How low can the power consumption of the Cat Feeder Unreminder go? Parker and Stephen discuss leakage current on this episode of the podcast!
What is the worst thing about the Analog Devices and Linear Technologies merger? The incompatible color schemes of course.
How do you know if an electrical component is inexpensive? When it says "Cost Effective" as a main bullet point on the datasheet!
Parker is an Electrical Engineer with backgrounds in Embedded System Design and Digital Signal Processing. He got his start in 2005 by hacking Nintendo consoles into portable gaming units. The following year he designed and produced an Atari 2600 video mod to allow the Atari to display a crisp, RF fuzz free picture on newer TVs. Over a thousand Atari video mods where produced by Parker from 2006 to 2011 and the mod is still made by other enthusiasts in the Atari community.
In 2006, Parker enrolled at The University of Texas at Austin as a Petroleum Engineer. After realizing electronics was his passion he switched majors in 2007 to Electrical and Computer Engineering. Following his previous background in making the Atari 2600 video mod, Parker decided to take more board layout classes and circuit design classes. Other areas of study include robotics, microcontroller theory and design, FPGA development with VHDL and Verilog, and image and signal processing with DSPs. In 2010, Parker won a Ti sponsored Launchpad programming and design contest that was held by the IEEE CS chapter at the University. Parker graduated with a BS in Electrical and Computer Engineering in the Spring of 2012.
In the Summer of 2012, Parker was hired on as an Electrical Engineer at Dynamic Perception to design and prototype new electronic products. Here, Parker learned about full product development cycles and honed his board layout skills. Seeing the difficulties in managing operations and FCC/CE compliance testing, Parker thought there had to be a better way for small electronic companies to get their product out in customer's hands.
Parker also runs the blog, longhornengineer.com, where he posts his personal projects, technical guides, and appnotes about board layout design and components.
Stephen Kraig began his electronics career by building musical oriented circuits in 2003. Stephen is an avid guitar player and, in his down time, manufactures audio electronics including guitar amplifiers, pedals, and pro audio gear. Stephen graduated with a BS in Electrical Engineering from Texas A&M University.
Special thanks to whixr over at Tymkrs for the intro and outro!
Hello, and welcome to the macro fab engineering podcast. We are your host, Stephen Craig and Parker Dolman. This is episode 237 237.
forgetting their 255 We're coming for almost almost. We have we have something kind of playing for 255.
We have we have a lot of like random plans for. I think we know it was funny because I was I was. I was thinking a few weeks back, I was like, Oh man, how cool would it be? If Parker sent me like the original files for our intro music, and I would just make an eight bit version of it. And then I was like, wait, no, it's already a
we should do a remix of
it, though. Yeah, yeah.
That'd be we have a couple of weeks. Or Yeah. Or half a year. It's like what?
Oh, that's a long time away. Actually. Now that I'm looking at it is what?
Many 13? No. 17 episodes away. Right.
Yeah. So 17 weeks?
Yeah. That's quite a bit of ways.
Is that next year? I don't know. I don't know yet. COVID has got me like I just, I was thinking about it earlier. And it was like, man, it's August. Yeah, it's August. Like, let that sink in.
You gotta remember back in February, we're, we were talking about how COVID was just influencing supply chains for electronics. Yeah. And now we are bunkered in our houses from it. So. And that was only 10 months ago. No, eight wants to go. It's already
you got COVID brain? Oh, all right. So Parker, either your cat is getting ridiculously fat, or it's not getting fed. So how are you fixing?
Okay, so the cat feeder under reminder, I started designing all the parts and symbols in Eagle. That's about as far as I've gotten so far. Because all these parts I've never used before. So I'm designing them an eagle right now. Is it like almost everything new? Besides like resistance aggressors? Yeah, pretty much like all the ICS I've never used before. And I typically, if I have like a, say, a QFN 16. I will generally not use that one for this part. If the manufacturer recommends a certain footprint, I will use their footprint. I don't go Oh, yeah, I have a QFN 16 I've used before, but now I'm going to use their specification.
Okay, so given that situation that that leads to some interesting. Documentation control. So when you do that, do you name that footprint, the part number or the like, say it's a TI part? Do you then name it? Ti QFN 16.
I name it the part number. And then like underscore the footprints. So when I scroll through the list, I can go the QFN is right there. I don't have to like memorize the part number. But that way when I because I have like, I probably have like 40 Different QFN sixteens, for example, in my library, they're all mostly the same, but they all are probably slightly,
I wouldn't be surprised if like 39 of them were interchangeable.
Oh, I mentioned that you could just interchange all of them if you wanted to. But you know, I always go even if it's like the same tea I won, you know, it's like QFN 16 is just a suggestion of what the package is what is like, like from what what JEDEC standard, right? And but we know that manufacturers will deviate from that they don't have to adhere to the JTAG standard where the QFN 16 Is that can always be different. We hate them for it, but they can do it.
They're different just because they want to be yes.
So yeah, that's a lot typically do. I just, it's just easier. I don't have to worry about, oh, they slightly changed the, you know, they're QFN slightly different in size or whatever, and it's completely borked my board this way, I just I eliminate that kind of problem.
And, you know, that's always been an interesting concept because I have a particular way that I handle things when I look at my libraries and my PCB software. And I've always been the kind of guy where I name things based on its part number such that a unique port number has its own symbol and its own footprint, and we're done. That means more work in terms of, you know, making a lot of things because I don't make like a default SLIC package or footprint or something like that. You know, I may use a generic one but But similar to you, if a manufacturer says like this is associate, here's the mask pull away, here's all this other stuff. I'll make a custom footprint for that. And I like doing it that way. But I've certainly worked at places where they're like, no, don't do don't ever do that. Like, yeah, that makes our library dirty, or whatever, you know,
there's definitely different schools of thoughts. My only takeaway from it is ever since I started doing it this way, like a decade ago, I've never bought a board and the footprint was wrong. Yeah. Because I've always made sure that I've always had parts of it, I've never had a problem with parts, not funny now. And so it's like, clearly that does work.
You know, at my first job, the engineering manager put an arbitrary rule in place that we all followed any new design, you start from scratch, that includes your footprint library. So like, so a per product, every product has its own library of footprints, and you like, it doesn't matter, if you're using a processor over you make a new footprint for it, and it's part of that library. That way, if you ever want to look at like what footprints are used in that library, you just open that library, and, you know, those are the only ones there. Now, there was a little bit of, you know, cross pollination, like what you do with with, you know, unknown six or three resistors, there's no point in remaking that right. But anything that that wasn't a passive component that wasn't, you know, like a regular component fully made from the ground up,
I have seen some documents on chain actually changing the O six, or, like, let's say you had a chip component, like an O six, or three, changing the pad size, depending on the height of the part. So you get your Phillip angles on the on the solder, right? I haven't looked too much. I just seen that in passing, I've never looked pass, like seeing that as a as a heading, basically, and probably an article somewhere. I don't know, if people actually do that.
It sounds like you're getting pretty anal, if you're, if you're shaving off 1000s of an inch of pads, just to, I don't know, to get I've seen enough boards under a microscope to realize that, like, I don't know, the Philips angle is is not something that's like, if you get 10 boards made by 10 Different manufacturers, you're gonna get 10 Different fill in angles, you know, from the same board, because their process is slightly different.
Well, yeah, and when you think about how it's, you know, it's a stencil that goes across the board at squeegeed. And so you will have variations, especially, you know, with a squeegee, different pressures across the squeegee, how far it's dragged already are and how old the paste is, even when you're still in the working cycle, or lifespan of the the pace on the machine. It's still like, at the very beginning, it's like, you know, it's like soft peanut butter. And towards the end, even though it's so perfectly fine, it started turning into more like gummy peanut butter, like attacking, the tackiness goes up. Yeah. It's still perfectly usable and worthwhile within the lifespan of the paste, but it will change how much paste goes through the aperture and gets left on the board. Well, and the sensor is going to pull off.
Exactly. And you remember this from macro lab that you implemented some changes on it. Sometimes somebody might forget to pull the paste out of the refrigerator at a particular time. So when you go to run your first board, it's a few degrees cooler than optimal, right? So it's, it's more of your, your thick peanut butter.
Or it's or it's like ice cream.
You know, we've had an interesting situation at work. It's not an it's not an issue. And in fact, this is one of those situations where like, we solved it by just like trying it and trying it a few 1000 times and it's like, well, it's okay, so we we reprofiled or have and we completely redid our stencil printer. And like basically we revamped our line in general what I mean by that is just like maintenance, it wasn't that it's not that it was bad before but like we fine tuned everything. And every part solder is fantastic. We get some of the most gorgeous solder I've ever seen except for one part. And it's a one and 4148 diode. That is a small guy it's like the size I don't know 402 I think SOT 523 package when it goes down on our apertures in our design, and when it goes through our reflow oven it slightly tips like it kinda like it doesn't angle like rotationally it kind of like a what is that in? Tombstone mean? Well no, not not tombstone in the long direction in the short direction it kind of like leans over. If you look at under scope, and boating terms of healing over there we go. Yeah, yeah, it heals over, not like violently, but every single one of them does it by a by a handful of degrees. And if you look at it under scope, and apply like IPC knowledge to it, it still passes, it just bothers the hell out of me that that are not flat. But we're not done a minor, we've done a few 1000 of these, and they're all fine. And they all work. It's just, it's like, gosh, every other solder joint is beautiful. It's just for some reason, this particular package, that particular paste we use and the reflow profile causes them to just right, just to the side a little
so this is a I think those packages are the leads are flat with the bottom of the park, Corona J lead right there just That's right, completely flat. Yep. So I've messed with those before and had the same problem. And I fixed it by changing the aperture on our stencil to be a kind of like a reverse homeplate pattern. Yeah, like a flag kind of thing. Yeah. So like, the part of the pad near the part doesn't have any paste, and then it angles out at a V. And so the, the part of the pad, the farthest away from the part is fully covered. It's like kind of like think of like an inverted V basically that way. And that works pretty well and keeps it from tilting. Yeah, same. I think those kind of park does always does that.
I think that part it has like a it has a slightly rounded fat belly on it underneath. And so like as soon as you give it the ability to kind of like rock and roll, it just picks a direction and leans a little bit. Yep. It's just annoying, because in like a flawless board, you kind of see a little angle, and it's just like, Ah, this could be better.
You should give that a shot next time. You have a chance. Yeah,
yeah. I mean, the thing is a one in 414. We use hundreds of them every day. Yeah. So it wouldn't be hard to test something.
Yeah. But that's what I ended up doing on that one product. And it seemed to work really well. Cool. So. So what I will do with the cat feeder, and reminder is I'm going to be posting basically screenshots on my packages and stuff I build build for it. And it's been a while since I've built new parts. I think the last time I built parts was for the pennant tour, which was last July. not know to Jeyes guy at this point. Really? Yeah, it was in your basement.
Man, I'm surprised that it's been that long since you've done a board.
No, I've done board since then by had parts already. Oh, your libraries have already already been? Yeah. Yeah. This is the first time I've had to design new parts since then. Which is nice. This is actually what I wanted to do. I'm like, I'd like building parts. It's fun. I think it's fun. Yeah. And then everyone else looks at you weird, but that's fine.
I don't know, I like it. It's kind of relaxing.
Relaxing is not the term I would use. But I like it. Especially like the symbols because like you've seen my symbols, there's a certain way I like to design them. So
I hate it. When symbols have crossed wires, like bothers the living hell out of me, I will go change a symbol to make sure there's no cross wires. Or like, I'll like
ground and power. I will clump those together where they make sense. And put them on one side of the part and put signals on the other side of the part. Like you got to think about the readability of everything because you could just be an absolute Psycho and put pins anywhere and it will work right but like you gotta also think like someone might read
this. The psychos are the ones that design the symbol exactly like the layout is Oh, God. Those are the psychos that's that's, that's awful. That's when I first started designing. That's how I did because I didn't know any better. Sure. The first couple of boards like this is awful to read the schematic. Yeah, and then it kind of switches, the biggest thing I've started doing was is putting my power pins. So it's a it's a IC, like it's got three points or Vcc and VDD vssi. I mean, I usually will put those at the bottom of the part. Like if you think of a rectangle at the bottom, mainly because let's say it needs four bypass pins, bypass capacitors, because as for power pins, that way when I put on my schematic, I put the four bypass capacitors underneath it and then I can connect them together really easily at the bottom there. And that doesn't clutter the top of the part which has all the metadata like the part number and the designator and all that stuff. tend to like that a lot. So I'm probably gonna keep doing that.
Yeah, I'm a big fan of, if a part has a bypass capacitor, then you draw a line from the part to that bypass capacitor. And then when you lay out the board, you lay out that particular bypass cap with that part. I don't like this whole thing. I see it all the time where like, on the processor page of a datasheet, you see the processor. And then down in the corner, you see, like 50 bypass caps, that just 3.3 and ground on the bottom side. And it's like, yeah, it makes sense. I get it. But I also like to see like, this one goes to this pin, and this one goes to that pin. But it does, I, understandably, it makes the processors actual symbol, a little dirtier. But that's where you go, and you change the symbol, such that it doesn't end up looking terrible.
Yep, yep. Yeah. Yeah, I like doing it that way, mainly for it's down the road. Years later, you have to debug that board again. And you go, Oh, what's this part? You know, yeah. And you know, where stuff is actually routed, correct. How it should be routed, just easily to documentation.
It's funny, because I remember you said, You've changed a little bit of your ways, in terms of the way you've ground things, or show grounds on a board to something that, you know, I was mentioning in the past, in terms of readability on schematic, but I've also changed a bit of my ways into the way that you've mentioned, where like, I used to be a stickler where like, you didn't have net names, there wasn't stupid things like a little pin hanging off of a schematic and then a word, and then it just connected randomly on a schematic. I do that a lot more. Now.
My entire boards are schematics, that though,
to some degree, I think that is reasonable. And like, okay, so if I say I have a nice analog op amp circuit, there's not going to be any net names anywhere on that you're gonna see every wire that connects to everything. But like, if I have an eye, I do see bus like SDA and SDL. Like, I don't need to draw wires for those everywhere SPI like, I kind of reserve it to like digital buses, those can have names, or if I have, say, like a 40 pin connector. And I know that there's ins and outs on that. And that's just a board a board connector, I might use just net names on that to distribute everything. But if you have to, like look at a circuit and understand what it's doing, as opposed to just like figure out where it's routing. I'm not using names on that you got to have wires. No, I agree with that. Like what my analog sections are typically, you can see the traces. I still named the nets, though. I've still named nets. Yeah. Yeah. But yeah, the like if you're drawn out a op amp and stuff. That's all drawn? Well, I guess what I mean, by naming nets, like nets still get named. It's just, they're not connected purely by the name.
Yeah, I get what you're saying. Yeah, yeah, you can follow the trace on the screen
exactly goes. I think I think that's critical for analog circuits. For digital stuff, it makes a lot, a lot more sense. Because it's connected to a lot more things. Yeah, to just have it connect by name.
Which kind of good rule of thumb, if you're, if you're joining up an engineering team, like, figure out how the rest of everyone does it, and go along with it or recommend a different way that might make things cleaner. Instead of just trying to be the wild card and do it whatever way you'd like to do.
I just remember the first time you looked at one of my schematics way back in the day, and I was doing the you would name the net for everything, including everything, but that was everything. stuff. We're still connected, right? Like, but maybe like 10%, but I didn't have it's still 10% But now Now I have a little symbol that shows that that's a ground and I have a little symbol that says this is a power line. Hell yeah. I mean, a one.
You know, so I love coloring schematics. Like I have colors all over my schematics. And it helps so much. And, and having like, default colors that that extend across schematics makes things readable so fast, especially in diagnosis. Like if you got somebody hand you five boards and they're like here, do you know figure it out? Just being able to read a schematic. Understand where your power is, understand where your ground is. It makes it so much easier. So I don't know like, the more information you can provide the better in my opinion. Yeah. Cool. So So last week I was talking about a small like side project of kind of monitoring and measuring some some tube activity with a with a little board I have called the 16 and 16 out. And
oh, and this is you had the voltage reference to
write what that what that special voltage reference. So I haven't I haven't built that up yet, because I've actually feature creep a little bit, but in a, in a really good way. So one of the things that I was looking at is oh, so I have to interface between high voltage and let's just say an Arduino. So let's say I have five volts available to me, but I want to be able to read 500 volts. So I picked up a nice little precision voltage divider off of Mouser. The manufacturer is called Cat Ock, CA, D, D O CK. And the part number I got is the 1776 dash c 681, which is for
the most freedom I've ever seen in a part number ever.
That's just empty. So this, this resistor comes on a ceramic substrate because these are, these are precision resistors that this particular one is a point 1% Absolute tolerance, and then the tolerance between the internal resistors inside is point zero 5%. So it's pretty damn good. It comes with like a $14 price tag on singles, but it handles up to 1200 volts. And it allows me to do divisions of 1/10 one 101 1000. And I think there's one more beyond that, but I only want to go 1/100 Because that can take zero to 500 volts and make that zero to five volts. The cool thing is this is on a ceramic substrate, so it has really low temperature coefficient. And if you're willing to fork out a bunch more money, you can get these things down to like two ppm. So if you're if you're wanting to measure high voltage down very rapidly, yeah, very accurately but but easily with with like a Arduino or something like that, you can just slap one of these guys on there. And the total resistance of this particular guy is 10 Meg. So it's not going to provide a massive load to anything. That's a very high impedance. Yeah, really high impedance, you know, you that comes with a little bit of sacrifice and noise. But you know, it's not going to be a big deal for what I'm doing. So so this thing is pretty cool. So I can just plug this directly in high voltage in and read zero to five volts out, and Bob's your uncle on that really easy. So the original idea I had was reading some voltages off and stuff. So that's honestly I've got that solved. That's pretty simple. In fact, with a, my 16 and 16 out, I can get some pretty damn good accuracy, even with the 100 times division. The one thing that I started thinking about is like, Okay, well, what if I wanted to measure current width on the high voltage side? And at first I was like, Okay, this should be a pretty simple task. And then I really started digging into it. And surprisingly, it, it's not that simple. So I spent a good chunk of the week devising up a couple of solutions I had for how do you measure very low currents, and we're just talking DC here, very low currents in a high voltage situation that is load sensitive. So here's the first thing I came up with is like, Okay, well, what if we do this with just some resistors? Right, okay. So let's say you do the classic, you have a current sense resistor with shunt, a shunt, you're trying to measure voltage across it, right. Let's say you're talking about 500 volts, one milliamp. Well, if you're if if that that high voltage is load dependent, you don't want your current source to be or your current shunt to be really high resistance. Let's just pretend you're going like one ohm or something, which is pretty big for a shunt, right? Well, one ohm, if you wanted to do like a voltage measurement across that, you'd have to use two of these Voltage dividers one on the top side and one on the bottom side, right. So you would take both of those values and you're dividing by 100. So you're taking a small current shunt, which is producing a really small voltage to read the small current and then you're dividing that once again down by 100 on both sides to be able to read it. I did some of the calculations to get in the range I'm looking at which is zero to 200 milliamps in that in, you know, a one ohm range on the high side, and I'm buried down in the noise Like, even a 16 bit A to D is not going to give me really good resolution because I would love to be able to actually read, say 100 Micro amp to one milliamp current. So doing a resistive divider on a high voltage side, it does work. But it works a lot better if you can have your current shunt be a high value resistance, and you don't really care that you're dropping some voltage across it. In my case, I don't really want that. So I could probably do it, but my accuracy is going to be pretty garbage. So I started thinking like, Okay, well, how does a multimeter do it? Right? Because you can just shove a multimeter into 500 volts. Kids don't do that at home without
knowing without reading the ratings on your multimeter. But you But technically
you can do it right? And it works, right? You can just set your multimeter into reading current, and then, you know, pass current through it and you can read it. Well, the thing is a multimeter is floating reference, right? So if you're reading five volts across the resistor, and the common mode voltage is 500 volts, who cares? You're reading five volts, you're not reading as long as long as the meter is sufficiently isolated. Yeah, it's it's not just an exposed board touching metal or you.
Worry about as you Yeah, exactly.
So the multimeter idea is novel and unique. And it's like, okay, well, the thing that sucks about that is like, okay, so if you wanted to measure multiple things at the same time into an Arduino, would you have to like create a bunch of unique multimeter style circuits? That all read down to? It just doesn't make a lot of sense. Yeah, you
go to harbor freight. And you get those coupons with a free multimeter. Yeah, and now you have all the ones you need?
Well, okay, but here's the thing I want, I want one central microcontroller being able to read multiple voltages and multiple currents, all on a high voltage side, right. So you got a, you got a five volt system that needs to be able to read low currents on a 500 volt system side. So I started thinking, maybe it's, it would be kind of unique and cool to go with an isolation solution. So for basically make a whole bunch of different front ends, that all have isolated power supplies, and they use isolated digital communication. So I found a handful of chips. In fact, I found one chip, I'm sorry, I don't remember the name of it right now. And one of the reasons why I don't remember the name is because it's brand new, and it's not available anywhere. But I believe it was either analog devices or TI makes pretty sure it's Analog Devices. It's basically a purpose built chip that does exactly what I'm talking about. So you apply power to this chip, it has its own isolation gap inside the chip. So it sends power across the isolation gap. So has its own DC DC converter inside. And then it also has an A to D in the chip that you can read back voltage across the isolation barrier. And it's able to read up to like 600 volts or something like that. So you apply five volt power to it, it's isolated in its own little area. So you can, you can plug it directly into 500 volts, and then read, I think it was SPI data across the isolation barrier back. It's not cheap. It's like 13 bucks for one of these things. But but in terms of like being able to actually read low current at high voltage, this is a pretty novel little chip that does it kind of all in one. So I was hoping to be able to use something like that, because I was reading it was one of those situations where I found the datasheet. And I was reading through and I was like, oh yeah, this is great. This is exactly what I want. And then I go to try to find it. And nobody has it anywhere. And it's expensive. So it's like God dang it. Like I spent days trying to like figure out a good solution to this. And I find one it's it's no bueno. So, you know, if anyone has any kind of like cool ideas on how to measure low currents in a high voltage situation without having much of a burden on the circuit. I would love to hear it and love to kind of like play around with that. Because, you know, the isolation solution, I could always just make that I could hack that in but I would like to do something a little bit more elegant behind it. Especially if
you don't like the wall multimeter idea.
Well, here's one of the biggest situations I want it to be bulletproof. So in other words, like let's say that my circuit I Turn my circuit off. But yet I still plug it into something high voltage, I want it to be able to survive that. Yeah, I want it to be able to survive indefinitely in the high voltage situation, if it's on or off, who cares. And that's the one thing that's that actually makes it a little bit difficult because like, when these units are on, a lot of times, they're like a lot of these run on a on a trick where you can just apply a Zener diode across, across two terminals. And you can use like, I wanted to call it vampiric power, where you're actually powering your device off of the high voltage line. But that's another thing. That that I forgot to mention, I don't want to power my device from the high voltage, I purely want to read the high voltage. And so a lot of situations out there, like, if you look at high voltage multimeter ICs, a lot of them high voltage, that means like 48 volts, and it's situations like reading a, a battery voltage, like you have to, or you have 412 volt batteries in series and you want to measure that whole thing. Well, you can just powered off of those batteries, right. So like the high voltage side gets its own power, and then the low voltage side you power from whatever you're doing on the I don't want that kind of situation, I want to power from the low voltage side and purely read the data. And, you know, I just haven't really done that much before. And I was I was surprised at actually how difficult or how few options I'm finding on doing that. There's a handful of like, interesting models that you can look at, like, you can put specific diodes that have a very known leakage current. So they appear as like a really high impedance. But if you just read the current that's flowing through them on a low side, then you can there's a specific curve of the leakage current, you can kind of glean what the what the model that? Yeah, but like, I don't know, that seems like a little bit more work than I wanted. I like I don't want to be like putting lookup tables of diode curves into a processor. I'd rather just read a voltage. Yeah, so I don't know. That was that was a fun one, just because like, I don't necessarily have a specific application for this right now. It was more like, Oh, I'm reading voltage, what does it take to read current at in this same situation? I'm like, shit, it's a lot harder.
Yeah, cuz, um, you could you could do the coil method. But that's typically you need more higher currents than what you're talking about?
Well, and I'm talking about pure DC. So oh, you know, you can do DC same way I could. But that's another modeling situation, I'd have to know what the field is. And I'd have to know what my sensors field is, you know, things like that. And that's just v is equal to IR is such a nice equation, you know? So, yeah, you're right, I looked into that also, but I kind of shot it down pretty quickly, you could just do it. Current chance coil. If you figure that out. There's also ways to do it with a hall effect sensor I've seen, in fact, I've seen some people model, the inductance of a resistor with a hall effect sensor next to it. You can do that. But I don't like the idea of I have to characterize a component just to read things. Because then because then like, how many sources of error? am I cooking into my equation? By doing that? Because like, well, in order to characterize something, I'd have to use my meter. And my meter is only so good. And then what's the point of having a good solid DAC? If my meters probably not that good? So yeah, calibration is paramount. Exactly. Exactly.
So I'm excited for mainly for your you're testing the voltage, but if you can incorporate the current in there, and for your graphs, that'd be pretty cool.
Yeah, yeah. And I figured there's, with our Slack channel, there's a bunch of really lot smarter people than I unaired that have probably done this before. In fact, you know, funnily enough, I was talking with somebody on the Slack channel a few months ago, about doing current sensing in a in a strange situation, we came up with a handful of ideas just like this. I don't remember if if that actually went anywhere. But it was it was fun, just like, I liked that. The brain exercise of like, okay, if I want to do it, how would I do it? And and my first inclination is to just go to Mauser and be like, is there a thing that does this? Right? Like that just like automatically does it and the answer is no.
Because we're using a IC that does current sensing on our 50 volt line for the Pinball Controller. Yeah. And so, but that's a, I think how that works is it, you basically pass all your current into the chip is it's in line, it has a short inside. I think it's a shunt inside with a with a ADC on the on it. And so actually, no, it's an op amp because you get a voltage out.
And so you have to read that which which, frankly, is not particularly, and it provides your isolation barrier. It's not hard to find an IC that does up to say, 80 volts or 100 volts, it's hard to find one that does 500. Yes, yeah.
And like I said, this wouldn't be a big issue. If, like, I would be happy to use a current shunt and just do voltage reading. If I was reading things higher than one or two milliamps. Like if I was reading 500 volts, you know, half an amp. I mean, that that'd be a ton of voltage, but still like, Yeah, but still, like, that would be a lot easier. But my accuracy is down in the weeds. And that's where it's troublesome. Yeah.
All right. So I've also been working on another project from my garage. And so I'm converting, so this the thing I need to do, I need to convert my normal garage door opener. So it's, it's a style that you know, hangs the middle of your ceiling in your garage and has a track that connects to the door, pulls it open and closed, works great. But I need to convert the normal style to what's called a jack shaft style opener. So Jack's, the jack shaft of a garage door is the little rod that goes above it that's got the spring on it. That turns when the rolls up the cable and basically allows you to open and close the garage door without the entire weight of the garage door. This is like the cyst. So first of all, I can just go out and buy a jackshaft garage door opener. But they're like 600 to $700 and already have a perfectly good garage door opener. It's just the wrong style. Now, why am I doing that? Yeah,
I was about to say like, what's the need? Why is because
in like a week or two, I am getting a auto lift from my garage. And so that opener takes up two feet of vertical space in my garage that a car will hit when I put it on the lifts. So I need to move it to a jack softstyle This is actually pretty common with people that do automotive work in their, at their in their home garage,
you need that extra two feet of garage,
you need that extra bit of garage space. So you you go Jack chef style. I didn't want to spend you know, $607 in New opener, so I'm like, Okay, I'll just, you know, convert it. And so there's a person out there, I'll put the link that makes a kit, that's basically a piece of chain, a cog that goes on your jack shaft. And then you kind of just cobble it together and it fits up there. Right
perfect. I you know, I was I was thinking that you might just buy like a shaft, some pillow block bearings and a cog off of McMaster and just weld all that stuff together.
So I'd looked at doing it all by scratch, way more expensive, really find this person's kit, the kit was way cheaper than buying everything separately. And so I ordered all the parts I needed. And then this last weekend, I took my garage door opener down and was like, Ah, crap, because my opener is a belt style opener, not a chain drive. So it's a belt. So it's got a big timing belt that basically runs instead of a chain, we'll see
that's where you will the sprocket to it right, ah,
getting to that I have a Liftmaster Formula One is the brand or the model number for it. But I've taken a lot of these apart and fixed a lot of openers over my life. And they if they're like a they're like a Liftmaster Chamberlain or a craftsman they're all owned under the same company. And they all they just have a different outside box, or they're the same box with just a different logo on it. And so they all use virtually the same thing. They might have a bigger motor, but the draw how it drives the cog or in my case it was the pulley, the tooth pulley. It's all the same. All the same guts inside it's basically a nylon worm gear with a nylon. What's uh, what's the angled a gear called? Oh, yes, shoot, what
is that called?
But they're all the same. It's all the same drive, no matter if you have a third horsepower or quarter horsepower, half horsepower, all the same drive helical gears, helical gear, so it's all the same drive in there. So I pulled it out. And so I'm showing Steven Oh no, this is what this What the mech looks like. So those there's the pulley, you're such a pinball guy.
You call it a mech, and call him back.
So this nylon drive gear, so you're the big motor hooks up to here with a worm gear. I'll take pictures for the, for the
post. So once again, electrical engineers trying to be pretending like we're mechanical. What do you think the reason is that it's Nylon is it for like, where it's,
it's no, it's to allow it to strip out if something bad happens. Okay, because you have half a horsepower, scorching something, right? And given that the nylon gears really cheap, and this is easy to replace you like you don't have to really take apart the garage opener that much to replace it. Yeah, just take that shroud off. Yeah, the shroud the trowel, which has the lower sleeve bearing just unscrews and then you can just pop a new one back in. So it's actually very simple to fix these things. And the but yeah, they all use the same geometry. And so my first tried to do was okay, well, I'll do is I've seen this part before, but it's perfectly usable, the gears still in really good shape. I'll try to press the pulley off. Okay. And that's all I tried to do first tried to press it off, could not press it out. I tried heating it up with a torch could not get the press out. So it's probably it was heated up. And the shaft was cryogenically cooled. And then friction fitted together. And it's there one piece now.
You know, that's how they do aircraft landing gear. Oh, is it? Are they pressed fitted? Heat? Yeah, yeah, they're, you know, like you said, their temperature difference, and then they they press them together, and then they just basically fuse into one.
Yeah. And so my, okay, I can't get that apart. So I'm like, Oh, I can just bring it over. I took it all apart to just the shaft and the pole and the pulley tooth pulley. And I'm like, I'll just take us over my friend's house. And we'll just leave the pulley off, you know, easy, and then we'll just put a new cog on. And so I started looking at, okay, now you need to get the cog, so I can get what size we have to turn that, that diameter down to. And to get a cog, it's like 20 to 25 bucks plus shipping on McMaster. So I'm like, That's not too bad. That's not bad at all. And then I'm like, You know what, I wonder how much it costs just to buy the whole unit with the cog already on it. $18 on Amazon. So I'll take a picture is
Yeah, they look identical. They're identical. Except what? Oh, yeah. What was the bill? And what's the kg? Yeah, exactly.
And I haven't installed yet when installed after the podcast. But yes, it's identical. So $18.00 Work verse half a day on a lathe and $25
I guess. Thanks, China.
Yeah, so actually think whoever the parent company of Liftmaster Chamberlain craftsmen are for garage door openers, and making universal parts that work in everything. So I can take a belt drive model part and shove it into my my no chain drive model part and shove it into my belt drive unit. And so other things I needed to do this because there's some other things if there's another person out there that's listening to this that also wants to convert to a jack shaft style garage door opener, is you need a way to lock your door, the garage door. Because in the with a typical opener, your your motor is on a worm gear, and you can't you can't force a worm gear backwards, you have to drive with the worm gear, you can't drive the worm gear backwards. And so your door is closed with that gear sets so you can't open up your door. But with a jack shaft, you're only locking the jack shaft and so you can technically if you are like Hercules, and you can lift the door because you won't have the spring to help you. Because that's on cable. So the cable goes slack we've tried to lift it that way. But you know, three or four people can easily lift the door and then it's unlocked. And so you have to get a like a deadbolt style lock. So I got a electronic once called the Sherlock Garage Door Deadbolt. And basically it like it like it has a little unit that sits on your garage door opener. And then so if it detects vibration, basically the motors turning right, it unlocks the lock. Hmm, it's pretty cool idea doing it that way.
I thought I thought you were just going to do a deadbolt into the guide rails. Like
no that's what it is. It's a dead bolt that goes in the guide rails. It stops the wheels from moving or one of the wheels from moving right. Okay, gotcha. Yeah, that's how It works, but you need to have it to automatically open. And so you, you have it open by it detects the vibration of the motor turning and then unlocks. And then the other thing you need is because you're now moving your, your, your don't have a physical, you don't have a rigid link, that's a good way to put it, you don't have original link between your motor and the door anymore, because you have this cable is if the door is closing. And let's say there's something underneath the door, that doesn't trip the line beam. So it's like, you know, the person standing there or whatever. Or like a table or what is really not what it is, you're gonna get, like a torque sensor or something like that? Well, no, that's what the openers have is they have a torque sensor, but because they're originally coupled to it with the with, with the motor gear set and the belt, that event hit something, it says, Oh, we have a high torque situation on the motor reverse. That doesn't work with cable when when you have a flexible connection, because then the cable just goes slack. And then all the weight of the door goes on to whatever your you know, whatever it's hitting. And so you fix that with a cable tension switch, which basically it's a it's a on off switch that is under tension from the cable. And when the cable goes slack, it the switch opens up and it goes okay. The door is that rest somewhere. So I need to turn off and reverse
seems like a lot of work
a little bit, but it's way cheaper than buying a a jackshaft I think I'm only like $90 and all the parts versus the jack shaft openers like 700. So okay. You still need like the cable tension switch and stuff with the cable. The Jack shaft, same problem.
Well, okay, here's the part that I'm a little bit confused about, I guess. So the problem is you need some clearance in your garage, right? But will your Oh, okay, but you don't need clearance to get into the garage. You just need clearance once you're in the garage once you're in the garage, got it? Okay, because I was thinking like, Well, I mean, the garage door itself hangs lower than what you're talking about. But it doesn't matter.
Well hang out the garage door is not as low as the opener. But yes, is if you have a car all the way up on the lift, you're not going to be open up the garage door. Right? Okay, got it, which is fine. I don't mind the garage door in my garage is not the main egress. I have a side door. That's the main egress. So like you're not locked in when you have a car up on the lift.
So when does when did you live arrive?
It's supposed to ship sometime this week. And then it's coming from I think Alabama. So whenever it shows,
is it a kit? Or is it like fully assembled? It's full. It's fully assembled. Cool. So so you can just start lifting stuff right when it shows up?
Well, yeah, I got I got drill holes in the concrete and stuff and install it. Oh, yeah. Okay, like this is one of the first thing I'll do is I'll sit on it and haven't lift me up. Of course, right. Yep. And then lift. Then you start with a smaller vehicle you have lift that's in progress up to the largest vehicle. You got tested.
Right, right. Like that big gantry crane I had in the Oh, yeah, that warehouse. The very first thing I did was like, you know, strapped myself into the crane and drove around
Where's him suspenders and hook it into your back loop?
I mean, we yeah, we had some we had some nice lifting, lifting ropes. So just do a bowline on a bite. And then, you know, strap your legs in and go to town. Yeah.
Cool. So I tried out something that I've been wanting to do for a long time. So in PCB layout design, there's there's some shielding techniques that you can employ that I've done on a handful of boards because just I've been told it's good you know, just general practice I know that it's reasonable to do but I've never tried the situation of like, okay, what what does it like with and without the shielding technique? So basically what I did was I designed with my buddy we designed a full amp on a on a PCB, so everything is all on one board, all of your amplification stages, your power supplies, everything on on one board. So a variety of different signals are all in one location floating around. What what we did was we designed the entire board and then I created to ground pores. Well, I polishes not ground force, I created two copper ports, one on top one on bottom, that are not connected to anything but each other. And on a five millimeter grid, I put via stitching across the entire board. So the purpose is to make them equal potential as much as I can everywhere, and to cover every open space with some kind of copper. So I took both of these two planes on the top and bottom of the board. And I connected them to one pin of a two pin header. And then the other pin of that header, I connected to a mounting bolt that the PCB mounts with to a chassis. So basically, if you short that header, those two copper planes that are not connected electrically to anything on the board, end up being an extension of the chassis itself. So if you think three dimensionally like, it's like this rectangular chassis has now grown to be the shield also. And that's the whole point like not, it's not connected to ground, it's connected to chassis. So currents flow in the ground for all the associated circuitry. But any current that flows into whatever this plane is, flows directly to your safety Earth, on the on the chassis itself, because this is a mains connected device. Well, then
technically, it's coupled to the live line. Great wire. Yeah,
yeah. So so the the whole point of this was like, I wanted to see, does this actually work? Or? It works? Let's just put it that way. It works. But I mean, neutral? Not live? Well. Neutral? In the box? In the box? Yeah. Yeah. In the outside my house. So it's like, way up. But yeah. So the so the concept was like, Okay, how bad is it? If I don't short this, this shield? And how good is it if I do short this shield out to chassis? Or it could be the exact opposite? It could be you never know. So one thing I can report is that it's noisy as hell and it's awful. And there's a lot of interference when it's not shorted. Which, okay, great. Like, that makes sense, right? Like, there's nothing surprising there, there was a lot of 60 hertz, um, that was coming from all the all the transformers and everything, there was a lot of high frequency buzz, there was plenty of 120 hertz stuff. And I'm not talking about just like audibly, like, if you stick a probe anywhere on the board, you'd get crap. And a lot of that was because my rectifier was spitting stuff everywhere. Actually, I had multiple rectifiers on there, because I have multiple transformers and voltage rails on this thing. So if you think about it, if you have this sandwich of of copper, that is just not necessarily connected to anything. It's just copper. You've in a sense, you're capacitively coupling everything to everything, right? Yep. And depending on how the copper surrounds everything, and plays with everything, you have different values of capacitance everywhere. So like, you just have a mess on your hands. And it's, it's entirely unpredictable. Which, once again, none of this is surprising. It's just I've never, I've never spent the time actually trying it. And then now I was like, Finally, I'm gonna try this. And so as soon as you short that, those copper planes to ground, everything went dead, silent, everything is shielded from everything. And like, I mean, to the point where
from audio testing, I hear resistor hiss. Like I hear the background noise of resistors it's quiet enough like No 60 hertz, hum, no buzz, no hum, no nothing. So it's worthwhile for sure. Like, and as long as you maintain proper clearances to everything, I don't see a huge reason, in low frequency circuitry to do things of this sort, you know, it just, it helps that much. And if you do have a circuit that that isn't, you know, say connected to chassis ground, you can still do this as a ground plane, as long as you have one connection, it only connects in one place at whatever your main star ground is, which is typically right at your power supply. So you know, it's it's a solid benefit. Now, the the one thing that would be really fun to test again, is to do this exact same thing with the exact same circuit, but not via stitch the entire thing. So basically have the planes on top and bottom of the board where both of those planes only connect to each other at 1.1 Point as opposed to across the entire thing. The whole point of like the via stitching was to just provide maximum ECWA potential across the entire board, right? And so success. There we go. Once again, both situations not having it shorted and shorting it out. None of that is surprising. It was just like, it's been years that I've wanted to, like, just prove to myself that it does something and it does something, you know, that the biggest issue is like, if you if you're planning out a system like this, you have to, you have to be really careful with how your EDA tool handles that, especially, like, it works really well the way I did it, because I had everything tied to one pin of a header. But if you want to tie that to ground, well, most EDA tools are going to be like, well, everything now connects to all of these things, you know, yeah. And that makes it kind of crappy. In fact, you'd
have to make basically you have to make two grounds basically, and then stitch them together at that. That point.
Right, right. You're different nuts. I have not found a EDA tool that handles shielding at one point. Well,
yeah, it because usually, basically in EDA tools is if a net, a net, you can't have two nets connected. Right? As I haven't seen one yet either. That allows you to connect to nets with a virtual net. Basically,
I've seen one situation work for this, it sucks because it throws a DRC error. But as long as like, you know this is the only DRC error, then you're fine. You can indep trace, at least you can create a garbage footprint that is like say a no 603 resistor where you purposefully overlap the pads. And so it's just a chunk of copper at that point. And but it considers the two pads as two separate elements and to separate nets. So it would work but it would always say like, Oh, these two things are touching and as long as you ignore that, but I hate that I hate having any DRC errors.
i That's how I've done it in the past. I've done a thing I took a 12 six package and then I just took one and I just drew a big piece of copper in between in the footprint and I'm like well that's that's just how it's gonna work. Sure.
Yeah, it just it just works, right? Yep. Cool. Yeah. Just a fun experiment.
Yep. That was the macro engineering podcast. We're your host Parker dome and Steven
Greg. Later everyone. Take it easy.
What is the worst thing about the Analog Devices and Linear Technologies merger? The incompatible color schemes of course.
How low can the power consumption of the Cat Feeder Unreminder go? Parker and Stephen discuss leakage current on this episode of the podcast!
How do you know if an electrical component is inexpensive? When it says "Cost Effective" as a main bullet point on the datasheet!