Frictionless Spherical Cow Transformers

Hello, and welcome to the macro fat engineering podcast. We're

your host Mark Nolan and Steven Craig.

This is episode 239. So, Steven, I did not get to work on the cat feeder on reminder. I went fishing.

Hey, that's a reasonable excuse. I saw some pictures from that. Those were those are some. Those are some big fish.

Yeah, we caught a I caught a 30 inch black drum. And one of my friends caught a 41 inch black drum, which is That's enormous. Pretty big fish. Yeah, no, that's

huge. Yeah.

And this is by the way, we were fishing in the Gulf of Mexico off off the coast of Galveston in the surf. Like wait fishing, so a boat or anything. And it's always fun catching something that big. Because you're reeling it in it been like chest deep water, and you're like, that's about 30 minutes. You're like, I really don't want to know what's on the other side of this line at this.

Yeah, I've got some big red fish out there. And it's fun because they fight for a long time. Like fishing out in the in the surface. Like it's entertaining. It's not just like, you know, pond fishing, which don't get me wrong. I like a good pond fish. But out in the surf. Like you got to do some stuff. Yeah.

And it's because you know, basically, after, after about a 20 minute fight, you're like, okay, whatever this is, I'm gonna throw it back anyways, because it's gonna be too big to keep and eat. Right? Yeah. And yeah, so you're like, what's it going to be? It's going to be big shark. It's going to be Yeah, so they can catch a shark on on that day to throw it back. was a two foot blacktip shark. And those are always fun. So people that are swimming out in the ocean, there are sharks everywhere. And and a hurricane a day away. And a hurricane day. Why yes. So yeah, if I'm, if we don't have a podcast next week, it's probably because I could not record anything because of a hurricane.

I didn't even think it but we were talking about the hurricane right before this. But yeah, you're right. Maybe Maybe I'll do a Steven only podcast. Well, we

did one during Harvey.

We did one like three hours before Harvey hit.

Yeah. And then we did one in the aftermath that basically we did we do that at your apartment? I think we did do that.

You're the first one. The one right before Harvey was done in the bomb shelter. And then a few hours later, the bomb shelter was no longer. Oh, and then the next one we did at my apartment? I think I was just I think that was like right after like the roads were like partially drivable? Yeah.

It should be things should be fine. If anything, I'll just start the generator up outside and run my, like recording equipment. Or bring them recording equipment into macro fab to do because they we usually don't lose power there. So Steven, what have you been? Not not doing?

I've been trying to do things attempting? Actually, I wanted to call out. So not last week, but the week before, I'd mentioned looking into solutions for measuring low currents in high voltage applications. Or how do you measure basically, how do you measure current in a high voltage environment. And it's pretty cool. The Slack channel started, you know, jumping up on that so and, you know, kind of given a bunch of ideas. And Tony underscore W recommended a really cool I see that I want to showcase here. It's called the ACS 37002 which is made by a company called Allegro micro. So this this IC is kind of purpose built for doing exactly what I was talking about. It has an integrated Hall Effect sensor that can be ground referenced in very high voltage applications, which that was kind of there's so many ways to measure current, but having the stipulation of it being ground referenced, and then being able so high voltage on one side of my current sensing and then low voltage to a processor on the other side, that makes it incredibly difficult. Like that's the hard part to get by and this ice basically does that. So it offers 4.8 kilovolt RMS of isolation voltage in like a T sub 16 package where one like the left side of the package is dedicated to high voltage and the right side like it's kind of cute the way they worked it out. But so you can power it of up to 6.5 volts so it can be on your 3.3 or your five volt circuits. It has a negative 50 to positive 50 amp sensing range. So this thing is who monster like. Okay, so I think the way they have it set up there. So on a T sub 16, there's eight pins. Let me let me, let me look up the datasheet. Just so I'm not talking about yeah,

there's eight pins of it are set up for the high voltage current side.

Right, right, right. So you have four pins dedicated to one to the to one side, and then forward to the other side, across that one small resistor. So you know, what they they do whatever magic inside that chip to provide 5000 Almost volts of isolation, which is kind of ridiculous. So the thing is this particular chip, because it is meant to measure a high current, it may not be the exact right choice for my application, because most of my application, I'm looking for moderate to high accuracy in the milliamp range at 500 volts. So the thing about this chip is, you get a couple of options with your gain selection, you can you can do some magic with pins and adjust gains of things. And you end up getting somewhere in the range of 30 millivolts per amp to 60 millivolts per amp, worth of output voltage. So this thing just basically spits out an output voltage that's proportional to the current that's flowing through it on the high voltage side. And that's fantastic. And all but if it's only 30 millivolts per amp, or let's say on the high gain side, it's 60 millivolts per amp, if I wanted to measure one milliamp, that would be 60 micro amps per milliamp, or sorry, 60 micro volts per million, right. So that is, we're pretty low in the voltage range there. So you know, solution to that might be to apply gain afterwards and bring it up. But the problem is, this chip also doesn't have spectacular voltage offsets, it's in the millivolt range. But in order to get 60 micro volts per milliamp up to where I would need it to be, I'd have to apply a gain of 1000 or 10,000 or something, well, then I'm adding a bunch of noise. And then I'd also be applying gain to an offset which I'd have to adjust somehow. So maybe I could have a trim pot that would allow me to zero it out and things but I don't know it gets a little difficult. Because, really, I would love to read the range of zero to 200 milliamps with a resolution of point one milliamp. So really straight off this chip at its maximum gain, I'd be looking at six micro volts is what point one milliamps, which you can read six micro volts, but if anyone's tried to do it, it's tough, right? If not, not the easiest thing. So, you know, I could add gain to it. So I was really excited when I saw the chip, because it was like, Oh, it solves all my issues. But damn it, it doesn't. It has crappy gait. And for my application, I mean, it's great for reading amps of current. So, you know, I suppose if you were doing like,

I don't know, like motor, the battery monitoring or something like that,

or like motor coils and things like that, that get that have like high spikes and things like that you could do that pretty easily with a chip like this, especially because that's 5000 volts of isolation, or almost. And so yeah, there's a lot of cool aspects of it. So I want to look a little bit further into it, because maybe I could do some, like offset calibration stuff to get rid of whatever offset but still apply a boatload of gain to get where I want and then you know if I design a really low noise circuit, maybe I can do that. You know, in fact, I haven't even looked at what this chips noise figure is,

or what its actual what its resolution is.

Right, right. I don't know. I mean, it's it's analog. So its resolution is its resolution radically infinite. Theoretically.

That's a throwback to like episode three.

Yeah, when we found a data sheet that had a potentiometer. And it was talking about rotational accuracy. Yeah. So yeah, I don't even know what the what the noise is. I haven't looked that up on this. But, you know, if I were to apply 1000 times gain or whatever to it, this this thing's noise floor would set the the accuracy, the minimum accuracy, right? Because I'd be boosting the hell out of noise. So so I'm gonna keep looking at it, because it might be it might work for my solution for my situation, but it's one of those things where it's like, I hate it, because it's so perfect. In some ways. It's so not perfect in other ways. But when it comes to doing what I'm asking for, I think that's kind of where I'm at. I have to compromise somewhere. You Okay, so I did look at, I did look at another solution. And this is also a compromise, but it might be a compromise, I'm more willing to take. So, unsurprising I'm trying to measure voltages in, in tube land, right, which are, which are high voltage stuff. So I'm trying to measure voltages around a Pentode. Tube. And a Pentode tube has two major conductors that I want to measure. One is the anode and one into the screen. And the way these tubes work is you have current flowing into the anode, and you have current flowing into the screen, they combined together, and they both, you know sum together come out of the cathode. So if I read the current at the cathode, I get the total amount of current that flowing through the tube. But I don't know how much of that is the anode. And how much of that is the screen, I can guesstimate. But I want to measure that that's sort of what I'm going at here. Because in order to properly bias these tubes, you want to know how much heat just the anode is dissipating, so you have to know the voltage at the anode. And you have to know the current at the anode. If you're just reading the cathode, your numbers are going to be off because they're skewed by however much the screen is. But if I can measure all of that, I can get an answer. Now, here's the thing that might make this work out. Well, for me, there is a resistor in series with the screen. And it's a known value resistor. So I could put resistive dividers on either side of that screen resistor, measure the difference across it, and it's usually high enough that I could get plenty enough voltage to get accuracy out of it. So I might go that route, where I just do resistor dividers, read the voltage on either side, and then I know how much current is going through it because I can just divide by whatever the resistor value is to the screen. The thing that sucks about that and where the accuracy goes kind of crappy is I have to assume the value of the resistor on the screen. In my calculations, I just, let's say it's 470 ohms, or one kilo ohm or something like that, you know, typically they're big two or three watt resistors. And I only put like 5% resistors in there. So the tolerance of that resistor is going to dominate the air of my whole system there and I just have to be okay with being five, excuse me, 5% or something like that. It's okay, I could do that. The other thing is, in the cathode, the cathode is virtually never at anything at any high voltage. In fact, typically the cathode is just grounded. So what I can do is I could put a one ohm resistor in the cathode, I could just read across that resistor, and it's only going to be in the it's not going to be much voltage, let's put it that way. So I, what I can do is just get volt, two amp relation off of a one ohm resistor, subtract what I read off of the screen, and since I know the screen and the anode combined together, if I subtract the screen current, I know my anode current, and I can back calculate my power dissipation of my anode. Bob's your uncle, right? So the question is, do I want to get this magical, super awesome, special chip and apply a boatload of gain to try to get to try to get like a direct voltage off of the anode that I want? Or do I just assume 5% error? And do it the old school way of just putting resistor dividers and measuring directly? Like, there's a lot of trade offs in both directions? You know, I

would say calculate your error with that fancy chip and see if it's somewhere near 5%.

Right? Yeah, like just do the calculations of both find out which one is worse? And go the other direction? Yes, that might be what I go for. And honestly, this ACS 37 002 Chip, I was like, Oh, this is super fancy. And it's like, buy this company that doesn't seem like you know, one of the big guys. So it's got to be expensive. And it was it's like four bucks, which you know that that is expensive in general, but I'm not making like 1000s of these. I'm making one. So at first I was like, Oh, this is gonna be like a $20 chip or something like that. So I don't know, it's pretty cool. I like it. I actually ended up designing a pretty cool

connector that is based off of the tag Connect system, such that I could have like a bias test system in my guitar amps, where I just plug in a pogo pin connector into my PCBs, and I get anode screen and cathode voltages for every tube in my system. And it can go off to a box that talks to my computer, and my computer can say, Oh, your anode is dissipating XYZ, which is, you know, 50% of maximum and you need to be set for 70% of maximum. So adjust your bias until you're here. So I don't know it's it's it's a really Really roundabout and long winded way of doing something that you can do with a multimeter, like pretty easily. And like the accuracy of these things is like you only need to be accurate to 10 or 15%. So I'm going way overboard. But this is this all spawned from the idea of like, Oh, that'd be really cool. How do I do current measurement in a high voltage situation? And it's just like, Oh, crap, this is a lot harder.

So that's what I haven't been doing.

I was doing a quick Google on on. I just typed in high voltage, low current current sensing, right.

Yeah. You find something good.

It's interesting. It depends on the company what they determine high voltages

that yes, because because most of the time high voltage is like, above 30 volts.

Yeah, that yeah, that's what I'm finding, right?

It's really difficult. Okay, so high voltage applications, most of those chips that you see are like, multiple batteries in series with each other and things like that. They're meant for like, you know, for 12 volt batteries put in series and you're like, Oh, you need to measure high voltage 48 volts. And I'm like, Well, okay, that's cute. Let's go that times 10.

This one, up goes with the 400 volts.

Almost there, you add another 200 volts on that, and I'd be happy.

And this one's only goes up to 10 amps. So yeah, it's even better. Slow range.

Yeah. If you found something, I'd be surprised because I've done a bunch of searching and, and the slack hive mind has done some searching for me too.

So I'll take a look after the podcast. Hey, if you find the magic chip, I'll go for it. Does my Google fu

so what you've been up to

that garage door opener, I finished the hack. Ooh. So the hack was the turn a standard garage door opener that you know is in the center of your garage and it connects via track to your door to what's called Jack staff style, which operates by rotating this the the shaft that goes across the holds the big spring that can kill you and lifts it up by the cables. So I got a some chain, some cogs put it all together. And it actually worked pretty well out of the box. The only problem I was having was the because I how I had to mount my my opener is the opener thought everything was reversed. Because when the opener was closing, it was actually opening the garage door. And when it was closing the garage door it was actually thinking it was opening it. And so the logic on the button was different. Because like usually you when you press the button when it goes down and immediately goes back up, and then you press it again to stop it.

We did you add another cog in there and do some code inversion.

No, I did something a little bit smarter as I just opened it up in verse the wires in the motor.

That's the electrical engineer solution to it.

Exactly, instead of having to reverse the direction or flip the chain around or whatever. It's actually that was one thing I wanted to talk about real quick was reversing the direction on AC single phase motor. Because this thing has I was like, Oh, it would probably be DC motor, right? And I'll just open up and reverse and I opened it up and there isn't a big rectifier or anything like that, that could handle a half a half horsepower DC motor in there. So I'm like, oh, it's got to be a AC motor. And I'm like, Oh, how do they flip it around how they reverse the direction? Because garage opener opens up both ways. Well, so it's got a split phase motor in it. That is. So I basically started researching how AC motors works actually never thought about it. And usually a single phase AC motor operates with a capacitor, and so you have a split phase winding. And then that capacitor basically shifts the phase of the AC on one phase by 90 degrees. And that's how you have to rotate. And so basically looking at this, this motor is designed with a split phase being exposed. And then both of them go to a capacitor. And then the there's relays on the control board that flip what windy is connected to the capacitor. And so that's how it reverses correctly. Now all I had to do was just flip which which ones plugged into the capacitor which way so I just reversed them. And that actually worked really well except that when I did I put it all back together and then the limits which is that so it knew how far to go. Didn't work so I It's hard to switch those around too. So

obviously, if you switch them at att, did you rewire them? Or did you switch them on the board?

I, so they everything in there is like speed connected together. So I could just unplug stuff and then plug them in different ways. Yeah. And so yeah, if you're doing this and you need to reverse a motor and a garage door opener, you switch it on the capacitor for the motor, and then you have to switch it for the limit switches. Because then up is up and down is not down. Yeah. Kind of confusing. It works. And I did put in the, the switch for the cable tension. And that works too, I was able to test that basically put a chair in the garage door. And when it bumps it, the cable slacks, the sensor goes off and it opens up. So it works great. I think all in I was $110. That's pretty cheap. Versus Buying a specially built Jack shaft one which is 600 ish dollars. Now there is exposed chain and the big cog. So if you're like eight feet tall, don't put your fingers up there.

Don't get your hair cut in.

Yeah, we'll get your hair caught it. But you know, it's like, I'll keep it that way. And if I move out of the house, I'll just put the original one back together, I kept all the parts. It's all reversible. So it's not that big of a deal. Pretty good project, though. I was pretty happy. It all worked out in the end.

So the ever had transformer manufactured for one of your products or projects?

No, I have not.

So I've recently went through the process for for a project I'm working on. And I've done it in the past. And I realize we kind of haven't talked a whole lot about that, like

so is it? Was there a specific reason that you couldn't use an off shelf? One?

Ah, multiple reasons. And I honestly I think that's that's the crux of the transformer matters like getting transformers made for yourself was because you have a bazillion reasons why you would want a custom transformer manufacturer. Particularly the you

can just call up Ross and be like I need 1000 of these transformers made.

I could I get them in a decade. The reason why I went with this transformer is because go figure the off the shelf transformers didn't have the tap I wanted, they didn't have the voltage I wanted. In fact, we designed this entire project around two transformers one that did all the things we needed. And then an extra transformer that is basically that one extra tap that we wanted. And the whole system works great. So it was always our plan to take the two transformers and basically crushed them into one and incorporate in the big transformer the tap of the lower transformer. Gotcha. So, you know, when it comes down to transformer manufacturing, the what I've done in the in the past, the handful of times I've worked on mains transformers, but also some some of the small I call them yellow tape transformers. You know, those little brick ones that you see in switch mode power supplies that have their black with that very distinctive yellow tape around them.

I wonder I wonder if that's the same tape. China companies wrap their boxes in? Yeah,

it's a little too yellow because the Shinzen tape is like, I don't know, it's like a milky brown amber color. And, and those SNPs transformers are like safety yellow is slightly difficult. You're right. Yeah. Yeah. So So one of the things I I've always found really funny is every time I've had Transformers manufactured for me, it's always been different. It's like there's not like a, it's not as easy as other processes. But it's also not as hard at the same time. Like what I mean by not as easy as in every manufacturer I've ever talked to to get transformers made has different ways of doing it. And you have like they're not standardized. I've I've certainly worked on with with some manufacturers where I wrote up a nice spec and did my own drawing for things and shipped it off to them. And then we work together to modify that. I've had others where they were just like yeah, just send us like, what voltage what current you want in an email, and then they made the thing I just did one recently where this guy had a a Word document and like you You don't even you didn't even tell this this person what voltages and currents you wanted. You'd tell them what what it would connect to and then they would determine it for you. And then with with the switch mode power supplies, I've had some wear Like I spent half a day I went over to a transformer manufacturer with my product, and we went over the design and what it would show like, it's never the same. And I think that's what's kind of like special about transformers. It's just not ever easy. Unless, unless they're unless you really just like, a lot of my transformers have been for tube applications, but I've done a handful of others. For other situations, the tube transformers are generally pretty straightforward, because they don't change very much from design to design. You just say, oh, I want 300 volts on this one, or I want 350 volts on this one. And most of the time those guys have a fancy little calculator where they just plug things in, it tells them how many laminations to put in there and what size

they they stole it from browsers website. Yeah,

yeah, I bet you Ross's transformer document is more intense than most of these. So here's the thing, what I kind of wanted to mention, has been my experience with transformer manufacturers, a lot of times, they hire engineers that are very well versed in transformer manufacturing, but also the loads that they attach to. So those guys are really, really good at that. So if you're having questions about your power supply, or designing something around it, and what kind of loads you're having and whatnot, most of the time, I would recommend, contact your transformer manufacturer and see if you can get in contact with one of their engineers and just started a dialogue about that. That's, that's a really great way to start. And in that dialogue, you start to create the specs for what you want. Because like, you can, you can basically, let's put it this way, there's there's sort of three major things that you get to define a transformer and the main use, the manufacturer gets to sort of dial in the rest of the specs with the transformer, you get to pick what voltage you think the tap should be, you get to pick what the current is on that tap. But you also get to pick what's called regulation. So a regular regulation figure for a tap is usually in percentage, you don't do it in volts or or amps, you do it as a percentage. And what the regulation

what it is a measurement of is the voltage at that tap, if you had zero load on on the tap, minus the voltage at maximum load divided by basically, you're getting a percentage of how much the tap changes. You get what I'm going out there? Yeah,

I'm getting what you're saying.

So, you know, like, I bought, I

bought, we call that the same thing with solar panels had the same thing. It was the man What was that figure called? Keep going, I'm gonna look at it like an efficiency, not not efficiency, there was a fishery, but it was a it was a ratio between what open open voltage was and what voltage at at maximum power, efficiency, and, oh max power, there was a ratio for it. Very similar.

Yeah, it's basically exactly the same thing. So you have to you have to know, in general, what your load is going to be like, on your power supply. And then you have to know how, what what kind of variation you're gonna see in your load on that transformer, and then you can spec a regulation figure based on that. And, you know, if you went to a transformer guy, manufacturer, and you're like, hey, I want 1% regulation on all my taps, they're gonna laugh at you and say, you know, get out of here, because like, something, something like that, it's going to be really difficult for them to achieve. But if you say like, Hey, you know, I want my my transformer to be able to handle 12 to 20%, regulation, something like that, that might be pretty loose. In fact, like, if you go to Mouser, and you just look at, you know, whatever 12 volt transformers they have for just like whatever use a lot of times those are 25% regulated, which means if, if you're not driving them at their full load, the actual voltage that's coming out of that transformer is going to be way higher than you think it is. And you need to take that into account because you're now going to start dissipating a lot more heat in your circuit that you weren't thinking like, if you spec a 12 volt tap, it might be spitting out 15 or 16 volts AC, if you know so, at the same time, like if you're trying to spec a transformer, you don't want to spec like just gobs and gobs of output current because that really affects the the regulation like a 12 volt tap at five amps. And say you only needed like half an amp worth of load on that you're going to be way over 12 volts because you're not loading the transformer down. And it's regulation figures gonna make it go through the roof. So these are all things to kind of keep in mind when when getting a transformer made for you is you got what voltage you you basically want how much current that that tap is going to need to provide basically at a maximum and then how much regulation you You're, you're allowing that tap to go up and down based on the load. And half the time, you can just, you know, put that in a spec. I mean, assuming that you don't have like, agency regulations or anything like that to worry about, you could just throw that in your own spec, hand it to your transformer manufacturer, and then have a conversation on how that all works out. You don't have to worry as much about like, what, what grade steel they're they're doing or what enville They're putting on it, or what how they varnish it or anything like that. They usually handle all of those nitty gritty details, you just define the

higher specs on them, you're treating the transformer as a black box, and they build you the black box.

Yeah, yeah. And in the so if you've ever tried to do a linear power supply where you where you wanted to hit a DC voltage, like spot on, it's not the easiest thing on earth. It's and I'm not talking about with, I'm talking about unregulated, say you wanted to put a rectifier put a capacitor and put a load on it off of a transformer tap and you wanted to be within 2% of your, your voltage. That's not the easiest thing on earth. And it's dependent upon so many things. And actually one of the things it's dependent upon that a lot of people don't take into account, it's actually dependent upon the source resistance of the coils of your transformer. And most of the time, you don't know those, most of the time, if you look up a transformer datasheet, you don't get that. So that's where that dialogue kind of really helps out with with talking with a transformer manufacturer is a lot of times you can you can approach them and say, hey, I want X voltage at the end of a rectifier, on my cap, let's design a tap that that provides me that. And I know my mains are going to fluctuate by this many percent. And I'm willing to accept this much load regulation. Therefore, I'm willing to accept whatever range of DC values that are show up there. So it gets way more complicated than what we learned in college, you know, where it's just like, yeah, you have this AC, it goes through a rectifier and there you go.

Now your transformer. Think of it as a spherical cow that has no friction as it moves through the air.

Yeah, for sure. You know, one of the things that I think it's funny, in my experience, designing mains transformers has been a little bit more difficult than switch mode transformers. Because the switch mode transformers are a lot more

unified, shall you say? And they're made in a lot more bulk. Let's put it that way. So most of the time with with switch mode or flyback transformers, you can you can just say like this tap this many turns this gauge wire because those matter a lot more in switch mode power supplies than say, you know, the the other specs I've been talking about with mains transformers. And, and it usually comes out pretty well. And, you know, there's other knobs you can turn if you want to get really deep into it like what like I mentioned with the Grain Oriented steel and do you want em six? Or do you want this or that or and you can ask for like the BH curve so you can see the saturation and when the transformer is going to kind of crap out and things like that. But you know, I for the most part, those are those are academic and educational. Because if you've specified your your circuit well enough, then those should be covered. But within your specs, you know, most of the time you don't have to worry too much about that. So I don't know. I I find specking transformers to actually be pretty damn fun. And

hopefully that's helpful to someone

you're just going crazy for not talking to people at work for like the past eight months.

We you know, we don't deal with transformers too much at work. We try. We try not to touch switch mode power supplies that use flyback transformers and crap, because those get those get really hairy really fast. Yeah. So although like I think I always thought those were those are interesting because my my entire experience with that has been less about defining actual voltages and more about you, you worry more about turns ratios with switch mode power supplies than you do saying specific voltages because you're worried about the ratio of what's going through it. Whereas like with a mains transformer, you can assume a particular voltage through it and you can assume a particular voltage and you do the same thing with a switchmode supply. But the ratio matters more because you have a controller on there and the controller is just going to switch on and off the flyback transformer and it will do the adjustment for voltages. You just want to make sure that that ratio is within its happy zone.

Yeah, within the the what the controller expects

write exactly which nowadays like, you don't have to design your own controller, you just go to TI you find a flyback controller, and it says, produce a transformer with this many turns here, and that many turns there, and bla bla bla bla bla, and that has, you know these things, of course, I'm making it more simple than it is here. But, but a lot of times, that's what you end up with. And then you go get one of these yellow tape transformers, and you're good to go. Yeah, and here's the thing. Remember, earlier, I

said, one of the reasons why we were doing or I'm even talking about this, because I took two transformers and basically shoved them together and turn them into one transformer, the cost of buying those two transformers separately, versus getting five of one of them made, it's cheaper to get five custom transformers than it was to buy five of each, and then put them together. So that's another thing like, if you need a custom transformer, don't be afraid of asking their act, they actually end up being cheaper than you think, especially when you get up in quantity. But I mean, I'm talking about five here was cheaper than just buying off the shelf.

Interesting. I wonder if the setup costs and design is just so much cheaper, or such disproportionate to the actual labor actually building transformers?

Well, here's another thing to consider. And it sounds like that's true, it is true. And and one thing that if you buy an off the shelf transformers, a lot of times, you'll get a lot of taps that you don't care about, or you get a lot of extra wires. Because off the shelf transformers are designed to be a lot of flexible, universal and flexible, right. So you'll have universal primaries, or you'll get different taps on the output that you can parallel or series and stuff. And then you end up having to deal with all these extra wires in your design, that they're great for prototyping, but you don't need it in your final product. So in just taking two transformers and putting them together, I think we eliminated it was between eight and 10 wires that were just extra taps or configurations or blah, blah, blah. And we took a total of like 14 wires for our circuit and brought it down to like, five, I think, because we just didn't need that many. So and and we saved money on it. So think of all the labor for doing the wire and the connection and stuff. Assembly. Yeah.

So if you keep just leave the taps just floating around under the chassis, yeah, you

gotta heat shrink them, or you got to you know, cut them and heat shrink them, or you got to tie them up somewhere or you got to do something with it. Because they're, you know, they'd be happy to conduct if they touch something

would be a shame if we just aren't over there.

I've had that happen. That's not fun. Yeah. Actually, in this situation. It's funny, what we did was, we actually unscrewed the transformer took the end Bell off, cut the wires at my bobbin heat shrunk them and then put the put the N Bell back on it. And this is purely for prototyping, but we wanted to, like, we wanted to pretend like those wires didn't exist. So we cut them at the actual coil, and got them out of the way. So we don't even have to look at them. Which is it's that's so much work, we could have just heat trunk and shoved him in our chassis, but it would have made our chassis look ugly. So no, yeah. For one off, that's fine.

So completely 180 Now we're going to jump over into programming land with zeros and ones and computers. So I don't know if I talked about this before, maybe we're gonna have, if not, we're gonna talk about it again. So I've been working on some scripts for work for our engineering team, and, and for our sales team. And I've do a lot of my scripting in Python. And I kind of wanted to like make a self contained application. Because the great thing about Python is very easy to prototype something together. The downside to Python is it's an interpreted language. So you kind of need you need Python on your system installed. And then that Python environment, it's called an environment, that environment has to have the credit modules installed, so that everything runs correctly. And so something that should be simple ends up being confusing and complicated. Yeah, it's not like I can hand you this text file, Steven, and you wouldn't be able to run it. You know, x, let alone if you had Python installed. You wouldn't run it because you'd have you'd be missing probably a module.

Well, yeah, quick tangent. I want to throw this in there for anyone who is doing manufacturing or thinking about getting their product manufactured. I cannot tell you how often I get emails where a customer is like, I've got my whole system set up and everything's ready. Oh, you You have to do is download Python. And and then you can then you can run my program and test never once has that ever worked ever. And the problem is now I've got a computer that for this customer, I've got Python 2.7 installed. And for that customer, I've got Python 3.3. Because it only works for that. Like, if you're going to set up a system for your manufacturer to test and program your stuff, make it 100% Bulletproof and tell them everything they need to know.

Yes. And so this is actually one way you can do that is I found a piece of software called PI installer. And you run this inside your I'm using like pi charm, which is a, a integrated development environment IDE for Python. And you can install pi pi installer into that. And basically what it does is it compile compiles and quotes your your Python script into a whatever platform you want to target for now it's not cross compiling. So like if you're on Linux, you can't make a Windows exe. The only way it works is. So if you want to make a Windows exe, you have to be on a Windows system to make that work, which is fine. I mean, I got pi charm, I do most of my development on a Linux platform. And then I just bring it over to Windows. And I just make sure that what modules I use are cross compatible. Usually they are the only ones that usually aren't our like maybe like system level stuff or our gooeys graphic user interfaces. And so basically, that was my way of making it so that non technical inclined people on our team can use these things now. Because it used to be Oh, Parker, go run this script. And that happens like 100 times a day right? Now I can I can go okay, here is a tool for you. You can run it whenever you want to run it now. It's great. And so going back to that the GUI the graphic user interface is if you're going to be doing this and so you want to work anywhere, right? You wanted to work on Linux, you want to work on Mac, you want to work on Windows, is you need to pick a GUI that is cross platform. So I was using to tinker tinker. It's actually what I used for the trophy.

Okay. k i n t er to condense

to Kinter? Yeah.

Are you sure that

it's that's how it's spelt?

Yeah. Okay. And it's

a it's a cross platform Python GUI.

Oh, it's TK interface is what it's short for. Okay. Yeah.

And it is a little weird. Just because it's one of the older Python ones, but it's very well documented. And it works really well. And it's cross platform. And I've used it before because that was the I use a version of it for the Raspberry Pi. Trophy that we gave away. Which I wonder if that still works.

Who knows? We can find out the weather on Mars. Yeah.

But b I've tested to keep her on Windows 10 and a boot to 20. lts works great. Yeah, I really like it. And I would highly recommend if you're doing like a Python script for the help like programming your product or anything like that, try seeing it pi installer will help you out in that regard and in packaging everything together. Because you can also the cool thing about it is you can make it if you're a Python script is just a command line scripts. So like you run Python, Parker script.py, or some flags. If you compile it, you can it will actually still work that way. It'll be Parker script dot exe, and you can still pass it flags through that exe wrapper, which is pretty cool. But then the thing is, I can give that exe to you, Steven, and if you trust me, you should be able to run it on your computer and not get a virus. Yeah. That was the best thing I was like. I was like it was like nine o'clock or 10 o'clock at night. I'm like, hey can engineer just download this and test it. One download it and my computer says this is not trustworthy. And like I just yet still worried about it. Don't worry about it came from my computer. It's fine.

You know, okay. Oh, A small tangent, we've we've mentioned this before, but I feel like it always needs to be to be mentioned, if you're developing a product, and you want it to be produced at a contract manufacturer, and it requires some sort of test, or program, or both. The assumption is, and this is a very safe assumption, they only have windows at that office. If if you want them to test something on a Mac, or Linux system or Ubuntu, you provide that system, like, you can only assume that Windows is available, and don't get upset if you provide them some kind of weird Mac thing. And they're like, well, we can't do this, like, I'm not, I'm not trying to sound too butthurt about this here. But if this is one of those situations, where I've dealt with it enough, where it's like, dude, like, if you want us to do something that is outside the norm, you provide it and then everyone's happy and provide instructions to

it, the the issues I started running into is people will send virtual machines, or like live installs of like a Bucha, or whatever, that has all their stuff installed, which works great. Except, you know, oh, the USB hub chip works slightly differently in this computer than your computer. Right? And now we have issues, right? So I actually suggest to our customers is when they're developing this stuff is make it work on a Raspberry Pi. And then just give us that platform, right? Because then you'll because instead of having to provide, you know, a 300 foreign dollar computer, and having to ship a couple, you know, like 20 pounds. Now you're just shipping a little tiny, $30 box that's got everything ready to go on it. And then also don't assume the internet's great on the other location.

Yeah, yeah, for sure. Like if you if you need to, like store and record data, and then send it to something, like, come up with a way that doesn't rely on the internet being flawless all the time. Yeah.

The the way I've like to do that is I use a 4g hotspot. And then what so I've done is have like a Raspberry Pi connected to a 4g hotspot, and then the Raspberry Pi is actually running like a VPN that tunnels through because then you're like, okay, no matter what, as long as it's got 4g connectivity, you're going to get the same tunnel. That seems to be this. It's also safer for security and all kinds of because you're running encrypted over the air. This podcast brought to you by Nord VPN. No, we're not. Not at all not. Now, Nord VPN, if you don't want to give us a lot of money. Yeah.

You want to know a trick I've done in the past that you know, kind of solves that you USB issue. We were we were developing a Windows software app that basically connected to one of our products. And we didn't know like it could have been on any computer out there. It could have just been whatever out in the field, any version of Windows. So what we ended up doing was was writing a a USB port scanner that we would basically open up all of these USB ports. And we will go one by one and just basically send a signal out on the USB port to our and basically say Hello, are you there. And if our device detected, it would be like Hi, I'm here. And then our software in windows would be like This is now my USB port. It would find whatever it is such that you didn't have to plug in go to Device Manager find out Oh, which one is my device plugged into? It did that automatically. And that made it work on any Windows machine without significant problems

unless they hit the 256 number limit for serial ports? Well, we

weren't creating serial ports. We were just pinging them.

Yeah, no yeah. What I'm saying is that they plugged in that many devices well okay, which happens when you're programming a ton of of comport? Oh, yeah.

Yeah. Where are you like a new device found? I'm making a new serial port.

Yeah, but then Windows only go only has an eight bit number two for those comport so you hit 255. And then you have to go in the registry and knock it back, knock it back and then reboot the computer registry?

Right. Been there? Oh, yeah. Yeah, but but but in terms of just like, honestly, doing a handshake saves so much trouble with with customers calling us being like, well, it doesn't detect our device. Well, you can't just plug it in anywhere and it knows unless you have something like a handshake going on, you know?

Yep, yep. Let's wrap this thing up. Let's do it. That was the macro fab engineering podcast. We're your host Parker, Dolman. Steven Greg. And pineapples. Totally okay on pizza. Hell yeah.