MEP EP#217: Frequency Stability Dependability

Podcast Transcript

Host 1

Hello, and welcome to the macro fab engineering podcast. We are your host, Stephen Craig and Parker Dolman. This is episode 217.

Host 2

So this is what second episode since are locked down?

Host 3

Yeah, yeah, I you know, the thing is, okay, so I'm considered. And I'm using air quotes here. For those who can't see, I'm considered a, what's called required employees. So I'm still going into the office. And, and you know, I'm a basement rent so I go to the office and then I come home and I go to the basement, and then I work on projects and stuff. So not a whole lot has changed for me.

Host 1

I mean, really seriously, kind of hasn't.

Host 2

I've been working from home. But it's like, when I'm when I because I used to work. I work from home on Tuesdays before this for the podcast, but just like, my life hasn't really changed. It's like, I don't like going out to bars and stuff. So I guess if you have a social life, your life has changed a lot. What's that? Exactly?

Host 1

I'm an introvert and engineer.

Host 2

Yeah. I've been working on projects and playing the new Doom game and so good. Yes, it is very good. Doom internal injected into my veins.

Host 1

I picked that up on Friday, and I clocked way too many hours on it over the weekend. Way too many,

Host 2

Well, hopefully not enough that blocked you from doing projects. Is that right?

Host 1

I mean, maybe I did. No, I got some stuff done. Like I was it was Sunday night, at like nine o'clock at night. I was like, Oh, I actually have to accomplish something this week.

Host 3

So, okay, so I've actually got two new designs that are going through at work. I wish I could share what they are. But their top secret right now. But I can tell that they are both highly dependent on frequency stability. So they they they do something that requires that. And when it comes down to it,

Host 1

Getting getting frequency tested properly is not necessarily the easiest thing. And traditionally, we've used the frequency counters on our scopes, because those have been good enough for what we do. I mean, for audio purposes, that's most of the time, that's enough. But I actually ended up purchasing a one of the Amazon wonders or the Chinese wonder frequency counters, you know, those like sub $100 frequency counters the 3156. It's basically if you go and search for the 3156 frequency counter, you find the exact same frequency counter with like 18 different brand names on it. So it's like it's clearly made in one location, they just slap a different sticker on it. So it's actually a good frequency counter. So because I've been testing some some of our oscillators on it. In fact, I had a client project where I had six different oscillators to calibrate for this person. And I was able to calibrate to eight octaves of tuning on an eight octaves positive and who knows how many negative, I wasn't really worried about the negative. But this this frequency counter was great for that. It's kind of, I don't know, it's got a little bit of wonky UI features going on. Like you have to like scroll through like configurations to change modes and things. It's really, I don't know, it's cheap. Let's just put it that way. It's really cheap. It has one unfortunate thing that

Host 2

I'm actually looking up them on Amazon as you're talking. And speaking of funny brand names, this is my favorite one we've come across to fall far. Cooler Toronto with a que

Host 1

Koolatron Yeah, makes frequency counters. Yes. Is it Wait, isn't there a person on our Slack channel name? koolatron. Could be is it same? I'm pretty sure there is. Yeah. Like an active member

Host 2

In that name. So that's awesome. That I mean, yeah, maybe maybe they're selling maybe they're to Amazon reseller. Yeah, that'd be cool.

Host 1

Give us a discount. So get this. So the frequency counter is great. It has a knob for the gate length. So your gate length is basically the amount of time that it's sampling your your frequencies over and depending on the position of the knob you get higher or lower resolution. I'm not in a rush right now. So I have my gate length maximize so I get as many decimal places of resolution as I can just it helps out a lot. The the thing that kind of sucks about it is there's no there's no real averaging or anything that you can do with this device. Nice. So like, if I'm measuring, say, I'm measuring 50 or 100 hertz or something like that, I can get four decimal places of accuracy. And, you know, for most applications that you don't need that much accuracy, but I'm doing development right now. And I want to find what is the overall accuracy of my design, such that I can back it off and find out what my tolerances of my components are, I can make things cheaper, right. So I right now, I do want a ton of accuracy in this. The most annoying part of everything is the ability to not have averaging because the numbers do bounce around a bit. And the with your gait time turned all the way max on this device, it's 10 seconds, sampling time. So you find your, I don't know, variability every 10 seconds, it's really annoying to wait that long. So I would love to hack this device and somehow add averaging into it. I just came up with this idea, like two hours ago, because I was calibrating something. And I was like, man, it would be really nice if I could average on this, until I kind of wanted to open the conversation about what's a way to like crack into a device like this, and hack it such that I can get averaging. So I searched for schematics of it. And nothing really shows up, there is a guy who has a website that just has some like gutsche photos and a list of all the ICS. But they sit on the website, they're like I didn't trace out anything. So they just have like a list of ICs. So I don't know exactly like the What the What the, the baseline function of it or how it works or anything. But what I'm kind of thinking is, okay, so there's a brute force way that I that I could really just get the values out of this. And it's, it would be really annoying, but it would work, you could, you could just tap off all the lines to the display, right, and then just back back, translate it and grab the numbers, right. But that's a ton of wires. And you'd have to have something that would be able to read all of those lines and work it all out.

Host 2

You could if you were doing it that way. It's probably a spy screen, so it wouldn't be that many wires.

Host 1

So on, I don't think it is because well okay, so there's two boards in this device. One is, quite obviously the actual frequency counter. And the second one is like a display board. And on the display board, there's a microcontroller. Well, there's a microcontroller and each board so in between there's a small like ribbon cable that I don't know I think it has 10 wires that go in between so they clearly the PIC microcontroller on the on the front board is handling all the display action. So it some the the the actual values are getting passed in between the two boards. I don't know if that spy or whatnot, it might be and that might be the easiest way to kind of hack into this is to just sniff those lines and see what's traveling in between.

Host 2

Is this the one with the it looks like it just has a LED readout.

Host 1

I mean it's a yeah bunch of seven segment displays.

Host 2

Okay, so that is this is one I think you get the gain express.

Host 1

Gain Express. Oh no, like mine starts with a why like, I honestly I don't even know how to pronounce the name of the company. But I mean if you if you look these things up, I swear to god there's like 20 companies that do the exact same thing with a sticker on the front. And the sticker is quite crappy.

Host 2

And it's like Victor Yeah, power locks tech Pro.

Host 1

So yeah, that's kind of what I buy

Host 2

Mine cooler Tron though. I want a piece of equipment. That's it's cooler.

Host 1

Oh, yeah, gain express. There you go. Let me see if I can find the one that I got. I mean, I bought it from Amazon few months ago. Not Victor. There's more than one Victor is out there.

Host 2

Oh, yeah. So you can totally that's easier because I was looking at the one that's had an actual like TFT display on it. Oh, that was like man. Yeah, you'd have to like it. You'd have to read that and then figure out where on the screen you're pulling the data from. So this is a little easier. Exactly. Yeah. If you really want to do with like the convoluted way that wouldn't require any opening up the box, you could get a Raspberry Pi with the camera module and then just envision and do a vision capture for it. You could actually probably get that working in like an hour

Host 1

That's you know, oh here it is. Why I T and Sen. Yet Tenzin is the one that Oh yeah, I'm

Host 2

Actually just click that one.

Host 1

Yeah, that's that's the guy So yeah, like, I don't know. Okay, so what are some other ways that could perhaps, work for this? Without? Okay, you could always just make a frequency counter in a frequency counter, right? Yeah,

Host 2

I think is there's different ways to go about it. Yeah. Is, depending on how much knowledge you have to reverse engineer from the device. Yeah. Like the I don't care about anything inside the box idea is a vision system that looks at it and just gets the number. Sure, right. That would be like that, well, you don't have to know anything about what's inside the box, then there's one more step would be, like hacking to display what the output is. Yeah.

Host 1

So hacking the display is the most like surefire way. Because you know, what the display showing? So you know, what the data should be? That's probably like the easiest. But look at how many segments you'd have to read in order to get an accurate number.

Host 2

It wouldn't surprise me if there is a segment display controller on it. And you usually talk to those with spy. Yeah. And so if you could look at that board, I don't know what pictures or ICs are on the same line apart. And yeah, and I bet you if you found what I see is actually driving it and found the datasheet, you would just you would just go on the front end of that chip, which would be whatever is talking telling it what the display, read those couple data lines, that's gonna be spy, probably it's gonna have a debt, one data line and a clock maybe? And

Host 1

Yeah, that's Yeah, I think I think you're right. I think that's probably the easiest way I can probably just get an Arduino to sniff those lines.

Host 2

Yep. And just spit out on serial? Yeah, yeah, that's cuz I would

Host 1

Love it. If I could take, I don't know, 20 readings and average them, then I think I could get some really good information off of it. You know, another thing that I really would like to do with it, is instead of averaging, I was noticing on the device that I have today, if I make an instantaneous jump and frequency, so if I send a control voltage to go from one frequency to the next, it does the jump, and then there's a little bit of drift afterwards, I would love to plot what that drift is. So, you know, maybe, actually, you know, it'd be really cool is to have an Arduino where you press a button, it sends the control voltage, or at least talks to something that can send the control voltage, and then starts plotting all of those lines.

Host 2

Yeah, actually, that would be great. With some of the more recent Arduino stuff like it has a built in plotter, like, you know, has that built in serial terminal, right? It's actually a built in plotter, too. Yeah. So you could plot your frequency over time graph.

Host 1

I would love to see if that's something predictable in in these devices. Also, it'd be kind of nice to just have an Arduino take, I don't know, 1000 points of data over an hour and see what happens to the frequency as it just sits. So. So I'm also going to be doing a bunch of temperature testing here soon. So you know, put them in a box, heat them up, let them soak for 20 minutes, and then take a bunch of readings. So it'd be nice to have this for all like I probably have a few weeks of testing with all of this stuff going on.

Host 2

Yeah, crack it open. See because it is using some icy to drive that whole display. And it's probably, and it's probably talk to you by spy.

Host 1

It's pretty, it's probably a Singular Control IC that controls all the segments on this thing. So yeah, you're right, that's probably the best way and then you can just read the datasheet and sniff the lines. And you've

Host 2

Got a logic analyzer. Yes, I do. Yeah, there you go. You would, and I bet you, you can use the generic Arduino library that's like a spy in and just output it onto your serial terminal, and then figure out how to decode it into what it should be displaying.

Host 1

Right. Right. And the whole thing about the display is that it changes decimal points, that's going to be the hardest part. Because it the decimal point shifts all over the place.

Host 2

No, so it's trans. You'll know that by that data, though.

Host 1

Right, but I'm gonna I'll know that by the data, but I'm gonna have to interpret that, you know, oh, yeah, it's not much of a problem. I'm saying like, it's easy, but that's probably the hardest part. So the Yeah, I liked it. That's, that's a good that's a good way of doing it. And I also just go by a frequency counter that has averaging.

Host 2

Well, actually, this could be more powerful, because now you have a programmable device that is, quote, unquote, hooked in to the output of your, your test equipment.

Host 1

Yeah. And you know, I've seen inside this box, there's so much room in it that I might just sacrifice an Arduino and put it in there and drill a hole in the side and have like a little USB thing. Greg just plug in. And then now I have, oh, what's it called? data memory. So

Host 2

Cool. Let me know how I'll be expecting at least figuring out how to read like, if it's spy or not by next week by next, I'm actually super interested.

Host 1

Like, I thought about this, and I was like, This sounds like a Parker project.

Host 2

Yeah, if you if you figure this out, I'm probably going to copy it. Oh, cool. Yeah, I'd love to have like, really good frequency counter, because usually, I do a lot of record recording of stuff. And that's why I don't really have a frequency counter. Because the ones that can actually record are a little pricey.

Host 1

Yeah, they get up there. And I mean, like I said, for this thing, I was able to, to, excuse me calibrate my oscillators to eight octaves, which is a quite a bit, you know, for, but I mean, my frequency range is a 10th of a hertz to 20 kilohertz, for the most part. So I'm using a very, very small portion of this thing's capability because it goes to 2.4 gigahertz. So the thing the thing about this, this guy that's, that's important, and really important with any good frequency counter, is that they have an urbanized crystal in them. So you don't you have to turn them on. And you have to let them sit. Because like, if you start taking measurements, it's going to be off, they have to come up to temperature, and they have to like sit and soak for a long time. So I usually show up at work, turn this thing on, and then Mike start going like two hours later.

Host 1

Hmm, interesting. Yeah. Cool.

Host 1

Okay. Well, I guess I have a little mini project to do for next week, figuring out what that chip is. Yeah. Yeah. Cool. Baby steps, baby steps.

Host 2

You can't knock a project out in one week. That's just not the right way to go about it.

Host 1

No, no, that'd be unfair. It'd be unfair to you. Cool. So I've got another simulation that I got. For every one up on my GitHub, I actually had some questions about where my GitHub is. So github.com/analog en je, that's where you can find it, we'll put a link up to that. So I've actually been using this for a while and Lt spice, and I wanted to share this with everyone. One thing that I found cool about LT spices, they make arbitrary sources, really easy to use. In pea spice, I've found them to be really difficult, if not just completely non existent. So an arbitrary source is a source that you can apply a function to. So that makes your circuit simulation incredibly powerful. Because a lot of times, you might have sub circuits or whole chunks of circuits that do a particular thing, like I don't know, take a whole circuit where its entire purpose is to perform an X exponentiation on an input, you can just use an arbitrary source and plug in that function. And there you go, you have you just exp, whatever. So instead

Host 2

Of having to simulate all those components, you're just saying, this performs like this.

Host 1

Exactly, exactly. And the cool thing is, the input can be whatever you want it to be. So you can just say the input is the voltage on Node XYZ. And it will take that and what's great is it's not a static thing, you can have this be in feedback loops, you can have them be control signals, you can have them do whatever you want. So in this simulation that I have here, I actually have two different circuits where I have one source that is producing a negative to one volt to positive one volt triangle wave. And then that is being used inside of a an exponential voltage, arbitrary source and an exponential current arbitrary source, and then I have them in some op amp stuff that you can play around with. So I've actually used this quite a bit in simulating VCA circuits. So that's Voltage Controlled amplifier. So you have you have an amplifier with a gain can be adjusted both via the voltage from a different portion of the circuit. So like, take a compressor, compressor circuit, where if if the overall energy of of your signal reaches some kind of threshold, then it automatically reduces the game of the circuit. Well, you can actually simulate that by using arbitrary sources as your gain control blocks. And if you want to use nonlinear functions, like exponential or logarithm or something like that, you can actually pull it off. So this is a really powerful technique and the LT Spicer has made it really really freakin easy. So I'm super impressed with that, mainly because I've been using piece bys for years and years and doing these kinds of things have always been really difficult. In LT spice, you just plop down one of these symbols, voltage or current or whatever. And it just says f is equal to, you know, or I'm sorry, voltage is equal to f of, and then plug in your, your function that you want. So super nice. The cool thing about it, too, is that you can embed these in sub circuits. So we have some voltage controlled amplifier ICS that we use at work. And I ended up creating a sub circuit for this voltage controlled amplifier that I can then just plug in and embedded within the sub circuit is one of these arbitrary sources with a function in it because these particular VCAs read a voltage on a control pin. But then they control the gain of a current flowing through x in an exponential format. So it's a voltage controlling and exponential current. So it's a really nonlinear, very weird kind of thing. But it's super easy if you just have functions available, and so works out really, really well. And I've used these to simulate Well, amplifiers, but also Voltage Controlled filters, and like moving filters and things like that. So give it a give it a look. That's github.com/analog EMG. And you can check out the simulations page, along with all the other ones I've done this year.

Host 2

I think last week, I forgot to put the link up and I had someone or I think last simulation of the week was three weeks ago. Yeah, it was a little while ago. Anyways, I forgot to put the link in there. So my was my bad.

Host 1

That's all good. So what you've been up to burger,

Host 2

I've been doing lots of work on on my 3d printer and kind of like reverse engineering components and stuff like that. Kind of like learning more about that. This is kind of like an extension of my Autodesk fusion 360 adventure.

Host 1

Have you tried the slicer in 360 yet?

Host 2

No, I have not tried to slicer yet. I was going to try it. And then I ended up having to basically run my printer for two weeks straight. And yeah, I haven't. I've made some like it works right now perfectly. So I don't want to mess with it. So it's like, now since I'm like kind of done printing that whole two weeks span. I'm probably gonna

Host 1

Quit Quick, quick side story real quick. Yeah, a buddy of mine worked at HP. And he worked in the server division. And he was writing assembly code for a lot of the low level stuff that happens in the in the server side of things. And I mean, he was fresh out of college. And he started looking through all of this assembly stuff. And he was just like, Man, this is garbage. This code is terrible. He's like, I could write 10 times better. And he's like, I already have written some routines that are a lot better than this. And he goes to his boss and was like, hey, you know, like, I would really like the opportunity to to rewrite these things in his boss was like, no, no, no, no, no, it works. You don't touch it. Don't ever ask again. Like, like, Crap spaghetti code, but whatever it worked.

Host 2

Until it doesn't work. I don't know. Like, I

Host 1

Would think in HP server. It has to work all the time, right? Sure. Sure.

Host 2

That's what they want you to

Host 1

Think. Yeah, right. They will they got to me then. Yeah.

Host 2

So I'm at the point where I'm going to start doing some 3d printer improvements for my Monoprice ultimate maker. So stuff that already has that. Actually, it had this when you were still here in Houston. It has a internal Raspberry Pi that runs Octo print, which is like a print server. So you can just drag and drop from your browser, the files and you control everything through it. It's really nice. Not having to basically mess with the interface on the printer.

Host 1

Right? Because it's like one dial and like a push button and a push. But yeah, well no, the dial is the push button, right?

Host 2

Yes. Yeah. So you press it and the rotary encoder knocks notches over and you've selected something else.

Host 1

I hate that so much. It's the worst.

Host 2

So yeah, I got OctoPrint running on it. I have a micro Swiss all metal hot end which allows you to run hotter temperature stuff. Because the stock one has a PTFE or Teflon liner. And when you heat up that liner, it doesn't like it because it gets really soft and gooey and started gumming up your your flow.

Host 1

I mean we are talking about 250 Celsius. It is pretty hot. It is

Host 2

Pretty hot. Yes. And it has an enclosure and it has what's called a BL touch. Actually it did not have the BL touch when you were here. That's something that you'll find really interesting. It's like a a Mechanical pin that can will touch the bed and the moment it touches the bed it like it triggers I guess a lever or something inside of it. And so it knows that it basically it's zeroing

Host 1

Yeah, right? Right. So does it have like a switch that it sends to the? Yeah, okay. Yeah. So I, I actually had a, for the CNC that I got, I put a, gosh, years and years ago, I paid a lot, I shouldn't even say paid. He made me for free. I, the guy who ran the lathe at are the first company I worked with, I was like, what kind of accuracy can you get? And he's like, I can promise you four zeros. So it was like, make me make me a piece of steel. That's one inch to four zeros. And he totally did it. So I have a one inch little block. And, and I use that to touch off with my, my workpiece.

Host 2

Yeah, so it's that same but but as I have it set to do auto bed leveling. So instead of having to basically drive the head around and figure out what your zero is, by by calibrating it with usually a piece of paper by sliding it, I'm gonna go, yeah, that feels like good enough about the same write, it actually just drives around the bed and then touches all the points and then it maps that out. And so it has what level is. And so actually, when it's driving around, it's adjusting the Z. That's so awesome. Yeah, it's really nice. And basically, you don't have to ever touch the bed ever again. So it's got all that and a bag of chips. So what I want to do is improve like the cable management, because it's got this like really fat wide ribbon cable that comes off the back of the printer and then down into the enclosure that goes to the head. And that has most of the control signals and all the power and all that good stuff. But it's kinda unruly, because it's wide and flat. And I'm hoping to see if I can replace it with kind of like a just a normal round cable that can actually put through like a some cable management.

Host 1

How many? How many signals is it?

Host 2

It's a, it is a two by 10 connector by I don't know how many actual what a power ground, someone's got a pin out online for it. So that's going to replicate it into like a cable. And I was actually going to use your idea and just use like, split loom. Yeah. For the because I don't want an E chain, because each chain would really actually make sense for this because it's like, the head drives around and everything comes off the top. Right, right. And so I was gonna use loom like you did. And all the BL touch signals are not in that wire, that that two by 10 wire connector. And so those are just kind of like taped together, it looks like your old CNC machine used to be where all the wires are kind of just dangling off the head. And just like hell yeah. And it's just like, it doesn't move enough to where it needs, like protection, I guess. But I want to I want to actually aluminum so it makes it look nicer. And then I needed that fixed bed, which has been working really good with the BL touch after a year and a half because I basically I printed the standoffs because I because I guess I should backtrack a little bit. The stock bed has springs on it. And so you adjust screws on the bottom to adjust the angle of the bed so you can make it level. When you get a BL touch or any kind of like sensor that's detecting auto bed leveling. You want to fix bed, so it doesn't flex at all. Right? And so I actually printed standoffs, which worked great, except that I just noticed that one of them is actually starting to melt. So the bed is hot enough to deform them basically.

Host 1

Oh geez. Well, it's 100 bed right?

Host 2

It's a heated bed. Yeah. And beds actually at 110. See? Oh, geez. Yeah, yes. My bed can boil water.

Host 1

So the level the level is not level.

Host 2

No, but it prints fine because that has a BL touch. Right?

Host 1

It detects it automatically compensates. Yeah.

Host 2

So I need to get those 3d printed parts out and probably switch them over to Oh, something that's got a really low coefficient of thermal expansion. So I was not going to use aluminum aluminum would be easy. Maybe I'll probably just get steel spacers ceramic ceramic. He wanted to be stable. I wonder if I can get three quarters inch tall ceramic stone, stone granite.

Host 1

The so the CNC that I operate at work. I'm so spoiled by it, but it has. It has its own coding language called simple and so it doesn't run on G code, it can. But the simple code is so much nicer. And everything is just function commands in it. And it's all it's basically like C, effectively, but the Z probing like you're talking about on yours, I, that's just a command on mine. So whenever I go to, like, cut a new thing, I throw down my material, I find an edge. And then everything else is just done. I just load the program and the program already tells it like move to this corner, probe this much with this x y grid and things. Oh, it's super nice.

Host 2

Yeah, that's actually built into the firmware is there's a G code command that you run. And he goes, Okay, I'm gonna run bed leveling and then figure out what the bed?

Host 1

Yeah, that one G code that you run opens up like a bazillion other

Host 2

Lines of code? Yes. So I want to get the I need to fix the beds. spacers. I don't know what material I want to use make it out of yet. It's actually Saran. If I can find ceramic. That'd be really good. Steel would probably be fine. Steel is?

Host 1

What would the rule of thumb is 1,000th of an inch for every 100 degrees. Fahrenheit. I can't remember.

Host 2

So yeah, that should be fine. Yeah.

Host 1

I mean, if you're talking about two thousandths of an inch, your bed is not that accurate? I bet.

Host 2

No, the aluminum plate that it is made out of probably works more than that. Yeah. And then I also need to upgrade the fans. Because about halfway through this podcast so far, the printer finally stopped printing for the day, and shut off. And I could actually hear myself think now. The this printer originally came with sleeved bearing fans, which are fairly noisy fans to begin with. But now it's probably got, I don't know, several 100 hours, maybe 1000 hours of prints on them on this machine. And they're getting kind of noisy. So I'm just going to go through, I got to open up the printer up, find all the measurements for the fans, or some ball bearing fans on, you know, Amazon or eBay or whatever, replace all those guys. And then I started looking at how do I make my prints faster? And not just like printing faster, but like, hey, what about making the setup? Basically it you know, heating up and doing the auto bed leveling? Because it does that every single time? How do I make that faster? And so I started looking at the like what it does, when it first turns on and starts the heat up everything well first, that heats just the bed up. And that takes from cold for room temperature 210 c takes it eight minutes, because it's a 24 volt heater on it, which isn't actually that bad. There's most bed heaters out there can't even do that ever. Like they can't even get the 110 See, let alone in eight minutes. So I'm like, okay, that's not too bad. And then the hot end, which is the extruder part only takes six minutes. And so but it waits for one to be done and then does the other. So I'm like, Okay, is there a way I can make that happen? Same time. So I looked it up. And apparently you just inject some different code in your starting commands that your slicer puts in there for the G code band that fixes that problem. So now I basically almost half my startup time, which is pretty good. And then I'm also thinking about improving that even more. Basically, if I can make the bed heat up in the same time as extruder. That's like optimal, basically, they heat up at the same time. And so I'm looking at putting in a 120 volt AC heater for the bed. It's 500 watts. Yeah, one minute to get that hot. Actually, yeah, it's about a minute and a half to get that hot. And actually, I would actually code the firmware to

Host 1

It's just a relay that connects it right to mains, right?

Host 2

Well, yeah, you actually use a solid state relay that so use the MOSFET driven output that is normally doing the 24 volt output to the original bed and you have that drive in a solid state relay that's turning your 120 volt on and off. Put that on there. But you also probably should put a thermal fuse in it.

Host 1

Yeah, you don't want to just running away.

Host 2

Yeah. Because you're, you're connecting drug payments of like, what if the solid state relay fails? That's like, that's the worst case scenario.

Host 1

What is the what's the common failure mode for solid state relays? Is it is it closed?

Host 2

What happened to the one on your brewery?

Host 1

That one? That one was open and when did not short?

Host 2

Okay. Okay. Yeah, hopefully the common mode is open.

Host 1

I would hope so. Yeah, but you might, but who knows? Maybe not it could it could

Host 2

Be fused, it could fuse together inside there. So I'm gonna put a thermal fuse in there. I don't know what temperature to rate that thermal fuse for. So if anyone out there knows, let me know. I've actually tried searching for hot. Yeah, that actually, people don't do that. Scary. Yeah, no, it's just like, you know, I kind of, especially since I actually leave my printer alone while it's printing. I know that's a big no, no. But

Host 1

Did you go to work and have it printed? Right? Yeah, yeah, yeah, you need some protection on?

Host 2

Yeah, so I'm gonna put the thermal fuse in there. I do have a, a smoke detector. But it's like, well, this smoke detector is going off its garage is done. Like yeah, it all goes up. Yeah. Also, I can turn it off remotely. So that's cool. OctoPrint allows me to can you do it from your phone? Yes, because you just hit an IP address, and then log in. Okay. Yeah. So you can do it technically through your phone. So yeah, if there's,

Host 1

You know, okay, so if you're doing this, you know, you need to add a little webcam on the head, too.

Host 2

I have a webcam that sits off to the side that looks at it. Okay. Yeah. So basically, like, if I'm printing it at home, I just have it on, like, a different window at work. And I can just like, check in on it every couple of minutes.

Host 1

Dude, why don't you hook up a particle to the webcam and have it tweet during prints? So every, I don't know. 30 minutes, you get a tweet of progress.

Host 2

It's on fire you.

Host 1

I wouldn't be awesome.

Host 2

So. So if anyone out there has recommendations for the thermal fuse, let me know. Because I don't know what I probably need to actually take the temperature of this bed. Like when it's running, and actually the silicone 500 Watt one, like what's the its temperature when it's ramping up? Because like, because this is gonna be mounted directly to the heater on the bottom side. So it's going to be hotter than the top of the bed.

Host 1

Do you have a thermal camera?

Host 2

No, I do not. I thermal cameras my finger.

Host 1

It'd be nice to see if your bed has like ridiculous hotspots, see if it like varies by 2030 degrees?

Host 2

It probably does probably, but I mean, it's just quarter inch aluminum. So it's not too much heat spreading through that. But actually, the silicone heater is gonna be better.

Host 1

Why don't you? Why don't you replace it with cast iron? This machine gas? Yeah, just flat cast iron on the top. And rough on the bottom. You should cast a new table. I would? Oh, that'd be so cool.

Host 2

Way too much. So yes, if any people will have suggestions for the thermal fuse, let me know. And then another one is, I probably need to take apart the rails and clean them. Because I've actually never done that I've I've lubricated them a couple of times with just like three and one. Let's go motor oil. But I know that's not what you're supposed to use. Because it gets a little gummy. So if anyone has suggestions for what to use for lubrication, Stephen says white lithium, but I don't think you're supposed to use those on these kind of Rails because it's a sleeve bearing. And I always think that get little gummy

Host 1

White lithium, well, okay, white lithium, or three and one is my go to for like, if anything if I need some, okay. Like, if it's a chain, I'm going to use three in one. But if it's like two faces that rub against each other, I'll throw some white lithium on it. Like that's sort of just like the first thing that comes to mind.

Host 2

So I think Mike Williams, who's been on this podcast before he suggested sewing machine

Host 1

Oil. What's special about that?

Host 2

It's for sewing machines. I think it's a really lightweight oil. Yeah, like a really low viscosity. thin oil. So

Host 1

I wouldn't Oh, you know, if the bearings are linear, they're linear bearings, right? Yes. If you go to McMaster, there might be some information on that, you know,

Host 2

Oh, what the? Lubricate the morth Yeah, yeah.

Host 1

Or you could probably go find some data sheets for him. We, the the linear bearings on the CNC work that we have. We have some like super, like crazy expensive German grease stuff. But I bet you it's just like regular grease with a German name on it. And it looks like ear wax that you have to like squirt in all the little injection points,

Host 2

You know, zerk fittings.

Host 1

Is that what they're called? Yeah, that

Host 2

Sounds right. Yeah, yep. zerk fittings.

Host 1

They are a pain in the butt to get to our Zerk fitting well, you

Host 2

Might not have the right fitting on your grease gun. No, I got

Host 1

The right fitting it's still it's a pain in the butt as in like the gantry has them in weird places. Ah, Ah, yeah.

Host 2

I wonder if Zerk fitting is short for something. Oh, it's named after the person who came up with it. Oscar your least sees zerk pool

Host 1

I would have fitting named after me something like that would be you know, okay, back in college we had this thing about like if you had a if you had a unit named after you, like a Dillman would not be a good name for one unit nor would a Craig You know, like it just doesn't really doesn't work too well.

Host 2

No, no. What would you say you just want to fit in? So Craig would be a fitting?

Host 1

Yeah, no, I could I could see I could see a Craig fitting being something at at Home Depot, like you divide that and it would be like, I don't know, there'd be like a one of those plastic tags on it with the K and like, you would just have to know what's a Craig fitting, but it's just it says K on the Side, you know how they have like, the M and the F for for all like pipe fittings now and stuff instead of actually knowing things. They just say like MNF on the side or whatever. Dr. Cutler. There is a pocket screw company that that does creg but it's KR e g, instead of KR AIG. So, of course, I bought that so I can have a Kreg pocket screw. One day, I'll have something named after me.

Host 2

One day. You're creating amplification. Yeah, that's true. Yeah. Yeah. Oh, so what's been spurring all this 3d printing? No, I've been doing for the past couple weeks is, I've been doing a lot of kind of like reverse engineering, building components from my wagon, like a column, a steering column bezel that has switches and stuff on it. So I'm like measuring, trying to figure out like how to properly measure things. Because in SketchUp world, how I'm used to, is you kind of just drew it and then I'd print it and be like, Okay, I need to adjust this, this and this probably and just kind of fudge it right? And then you tweak it enough. And then you get something well with Autodesk fusion 360. It's like, I want this to be one inch by three

Host 1

Inches. Is not a tweaking software. No, it's parametric.

Host 2

So I'm like, Okay, I gotta get parametric with how I measure things to draw them. And so I'm like, Well, I only have a six inch digital caliper. So I'm like, Okay, I need more tools. Was good sign right, I need more tools. Yeah. So I got a edge gauge, which is a I got right behind me. It is a piece of plastic with lots of little tiny pieces of plastic. But when you push it up against a surface, it it and you imprint the profile into it. It's kind of like those oh man back in the 90s you had those actually probably even earlier 90s But why? Remember those little nail toys the nail toys that you would press your face into intimate imprint in your face?

Host 1

The funny thing is okay, good eye like I don't think they ever really intended that those things are supposed to be pressed into your face, but everyone on Earth press those things into there.

Host 2

And that's why we got Coronavirus now it comes full circle. Yeah. So this is basically this is a way to get profile weird edges and, and I think it's mainly used for like carpentry. So you can like copy a piece of molding or something like that. I used it to copy the folds and a piece of sheet metal that's on the wagon, so I could make a 3d printed part that would perfectly fit over the contours. And I actually got it in the first go. I was actually really surprised. So how I did that was I you know, I just pressed it up and got the profile and the the edge gage actually has a a a centimeter ruler on it basically it says that has millimeter divisions. So you can take a picture of it, bring it into your CAD tool, adjust the scale of the picture so it matches up with your your one millimeter grid, and then you can actually just trace it the profile and then extrude that out, printed it up. I basically just made a really thin one printed out and it matched perfectly. I'm like well I got my profile now. So that was super cool that that worked like out the box like it was kind of amazing that got on the first try.

Host 1

You know Steven galena which was a guy that started out at macro fab with with myself. He and I used to build guitars and And we, we got one of those profile gauges. And we measured an old 70s Fender Telecaster the back of the neck, because the back of those necks were done by hand. And so like the the radius from the heel up to the, the, the headstock is not continuous and it was done by hand. So we did every three inches, we took a profile and then brought those into I was using Rhino at the time, and then we, we we just kind of like skinned the neck and then cut that on the CNC and it was perfect. Nice.

Host 2

That's actually when you look these up on Amazon. One of the applications they say it's for looping. Yeah. I wrote. So I really liked this tool, especially since it was like $10, which I'm like, it's just amazing. Hazard fraught thing. No, it's a Amazon wonder. Yes, Amazon $10. Wonder, let's see what the what the title of this one was.

Host 1

Koolatron Oh,

Host 2

I wish fn I

Host 1

You. Okay,

Host 2

Now. Sure. Great. That's this tool. And then I bought a, a dial indicator, which are usually very expensive. But this one was like $30. And before I want to use it for I'm not taking precision machinists. measurements at all, I actually really wanted this just so I can do radius measurements on device on on. I guess hard objects like a coke can or something like that. And so I actually got down on cater and I printed a attachment on it. That is a radius gauge attachment for dial indicator, basically, it's two prongs, that you know, the distance between them. And so when you put it onto a round object, you know, you basically make an arc that you're measuring from, and you throw that into a calculator, it spits out what the radius is. And I it's actually within like, a quarter of a millimeter. Oh, that's pretty good. Point to five millimeters of so I was like, Okay, I'm, that's perfectly fine for replicating stuff, right?

Host 1

Well, that depends if it's a plastic that goes on the outside of a of your car. Sure, that's fine. But if it's the shaft that goes in your engine, that's not so good.

Host 2

No, you're correct. This is this is for reverse engineering for building plastic stuff, right. works perfectly fine for that. So I like this tool. What is this brand? Neo tech us and then I got a angle finder. Digital angle finder. Which is this one I don't like it's really cheap feeling. Let me see it.

Host 1

Oh, geez. What?

Host 2

Yeah, so you as like, we're an audio podcast, right? And I'm showing something to see. Oh, of

Host 1

Course. Yeah. Oh, so you have a Yeah, you get like a little finger that sticks out of this device. And. Okay. Yeah. And then you just that probably makes more sense. Showing it that way. Yeah. Okay. So you can like I guess put it against two objects and see the angle in between them.

Host 2

Yeah. So this is for like measuring. Like, whatever an angle is cut to, right. So I use this. I don't really like this tool. This is a general brand one. It was like only $20. I should have probably spent 30. It doesn't really hold its zero really well.

Host 1

So it's not repeatable?

Host 2

Yeah, it's not very repeatable. It is good enough for what I'm using it for now. But I'll probably get a different one. Probably spent $30 instead of $20. But I mean, it didn't come with its own coin cell battery. How cheap is that?

Host 1

Oh, that's garbage. Yeah.

Host 2

And so one tool I'm looking at getting next is a good flatbed scanner. And I'm thinking about just like going into like a Goodwill or something and buying one a used one.

Host 1

That's probably good for like five bucks. Yeah, cuz I don't really need

Host 2

Super high DPI, or anything super crazy. I just need a black and white scan. Excuse me, of whatever I'm trying to replicate into the 3d realm. Well, 2d 3d realm into the computer. So it's been pretty fun. I Like I have to post pictures by made like a bezel that fits underneath your steering column and a wagon that switches and stuff. And I had to like make sure it would fit the openings and the mounting years would all line up and stuff, it's kind of hard to explain without just showing a picture. So it'd be on the blog, and then I made a rear wiper rest that fits on the back of my tailgate on the wagon. And that one I'm super proud of because the stock solution for real, so wagons were built from like 67 to 91. Okay, long time never didn't change the body at all, they just changed some of the trims over the years. And the rear wiper only existed for the last three years of it. So from 8990 and 91, they were the only ones actually had a rear wiper. And so they kind of like cheaped out a lot on it. It's kind of like hack together. And include it totally Yeah, I don't know why they haven't added it then at this point. But the wiper, instead of having a little holder that the wiper would rest on, right, they could actually just put a piece of clear tape on the paint, and it just rests on the paint. Hmm. So it's really ghetto. From Factory, so I actually printed a little rest. And that's that was using the gauge finder, the edge gauge for and so I get the profile the sheet metal, and I printed out a little rest, and I tweaked the shape of it. So it would when the wiper would come down, it would hit the rest and then pull off as it went down farther. Works great. I had to get some pictures of it. But the my favorite thing was I was building them or printing them with straight sides. And it just looked like a piece of plastic on the I mean is a piece of plastic. Right? But it just looked like kind of like out of place. For some reason. While I was looking at it. I'm like, man, it just sticks out. Yeah, and sure. It's a yeah, look at it. And I'm like sure it's a piece of black plastic on white paint, white sheet metal paint. But I'm like, it still doesn't look that good. And so I was looking at other ones like how do they how do they hide it? Right? Because it's this is black lump of plastic on the back of a tailgate. And one thing I noticed is all of them are like kind of like cone shaped. Right. And so I put a taper in it. So I made it like a trapezoid. And then I put that one on. And I'm like, oh, it looks great. Now, I added a draft angle to my 3d printed object, right? Yeah. And it looks like it should be there. Now. This is weird how that

Host 1

How it's ingrained in your head that plastic should have a draft angle?

Host 2

I guess. So. That's why Yeah, it has to be that way. Namely, that's the solution to you. Because you've talked about this before we're three 3d printed objects always look like they've been 3d printed. Yeah. And I don't think it's just inherent to like the, the layer layers. I don't think it's that because even if you'd like sand it down. So it's still kind of looks like a 3d printed object. And it could be that, that they actually have 90 degree walls, we actually have 90 degree angles on them. And you're so used to plastic not having that.

Host 1

Maybe. So maybe you can design for not 3d printing. 3d printing.

Host 2

Yes. Well that like have a how to make your 3d printed object not look like it was 3d printed. Because isn't that I mean, this piece of plastic it has the I didn't do any post processing. So it's got print lines and stuff. Looks great. Yeah. Yeah. So it was it was on the

Host 1

Scene it in this final resting place. He needs to post a picture of

Host 2

It. I'll post a picture of it yet because I put it on it last night. I didn't take a picture of it. Yeah, it's just like I added like I just added like a 10 degree draft. That was it. Oh, 10

Host 1

Degree is a lot but isn't most draft angles are like, one or two degrees. Two degrees, I think is sort Yeah, but I, I could do that. Yeah, but but that's not even. Like, if you're standing a foot away from it. You wouldn't see two degrees. No. Yeah, but you'd see 10 Yeah, probably have 10 degrees. The whoever's injection molding your plastic would be like I love you.

Host 2

The prints are the mold just pop right out. Yeah.

Host 1

We don't even have to have ejection pins on these things. They just know out.

Host 2

So cool. Cool onto the RFO.

Host 1

So this week on the RFO which is the rapid fire opinion. Oh,

Host 2

One more thing. Yeah. Back to that. So if anyone out there has other reverse engineering tools or tips and tricks Bring it up. I actually kind of think we should have someone on the podcast to talk about that topic. That'd be cool. Yeah, like, going like maybe going through the whole process of like, okay, you want to you need to reinvent reverse engineer and enclosure from a customer, because they don't have the designs because some other, you know, a factory that they're moving away from, or their place they contracted is withholding their design or whatever. Like, how do you go from like getting a you know, it needs to look like this to something that, okay, now we can build, you know, a million more of these,

Host 1

You know, on top of that, as a secondary to that, it would be really fun to get a metrologist on to talk about how to measure things. Like how, like, when you put calipers on something, how do you know what you're measuring, like the calibers tell you something? But is that right? Are you doing it? Right? Yeah. And above and beyond that, like, great example, I deal a lot with Hammond enclosures, which are cast aluminum, and are injection molded, injected aluminum, basically, they have draft angles. How do you measure a surface that has angles and everything is curved? Like you have? No, you have no zero point? You have nothing to measure off of? Yeah, but then how do you like actually pick up an edge on that and stuff? So I don't know. That'd be really fun. If anyone knows a metrologists, please hit us up. That would be really cool, too. Or something, someone to talk about reverse engineering. That'd be awesome to get on the podcast.

Host 2

Because the we've talked about reverse engineering, electronics and a lot of times, but I think a more mechanical approach would be kind of cool. Yeah, for sure. Actually, having an in depth one for electronics would be cool, too.

Host 1

I mean, this whole podcast has been dedicated basically, to hacking and reverse engineering here.

Host 2

That's true. But like if you need to replicate a circuit board, right? Like, how do you do that?

Host 1

Dave Jones had a had an interesting video a few years ago about that. And it's worth, it's worth looking up. Basically, basically, he said that it boiled down to like, well, you know, good luck. You're gonna be here for a few hours just staring at like this trace connects to this trace and this trace that trace and like, yep, that's how it goes. I didn't you guys get a fancy X ray machine at the Fed? Yes. So you can do some real reverse engineering with that.

Host 2

Yeah. But we don't need to. We're scanning stuff that we build.

Host 1

So yeah, but I'm just saying like, that's, that's a really useful tool.

Host 2

Oh, yes. Yes. Because I guess you can get the metal traces. And now you have an image of it. Well,

Host 1

Multi layer boards. That's what I'm getting at. Oh, you don't have to guess on those?

Host 2

Yeah. Because I guess on a two layer board, you just store it in the scanner.

Host 1

Yeah, yeah. I've actually, in fact, I'm holding one up, I have one just lying around here. I bought some boards from a guy a little while ago. And I was I was curious about improving on his layout. So I scanned those in and then started tracing them. And then I just stopped because I was like, man, you know, I could just do a better job if I did from scratch. Seriously, like, I liked this guy a lot. And and I'd like his board designs are good. But it's just like, there's so many design decisions that were like, I was cutting a ton of corners, that I was just like, You know what I might as well just do it from scratch. So

Host 2

Cool. Okay, now on

Host 1

Yeah, then now the RFO. So we've got one hour for this week. And I put it up here because I just thought this was really cool. So it's a Hackaday article called a NES motherboard for the open source generation. So there's sorry, I apologize. I don't remember the person's name. But they created a full layout for a replacement PCB that goes in your Nintendo Entertainment System, the original NES. So you can actually just download the files and get your own NES motherboard created. And you can replace the guts of your old Nintendo, which is kind of nice, actually good because they're starting to get pretty up there in age. And I know mine has had trouble in the past, firing up. And that's not just the car. It's like my actual motherboards not doing too fantastic. So it's cool too, because this is only a two layer board. So this would make in my opinion, a really great project for someone starting to get into electronics, and they're getting past that like oh, I got an Arduino now I want to do something more intense and learn a little bit more like you can get the the entire Nintendo schematic it's it's available, which is cool too, because it all fits on a single page. So you could trace out the entire schematic and get this entire board made. And what's cool is everything except for two parts is readily available. Readily meaning like you probably have to do some eBay searching and things like that. It's not like stuff you can get from Mouser but I Almost all the chips are available. And you can still buy new old stock processors for the for the NES. But there are also newer made CPUs, like the UA 6527, which can, you can still purchase that. And that's a drop in replacement. So go check this out. One of the things that's interesting, though, is some of those replacement parts like that CPU have, they work? Pretty much 100%. However, the PWM, or the duty cycle output for the square wave sound channels is reversed. So whenever so the original Nintendo had five channels of audio, and two of them were square waves. And they only had four options for duty cycle, which was, like 25%. Or no, it was like 12.5 2550, and 75, something like that. So this particular CPU flips those numbers. So when it says, play at 12.5%, duty cycle squarewave, it's actually playing a 75%. And so like, it doesn't change the pitch of anything, it just changes the timbre, which is really interesting. So I have a, I have a website where they do comparisons of like, what is the original Mario Brothers theme song with 12.5% duty cycle versus 75%. And it's one of those things where like, if you're not paying attention, you probably wouldn't notice it. But if you play him side by side, you're like, oh, wow, that's actually makes a pretty big difference.

Host 2

Couldn't you easily fix that with a little amp an op amp circuit?

Host 1

Could you? I don't, I don't know if you would? Yeah, you probably could.

Host 2

Yeah, you would just do an inverter? Yeah, if

Host 1

You if you just invert the square wave, right. If you invert the square wave, then you could fix it. Yeah, actually, you could probably fork this guy's design, add an inverter in it and just have a CPU fix. Hey, that's a good idea.

Host 2

Hey, someone take my $100 idea and make that happen.

Host 1

I like that. That's great. Yeah, I think the only thing is, I don't know if there's actually

Host 2

No, it's even easier, you can probably actually just use a inverter IC.

Host 1

Probably you don't even need an op amp. You just have to make sure that the output levels are right. Yeah,

Host 2

I mean, it's square with its digital, right. But I want to zero.

Host 1

I don't know if the if those channels can be used for any other functions, though, if there always were. Because if they're not, then that doesn't work. Right? You have to know if they're square.

Host 2

But if it wasn't, if it's only square, you can use the inverter. If it's not, then you would have to use an op amp and just invert the signal.

Host 1

You Yeah, I don't know if that would screw with any phase if you have one of your channels out of phase. That might? I don't know, I'd have to I'd have to see. I mean, I doubt it. I mean, everything is really simplistic with the NES. It probably doesn't give a shit about phase probably doesn't. So yeah, check that out. There. One of the chips that is not available is like the lockout chip for location and stuff like that. But there's plenty of workarounds. And there's other IC boards that you can build and make and plug into that socket.

Host 2

I think my lockout chip on my Nintendo I have it disabled. So

Host 1

Is that just like shorting two pins or something?

Host 2

Yeah. Basically, the lockout chip thinks it's all groovy all the time.

Host 1

Do you have any, like Famicom games that you played on it?

Host 2

No, I have it so that you don't get the blue flashy screen a lot. Oh, got it. Yeah. Because usually most time that it's failing is when you get that blue flashing screen. It's because the connection between your cartridge and the lockout chip isn't good, right? So you disable that and that just disables basically half of the issues that you get with old Nintendo's. And then so basically, when you boot it up and the connections not good, you just get garbled stuff on the screen. So then you take the card out, wiggle it a bit, put it back in, blow out eight or nine times. So why don't we Why don't we like retro gaming? I don't know.

Host 1

You know, I actually went and played I don't remember what game it was. It was some NES game the other day. And I started playing it and then I put it down like five minutes later because I was just like, This is so damn hard. How did I put up with this when I was a kid like I put up with all that you had, right? All like they they made games long by just making them ridiculously hard.

Host 2

Mm hmm. Cool.

Host 1

Well, that was the macro fab engineering podcast. We were your host, Stephen Craig

Host 2

And Parker Dolman.

Host 1

Take it easy.

Host 2

Later, everyone