Relay manufactures hate this one simple trick that makes your “sealed” relays last longer! Except TE connectivity who has an note about this relay feature.
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!
Welcome to the Mac fan engineering podcast, a weekly show about all things engineering, DIY project manufacturing, industry news and Python coding. We are your hosts, electrical engineers,
Parker, Dolman and Steven Craig.
This is episode 316.
So I got a new project. Well, I say New. This year, I've been going through the list of like, what's the oldest project finish that what's the next oldest project finish that can go down the line. So in that series of projects, I've got one that I've been wanting to do for a long time. And it has something unique about it. It This project involves star grounding. And we spoke a little bit about grounding last week, and we weren't going to talk about this topic last week, but we pushed it to this one. Now, we've talked about grounding in the past about different methods of grounding in your circuitry. And for the most part, I think, with the projects that Parker and I work on, we do, like plain grounding, or, or even like split or broken or cut planes, but on our PCBs, it's mainly you plane or we, I think we've used the words plane and plunge before where it's, you know, take apart, go to a V drop down to your ground plane. And typically, that's
why one layer is essentially your ground plane.
It's your zero world, your reference plane. But But I kind of so on this new project I'm working on, it has brought up an interesting conundrum. And and the what I wrote in our show notes is how big is too big. And what I mean by that is, let's just pretend like you had a board, that was that you had a ground plane on it, right? But that board is 20 inches, 40 inches, 100 inches wide like, it starts to get to the point where like, that ground plane on the proximity of things or the distance you have to travel across that ground plane really starts to matter. And in this project that I have right now, it's sort of working out that way. So this project I started a long time ago. And it's a kit that came as part of a forum that I was involved in many years ago, where the basically we all purchase one, one gentleman develop this board, and we all purchased this board. And basically reference designators are all over the board, but no values, you got to populate it with whatever you want. So you could sculpt it to be however you want it to be, there was a few things that were predetermined, just because they weren't required for like housekeeping in the circuitry. But most of it, you got to pick yourself. Well, one of the things that was predetermined was General ground zones on the board. In fact, there's nine architecture, right, just overall like layout concept. There's there's nine zones on this board. And this board is, by the way, enormous, it's like three inches tall, but it's like 20 inches long, it's this huge like hockey stick of a PCB. And it maybe maybe not 20 inches, I might be exaggerating there, but it's not too far from 20 inches. So each sub circuit on this board has its own ground plane. So this this is planes. But the each plane service is only it's a handful of circuits that attached to it. So then there's nine individual like squares of copper basically on it. And then each one of those planes has a pad on it, that's just call it a ground and then a number. So ground one through nine on it. And then the whoever builds, it gets to choose how you ground each one of those. So you what you could do is you could run a wire from ground nine to ground eight, and then a wire from ground eight, around seven. And you could daisy chain them all the way down the line, and then pick any of them at any of those locations. And you could run a ground to the chassis and then everything is grounded, right? You could you could also say pick ground one which is one far end of the board and ground that maybe you ground it near where jacks are or other external components on the on the chassis. It's sort of like pick your own adventure or choose your own adventure with it. But one of the recommended ways of grounding this circuit. I've always wanted to do but I've never, I guess you could say never had the balls to do it. In other words, never like never been willing to invest all the money in a design to try it. But since I don't get to pick this design, like I'm gonna give it a shot. I've never actually done star grounding in this way. So the way that the the recommended way that this circuit is built is each one of those ground planes has an individual wire that is attached to it, and then sent all the way to one ground point on your chassis. And technically, when you're doing like a system design, especially with mains, you're going to have two connections to your chassis and to only, you're going to have one near your power inlet where your mains enters. And that should have some kind of Loctite, or star tooth washer or something that digs into the chassis. And that's your protection Earth, that's your, that's your safety ground, that makes sure that the chassis can never become live. And the user can never touch something that becomes live, because if anything does short to the chassis, it'll blow the fuse from current flowing through that. So that's your safety ground, that's one of your ground connections that happens in your system. The other ground connection is your reference for your circuit. And that is what I'm talking about here. My star ground is where I run an individual wire from each one of these planes to this star ground. And I'm actually gonna build it that way. Because it's outside of my normal way. But I really want to try the performance on it. And it's got, it's got me thinking a lot about this zero land, this ground world, that most of the time, it's not that I don't pay attention to it. But like, once I've established my rules on it, like I connect to it. And that's that, right, it just, there we go, there's my return path. But so in this situation, the reason why I wrote how big is too big is in this particular situation, because my PCB is so large, and I have a star point at one point in the chassis, some of my ground wires from these ground planes are going to be in many, many inches long. In some cases, some of the grounds are pretty far away from my store ground where it actually exists on the chassis, upwards of 1015 20 inches. And that's a long way for ground current to flow. And because each one of these grounds is individually separated, but power flows between the circuitry on the grounds, that means current has to flow between multiple of these ground. So So in some cases, I've taken ground loops, or the ground path current in this circuit, in some cases, I've taken it from what could be an inch or less, and made it 30 inches. And you know, like gut feel is like, that's not good. Right? Yeah.
You I foresee that being basically a big antenna on your, in your in your ground return or your current returns basically.
Right,
right. Now, how bad is that? Depends on the impedance, right of your of your circuit return?
Well, it depends. It depends significantly upon what frequencies of interest the whole thing does deals with, right?
Yeah. Like you could think of like one thing I was thinking of was, you mentioned like your, your signal current to says like how much current is like the, if you had a lot of current flowing between these sections by passing one to the next section, then you get a lot of current on those ground returns, right? Because they have had to come back the other way. Thus, you start getting if you go from one inch to 30 inches, and you say you're drawing an amp, now you're talking about accurate, significant voltage differences between different stages. They're both references. Right, right. Well, so it really depends, I think, is is on, like how much current is actually flowing? And is that actually going to be a does actually matter?
Exactly, exactly. And that's, that's the whole fun part about this project is, like I said, I kind of had, I had kind of always been averse to doing it this way. But because this project is set up that way, and I don't really get to choose, I want to give it a shot and just see how it works. And I'm actually soldering the wires in such a way that if it doesn't work, I could change it to something else. In fact, I was I was lucky to get a 3d printer recently, after we had discussed on a previous episode, like should I get a 3d printer? I was able to get my hands on one. Thanks Chris Craft for that. And so this this 3d printer is great for printing little brackets and things like that. So in terms of doing wire and cable management, I'm making little brackets that will help quite a bit with doing this and make it really easy if I want to switch topologies, let's put it that way. But, you know, there's one argument about if you're doing daisy chain grounds, so in other words, you have, you know, you have one part of your circuit feeding another part and then further down the line, it feeds another part and you continue that however many stages, you have the impedance of every ground return some in that style of grounding, and that. So if you have significant currents, then something further down the line can impact something much further up the line. But if you do true star grounding, where they all returned to one point, then it separates each portion independently. Such because your ground recurrent is for each stage returns on one wire, as opposed to all the wires summed together. So it's a totally frequency dependent. If he asked me, I think lower frequency circuits may actually benefit from something like this, I wouldn't ever dreamed of doing something like this in something that has like a switch mode supply or something that was high speed digital, but if you're working in lower speed analog stuff, I can see the benefit from it. Because Okay, consider set steady state, like, okay, obviously, this is an amplifier, because that's, I do like 90% of that. But so consider steady state, that's the situation where I want the least amount of interference and the least amount of noise. So in steady state, we're talking about dc currents on all of these things. So if we're talking about DC, across 30 inches of wire, that's already low enough impedance, if that provides me lower steady state, no input signal noise. Well, that's preferable in that case, because then much more quiet as a whole. Even if it picks up more noise during operation, in other words, during a signal flowing through, say, those act as a bit of an antenna, well, if I'm rejecting those frequencies anyway, during the dynamic conditions of having a signal flowing through it, I may be able to, let's say, deal with any negatives that come with it. So I'm just excited to try it. Because it is, it seems counterintuitive to everything that seems right. Like grounding to me always seems like get to my my ground as quickly and as short as possible. And provide the least amount of interference to there. At the same time, try to establish as as good of a reference as possible with my ground. And this seems like it meets some of that criteria where my reference is, it's going to be great because everything is referenced to exactly the same point. So every so if Well,
not really still, though.
Well, that's the whole point of star grounding, though they are because they're all the grounds meet at one individual pins
on it's all the wires of the same length in yakugaku.
Oh, okay. Yeah, no, no, I see what you're getting at. Yeah, what but okay, in a situation with higher current, I think that, of course, that comes into play, because now you're talking about voltage drop across the wires. However, even in this situation, if you if I had one of my stages, that had higher current and I had voltage drop across that wire, it would only apply to that stage, because everything else has its own wire to it. So if voltage drops more across that heavy current stage, none of the other stages, shall we say, see it. Whereas in the daisy daisy chain situation, depending on where that that heavy current stages in the string of things, it could affect other things. That's where the shared impedance versus separate impedance is really attractive in this situation. So regardless, all all this this boils down to is nine individual chunks on one giant PCB are all going to have nine individual ground wires that all bolt to the chassis at one point. So I'm really curious to see if it makes a difference.
Make all those wires the same length.
That's going to be really okay. So the hardest part about that would mean that some of those wires will just have bundles of extra loyal Yeah, at the same time, I'm putting, I'm strategically putting my Starpoint to be closest to the most sensitive parts. So the most sensitive parts have the shortest wires, the least sensitive parts have the longest, least sensitive being like the input rectifier and whatnot. And it's important to note that the way that this circuit works is it's not every single ground in the entire circuit flows to the star ground. So each circuit has all of its own individual ground return path, so like is contained It's contained, right? So there's a capacitor that feeds each chunk of the circuit. So signal currents don't flow through the ground return wires, signal currents are contained within their own little loop. So in terms of their ground return path, it's actually short because they're flowing within that little ground path in the chunk. It's just the interstage. currents
will flow the chunk the chunk, the chunk the
chunk around. I like that. Yeah. Yeah. So So perhaps there are some tangible results from this. I'm wondering if I get lower noise, lower interference and lower chance of oscillation. Because of this. I don't know the all the reasoning behind it. But I've been told that you get better better radio frequency rejection by doing it this way. But I don't know if there's empirical data, I could give them something like that.
I always like thinking about these kinds of problems. Similar to those problems you get in in college where like, if you have a one ohm resistor array that's infinite. What's the resistance between A and B, right? Yeah, like biggest thing of the problems like that. So it's like, what if you had the star ground, right? Yeah. Then what if you made the, the wires infinitely short. And the star grounds infinitely big, which is basically what a
that's a ground plane, that's a that is, is a really simplistic way of looking at a ground plane. Yes,
that's, that's like, when you say Star ground, and you just turn one knob the other way, and the other, not the other other direction. But for the for a star ground, you get this plunge in plain style.
Exactly. And I have a circuit that I'm dealing with at work right now, that has, I don't remember how many components it has a gazillion components on it, and lots and lots of different varying currents and lots of different varying power supplies, I have six different power supplies that are all doing different stuff, it's got digital, it's got analog, it's got a bunch of channels that need to be separate from each other. And you know what, I did plunge in plane, I did a big old ground plane, and there's excellent channel to channel separation on it, even though I've got current flown all over the place. Now, here's the thing, I did chop up my ground plane a little bit, I knew one area of the board that had extra current. And I saw I, I routed all my power traces a particular way. And I made sure that return paths followed those power, such that those were not near anything sensitive. And then I actually chopped up my power planes to flow individually to all of my channels. So I couldn't have, I couldn't basically have power flowing from one channel to another channel, they were in their own buckets in a way. So I did a lot of work making sure that my power was routed in chunks, but my grounds I actually let them flow wherever made the most sense for them. And the separation is fantastic on it. So that's a situation where like, yeah, the wire length is really, really small. The star is really, really big, really, really big. But the question is, how big can you make that star before it stops being a star?
Yeah. Interesting question.
It is, yeah, you know, and actually, so I've seen a lot of this is not a good way of doing things. But I've seen a lot of people use the chassis as the star, just drop on and bolt to the chassis wherever you can. Because, hey, it's a big huge monster chunk of steel, right? You can just show you know, plug to it, and there you go. But that usually ends up with bad, really poor results. If you get
that. That's actually how a lot of old older cars and a lot of people who work on cars, that's you consider the chassis as your ground.
Yeah. And you allow current to flow through it
through it, which is fine ish.
Well, let's just put it this way it works, but it's not ideal.
No, it's not ideal. So what I do on all old cars is actually kind of, I do kind of like hybrid, basically, if, like, if it's under an amp is my rule, but under an amp, it goes the chassis, okay. If it's over an amp, I'm going to have a return wire that goes back to the battery. Yeah, yeah, it's right I usually will do is I put a terminal like a ground terminal thing and the engine compartment and put one in the back of the car. And so then a big wire that connects them together and then you know, onto the battery Guess what, I never have any electrical problems that everyone else complain about all the cars,
I had a fun,
but it's going to you're kind of like each area is kind of like localized, so to speak, yeah, then you have a star ground, which ends up being your battery post, right. And then today, and then
in a lot of ways, like whenever you purchase equipment, like your computer, or your monitor, or the light bulb or whatever, everything that plugs together, they are being star grounded, right? Like everything like your computer, and your interface and your monitor, they all get grounded at whatever strip you plug them into, or the wall, like they're starred together at that point, right. So I mean, it does work. But even then you can run into issues. Because if you have significant current flowing between those two, that has to flow down to the star, and then back, you run in with some, you run into problems with that. So I'm just doing the same thing, but inside my chassis. And you know, I think I had a problem with my first car at a Nissan Sentra back in the day a little, little tiny guy. And if you had the radio on and you accelerated, you could very very clearly hear noise increase with the RPM of the of the engine.
Oh, you get a little ignition ignition noise on your on your radio antenna.
Exactly. And I would not be surprised if if I ran a dedicated ground from the radio to the battery negative terminal. I bet you that go away?
Yeah, probably go away. Yeah.
So yeah, circuit grounding is. I like this situation where it's like, okay, I don't get to choose someone else chose for me, and I'm choosing to make this project. So I play by their rules. And other people have built this have reported positive results. It's just, if I'm going to spend the money doing a project, I usually just don't go this route. But if this has good results, maybe I'll start to incorporate that in future designs. Yes, probably not.
For like The Pinball Controller, I designed the current one printer. So how it is is basically anything it's, I have a it's four layer board. So signal, 3.3 volt ground signal. And it's all divided up for like, because there's like a 12 volt line is based is 12 volts and five volts is also got to come off the board to do other pinball stuff. But when those come off the supply, and those go to the connectors, and they hook up to like MOSFETs basically, for example, because you have to control some lights. Well, that 12 volt return actually have its own ground return for that. 12 volt. Yeah, because I want that to also go, basically, because like your power inputs on one side of the board, and then you have 11 inches of board where you're picking up that power on the other side, right? Well, I don't want that 12 volt ground return to also just mingle with the logic level return stuff, the 3.3 volt stuff. Yeah. And so it has its own ground, even though it's referenced the same. So actually, it's like a star ground on the right side of the board. So it goes so it has a 12 volt line, it goes away cross. And then the returns of those basically those MOSFETs I guess it'd be the at the gates, not the door Sidra drain,
right? Well, no, the other way around is probably the source. You write
source. Yeah, source. So the sources of all the MOSFETs are connected together, and then it has a trace that goes all the way back parallel to the power of that 12 volt. All the way back on the side. Yeah,
yeah. So it works really well inside the cabinet of a of a pinball machine. Is there uh, is there like, a designated star ground that everything gets connected to?
Like, the rails and like the legs and stuff? Yeah, they're connected to like the ground pin on the, you know, hundreds.
Okay, so like, okay, so take in a pinball machine. Okay, I'm just guessing here, but you've got you've got some kind of thing for the speakers. You got to you got to an audio and for the speakers, you have something to drive the screen. Right. Then you have some kind of a computer system. Something to drive all the solenoids and all that stuff, and lights and things and that's your board. solenoids and lights and yeah, and all Macs and things like that, but low level control. So those four different sub systems, how do they all get grounded together? So audio amp screen drivers.
from computer to our board, it's grounded through the USB connector. Okay, cuz that's how they communicate. Yeah, yeah. And then There's a 48 volt, and a 12 volt and five volt power supply. And those are connected on the AC side, of course, but then they're all their outputs are also connected to low grounds are connected. They're
all tied at the power supplies, right?
No, we actually connect them. The five volt and 12 volt are like the same power supply. Yeah, so they have a shared ground already. But the 48 volt we actually connect those on the board itself. Okay. For that's for better referencing, basically, because like, the closer you can connect those together on your MOSFETs on your gates, because we don't have we don't have isolation there. Yeah. So we want to make I basically want to make sure that that reference is as close as you want that to be your lowest noise one on the board. Yeah. Because you could, there's a, like a zero ohm resistor, you could depopulate and then connect on the power supply side. If you wanted to. generally not a good idea.
Yeah. You don't want you don't want anything. Right. Like that would be really bad. Yeah. Let's see. It's interesting, because it seems like a hybrid approach is new. You're not doing a star ground like things are inheriting ground.
A little bit. Yeah. Yeah.
Which I suppose works. Right. Because I mean, like a most of pinball is digital. It's digital, low current, and some stuff high current.
Yeah, basically. Yeah. It's not too big of a. Yeah, it's like even an older games and stuff like that. Like you all just stuff like the person touches is basically goes to a, it's like the chassis on an on an amplifier. It all goes through the center prong on the Yeah, the safety ground. Yeah. Because you have that, because that's what the person looks like all the metal rails, the coin door, all that stuff has lit up. That's non negotiable. You have to have, you have to have that. So that's that's, that's like, chassis ground for a pinball machine. Is that part? Yeah, it's because it does the same function. Whereas everything else is like, I mean, it's dc side. So whatever makes the most sense, where it makes much more sense how to ground it. Yeah.
Okay, that's interesting. I would almost think that your board, the one that controls all the actual switching, that having the quietest ground would make the most sense. And then, and then the computer board, inheriting its ground from your board? Which, which it sounds like that's similar to what you already have.
Yeah, I guess so. Because, remember, it's also, it's, it's only a USB connection between the computer and the board.
Does the competitors the computer inherit power from your board as well? Or does it get separate power?
gets its own separate power as its own AC power supply?
And then how is that not a ground loop? It? Oh, it has its own floating supply? Or is that supply also grounded?
Oh, I'm pretty sure that's all grounded. Yeah,
that's interesting. Because then I'm curious how that isn't a ground loop. Or if it isn't, probably doesn't matter, I guess. Because if you have your power supply for your computer board grounded, and then you have your power supply for your board grounded, and then you're connecting those through a USB. Now you got a ground loop between that whole system.
But you're technically yes. But you don't want you don't run. So in that situation, you don't run. You're not running bus power. It's just you're basically, yeah, you don't have any current flowing back on your USB, or on your USB V bus and ground lines. There's no current there.
Well, but that's the whole thing about ground loops. You don't get to choose where the current flows. Because now the current has two paths to go. It chooses whichever one
that's why you put a little little ferrite bead there. Yeah, you choose.
You get to coerce
the current AI coerce. Yeah, exactly. Yeah.
Well, you know, I'm sure it's really it's, that's also got to be frequency dependent. So if your ground loop noise is high frequency, it's not going to pick the longer path, you know, it's not going to or it's going to pick whatever path works best for it. And if there's a beat in the way, yeah, it ain't picking that.
Exactly. I mean, that's the kind of so when you look at a lot of USB designs, yeah, like the shield. So technically, the shield is owned for I think this is it's mentioned, if I recall in the USB spec, but that doesn't actually tell you how to do it. Because they say Well, because it
because it's choose your own adventure with the shield.
Yeah. Because what they say is the shield because the main thing is the shield on USB cable is like, okay, the shield should only be connected on one side. Okay, because that's how it was kicked on both sides, then you have basically a ground loop in there like in your cable. Yeah, that's supposed to have that. Right. So they, they basically say, okay, shield only one side and you go, Okay, what sides should be shielded? And they go, good luck, whatever. Yeah,
no, how do you know that what you're plugging into is the side that does or doesn't have that? Exactly. You don't? So the trick
is for little fear at be there. Yeah. And so if you're the side, that's, that's not supposed like, say you plug into your computer and your computer is the side that has it connected? Well, then now you put a little bead there, and now you effectively curb stop that ground loop from happening. Whereas if the other side is completely open on the shield, then on your side, you still at least have some connection there that will sink that interference noise down onto your circuit.
Yeah, yeah, for sure. I mean, it's important to note that you don't necessarily stop the ground. Because it's it's frequency dependent. You stop high frequency crap from or you. Like I said, You coerce it to cause a different path? Yeah, yeah. I'm curious. What, what, generally, what b Do you use? Or what? It's like 100 over 100 megahertz or something like that? 600 Ohm. Yeah.
So that's what it is. It's about 100 ohms at 100 megahertz. Yeah. And like, if anyone ever opened up one of my designs, it has USB, it has that. And it's not something I ever. I don't remember where I picked it up from, I think I did see a design that did that. And I was having problems passing the FCC at one point on like the USB stuff. And I saw and I was at that point, I was just connecting the shield directly to ground. And we were having problems with that. And I saw this design that basically gives you a little decoupling while also still allowing it to be connected a little
bit. I think that's a good word for decoupling your ground in a way. Yeah.
And cut the trace, put put a this bead on there, and boom, it was
perfectly cool. So that's been your beat of choice.
And I just did on every design, because ever since I started, I don't I haven't had a single, you know, emissions problem. You know, okay,
that right there. I got a little bit of a tangent. But But it's funny. That right there is one of the hardest parts about teaching engineering and get hang on real quick. I was actually talking with somebody at work, who's super eager about learning electronics, this, this guy is awesome. Like every day is just like God, can you give me some more information? I read through you what you said yesterday, let me let me hear so. And he'll pull up schematics and be like, What is this part do? What does that part do? And I'll walk through it. And like, so many times he'll point at things where he's like, why did you choose this? Why does this there? And it's like, I can't exactly tell you why it's there. Because the circuit might function if it wasn't there. But like it's there for protection, or it's there for like, two for Parker's situation where it's like, oh, it's gotta pass FCC, or it's gotta be blah, blah, blah. Or like,
well, it's good to know why. Yeah, of course. Why is the ferrite bead there? It's to decouple the shielding because one side of you USB has to be the shield has to be connected
to ground. Yeah. Yeah.
But if you on the other side is also connected. That's bad. Because you so you have to play this middle ground. Now. Could you spend time to figure out what frequency like what's the best be the put there?
Sure. Sure. Right. Yeah, let's use 100 ohms, and not 120 ohms, or six or
a different frequency or a different frequency, you could ask you, the best thing is to figure out what frequency you need to reject. Yeah. And then pick that frequency. And then I want to know how you would figure out what impedance either you might have to just try, I guess you can simulate it and figure out what and then sweep the impedance value. That's how I think that's how you would sanely figure that out. You can probably calculate it but to be
honest, I think I think a lot of people do it the way you did where it's like you're at FCC testing. You failed. Solder one in salt. You didn't fail. Cool. That's it. That's the one Yeah.
Well, it's we talked about This before where like, why 0.1? Micro Fred as like a decoupling cap?
Uh huh. Yeah.
Where is I can't remember,
they think there is an optimal value, but it would take more work to find
it. Yeah, it takes more work to find it. And point one microfarad is so overkill for it. That it's fine.
Yeah, yeah. But I saw I saw, I've, we talked about this in the past, but like, there was some article we were reading where it was like, you can go up to, like, you can go to the, to the moon with your bypass caps. And it's sort of like, of course, there's way diminishing returns at some point. But point, one 110? Like, they tried a bunch of different values. And it's like, everything worked. Yeah, yeah. The,
I think what's important on bypass caps is the distance between the voltage, the voltage rail, and your ground that you're trying to basically bypass is really close together. So a smaller value is more beneficial. They're, like, if you put a big, you know,
a 12. Think
about the extra thing about the the thing Who else think talked about earlier is like, tweak the knobs the other way, think about what happens, think about if you put a ginormous resistor down as your bypass cap, now your pads are freaking ginormous away. And now you have these big problems with how far the current has to go.
Yeah, you know, in my, in my layout techniques, I've actually started now, I'll lay out my board outline, because most of the time that's already pre fixed for me. And whenever I place down a part, an active component, the very next thing I do is place with bypass caps. And then then all of my analog or digital stuff has to conform to where the bypass caps are
i That's why we do it is the place your component where you might think is the best spot to for it to be. And then the next thing is all the bypass caps that that chip needs.
And the next thing after that is the vias that service power and ground. Yes, yeah. Exactly. Then then everything else?
Yeah, and I typically do this is also way overkill to is, if a chip has multiple VCC and grounds depends on the layout, usually one bypass cap per VCC, or ground pin combo. So it has every chips got a VCC and a ground, like voltage and ground. Yeah, so you have one, sometimes you get, like, two grounds and two power pins or multiple, like microcontrollers. That's common. And so you get share those, but if they're far enough apart, I'll start adding more bypass caps 100%. Yeah, the filament. So like, if one has one VCC, one ground and then another ground, but the other grounds like somewhere else? And it's not like a logic pin, like a, like a clear pin or something that's just tying the ground just because you don't care. But it's actually a supply pin. Yeah. Then I'll put also a bypass there.
Yeah, yeah, I think I think a good rule of thumb is just to start with, if it's a if it's a supply pin, it gets its own bypass cap, the one situation where I'm not maybe not one, but one situation that I would consider sharing, like, pins with with a power is if I know one is extremely low current, like, if it's a reference pin, and it's really close to another power pin, I'll tie those together, that's fine. Because it has such low current, that it's that the reference pin is just reading the voltage kind of thing. Like, like z battery. And you'll you'll see those on microcontrollers, like a battery sense pin or something like that. That's going to have so low current if it's going to be negligible. I'll tie that to another power. Yeah, it's got a mega ohm input and P Yeah, right. Exactly. Yeah, if the layout supports it, it like it's not worth the extra penny to put a bypass cap down on it. Exactly. It might even be worth to just drop a via and put that to power and not even bypass it if it has no impact on the microcontroller.
So do we want to talk about Python? Or do we want to keep talking about like layout stuff?
I mean, we're like 40 minutes deep, we can push Python. Yeah,
well, I'm instead of an Python coding. It's gonna be belt and layout. Design. Yeah, because our next topic is coil cap. Coil cap. That'd be cool. Actually. Coil coil craft has elevated inductors and put elevated and like quotes So I found this on Twitter. And it's basically think about an inductor. And so it's got, like a surface mount an inductor, let's say a wire around one that using like a switch mode, power supplies and stuff like that. Yeah. But instead of like a normal J lead, the J lead is taller.
It's got it's got legs, it's got legs. Yeah.
So and so yeah, so basically, the bottom of the coil doesn't touch the PCB, just the jaylene stew. And what that allows you to do is when you can do in switch mode power supply, such as now, you can put like your feedback caps and resistors and stuff like that, or maybe your diode, like under the package and even make your loops even shorter. I saw
some images where they put the controller under the the inductor.
Yeah, so that was another thing is there's a company apparently there's a couple other companies, but this is when I found a company called toric. Semiconductor makes a combined switch mode controller and inductor like package. So you put the whole thing down on your PCB. It was kind of a cool idea.
The part like as soon as I saw that, like I had, there was a bunch of thoughts going through my head where it's like, I've read so many data sheets, where it seems like the rhetoric in the datasheet would make this like, don't do this, like putting putting the controller inside the field that the inductor is generating, or is just part of the inductor seem like, okay, so if the if the controller IC is adjacent to the inductor, that's one thing, but you're putting it in the plane of the inductor. That seems wrought with issues. I don't know if I can necessarily speak to them. But just like from a gut feel, yeah, sure. That's great. You've, you've shaved off a millimeter on your loop. But have you introduced more issues by putting it inside the actual field of the inductor? I'm not sure.
Apparently not.
I guess not. Right. Yeah.
Apparently used to get better performance.
That's kind of legit. The one thing that's interesting is like, okay, cool, you can do that now. Now, how do you how do you, I guess, relay that information to your contract manufacturer, because now they have to have some really unique ways of assembling your board.
So look up part number XC, l 205. By torex, if you just want to look at the data sheets, but yes, so that was my next thing is, so we have x y RS, which is x, location, y, Location, Rotation and side. We're gonna go 3d x is just sent me these 3d. So we have to go 4d.
Yeah, cuz because sequence of policemen.
Yes, and I, because usually, that is something that most contract manufacturers have to pay attention to? Well, mainly with older machines, because your your head and your nozzle on the pick and place is short, on the older like a GSM, like our old GSM at macro fab had a really big head, and a really short nozzle, which is great for high speed, because it's very rigid. But you can't place small parts next to big parts. In any order, you have to actually have to pick an order. And usually you put smaller parts down or lower in height parts out first, and then you put bigger parts up.
You can also with with machines, you can run into parts, like physically collide with them. Yeah, yeah. So that's how I was Yeah, explaining it. Yeah. Our machine, we have to pay a lot of attention to that. Yeah. Whereas
like, my chronic, has really skinny tall heads, which, generally you actually don't really run into that problem a lot. So you can if you put a ceramic passer right next to electrolytic capacitor, yeah, you're going to have problems and you have to put the order in correct. But most pick and place machines nowadays, like part of making the package is you putting in the heights, and it automatically does its own collision avoidance.
Absolutely. Yeah.
But which actually would kind of solve this problem where I'm actually thinking about because the inductor is gonna be higher. So of course, it's going to place it after that after the parts underneath.
Yeah, I guess I guess it has to, it's just I've never run into a part that goes over a part except for we have one product that we place a shield on.
Yeah, a shield goes over. And I would guess if you I would tell your contract manufacturer this is what you want. Yeah. because they are going to be like, wait, what? Yeah. I've never seen this kind of parts before. Yeah. Like, I'll actually I'm Penetrator, I have surface mount components underneath through hole components. Because I want my fuses to light up when they were active. Oh, right. Yeah, backlit my fuses looks cool.
You mean you don't you don't have them on like a breed circuit that makes them like soft glow at different rates? That would be cool. That would totally use it,
to control them with the microphone. Yeah, I think we're pretty much maxed out on IO on on Sen. So Fabio in chat says can't do aeoi though, would that be a problem? Um, it really depends on the technology that you're sticking underneath the inductor or underneath the part. For instance, if it's just pat, like, let's say it's just returns, and your controller and your feedback loops for surface mount. Excuse me, for a switch mode power supply. You probably don't even need a why those, like 99.999% of time it's gonna be fine. Because the stupid Powerparts Yeah.
You always have X ray available to?
Yeah. Do you have X ray, if it's that important? I guess
the development thing is the part that sucks, though. Like if you're trying this for the first time. Oh, you're getting the probe points. Yeah, good luck. Yeah.
Yeah, that would be tough.
I would I would almost okay, if this was the first one coming off the line, I think what I would do in this situation is telling my cm don't populate the inductor. I'll solder on all my little wires to be able to access points all soldered the inductor on I can validate. And then in the future prototypes once I had it right, then you place all of it. Yeah.
The more I add something,
this just seems like if you need the most altra compact?
Oh, that's, that's what I was going with is, I would see this design, like, why would you use this design? Why would you use this design, one to minimize board area makes us more compact. And that would lead you to either smaller devices, were going to be really hard to aeoi to begin with. Because how the density is insane. Yeah. And are to lower the cost of your assembly. Now one thing I was actually thinking of, that's probably actually more difficult for your CPM. On these kinds of vices is heat shadowing. Yep, during reflow. That is one thing that's going to be interesting to try to handle especially if you have a lot of if it's a simple board, like a compact board and running like just qf ns, and like a couple of couple bleeded components, that kind of stuff, it's probably gonna be fine. But if you're running like a BGA technology next to this thing, yeah, that's where you might run into issues.
Yeah, you know, I've run into heat shadowing, with simple components, like, like electrolytic on SMD stuff. Like, this seems much more difficult than that. So if you if you packed a whole lot of components as close as you possibly can, under the, it's like just a giant brick of parts that are not going to they're gonna get, they're gonna not get convicted heat, they're gonna get basically radiated heat. And that's got to be enough to solder them. And I bet you it's not in a lot of cases.
Yeah. So that's what's gonna be interesting. And that's actually a subject that I haven't really ran into, in one giving advice for or knowing a whole lot about. Because most of my designs are not compact or high density. I don't really have that problem of heat shadowing on almost any design ever designed. So it would be like, how do you? Is there a way you can design around that or is that more of a cm problem?
I have actually an example of it. I've recently designed from boards that go in an array that go through our reflow oven, and the first time I did it, I forgot about heat shadowing or just didn't pay pay attention to it. And I had improper soldering on some of my components. And the thing was, if you, okay, so think of a top down view of your reflow oven. It's like a giant rectangle with a conveyor belt, right? That goes left to right, let's just say left, right? I had a string of a lot of tall, aluminum electrolytic go through, but I had them perpendicular to the motion of the conveyor belt. So all the air that's blowing from the blowers on the inside of the reflow oven would hit the electrolytic and basically get shielded from the things behind it. So one side of all my electrolytic solder, well, the other side did not. So all I did was I rotated the boards in the array by 90 degrees. So now they conga line through the through the reflow. Oven, they all solder well. And that's mainly just because the hot air was able to flow evenly across on both TR on both terminals. Yeah. So that is something to pay attention to. And your CM should be able to tell you like look at an array and be like, this is not going to solder Well, this this is now in this situation with these inductors because they have two legs that wrap underneath them, I would want to rotate the board such that any airflow could flow underneath underneath it underneath it. So I wouldn't rotate the legs such that they blocked everything underneath it. That's just guaranteeing issues. That's
I added more knowledge to my toolbox for designed for DFM there. Yeah. Because if you let go, that's one thing if you see, if you have two of these parts that are 90 out from each other, it's probably not a good idea.
Right? Right. And then okay, and that right there is a situation that the CM cannot fix. That would be go back to the engineers and say can
penalize it. 45 degrees.
So each one gets subpar. soldered.
Yeah, subpar soldering. Yeah. I do like reading these app notes though. Yeah. Cuz they always have like applications. And they always say like, mobile phones, Bluetooth headsets, PDAs. portable game consoles. It's like, like everything.
Yeah, but but also like, everything that is small.
Yeah. What's the electronic dictionary?
Oh, that's, that's yeah, that's the there's their like, it's a dictionary that that you just hold in your hand with? That's not a phone.
Hmm. I've never seen it written out like that before. Yeah. I guess there's also like, PDAs in this thing. So this application setup is a little old. Sure. I wonder if that's just like a microcontroller manufacturers attempt at SEO? Maybe. Bobby is
you know, the thing about it is you
can use it in a smartphone.
The part that goes under the inductor, it looks like a QFN style part. It is yeah. Which that seems like even worse, because that itself is blocking heat. From getting
Oh, on this one. It talks about the MCL 205 torques. Well, that's all they've they combined the inductor and the QFN in one package.
Oh, okay. Okay. Okay. I was looking at another image where they were separate things. Oh, yeah. Muscle cool craft mills.
Cool craft makes like a scroll craft doesn't actually do semiconductors. Yeah, they made this style. So you could use your own inductor basic write your own controller and that kind of stuff. Yeah. But I'm going to try one of these Xel two Oh fives. My next project because it looks pretty awesome. You just drop it down, you add a couple of like, external capacitors and boom, you got to switch from a power supply in the size of a you know, so I see aid package,
man. Okay. This seems this is really interesting. The Okay, so this x CL 205. I'm looking at the first page on the datasheet. And they had the typical performance characteristics. And if you've ever spent much time doing switch mode power supplies, one of the things that sucks about them is if you need only a little bit of load current from a switch mode, typically their efficiency is garbage at low current. And low current is relative but like if you're dealing with a small thing, say let's 10 milliamps or less. Most of the time, you're sacrificing a lot of efficiency on that. This one has crazy efficiency for low like it. Their curves are saying like 70% efficiency, all the way down to like a 10th of a milliamp that's that's impressive. Because most of the time like that The curves are efficient, but you got to pull enough juice for them to be efficient.
It could be part of that is minimizing the loops. Maybe,
maybe. Because if so that's really attractive, those kinds of curves. And these are these are not as good because most of the efficiency, like when you do get out to higher currents, they're sitting like 80 82% which you can get switch modes much higher than that nowadays in the low 90s 9495. Yeah, yeah, but if you need but most of those ones that are like 9495 That's like at half an amp right? And most of my little circuits are not pulling anywhere near that
it's pretty cool idea Yeah,
I like it.
So let's wrap up this podcast.
Great. So that was the macro fab engineering podcast we were your hosts Stephen Craig and Barbara Dolman take it easy
Thank you, yes, you our listener for downloading our podcasts if you have a cool idea project or topic or layout PCB suggestion, let Stephen and I know Tweet us at Mac fab at Longhorn engineer or at analog E and G or email us at podcast at Mac fab.com. Also check out our Slack channel. You can find it at macro lab.com/slack Or come listen to our live stream. It's Tuesdays at six o'clock Central at twitch.tv/macro Fab
Relay manufactures hate this one simple trick that makes your “sealed” relays last longer! Except TE connectivity who has an note about this relay feature.