exactly. Know, every single thing, like every problem that I'm trying to solve creates a new problem that I have to solve kind of situation. And okay, so let's start with this, this schematic for this, since it's 50 pages long, which I actually don't mind long schematics, they're fine. But this one is brutal, in terms of they it was it was drawn, I think in in 2000, which isn't terribly long ago. But it was still back when schematics were drawn, such that they tried to fit everything within a page size, such that it could be printed, right? But But given today, like I'm on a tablet, or I'm on my PC, or I'm even on my phone, so I don't care if your schematic is 3000 pages long, because I have it in a digital copy. Now, I want your schematic to be incredibly readable. That's number one, for me make it such that it's so easy to read. But these these schematics, we're trying to shove everything in the page sizes, they start looping around, and they start referencing each other different pages and things and it just, it's a nightmare. I hate that. But I get it. That's how it used to be. And, like efficiency with paper matters. still does today. So whatever. But so I find out that this analog multiplexer is is crapping the bed basically, I'm, I'm sending it like I can, I can flip a switch on the front panel, and I can see that the control port is receiving a zero or a five volt signal. And therefore

the it's receiving the signal to switch and it's not it's it's doing that just

not switching. And on one side of the switch, it has a it goes to the non inverting terminal of an op amp that doesn't have a ground reference. And this analog switch is not switching so that that pin there is that like Giga ohms of impedance. So it's just causing the op amp that's post this switch to freak out because like I don't know, whenever some alien in a distant planet craps and like radiation comes around like it shows up as a voltage on this pin and makes the op amp freak out, right? Because we're at Giga ohms where we're at Giga when

alien craps does a butterfly flapping its wings

100%. You see you get it. So this switch. Here's the thing. That's funny, though, because like I'm seeing what's wrong with this. And I got super lucky on this because almost all of these chips are P DIP packages. And they're all soldered to the board, which Oh, they're soldered. They're not there. Except Except for these analog switches. They're socketed. So what's nice, I wonder

if if they go they had problems of lifespan?

Yeah, maybe maybe. The nice thing is though, like this board has multiple of these CD four or five, three, so I just swapped them around and found that the problem followed a chip so it's like okay, cool. There we go. That's really simple. But the funny thing is just to kind of get my ducks in a row, I was like okay, I'm gonna pull up the datasheet for the 4053 because the the I know what my logic coming in is but I don't know which switches activated for low and what switches activated for high. So I go to Mauser or whatever, and I find a datasheet for it. And I searched through this Texas Instruments, CD 4053 datasheet. And I swear to God, it doesn't say, and like, that's the baseline function of this chip is that you apply a voltage to a switch pin and it switches which signal flows through the chip, right? And so it has it has basically a common input, and then it that common input either switches to the X output or the Y output. And in this datasheet, they even have a truth table. And the truth table is wrong. Straight up. And I'm not trying to like I sat there for so long trying to figure out like, what all I want to know is if my signal is low, which is it going to x? Or is it going to why? I don't know the answer to it. And in this truth table, it literally said low, it says it says a something like y or x. And then if it was high, it said y or x, like if the information was actually incorrect in the in the truth table. And it kind of brings up a point where it's like, if you're writing a datasheet, for something that is simplistic like a switch, I think that's like the very first thing you should have checked is like, can you read the datasheet and actually see how this thing works? I actually ended up having to go to a different website where somebody had just explained the function of the chip. And I knew that like this, my signal was either being steered to the X output or the Y output, you know, that's all it can do. I just needed the actual truth table in there. And it looks like Texas Instruments has updated this datasheet and in 2017, or something like that. And they corrected the the truth table in it. But like, that seems really fundamental. At the same time, like

was the old data sheet you were looking at? Because I'm looking at the one on T eyes website. And that one's that one's correct. Yeah, was your old one like one of those like, like, from the dusty tomes of Xerox machine,

the Grimoire of 4000 series like ice chips. It wasn't one of those ones that's been photocopied 10 trillion times? It it is. But I think it was the one from 1998. I don't know like I you know, it's funny, I actually tried to find it again. And I couldn't find it before this, this podcast. So maybe I just like hit that one really unlucky link, the very first time I was looking for it. But But regardless, I think one of the one of the things about that datasheet, that was a bit annoying is, you know, I'm not going to design with it. So I'm going into looking at this datasheet or looking at this circuit that I'm working with, with the assumption that it does actually work like the design is functional. So my goal with going into this datasheet was to find all I need to find is the truth table, so I can see where my signals are being steered. And the truth table for an IC multiplexer switch was like, way, way, way down way past, like theory of operation way past all of the other

looking, it's on page

six, near the end, it's like writing the packet to Yeah. So,

um, but I wouldn't say I would like, like, yeah, it's passed, like, way down there. So like, the first thing you should see, because you always get the first page, which is like the marketing link the marketing page. But you also you do look at that, when you're selecting a chip, like, does this have the features I'm looking for? And then usually the second page is like Table of Contents, maybe maybe revision history, which typical ti ones. And then like the next page is pin pin out. Okay, it's like that. And then it's got, I think the problem is, is they have so many charts in here, that most designers don't really care about. And like, they have so many different ways of like, test, I guess this is an analog chip, they had to test a lot more like parasitics of the chip itself, you

have to know its function across like a range, you know, so the charts help in that in that case.

But the next thing you should see is like, how does this freaking work?

Exactly, like, Thank you. Well, like I was driving myself nuts last night, looking at this. All I want to do is see how your damn thing works. And like you buried that like way down there. That should that should be way up at the top. I mean, I honestly the truth table is not that big. It could be on the second page.

Yeah, I would say I would expect that to be like after the pinout Yeah. And then it'd be like a truth table and then like, all this stuff about like um Now it could be after Electro. Well, the problem is electrical characteristics of this chip is three pages long.

It's ginormous. Yeah. So

that actually just might be the problem is usually that would be in like the top eight pages. But there's so much other stuff going on with this chip. Yeah, that just gets shoved down there. Yeah, yeah. But like, they have like, three pages in here of, of like, we hooked the chip. It's like one of those, we hooked the chip up this way. And this is how it performed, then we hooked it up this way. And this is Oh, yeah,

in order to get the data for this chart here is how our test was set up kind of thing. And it's pages of that.

pages of that. I think that's because when you look at digital switches, they, I think it's the application because it's an analog, it's built as an analog switch. So you would you do need to know, those pair parasitics really didn't be characters really well. Whereas digital, depending on the frequency, maybe, maybe not, depending if you get really high frequencies. Yeah, you have to care. Yeah, but most of the time, it's like, well, as long as it's not terrible. It's going to work. Yeah.

100%. So yeah, I mean, luckily, I was able to kind of dig through it and figure it out. And it ends up being generally obvious, but the worst part about electronics, and engineering is obvious is not good enough, you have to know the answer to the answer and and like if I go to your, your datasheet. And you make a bunch of assumptions, I'm like, Well, this is an old IC, you probably know how it works. No, I don't like I want to, I want your data sheet to be very clear on how it works. Especially when I'm doing a repair job because repair job. I'm not trying to design with your thing. I don't care about all the characteristics. I want to go in there. I want to see how it works. So I can see if it's working properly in the circuit.

Now look at this page. One of the pages this 21 That sheets is interesting.

Do not do this on the stream. What's up with it?

I got flipped the right button on my stream thingy. Layout guidelines. Oh, yeah. And but it's like it's one page layout guidelines. And it's it is use curved traces instead of shame furred or corner. But it also has like, like, at the apex of the turn, the trace width is increased to 1.414 times its width. Which is Yeah, that's a corner like a triangle. No, a 1.414 is like if you bend it that like a rounded corner, it like widens out, right?

Yeah, that's that's the argument with curved traces is their, their constant width. Yeah, throughout the turn. This Okay, so I've seen

the apex turn the trace width is increased to 1.414 times it's worth this upsets the transmission line characteristics, essentially, the distributed capacitance and self inductance of the trace resulting in the reflection. Oh, that's the argument against a 90.

Yeah, yeah. Yeah. That's like the 90 is the worst. I'm using air quotes. Air quotes. And curved are supposedly the best word. I don't know. The arguably, maybe?

Well, whoever wrote this party back in August 1998. Was into it. It was into rounded corners. Yeah. Because that's actually the first time I've ever seen that data sheet. I've

seen those in Texas Instruments, data sheets, they put them in a handful of them. But it's but that's generic layout guidelines, not even layout guidelines for this. I see. It's

like that's the only layout guideline is like, is like, use rounded corners.

Yeah, yeah. It's just random life tips at the end of the day.

Random life tips.

Yeah. Yeah. And okay, so back

to foster meet in the in the with warm water.

Exactly. Yeah, that's, that's actually quite dangerous. So on that on that schematic. So I said that that schematic is 50 pages long, and that's true. But there's one aspect about it that I don't know why it's in here, but 15 of the pages, something like that. 1516 of the pages are doubling the actual size are now right. Yeah, yeah. Back to that synthesizer. 15 of those pages are actually duplicated. So the what the problem is, they don't say why they're duplicated. It's not like these 15 were rev a and these 15 were Rev B. You just go scroll through this PDF file. You You hit those 15 pages, you find some other ones, and then it repeats those 15 pages. And because they're all images, I can't search them. So I can't like compare the two. So I don't know if anything Oh, no, no, they're the same or not, don't ya don't even or it's like a Reb, too. Yeah, I think I think this is a really fantastic case of what not to do on how to supply schematics to people. So I've

opened up one schematic before, it was a really big one. And everything looked fine until I got like to the bottom, like four or five pages, and it was just all the bypass caps. Growth. And I'm like, this isn't the worst thing ever. Yeah. So So for everyone out there least this is what I like to see in schematics is components that get grouped together on the board, stay together on the schematic as well, generally. So when you draw like your part, your active component on your microcontroller, and you need like six bypass caps, and like the crystal, like everything the micro scroller needs to live, should be on the same page.

I agree with that. Yeah, I do it that way, myself too. Just because I'm op amps all over the place. And every op amp I have got to bypass cat for it. So on a on a single page, I will put all the power sections of my op amps in one section of it. And perhaps this is anal. But whenever I lay one layout a board, if if I have an op amp say you're 20 and it has C 15 and C 16. As its bypass caps, you better believe on my board C 15. And C 16. Are our 20. Yeah, I've

done it. That's how I want them. Yeah, yeah. So I don't I'm still I don't really like doing it. I mean, I guess I don't design really crazy like you design way more complicated boards and I do

you actually draw the bypass caps. Like you don't do multi part components, right?

Yeah, I don't typically do multi part quartz. But while I'm talking is like you usually use multiple sheets and stuff like that. Yeah, I'm still like one big sheet, I do separate everything out into like, I will draw a box around a section that's like this is the power filtering. And this is like my, for example, for my pinball controllers, like this is the section that deals with the solenoid drivers. But it's so one page, mainly for like, just searching is faster, like bigger monitors are easier to use than like a smaller monitor and just having everything right there is easier, at least for me. I do No, that's it makes printing that terrible, because you can't actually print the schematic out because it's so big. Yeah. So you can't do that, which is the downside, but I don't print anything out. So it's fine. I do. I think that's my next step in my EDA tool. Evolution, I guess, is to start using multiple sheets. Because it was only until like Steven said it was a good idea to like use symbols for ground and voltage planes, or, and rails because before I would label them, and they'll just be labeled, but they wouldn't actually have like a symbol for like the little triangles, or like the little bar with the flat thingy on it. I just never did that. And I'm like, it's labeled ground. Why wouldn't you it's labeled ground. It seems like I hate that. I actually changed that because of Steven.

I appreciate it. So and you know, it's funny, I do some of my stuff your way now to where like, sometimes it makes sense to just see the word ground somewhere. Yeah, yeah,

it depends. Yeah, I do label it ground still. But I have a ground symbol for now. And yeah,

this is weird. One of the things. You do multiple sheets, though. You know, okay, so like, I tend to put guaranteed like power stuff is on its own sheet. I'll I'll certainly do power on its own sheet. But I might do a lot of things on one really big sheet. And then subcircuits on different sheets. I used to be a lot more anal about very specific sheets with one purpose, each kind of thing. But now I've relaxed a little bit on that. However, like I said, power guaranteed is on its own sheet. And that just makes it cleaner. When somebody is like, Oh, what are my rails for this whole circuit, go to my power page, and they're very clearly defined there. And then like if I have reference voltages or if I have something they're like weird LED driver voltages or things. It's all on my power page. And then I use NET ports to distribute those around. But that brings up one point on that schematic. that I was dealing with, they have many, many ribbon cables that connect, you know, control boards to other board like digital and analog boards. And in the way that the way that the draw them in the schematic is they'll just have like the output of an op amp goes to a port that just has a number, it doesn't tell you the function, it doesn't tell you really the label and like the conductor, but no, they don't label the the connector, but I found out that there is like a schema on there where like the first, like, it might be like this op amp goes to a number for 19. That would mean the 19th pin, but not of connector for that just is whatever connector all the fours go to. So it's like sort of has some scheme to it. But it's kind of bad. But the problem is that like there will be there will be ports that like multiple ports that go to an op amp mixer. So clearly there's like, okay, these three or four or five things all are getting mixed together, but you have no idea which one is which, and they might have different gains. And so the way to find out is you have to follow it back to the connector, follow that connect it to whatever other page, follow that to what it's actually doing. Just to find out that that's not the thing you cared about. It's like, oh my god, you got to do that three times. Yeah, you have to do it three times, and then find out that you were looking at the wrong op amp. And then you got to go do it again. Like oh my gosh, it's so bad.

So meta Colin in Twitch chat says any anything is one page if your printer is big enough, which just means I need to get a big plotter. Yeah, I

was about to say like, yeah, a plotter if your printer starts to change its own name at that point.

Oh, well, I do. So my strategy on that is I generally have, I set up my schematics where the left side is inputs, and the right side is outputs 100%. And then that way, when you're in then everything separated out. So like. So like, you have power and that kind of stuff. And then kind of like in the middle, you might have your microcontroller because it's doing the in between inputs and outputs. And then all your outputs, like I'm putting McDonough, like all the drivers that cars have on the right side of the board. And then even farther on the right is connectors. So like all the connectors go there. So that way, when you're looking at the schematic, you can go okay, if I'm dealing with an input problem, it's on this side of schematic from dealing with output problem that's on this side. And you can see the flow as almost like a pseudo flow chart to and that kind of way. Again, breaking stuff out into sheets will probably work. But I've never actually had a problem with this before. Because I've never I've never had to print it if I have to print something out. You can't

like, yeah, yeah, you'd have to make images and like, chunk them out. Oh, that'd be awful. Yeah. Okay, so one thing I've been doing with board to board connectors. And for the most part, this works, because I'm just mainly dealing with two boards, it gets more complex with more boards. But what I will do is like, I'll define one of the boards to be like front and one to be rear or something like that. And on my front board, say like my front board is where all my jacks and my potentiometers and all my control surfaces are all draw all those. Over on the right side, I'll put all the board to board connectors that are on the front side, and they're facing all those components. And then I will mirror those and put the the board the board connectors on the rear board right next to it. So you see all of my connectors, you see which pins connect to which it's just really clean. And it's really straightforward. You could follow it without having to dig very hard. I prefer that as opposed to like throwing them on a different page. And then you're searching in between now I want my board to board connectors to be right next to them each other on a schematic if possible. I think that just makes life so much easier to read. Or either not life but schematics. No.

Alright, we're done. Yeah. CD 4053. Now,

yeah, I'm done complaining about things. For now.

I get the complaint about things next. So today had an Autodesk Eagle DRC problem.

What do you mean by DRC? Problem

DRC. The eagle DRC can't catch this problem. And technically, I guess it's not really a DRC problem. It's a dF M problem, which is in between DRC and DFM.

Well that's a good thing to note because those the DRC does not mean that your thing is manufacturable

Yes, that the 100% this is in between basically, I was having a, I adjusted some footprints on the pennant tar project to basically I knew I wanted to make them wider on some of the sock on the pins on the pins that I made the pins wider on some footprints. So I could use slightly different manufacturers for the shift registers, the 595 registered seven four hc 595 registers got lots of over the board. Some manufacturers like to make those T SOPs wider, some might make them skinnier. So I just made the pads really big, because it doesn't matter because they're bleeded components and you can just make them as long as you want. So if you want a board that's, you know, five by 11 you have the real estate to do that have the real estate to get with it. So I adjusted it and of course, then what the wider now and so they bumped into some vias, and so I moved all the vias around so I thought, the design out, it passes DRC passes everything good to go. And we're building like, honestly, like 1500 of them right now. And then I'm like, Hey, I have another OEM that's interested. And I want to order the same ones just give me a ballpark lead time, right? While this morning I go hey, there's a weird there's a there's a D F, m slash DF DRC problem with these boards. And I'm like freaking out and I'm like, freaking out about it and ended up not being a big issue though. Basically, an eagle the so they changed this recently. I don't know when it's gonna be last couple years, though. Where the, it used to be vias were always on top of your of your layers. So you would always see vias,

nothing over them. In other words,

nothing would be layered on top. But for some reason, the default view now, the top layer copper covers it. That's weird. It's really weird because it used to not be like that. If you click, I think it might have been the introduction to have a button. Now it's called single sided view. And it only shows like stuff on the top or stuff on the bottom, which makes multi layer designs a lot easier to handle. When you click that button, the vias pop up on top, so something is up with this view is some bug there feels like. So what happened was one of these vias is just under the pad, and it's like almost the same diameter as the pad. So you just can't see it. It's basically to DRC tirely buried in the bed. Yeah, it's entirely buried in the DRC passes, because it's the same signal. So saying it's Yeah, same net, I'd say Yeah, same that. So it doesn't care that it's copper over copper. And what's interesting is when you export Gerber's, and you let's say you use we use at Microsoft, we use a tool internally called fat 3000, which is like a just a big CAD software for analyzing Gerber's. Fat 3000 also doesn't detect this as a potential issue. Because what's happening here is basically you have a via and pad situation, which can be okay or cannot be okay, it really depends on the situation. And a lot of times is when a fabricator sees that they want to either have you move the the fear out, or you want to cap and fill the VO with like non conductive epoxy, so it doesn't. The idea is so that when you put pastes on it, it doesn't work paste away from your lead, right? The good thing in this situation, it's only one pad. And the in those pads are so big, it won't matter if it works a little bit away, it's not going to care if it's the holes are like 10 mil diameter holes too. It's not gonna be that much solder that looks way. But it's interesting that this escaped, I say escaped the eagle DRC factory, I've actually been contacted by a numerical innovations, like I have an open ticket with them to add a way to detect this, which is basically if there's a via or drill in paste, flag it as a potential because that's what's happening is is basically there's a drill in your paste file and then be able to flag it for review. But does any actually EDA tool detect that? Because I was actually looking at all teams and I didn't see this option in Altium either.

Well, I think what you said earlier, is exactly the problem with calling this a DRC issue. Sometimes some Do you want it? Exactly? If you're doing like thermal pads and you're trying to plunge down to the other side of the board, sometimes data sheets even say put vias. Yeah,

maybe is that? Yeah. Yeah. But you're doing that for a slightly different reason. Like for outgassing, and that kind of stuff,

right. But yet how to do an H will know the difference. Yes, knows.

Um, I mean, you could flag it as most you in that case, you should have vn pads or not. Okay, unless you specifically say that this V is okay. Yeah, I

think that that would be, I would be totally fine with that workflow in my EDA tool. So just before we jumped on here, I actually just threw together a quick board in dip trace to test if dip trace would detect this. And it technically does. But like, it's sort of tangential, it's so deep Trace has two versions of the DRC. I mean, it's, it's one version, but it has two different sections of it. One section is just your overall DRC. And then one other section is called net to net DRC. So you can have it where you're running two traces that are the same net, and you can apply DRC rules to them in case you want to space them by a very specific amount and things like that. Okay. And it has one checkbox in there that you can, you can war one like user entry field, where you can control what's called SMD, or, or SMD, two via distance, and it will check all of your vias to any SMD of the same net. Not the same that Yep. And and that will throw an error. But yeah, that's, that's checking for V. And Pat, it does do that. But you have to define it, and you have to turn that on. It's like regular DRC wouldn't give a crap about

it. Well, the thing is, at least you have a option to put that in. Yeah. Whereas I mean, because this thing with DRC, and checks like this, it's a computer, it's going to do what you tell it to do, right? The problem with, let's say, Eagle, in this case, you can't tell eagle to check this,

right? Right, you don't have an option, you see, and what I would do in dip trace is, I would turn that on. And I would check my entire board, including pads that I intentionally did, you know, via embed, I'd fix all the problems, and then I would turn that section of the DRC off and run DRC again, and then it wouldn't find any errors. So like, I have layers of DRC. With that you can go further, you know, with so there's a lot of situations that I actually use that situation, or that I use that in that mode in the DRC because, like I said, a lot of times I've run same net traces right next to each other, but I want there to be a space in between it like yeah, you know, we were talking about last week with like star grounding. And I do things similar to that with power traces where, you know, I want a minimum amount of distance between them because my my PCB manufacturer, you know, if I put one Thau in between two traces, and they're the same net, my most DRC is aren't going to care about that. But your PCB manufacturer is going to barf on that and don't care about it, they're gonna care about it. So this helps with those kinds of issues.

So what do we so in dip Trace has actually an option for this? I haven't looked in. I've only looked at the passing clots in Altium. And it didn't seem like Altium had this feature. Eagle doesn't. I gotta look at KY CAD. Yeah, but what I have this question is what do we do about this? What I'm doing with this as I'm just gonna, I just was like, let it ride. Like, those pads are big. We're using a seven mil stencil. It's only a 10 mil hole. So it's not going to end in the backside is like is or is tented it's not going to like flow a ton of solder through it. It's gonna be fine. It's a J lead component, so we can inspect it, but yeah, why

you're gonna be no big deal. I mean, you should probably fix it in the future. Rev. But you'll already did it already. Yeah, good. Yeah. You know, yeah. Here's the thing. If you tried to do like a Gerber check on this. Your DRC on Gerber's is going to have a hell of a hard time finding this. Because the way the way it looks like in Gerber's is. Pad versus traits versus plane versus whatever. They all show up as just the same thing offensively as QA polygon as a polygon, right. So so, so a Gerber search doesn't distinguish trace versus pad. So so you could you could look at your drill file and see where your drills are hitting, but it doesn't it wouldn't be able to know distinguish between elements. Yeah,

so Mmm, my suggestion to numerical innovations whose builds fat 3000, which is what we use as an internal tool. My suggestion was, hey, if a habit as a checkbox, we can check if a drill hits the pace file. Right? Right, right and throw a warning for that doesn't have to be an error, just like a warning, or, actually the great thing about factor Ks, we can set what levels they are as well. So when our engineers check stuff, or our automated tools checks off, it will only throw like errors. And then it's also warning so that you can just like, warnings are stuff that you can just approve, where errors are like, you can't approve it, you have to fix it. Whereas this would be a warning for sure. Like because it could be intentional, or it could not be intentional. And so that they said there were they would implement that. But it's not in place yet. Like this only got caught because of a eyeball looked at it and was like, that's not right.

Yeah, something's weird here. Yeah. At the same time, with that kind of a check. I think that would I think that would work. But if I'm nitpicking here, you're looking at the drill, and where the drill hits the paste. But it would be nice to see like the edge of the annular ring, not be in the paste. And that's that that gets way more difficult. That's EDA fix. Well, you could just put

like, A, you just put a buffer on it.

Sure. Yeah. Or I guess, if the drill, the drill has to be covered by mask, or the mask layer that's even further than the pace layer. Yeah, that can be

Yeah. So what I'm gonna do about this is, I'm gonna write a ULP. At the Texas,

that's such an eagle thing to do.

I'm actually going to look to see if someone else has written this before. But if not, I actually I know how I would do it is basically you would iterate through all your pads on your board, iterate through all the pads, and then iterate through all the drills inside all that. And basically make sure that none of them intersect. Actually very easy

check. Well, okay, but what if What else a buffer? What if a via is fully within a pad not like partially in a pad? When you do your ULP? You have to like defy me.

So? Yeah, so what's going to do is I'm going to basically take a pad and you know, the size of the pad, and I want to add my tolerance I want around that. Yeah. Yeah. And then it's gonna say, is there a drill hit inside of in that area? Got

okay. Yeah, yeah. Okay, so not intersecting the border, but like, anywhere, it No, I'm

just going and, and my tolerance I'm checking for is gonna be big enough. That it won't matter. Basically, if it's inside of it's bad. If it's if it's overhanging it, it will be okay. Yeah, it's hung myself to the chalkboard, because that means it's also one, a tool that I'm only going to use, and it's just to prevent this issue. Well, and there's also

something nice about it, too, you've been using Eagle for what, probably over a decade now. Right? And this is the first time you've run into it. So yeah. We, we actually ran into an issue twice. With dip trace, I don't know, month ago, something like that, where we had one project. And luckily, this has been contained to one project. But DipTrace just changed something on the schematic that nobody changed at at our work. And, and it happened twice. I think that I think both the changes happened at the same time. We just caught both of them at different times. But it got all the way to we had ordered the boards. And I had noticed I just randomly had to look at that schematic. I was like this nets just not connected. Like we've been building this board for years. It's a mature product, but like a net, just poof, just gone. And so we've been keeping an eye really heavily on dip trace just to make sure that like it. I think something got corrupted in that one file. It's not just one file. But it's also like, every time we ordered that board we're like, Is this the one that we're going to build like 200 of and then find out we have to run traces on or green wire them, you know, from no fault of our own making netlist? Yeah, that's true. Yeah. Although what dip trace allows you to compare any schematic to any schematic, which is nice. So we can we have old versions that we know are good, and we can just do an A B comparison against a known good so we've we've implemented that into our process. Now, like any new board we're doing for the foreseeable future, we just check against an old Rev. Just in case. That's the case. Yeah. That one's scary, though, if your EDA tool is just disappearing,

that is just poof and traces.

But like I said, it's only on one board. So I think something got corrupted there.

And this is just all this was was a different way than displaying the data to you now, because the data didn't change. It's just how I viewed and so when you when I was looking at it, I just missed the fact that the viewer was underneath the bad now,

that's the thing, man, that's the thing about electrical engineering. I mean, I it's the one I know the most the discipline I know the most, but like, we have to manage and maintain 1000s of things. And any one of them is enough to take down your product, right? Yeah, to derail everything. Oh my gosh, like you have to juggle so many things. And like we put a lot of faith in our EDA tools that it's all we do. Well,

yeah. We the fact that you are trusting it to output Gerber's.

Right, right. Yeah.

Which is why I really like ODB plus plus a lot. That file format. I'm hoping that Autodesk moves that way to put that in Eagle. So

don't they have an output already available? Is it I thought

they do? I don't know about it. Well,

I thought they did. Maybe I'm perhaps I'm wrong on that. I've, I've read someone's wrote, like an incredibly extensive ULP that. Oh, yeah, I bet you that exists. It's got to write. Yeah,

yeah, the I do like that is still my favorite thing with Eagle is like I can I can write my own tools.

Yeah, you can. Yeah, you can wrangle it to do whatever you needed to do.

Yeah. I do want to try to install my first EDA tool though. Free PCV. Does this still website went away for a while but it came back. Someone's ported it so it runs on modern Windows. So no schematic editor B. You give it a you give it a netlist. Oh, I write up a manual netlist and then you rock the board.

It has a schematic editor now. Oh, it does. Yeah. It's called the schematic constructor.

Constructor. That's pretty cool. I don't a lot of boards. You know, that EDA tool helped me get through college.

I think I use more AlTi? Board AlTi board, which was boring. Yeah. Yeah, I use that back in the day.

Yeah. But that that EDA tool I designed all my video mods for Atari is on. And I mean, that helped the help pay for my college. That's cool.

You know, I think it would be fun to go back and play with one of these less feature rich programs just to get perspective on what your software does allow for even when like it's not doing what you want it to do or things like you can go back and be like, oh, man, I actually have this pretty easy.

Yeah. Could be.

Interesting.

Should we go on to our Python talk or punt it for a second time? I think we've

actually punted it twice already. This would be a third point, which I'm totally fine with doing because I think we could spend a while on it. I think there's some good stuff in there.

But okay, so we'll, we'll punt the Python talk for a third time. But next. Well, next week. We have

Cody in two weeks. We'll start with the Python talk. And we'll start with the Python thing. Yeah, because there's some fun stuff in there. Yep.

So that was the macro engineering podcast. We're your host spark Dolman

Ed. Steven Greg. Later everyone take it easy.

Thank you. Yes, you our listeners and our Twitch followers for listening and downloading our podcast if you have a cool idea, project or topic. Let Stephen I know Tweet us at Mac fab at Longhorn engineer at analog EMG are emails that podcast at macro fab.com Send us the data sheets that you hate. Check out our Slack channel at macro.com/slack and also the live stream which is twitch.tv/macro Fab