Skippy Documentation

Hello, and welcome to the macro fab engineering podcast. We are your host, Stephen Craig and Parker, Dolman. This is episode 276.

So I have a guess a treat for you, Steven.

Well, yeah, I'm super excited about this. So I sent Parker a package over over the weekend. And he got it this morning, which, so, a handful of weeks ago, if you've been listening to the podcast, I've talked about this once or twice. I had some lm 338 regulators that were kind of suspect in terms of their functionality and made me question if they were counterfeit or not. And it was in your amplifier project that you've been working on for like 10 years? Well, yes. But working on for 10 years is not the right way, because you had parts on a shelf for 10 years. And I picked them up about a month ago. And and I bought these regulators, these, these are not all these regulators I bought, maybe a month ago, I purchased them off of Amazon because at the time I went to purchase them, I didn't feel like buying them from one of the other guys because I didn't feel like waiting the two or three weeks it takes to get stuff from them right now. So I purchased a handful of LM three, three eighths from Amazon through them in this project, turned it on, and it did not function the way I expected it to. It had some really odd results. So I went and actually went back to one of the big players and bought some, some confirmed good LM three eighths authorized distributor authorized distributor. That's a nicer way of putting it. So I

just slapped you the pictures Stephen says this is the first time Stephens ever seen them well. Okay,

so I put these new Well, before I even look at these pictures, I put the new ones in the in the project, and it fires up and works beautifully right away. So what I do is I take to the two bad ones or the two known nonfunctioning ones. And then two of the good ones that I had just purchased that I knew worked. And I shipped them off to Parker because he's got a fancy X ray machine. And so I'm looking for the first time here. Now here's the thing I don't, I don't remember the brand of the bad ones or the brand of the good ones. So I like to think is if they're a different brand, who knows if the die is going to be significantly different or not. But let's let's but let's take a look at these pictures and see

what so I think I'm able to stream these pictures soon. I'll post these pictures on the blog and on Twitter. So the first picture I'm showing here is the is the it's lm 338 underscore zero, Steven. Okay. And it's the

one with the one with the short cut legs. That's that's the one that we're I don't know, bad, I guess you could say,

yeah. So when you take a look at the component itself, like no, you can actually you can't really see like the features on the dies, because we just don't have the resolution with our x ray. But you can see the size of the die. And what's interesting is the die size is about I think I measured it, it's about a third the size in your first part versus your second part. Mm hmm. So that's really telling. And I think they're different manufacturers, just because there's Oh, no, like, oh, the package itself is slightly different, like the legs are different. And

yeah, like the geometry is is different is different. Yeah.

And then so I then took a close up and I jacked up the energy that you can, you know, the X ray energy, so you can like see more into the device. I don't know what the right term for that is. So that's the underscore one picture is the failure device. And what's interesting is in that device, the the plastic housing, or the encapsulation actually has some defects in it, you actually see like that darker spot almost like there's actually something in that plastic that is more dense than the plastic itself. And then the third picture is the same kind of shot but with the one that works. And what's interesting about this is like the die is not even square inside of the package.

Yeah, that's the that's the first thing I noticed. In fact, before I before I even thought like good bad. I see that the die is all like wonky inside of the package. Yeah, it's really

weird. Actually. Next picture,

one thing that's nice on the on the the good unit, you can see the wire bonds really well.

Oh, yeah. And you can also see it in these next photos, like three and four, where I tweaked the contrast and the in the tube voltage a little bit more. And you can really see the wirebonds in those pictures. Oh, yeah. Oh, yeah. Though, you still can't really see it in the presumed fake unit. They are there, it's just a lot harder to see them. And I think it's just the plastic is more dense than the other plastic.

Yeah, I was looking at one of the other photos and you can make them out.

And then I just, and then I took the X ray head, and I, I moved to a 45 degree angle, so you kind of get a side shot. So picture five is the the presumed fake components at a 45 degree angle. And the last picture is the good unit working. And it's just interesting to see it kind of a different angle, because you can see like the the actual 3d geometry or like how the legs go into the package and stuff. So that's kind of interesting. But they are definitely quite different inside that that component.

Yeah. And the thing about this that we can't necessarily like pinpoint or put our finger on is, we don't know the manufacturers of each. The only way to make this kind of like a one to one is if they were both marked the exact same. But the size of the dye seems a little suspect to me.

Yes.

And just the fact that like, I validated that they don't work worth a crap. Yeah, that is neat.

And then I took some pictures of me see if I can switch. The There we go. And then I took some pictures of like, Oh, I didn't send these to you, when you send these to you. It took some macro shots of like the the lasering slash silk screening

all Wait, I didn't even I didn't compare them. They both say national, or they both have the national logo on it.

Yeah, so I sent them to you ones a direct head on shot. And then one is a 45 degree angle shot. So you can really because when you look straight at it, you just get too much reflection to actually see like the laser and but at a 45 you can see all of it really clear. Definitely completely different fabs. Because that's, that's very interesting. I wonder what that J M 20. RP, and OCA KXKUG. Three, some date code or

code stock code? Yeah. The logo is, is different on them, even though I mean, it's recognizably the same in terms of like, you'd look at it and you know what it is, but it has different line width, and is slightly funkier on one of them. Yeah. What's

the most interesting thing between these two devices for me, though, is like the chamfering is completely different on both these parts. And if we assume let's say the left the left ones, the one that's working for you right now. Yes. Picture. Like if you were just to hand these two to me, without me knowing I would say the one on the right was the real one because it's got more features. More complex, it looks more complex. And the was one more thing that jumped out to me, I can't remember anymore, though. Oh, come back to me. By I just thought it was interesting that oh, that the the fake one. It's definitely it's not been. It's not a LM what three, three, what's the what was the small 1317? So he wants to it's not one of those just like paint it over and re lasered or like sanded down like it actually does look like that's the original surface for that component. Right, right. So I'm going to bet you that component actually was counterfeit from the get go. Is someone put a different dye in there. That is not a 338 dye.

I'm wondering if a quick Google search will show the LM 338 Die let's see here. Like because the the one that the Bad one in the X rays is a square dye. And the good one is rectangular. So if I do a Google search, am I gonna find rectangular dyes or images? Now?

We definitely know they're different. I don't have a way do we want to like send them off somewhere to get the kept for sure.

I think that'd be really fun. I would love to see, like, there. We know for a fact there's a difference. But we don't. The part that I'm not sure about is did I just get bad parts? Or are they actually wrong? The die size is completely different for sure. Yes. So I would say it is a different part. For sure. What would be interesting is to get the 138

and do an x ray that dye matches. Yeah, that would be interesting.

Yeah, yeah, yeah, we should we should X ray of fresh 317. And see if it matches the die of a of the battle in here.

Let me put that in our notes so that I can go in next time I order some parts. 138 to 220. Yeah. So that way, I remember to order those parts and then an x ray that and compare it and see if that dies, actually smaller, like the one that you have.

So I found one shot, that's just a Twitter post of the internals of an LM 338. But the style that comes in one of those big to three cans, and all the die in that is rectangular, not square,

not square. Now do the same thing with the 138.

Okay, yeah, yeah, let me see here.

Well, okay, so it's lm 317, is what three ones I've been looking for. The images of that look rectangular as well, not square. So who knows? Maybe you just got some random garfi. It could be different die. Yeah, it could be a different day, or it could be something entirely different. Who knows? Now we do have to get decat detail what it is. Because a lot of times you'll get

a bunch of there's there's dyes out there that do very similar things to LM, one through seven and 338.

In fact, okay, so I'm here, I'll put it in Slack for you, Parker. And I put a link up in the Twitch chat. It's a an Eevee blog, forum, link for voltage regulator dies. And there's a pretty good picture of an LM 317. Right at the top. It is rectangular, it's rectangular, yeah.

So what is what is yours?

Well, and what's interesting is okay, so if you look at this, this picture from the website, there's four bond wires inside. But if you look at the thing, you X ray, there's 12345678 bond wires on the inside of it. So that's interesting. I mean, I don't know why.

Yeah, I wonder why.

I mean, a lot of them are duplicates.

Probably. Oh, definitely the carry the current if they're, well, you got to think as those are inside cans, so they can probably get away with a larger bhanwar. Like he actually looked those pictures like the bond wires are ginormous.

Yeah, they're, they're pretty fat. You're right. Yes. Your hair looks like. Yeah, so far from this good one. There's at least three bond wires that go off to one pin. And that's got to be where the five amps is coming from. Yep. Yeah. Or go into I guess.

Cool. All right. So we will definitely order some LM one through seven just to compare x ray wise, and we'll just ship them off to someplace to get decat

Yeah, I'm very interested to see this. And then I guess we could just post that one Amazon link. Just say don't buy there.

Don't buy those. Don't buy don't buy electrical components from Amazon.

Didn't you just do that the other week?

Actually got cancelled. Oh, did it really? That didn't have any in stock.

Parker was making fun of me for doing it. And then he texted me like the next day. He's like, Oh, I bought some Yeah, no,

no. The thing is, though, is I didn't buy him from some random eBay seller, or Amazon seller. Okay, it was some random person, but I like looked at the seller. And I found their warehouse and they ran like a distributor thing. Got it. Okay, so at least it was coming from a place that sold parts. I didn't feel that bad.

Yeah, yeah.

But yeah, like it. I ordered them and it's sat for like two weeks in And then they canceled the order. I'm like, thanks. Good, like, call me or something. But okay.

That's, that's great. Well, okay. And like, and I think it's really funny because the entire reason I bought these off of Amazon was mainly because right now I'm super spoiled. Because when I lived in Texas, I could order stuff from Mouser at six o'clock in the evening and have it by 10 The next day for regular shipping. And even now that I've moved to Colorado, both Mouser and DigiKey. If I ordered from them, I get it in like two days, but

because of a lot of circumstances that are going on in the world for the past year or so, it takes like two or three weeks to get anything from Magic Digi key. And I was like, well, I'll just they're just, they're just regulators. They're Jelly Bean regulators. I can get them from Amazon. And now it's like, a month and a half later. Yeah, except that now you got half a jelly bean regulator. Right. Yeah. And I still had to go back to those guys and order good ones

are good parts. Yeah. Yeah. Cool. So I've been doing a lot of code recently. Oh, mostly with Python scripting, but talking to test bench equipment, like multimeters and power supplies and that kind of stuff. Are you making automated test fixtures and stuff? Yeah, doing more and more of that. And usually in the past, I would like leverage, like, that test program that came with that companies device or whatever. But this is I'm trying to integrate even more into programming stuff now. Just to reduce like, I, you know, me, I want to do like the least amount of repetitive work. So I will front load as much stuff as possible. That's

the whole thing about engineering, you will spend 100 hours designing a thing that saves you one hour.

Yeah, but that saves you that one hour, like 1000 times.

Right? Exactly. Yeah. It seems like you're doing a bunch of useless work for a very long time.

Yeah. Then you press one button, everything works. Yeah. Or happens.

I had a long chat with with an engineering manager about that a long time ago, where they were saying, it's really hard to evaluate an engineers work, because in a lot of ways they can work for an entire year, and get what looks like nothing done. And then maybe the next year, it's the most amazing stuff ever. It's just really hard to sit down and say like to put a mechatronics engineer and be like you did good or bad. You know?

I like how someone in chat just said Oh, no to my topic I'm bringing up

I think it's because I said oh no to you code.

So it's it's a protocol called es si p i or the Standard Commands for Programmable instruments. And if someone out there listening knows what that is, they might have winced a tad. Because it is, it's actually not that difficult of a setup. The problem is documentation on it is? Well, because so it's a standard, it's a it's also has the best like it's SCPI or Skippy, which is of course

it has to have some kind of goofy now.

And so basically what this is, is you send command strings over serial to configure and get results from test equipment. And the trick with it is there's a standard, so like, depending it doesn't really matter what brand test equipment you have, if it supports Skippy, you are good in theory. It seems that a lot of companies just have slight they basically say yes, we support Skippy. And then it's like, well, not really, or they have their own flavor Skippy, like chunky or smooth. For those that don't know, Skippy is a brand of peanut butter in the United States.

So like how universal is this? Like, let's say I was a guy who wanted to make a test program or a test hardware, and then I wanted to adhere to Skippy like how universal is it?

It's so far fairly universal. I've actually talked to today today was like my, like, I've been doing a little bit of research this week in yesterday or not yesterday, last week about this stuff. And today was like I sat down and wrote a lot of code for it. And I was able to talk to like four different manufacturers. And it's like, nice, the all their stuff is like 99% the same. It's not 1% gotchas. And then always mess you up, of

course that each one has, right? Yeah, yes, exactly. And is it just like a serial protocol that you

it's a command, it's a command protocol, like, measure current, DC. And then thing. So like that command structure is the same across anything that's support Skippy.

Got it? Got it. So there's probably, like a cheat sheet for each unit where like just says, you know, this command, does this, send this protocol or send these words and you'll get stuff back.

Yeah, you. And the, yeah, cuz in the PDFs are the data sheets, like the big data sheets, like the almost like the user manual, for your test equipment, that will have a section of like, here's all the Skippy commands. Now, if you are an engineer out there that works on test equipment, the most important, this is why I keep hammering with, like, when people are documenting stuff for their software side is please give examples. Let's like the number one thing, especially for like other hardware engineers, trying to figure out that don't live and breathe. Skippy, or any kind of protocol is like, Give an example that a engineer can just like copy and paste, and it works. Just so they have something to build off of. That saves lots of because you don't know if it's like a syntax error that you're running into or something else, or like you don't have something configured, right? If you're like, Okay, if I copy and paste this thing, it doesn't work, then it's not the command I'm sending. It's something else I can look into. Yeah, yeah, that's useful. Yeah, cuz that would be nice. Because some test equipment, like you send it a command, and it doesn't do anything besides doing it. But a lot of times, you don't know, because it doesn't have a display or anything. Right. Right. Some some devices will actually like, say on display, like bad, bad command or bad parameter or something on it. When you talk to it. Those are nice. Like, I think a BK precision, Digital Multimeter I've been using will actually let you know on the unit itself when you don't send a valid command. But like I have a 600 volt power supply, that it just doesn't return anything. Like when you send a command, it just doesn't do anything. It just does it. But it doesn't let you know it did it or it was invalid or anything.

Yeah, so is. Can you set up skipping to do streaming? Or does it have to be one command one answer?

So like, if you want to get I'm assuming you want to like, pull like voltage?

Yeah, it's like, to me the next 100 readings or something like you have

to continually overhaul it. Okay. Yeah, you can easily do that on your end.

Right, right. Right. I actually kind of, I think I think I prefer that a little bit in a test situation, just like, just to make it more simple, where you're just saying, instead of like, I'm establishing communication, and I want to break it after a certain point. Telling It Like, being very explicit, say, give me a reading, get it back, give me another reading, get it back. Like, I kind of like that, just to kind of keep everything organized in a way.

What would it be synchronous is the word you're looking for? Well, I'm

extremely synchronous.

And so what's the Skippy command? This is a question from chat. What is the Skippy command to release the magic smoke it is a set the high voltage power supply to 600 volt and then constant current three amp and then power enable

divided by infinity I am curious I would love to see this a picture of this 600 volt

actually looks like a one you rack mount.

So it's probably not a linear guy. It's probably a big no it's

a big switcher. Yeah. Cool. switcher.

That's awesome. 600 volt, three amp. That's, that's beefy.

Yeah, it's a pretty beefy power supply. The interesting thing about the is the first time I've ever done any testing that's over like 48 volts. Yeah. So there's more considerations of like safety and stuff, like making sure 300 volts, or 600 volts isn't like anywhere near the operator, right? Yeah. But it's actually when you if you have the current limit too hard, or set too low on this power supply. It can't ramp up fast enough. Or it can it just takes a long time because we enable the output. It's got a ramp zero to 600 volts. And if you current limit, its current limiting those, those output capacitors and that's very You'd think So all I've done is like if I need to ramp really quickly, I'll set the current limit high ramp and then set the current limb low. Once the after ramping is done. I wish there was a way to like it would just automatically know that it's got some slew ratings. And I don't know if you can basically set a slew in there, like, say, I need to go from zero to 600 in this timeframe. And if that basically adjusts the current limit to do that, though, we'll see.

That's cool. That sounds expensive. Let's just put it that way. The power supply? Yeah,

um, I think it was like 800 900 bucks. So for a calibrated piece of test equipment,

actually, that's not bad. No, that really isn't.

And the documentation is okay. On it, Skippy commands, ran into a big issue with it, where like, it requires a certain kind of line feed at the end. That is not in the command documentation part. It's like at the very end of like the Skippy documentation. He's like, like, when you open up your manuals for all these devices, like they go, here's the command structure of Skippy. So you can like reread it, and like almost all the same. And usually in that will tell you like if there's a weird line feed or carriage return that you have to give it to like so it knows that you're done given a command. Yeah. And it didn't have anything special. So I'm like, oh, it's probably just the normal line feed stuff. No. So it just would never respond to Skippy commands or USB in general. And then basically, after reading the entire documentation on like the last line, it tells you what it is. It's like, why is that not like the first thing to tell you?

We've been talking about documentation pretty regularly now both both here and in our Slack channel. And I think it's almost universal that people are people just think that documentation is almost universally bad. Like you've you see way more examples of bad documentation than someone being like, this is good. Thank you. This was easy.

So it's actually just jumped into that is the this is a question I posed into Slack channel and Twitter last week. And it's it is bad documentation worse than no documentation.

I saw a lot of people saying yes, to that answer.

Or to to the I would agree yes, as well.

I think we need to qualify, like what is bad documentation? I mean, no documentation. We know what that is. What is what do we mean by bad documentation?

Because actually, I would say there's two things that are bad. Well, two different categories of bad documentation. Because one is, you read it, it looks like good documentation. But it's not correct, but it's bad. So that makes it bad. And then there's also bad documentation, which is when you open up and you have no idea what's going on.

And Mobius strip T's in the in Twitch head is has it perfect, right? They're incomplete documentation versus incorrect documentation. Are you not given all the information to complete your task? Or are you given information that will prevent you from completing your task based on the documentation?

So the most of what I see is incomplete documentation. It's like, if you have more if you have innate knowledge about the product, and you get this procedure, or this documentation, you can perform it because you you can figure out the gaps in it. And the reason why that there's gaps in it is because whoever wrote it, just didn't think that was knowledge that someone who had no idea what this thing is needed. But the problem is when you get incorrect documentation because it on the surface, when you read it, you're like, oh, yeah, this looks complete. And it looks fine. That is the worst though, because you start actually doing the procedure, you know, like three months later, and nothing works right at all right? And then you spend four months rewriting everything

that you know, that's one of the Gosh, that's that's the worst. Like, when when a great example, I inherited a project a product at my my first company in terms of like, I was the manufacturing engineer over it. And we had a test program that was written in VB, whatever, like, bazillion years ago, it was not documented. None of the code was commented. There was nothing and it was held on one computer In the manufacturing on the manufacturing floor on a spinning hard drive, so, and that was our number one product, that was the product that we sold the most of. So it's like if that computer went down, it's kaput. Nobody knew anything about it, the engineer who developed the entire testbed was long gone kind of thing. And, and I was like, there's not a solution here. That is, that is nice, because it would take probably longer to go and dissect all of this than it would to go and just write a whole thing from scratch. But that's a great example of, there's no documentation, even if you have the actual thing. Like we had the program, and it was functional, but there's zero documentation. And it was like, luckily, we caught it. And we were like, Oh, my god like this, we need a backup software of this thing, especially if you're making multiple millions on this one product. And that's the safeguard you have is a spinner hard drive on a computer in manufacturing.

So yeah. So we if we could rank them, like, the worst thing is, is incorrect documentation, then incomplete documentation, and then just no documentation.

I want to throw a monkey wrench in there real quick. Because no documentation could be worse. And the reason why I'm thinking it could be worse is because there's the expect expectation that there should be some documentation whatsoever. So if you can't find that documentation, now you have to ask yourself, why is there no documentation? And should I go on this wild goose chase to try to find it? Because it shouldn't be there? And is the documentation behind a paywall? Or is it really hard to get? Or do I have to sign up for all this garbage to get the documentation, you might waste a bunch of time trying to find it. So that might actually make it even worse to have no documentation?

This is a really good comment from from the Twitch chat, which is craft labs saying I love digging through old code to figure it out. And he calls it digital archaeology. So, so why people wouldn't like that craft lab is it's not so it's not your job to understand or to figure out the code. It's your job to use it. That makes sense. Like, I look for my example is like I see, at macro fab. I see like hundreds of products in like a month. I should not I don't, I can't understand 100% of each product. Just impossible, right? So having good documentation about a product, so that I can build it correctly for them is very important. But most of the time, I don't. Basically if something has no documentation, I ask the customer, is there any digression? That's the great thing for me, at least Steven is when there is none. And I ask I just go straight to the person who gave me the stuff originally. So I don't have to go like looking to see if there's a paywall, or anything like that. And the weird thing, though, with no documentation, though, is at least I know that I don't know anything about this. I think there was a comment on Twitter. That is pardon my French for this, but with no documentation, you know, immediately you're fucked. I'm like, Yes, I like that a lot. Because that's exactly what it's just right

away, you know, yeah, you know, right away, and

you know, what you're getting to, and you can charge the right amount of money to basically build that documentation for that customer. Yeah. Whereas, you know, eight months down the road, you built a bunch of product, pull up the documentation, and you start reading it, and you're like, oh, no, no.

Well, and there's Okay, so there's different grades of documentation as well. Because, like, Okay, so there's documentation that just gives you information about the thing. And then there's documentation that gives you information on how to use the thing. And then there's documentation that gives you all the nitty gritty, gritty, nitty gritty details of the stuff about the functionality of the thing, what most of the time when we're talking documentation, we're talking about that last one we're talking about, we want the how the heartbeats inside of this device, um,

theory of operation,

right? Yeah. Like, I want to know everything about it. But take, for example, my testing team, we were just talking about a document from one of our clients today, and they were it was it was a testing document and it was basically Like, you know, power up the unit, and you're supposed to turn a knob, and some LEDs will do some things. And I'm paraphrasing there, but it wasn't that much more in depth than that. And the result was, hey, that's fine. That's what the customer wrote, that is a test that test plenty of things. You know, it tests that the processor is reading the pot turning, and it's it's doing something in relation to it. We don't have to know all the nitty gritty of what's happening in there. In that situation, that documentation was plenty enough right there. But then again, I've also had plenty of documentation where it's like, you know, just make sure it works. And that's always maddening.

So going back to, that was lowered attend a planned tangent, I guess. Because going back to Skippy is basically the kind of the problem I'm running into is with, like, if you're doing an embedded code, like one of the things you do is like, you make a serial port, and you like spit stuff out over the serial port. So you can read what like your microcontroller is doing and stuff. I mean, you can have a debugger and stuff to like, see memory values and register values. But a lot of my lot of you a lot of your debugging is like printing to your serial terminal, see what's going on. But going the opposite way when you're making like code on your computer, and then like sending out serial terminal commands to a device that you don't really, it's, I mean, like it's a black box, you're sending commands it sometimes respond, sometimes it doesn't respond and you don't know why. It would be nice if I was if there was a way to do like a comport like Sniffer, or maybe even a like a spoofer. Like it makes like a virtual comport that you you can send stuff to and like just reports what it's getting. Oh, I'm sure that exists, right. So if anyone out there knows something like that. That's like, preferably open source. But if it's a good enough tool, I would definitely pay for it. Because that would I'd probably save money like in time. And like the first time I used that piece of software. Yeah, like, basically, because I have a a piece of test equipment that doesn't have the communique. It doesn't support Skippy, but it has its own protocol. And so I can talk to it, and I can get it to, like I can get to identify, so it will spit back what it is. Most have like a command called ID N question mark, like identify yourself, basically. And that works. Actually, that's part of the Skippy command. But all the other commands don't work. And I'm reading its documentation. And it's in its like, serial string structures kind of funky. And yeah, it'd be nice to make sure I'm actually sending the right stuff. Because also like I also have to make sure Python is like structuring stuff correctly. Yeah, yeah. It's a Python is not a very Type Heavy. It's like, it's like type like your values, like type agnostic, basically. Right? Like they can be anything. Where unlike C C's like you made this thing char it is always going to be a char unless you cast it.

And I will throw 800 warnings when we compile. Yeah.

So I wouldn't be nice to have a piece of software that was like that. Well,

what I see what I think is might actually be really useful in that situation is just a thing that's listening. And as soon as it gets a command, it flips it around and spits the exact same thing back at you. So you can just read it, and you know exactly what you said.

Oh, like, well, that's like an echo. Yeah, but I'm more looking like my computer does it USB cable that goes into this device. It's doing a emulated comport and just a piece of software that's like looking at that comport and is able to just say what the

that's effectively what the buffer is just a sniffer effectively, yeah.

Someone recommended the bus part. But that bus pirate actually is for hardware, like the actual read hardware SIG signals. I don't need a box on my USB cable like it should be able to just read the registers on the computer to see what's going on. Someone's recommended Comm port monitoring.com somewhere I get my free credit report. I'll check that out to the podcast. I'm just going on with with the test stuff today because it's good. These came in and these are Our EZ Hook installation piercing wire clamp skew 8507. Yes, we're bringing back the tasty chips segment.

Except there's not a choice so what is, is this something you clip onto a wire and it bites in?

Yeah, so it's got, it's gonna be really hard to show over the stream because the resolution, but it's got little teeth in there. Oh, there it goes. Yeah. Oh, cool. So you squeezed down this the spring loaded clamp, and you put your wire in there. And then when you release it, it, the needles pierce the insulation. And so you don't need the strip wires or worry about or worry about, like insertion forces for like IDC style cables. So this is useful for like testing products that just have a wiring harness, just that you have to hook up to the wiring harnesses test. And so I found these last week. And they came in this morning. So they the the getting their Mobius strip T's that's a user in our in our Twitch chat is it's what are the specifications. And the specifications are really, really interesting in datasheet. Because the max current is one app, you're like, Oh, good, that's fine. And then max voltage zero volts. I'm going to guess the reason why is because the company doesn't know like, I mean, like, what have you put like a, a, a wire that's coded in like, let's say Kevlar. That's, you can't slice through it. Right, Kevlar. I will know if they make Kevlar insulated wires, but if they do, this wouldn't work on it right? Probably. So they probably can't. They don't know how well they're going to pierce through the insulation. So they don't really have a voltage rating. I'll put this way though. It goes into like a 1024 stud. So you can probably put a lot of power through that's done. The trick is going through the little teeth.

Well, I mean, really like the, the voltage rating of this thing is just whatever you put it next to like the installation of like, where it's at in the system, right? Yep. Oh, so this this comes from easy. hook.com That's, it is easy hook.

Yeah, that's the they have a lot of interesting, like, test hook up stuff.

Oh, yeah, you can get them yeah, you can get different thread sizes and everything. That's cool.

Yeah, but that company makes a lot of really interesting test equipment like fixturing and like clamps and stuff like that. Hmm. So I would love to see pretty excited these What was that

I would love to see what people have done like just see these in other applications to see what creative things done with them.

So I'm going to set up five of these and they come in color coded. So I'm going to set up five of these in a row and this fixture I'm making and what I'm going to do is actually 3d print a cover that goes over it all that just has a hole in it so that what you do is you would and these buttons are exposed just the tops. So you press the top of the button and then there's a hole that you just slide the wire in and so that it will be able to have a stop. So you when you slide the wiring it will only go in like a quarter inch pass the teeth and then release the button and it will clamp down and hold it or you now as you think it have a way of eventually like just put like a big like pneumatic actuator so you can like a foot pedal stop down actuates all these down and then you stick your wires in and then lift up lift your foot off the pedal and unlatches or latches depending on your point of view I guess huh?

Yeah they have a lot of stuff here uneasy hook these are cool I'm gonna have to go in look through yeah

it's they're pretty cool website got like cool product.

What is what is your application with that? You said you're testing harnesses What are you actually trying to test? Are you trying to test like connection or are you trying to test actual signals on there?

Oh, I'm pass I'm gonna passing I'm gonna be passing 300 Watts through this thing that's so I'm gonna be putting it through its paces. communication, communication and power. Basically. Nice. Should be fun. It should be fine.

This will be fine. Don't worry. Got it cool.

So I got one more thing before we jump into your adventures in plastic molding because I'm really interested this hear about that. And it's an update to emeritus. It's an update to the amplifier. So I got the gross coating or gross vinyl off of it and I started vinyl wrapping the enclosure. Yeah, looks nice. So this is the red, red orange ish. I picked kind of matches the original. I liked and I have a cream. That's gonna go on on this side.

Super cool man far do that tonight.

Looks great. And then stop electronics back in it and start rebuilding that part.

Yeah, I'm looking forward to that. That part will be fun. I mean, you're already playing stuff out of it. So it's basically

yeah, we definitely need to recap it though. And because it did shock me. Death cap got me

up. Do you know how many it's not particularly high voltage, right. What? I don't remember what the what the voltage is in that thing? Oh, 120 volt. Well, I mean, like the the B plus. Is it like 300 400? Yeah, it's around there. Yeah. Okay. Okay. No, I think it was like 170. Is that what you got tagged with? Or did you get hit with like, ground current flowing?

Um, I think I got hit with the ground current. Okay, back to neutral line. Yeah. It was very low, low current, though, because it did not. It definitely did not hit me as hard as as a normal wall socket would have. Okay. Yeah. So it was it was a very uncomfortable feeling.

Completely different from your, your sparkplug experience, right?

Oh, that was definitely the worst. No, I was the last time I ever got shocked was the sparkplug thing. And so this was the last this it's been five years since last time I got got a little little tingle in the fingers. So

I got hit by Mainz. Two years ago, I was I was working inside of our CNC cabinet at work. And I had not switched it off. And I was adjusting something with a screwdriver and I it but it bumped into it and it hit the mains. And like I got that little you know, means when you get when you get shocked by me, I don't look okay. I don't, I don't want to reduce it. It's incredibly dangerous. You could die from it there. I did my spiel there. But like when you get hit by means it kind of you jiggle a little bit like you feel jigglin Right. And, and I've been I've been hit by it more more times than I care to in real estate. But like it hit me and I was like, oh, yeah, that's means, you know, turn it off. Keep working.

It's not the voltage, it's the current, right? So fortunately, most of the time for wall voltage, this is just generalization. And no, was it Don't do this at home. Right? is generally your skin has enough resistance, so you're not getting a lot of current. But that can easily change if you're a little sweaty. So all right, let's uh let's talk I would say it's not the volts it's the amps. It's true, the voltage allows it to go into you, but it's the current that actually will make your your your ticker not working. So

let's, let's talk about plastic molding. Because I've been working on a project since about last summer. It's been it's been generally slow, but we've had Okay, so the premises we have a product at work where we use some switches that are some nice feeling tech switches that have an integrated LED in them. Little J led guys and stuff. They're nice switches however we found greater than we care to deal with failure rate on these.

Oh, is this? Is this going back to oh man was like a year ago is definitely pre COVID The snap switches? Yeah, yeah, it is that you were you're replacing tax which is with these snaps which is with an LED

it's it hasn't been over a year but yes. The but yes, read over a year. No, I think we started out last summer. We've been talking about it for a long time. Let's just put it that way at work. We've been executing on it a lot sooner than that. So the Yeah, so we've been, we've been working on on getting this, this up and running. And so I, so basically, we have these, these tax switches that we've been putting on our units, and we're like, so we're just not happy with their results we get we get higher failure rate than we want. So we looked into doing some snaps domes on on PCBs as a substitute. And basically, what it boiled down to is we I've developed my own plastic actuator, for this, and we finally gotten to the point where we're getting them injection molded. So I kind of wanted to talk about the process that I've gone through in the past two or three months to kind of give an idea of, for anyone who's not been in who's never done plastic molding, to kind of get a feel for what, what it's like, and, you know, everyone's mileage is different. But with this one company that I'm working with, they've been absolutely fantastic. So here's, here's sort of the breakdown of how it works. I met with one of their engineers, and I showed them a 3d printed widget that I developed. And I gave them my whole spiel about I'm an electrical engineer, not a mechanical engineer. So I develop the thing that in my mind will work. But I want to work with you to make it work a lot better, let's just put it that way. And so the the way it goes, is that I provide the model to the company, and I get an initial review. And that's that meeting that I get with those people, they then go and have a big DFM, just review of everything. And that DFM includes a lot of things that are like this, this particular facet or feature of your your plastic will, will work well this other feature will not here's here's a report of like shrinkage, here's a report of all the areas where we think that we can hold the tolerances you're asking for. And then we have a discussion on things like, Where can we put ejector pins? And where can we put like, the actual gates that inject the plastic into the part, those kinds of things. So you get this whole report, and you go back and forth on that. And once you've kind of agreed that like, okay, these are all the ways that we can do things, and I can accept all of this shrinkage here. And there, you've released a prototype. Now, the goal here with what I'm making is to have it actually manufactured in the United States, in fact, manufactured here in Denver. But the mold itself is made overseas. So the process goes, we make the mold, we get prototypes, we validate the mold comes over here, and then we run production here, just it works out a lot easier, saves time. Well, originally, the plan was to do the mold, excuse me do the mold here as well. But just it didn't work out that way. Everyone's so swamped right now that it's like okay, well outsource the mold, making it but do production here. So after we released a prototype, it's about five to six weeks, I would actually probably say eight weeks, from the time that you release to prototype to the time that you will actually have your first prototype in your hand, because they have to actually make the molds for it. Now, depending on what you're getting shot it you know the size of it, and the complexity of it can change all of those things. Also, the end result of what you're looking for, like if you're wanting to make 10 trillion of the year little widget, they're going to do different mold sizes are different number of cavities in the mold or whatnot. For ours, it's small, and we're we got a small mold. Because our initial run we want to purchase 100,000. And we'll be doing about 100,000 every year. So I got a small mold with four cavities in it. So for this initial prototype, we're actually machining only one of the cavities. So I get after about five to six weeks, like I said, probably closer to eight weeks, I get my first prototypes, I get to validate them, then go through everything. And I we make any changes that are necessary to either the material or when we modify the mold if need be. And then we do a second round of prototypes. So you can expect another two weeks there may be two or three, and maybe even a week beyond that for shipping and all that. And then I get my second prototypes. And once I approve all of that, then the mold gets sent over. So probably another two weeks or so there. So all I'm really getting at is from the day I I said yes, we're good. Like the day after we approve the DFM report. You're probably looking at two to three months before the mold comes back and you've approved everything so you know, keep that in mind. If You're like trying to get to production in three weeks, probably isn't going to happen like that. And, and so after that, you know, the the mold comes in, and then you have to schedule how and when production actually works kind of thing. So just a general idea here for this device that I'm gonna get. In fact, I've got some pictures up. I don't know if you're able to share them with Twitch Barger the pictures, I got pictures of the mold and then pictures of the first actuator that I designed the first shot there, see if I can. It's basically a small plastic. I don't even know how to put it, it's like a four leg device that has a spring on the inside a three arm spring on it with a center pole. It's all shot from opaque plastic. And such that we can shine an LED through it. So our snap domes that go on our PCB have a hole in the middle and we we install a 1206 rear mount LED under on the backside of the board. So we have contacts on the top side of the board and LED on the backside of the board. And then this actuator fits on top of that and and actually it's just the snap dome. And then we can illuminate the center pole of this. And the goal of this was to make an actuator that is single one actuator per button kind of thing instead of arrays because we want to use this on multiple products. And just based on what the product is, you can just snap it into, you know, different local different locations because we have some, we have some models or some products that just need one. And then I have one product that needs 93 in an array. So it's it's just, it's kind of crazy. So we also designed it such that it kind of has, if you look at the profile, or if you look at top down view more like it kind of has a little bit of a football shape, such that if you put these in a four up array, you can stick an m two standoff right in between them. So it allows you to screw together there. So overall cost here, just kind of as like a general idea to keep in mind if if injection plastic is what you're going for. Back to that the the mold cost in total is is 10 to $15,000. That includes all of the all of the ancillary things, all of the both of the prototypes that we go through the machining, and the shipping and everything. So the thing about it is, the previous solution we had with all of our tax switches, we estimate was about for just the switches with ended up costing us about $35 per unit, when we're making in quantities of about 200 for this product, by moving to a plastic solution like this, it's closer to 23 to $27 a unit. So with the snap domes, with the snap domes, and with you know placing 93 actuators in there, and so and

so besides the monetary improvements, basically increase it be able to increase your margins. It's also reliability, right?

Absolutely 100%. And so we found with those tech switches, we had to implement a new test in our production, just to catch for, you know, when you put 93 units down on a board, there's a chance that that one of them is bad. And you multiply that times 200 units, you know, you're talking about many, many reels of these switches going onto these units. We run into bad LEDs before they even get to our testing department. And so you we didn't even factor that into that additional cost that I put up there that $35 A unit, but we had to rework or switch or something rework the switch, but even the cost of doing a full extra test on every unit. So it's sort of been a heartache in that sense that I mean, that's why we're spending all the money going through and developing our own switch for this kind of situation. And the thing is like in our world, the feel of the switch is super critical. Super important, super important. And we want this thing to feel like the tax which I've spent a considerable amount of time like adjusting geometries and stuff just to make sure that it feels right. Because if you buy one of these new units after all of this is implemented, and you had an old unit we want them to feel about the same. Yeah.

The like to you when you say like failures of switches like just like after assembly. Some people might think that's really weird, but like at before Mac fab at Dynamic reception. We had a device that had five tax which is on it. So when you when you have a normal tax which and you're just sticking with like the normal six millimeter style tax, which those are pretty reliable, but the moment you go off into a slightly different kind of tax, which, like for some reason it feels like the defect rate goes through the roof. Because we were using tags, which is the head of really long stem, and so that they could poke through the enclosure. And like when I started there, there was boxes and boxes of these units that just had failed tag switches. And one of the first things I did was went through those all those boxes and just reworked those switches basically turned, you know, scrap into gold.

Yeah, but it took probably took a considerable amount of your time to do

mostly finding the one that was bad. But it didn't take too long, because you just nuke the switch off with heat. So Oh,

yeah, yeah, the thing is, like with our application, we have in that one product that has 93 buttons, they're in a grid of point four inches by point four inches, you can't just hot air one off, you're gonna melt six of them.

If you do these are they're pretty spread apart. Yeah. Well, I'm excited to see like, this whole saga of the domes, which is injection molding, like like the final frontier, kind of like picture of it all stacked together on the PCB. Now,

I don't have these in hand yet. They will. I'm supposed to get them later this week or early next week. And I've developed what I call the button masher. 9000 which is it's a it's a fixture that has some solenoids on it. And I'm just going to put some, some like soft foam pads to simulate a fingertip on and, and I'm just going to have the solenoids hammer,

you're not going to go to like, just like, get a finger, I bet you can get a finger on like eBay. I'm pretty sure

that's illegal, maybe you know, I could just go like defrost a pig's foot or something like that, and just stick some pigskin in there. But but my goal is I want to I want to just take these things to failure with these solenoids and just keep pressing over and over and over. My hope is I want to get to over a million presses on these because that would that would easily exceed what we're seeing from field failures in the on the tax, which is the tax, which is probably the I think they're ready to do a million. They're not we're not getting anywhere near a million out of them. And it's a lot lower it's way, way, way lower. So if I could get a million out of these, now here's the thing, the entire actuation of this device only moves 16 thousandths of an inch. So it's really really small movements in them. And, and it's supported all the time. So I think I think I shouldn't be able to get that.

Well my quick Google searching for finger pressing analog doesn't really come up with like a way to get a finger. So

maybe I'll just go buy a pork belly and take the skin off and then just make little rings of pork belly skin. I actually foam is probably good enough

for ya foams and it won't smell.

I'm sure after a million presses, the skin will just be like every actually, that's that's one of the things that I designed into my fixture, the solenoids. The goal is not to have the solenoid like hammer into it, like go like I don't want the plunger to be off of the switch and then have to hit the switch into actuation I want it to be making contact with the switch when it's off. And then it just basically pushes the switch into over it's over it's trip point. So yeah, it's not it's not completely retracted. In fact, I'm so I bought little solenoids that have those retraction springs on them. I'm just going to take the springs off yeah, gadget junkie you smell bacon. The thing about it is these solenoids each pull an amp to do to do their work. I think I'm gonna have to have some kind of forced air cooling on the solenoids because they're going to be they're not like I mean, I'm going to there's four of them. My goal right now is to do them 1234 And then just kind of repeat in that situation. So there so they have a 25% duty cycle because I don't want to just have all four just junk junk junk junk, you know, should make it plain music. Yeah, that would or yeah, do some kind of, you know pattern in them, mainly also because I don't don't want to have to have a power supply sit there and do 12 volts zero amps 12 volts four amps 12 volt syrup you know just bounce between that

yeah cool cool that's

gonna wrap up this episode of the podcast

I think so to Allah will post these pictures and hopefully sometime soon we'll have like some working actuators going yeah

so that was the Mac fab engineering podcast we're your hosts Parker Dolman and Steven Gregg later everyone

take it easy

thank you yes you our listener for downloading our podcasts and sitting and listening to us through our live stream on twitch.tv. If you have a cool idea, project or topic let Stephen and I know Tweet us at Mac Feb at Longhorn engineer at analog E and G or emails at podcast at Mac fab.com Also check out our Slack channel. You can find it at macro voices.com/slack