MacroFab Engineering Podcast #49
MacroFab Engineering Podcast #49
What separates good documentation from bad and which kind of application notes do you like? Also, Ancient Chinese Semiconductor 7-segment drivers!
Figure 1: Parker dons a welding helmet to prevent himself from going blind while testing some super bright LEDs.
Figure 2: Stephen’s giant resistor layout in Eagle.
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 hosts, Parker, DOMA
and Steven Craig
C. C. Steven hosts.
We were guests last week.
So yeah. So last week we at the end of the podcast, we asked our listeners to name our conference rooms. Yep. And so we got a couple feedback. You know, a couple of people will give us some feedback on Twitter, on YouTube and an IRC, Tony de di wire, he gave me some feedback on YouTube. And he wants to call them Mac minor and Mac major, which are basically in some Oss. That's how you basically record MAC addresses for different device IDs. So you have a a major value and minor value, which is like 32 bit and 16 bit. Okay. Interesting. I've actually, I actually had to look that up what Mac minor and Mac major work, because I had no idea what those were. They just
sound like they could be value meals at McDonald's. Yeah. So it's a good one. Yeah, it just kind of funky in that way. That's what I thought it wasn't first because I didn't know about that addresses or anything.
Exactly. Justin Knight, who is our production manager? VP guy, huh. He's he calls the smaller room the pagoda? Because it doesn't have chairs.
And we just put a table in it the other week. Yes.
Some people have been talking about calling hertz and megahertz. VCC and VDD for power and ground. power and ground would be a good one.
Actually, yeah, that's that's, that's pretty good.
Right. You have a good one during Mic check. Yeah. TX and RX. People were mentioning like SDA and SEL for I squared C communication. Yeah. Yeah. I think Chris church, who's our CEO wanted to call it like, name it after some, like electrical engineering laws or something like that. Whatever we call what Maxwell equations? Sounds cool.
Sure, the way but you had one, right?
Oh, yeah, of course conference, big conference face.
That's absolutely perfect. That's what you would get if you let all the internet you know, we
should. I wonder if we actually did put a poll up. How many people would actually answer the poll?
Where would you put it up?
I don't know, Twitter or something?
Let's, let's do it. I mean, I like what we have here. But let's get a larger sample size.
Yeah. Well, I'll put a poll up and see what people actually vote on.
I like Nick copper face.
And then yesterday, I was working on some customer boards, testing them. And they were had some super bright LEDs on them. And I got really tired of basically plugging them in and being immediately blinded when they turned on these things were ridiculous. Yeah. And so I eventually actually just grabbed my welding helmet and stuck it on. And it was bright enough to actually trip the the sensor in the auto darkening helmets to actually dim the lens. Yeah, yeah.
Which you got a killer welding helmet?
Yeah, it's got a big viewing area. Yeah,
I mean, it's it's an enormous it's like a monitor in your face. Yeah. sighs it's, it's the biggest I've certainly ever seen. And it's got all kinds of like, controls and stuff inside. The
best thing it has that that helmet has is it has a button. When you press it, it tests the battery. So you can so basically dim the screen. Yeah. And so that way you don't accidentally blind yourself because your battery's not, you know, turned on seems like
a really simple feature. That should be in a lot. Because I
think that was actually the first thing that happened. When we first started welding the rollcage. And the jeep. Yeah. Mike Williams, who's a developer at macro fab, he knows how to weld so we were doing it in the first world and make that day. He did that mistake. He did not turn on the battery in his helmets. And immediately just arc flash in his face.
He was he was like a foot away from the well. I mean, his face was Yeah, so So you're getting into welding now? Yeah, yeah. So local Makerspace TX RX here in Houston. You took a class
there, right? Yeah. Took a three hour MIG welding one on one basically. Yeah, it was pretty good. I would recommend that to anyone who wants to learn how to weld.
Yeah, going and going and taking the classes is certainly a good job and doing it at the at the makerspace that's, that's pretty cool to support a local place. So our MakerSpace here in Houston does MIG TIG and arc or stick Yeah. And a bunch of welding equipment have been showing up periodically. Now. Parker is getting into it.
Yeah, from Amazon. Yeah, they actually the welder's is a shirt up today. And Steven helped me carry it out to the to the Jeep was
fun. Yeah. So looks like you got a lot of fun coming up here soon. Yeah, we'll
see. It should be fun. Cool.
So got an update on the the world's greatest resistor or what I've been calling it the resistor resistor,
or Zillow, the greatest resistor and the word Oh, the grill. I'm
sorry, I got it wrong. That was that was from one of our listeners. He dubbed it the the greatest resistor in the world. Yes. So last week, we were talking about this Well, the last two weeks
where I don't, I don't always impede flow. But when I do, I am the biggest resistor in the world.
Just know, this stop no more that. So yeah, got an update. This is actually technically the first board that I've ever designed in Eagle. And I totally broke the rules. And I didn't do a schematic at all, I just went right to the board, and just start plopping down resistors. So, so kind of a follow up from from last week, this, this resistor is just a giant array of 40,000 resistors on one PCB. And so I basically just did a massive copy and paste job and ended up making a monster PCB with 40,000 resistors that are all in a series parallel configuration. So we had talked about making the the PCB dimensions, basically the same ratio as a Oh 603 resistor. And if we got lucky, because once you stack all these resistors on there, the height of this array of resistors is 10 inches. So to make it the same ratio, it's a 20 inch wide board by 10 inch high. And I basically made giant copper planes on both the left and right side to kind of look like the side of a resistor. And now it's this giant 20 inch by 10 inch board that looks like a resistor and it's just got a boatload of resistors
I wonder if we shouldn't make this into like three millimeter PCB material. Just make it stiff as hell yeah. Instead of making it because at 1.6 it'll be pretty flexible. Yeah, yeah. Yeah, let's do that three Moon there.
It's the board's gonna get pricey. We'll build it once. Yeah, it'll be fun.
It's already more expensive to just buy the resistors versus buying a precision resistor already. So why not just knock it out of the park, right with the bill material.
But it's the principle of the matter. We got to do this. We got to make this so. So I've already got my Eagle files done. And it looks cool. So some quick stats about this. Each resistor is a 10k resistor to 1% tolerance. And each one has a power of 62 and a half. watts or Milla milliwatts Yeah,
I'll say six, two and a half watts. Now that'd be insane. So if you just add all the wattage son there, yeah.
If you add all the wattage of each resistor, you got 40,000 times 62 and a half milliwatts, you get this composite resistor will give you two and a half kilowatts of power dissipation. Now, I doubt we'd ever actually be able to see wattage basically just add them up series or parallel, you just yeah, you add them up assuming that all the resistors have the same wattage and the same value. But yeah, in this case they are so you just add them all up, or basically take the wattage of one and multiply it times the total amount. Yeah. So two and a half kilowatt, which is killer. So the voltage, we would have to put across this massive 10k In order to actually dissipate two and a half kilowatts would be 5000 volts. I don't have a way right now of actually the only thing I have right now as I have a high voltage power supply that can go up to 400 volts. So I think it'd be fun to kind of hit it with a huge amount of voltage and see what
we could probably try to like make a a Jacob's Ladder style, you know, boost converter to basically pump a couple 1000 kilovolts into this thing.
Yeah, but it has to be able to output a ton of output current. I wonder if we could just find a transformer that'll do 5000 volts and five amps.
That's but 10,000 watts.
No, that's two and a half kilowatts.
Oh, yeah, I forgot
the resistor. Yeah,
that would be hon you say 5000 Watt. 5000 volts at what? five amps? half an amp. Okay, how about say cuz that'd be look insane. Yeah, no, that'd
be 25,000 watts. Yeah, and then see Parker's doing some phone calls. regulations you
can be pumping 21 amps into that transformer on the front end. Less you can get a 220 volt,
huh? So it's gonna be really hard to take this to its maximum that I think that's the Yeah,
this your mate basically max out a typical 120 volt 20 amp breaker, if you had a if you could somehow find a
Yeah, and getting a transformer that could do that would be ridiculous equipment,
what voltage does like a microwave use? Because it has that transformer in there
the magneto? Yeah, that's actually higher, I think that's like 25,000 volts. But it's not going to be able to put out that kind of current, I guarantee you that
we can still try hitting it with 5000. Like take some windings out. Just Well, the the thing is because people turn these things into welders. That's true. I run it backwards.
I wonder if you know what I just got rid of a microwave, too, I totally should have brought it in. So this resistor can actually handle 10,000 volts. So we can we can always go pretty high on that. Give that a shot? Well, on top of being able to handle this, the interesting thing is the temperature coefficient of the entire resistor is actually it doesn't change. So the temperature coefficient of the whole resistor is the same as the temperature coefficient of any one resistor on the board.
So how does that work? Because is it because it's in parallel in series? Yes. Okay. Because you have to in series, you have to add the parts per million?
Well, they would both basically go up by the same amount. Yes. So you end up resulting with the same. Now that temperature coefficient only works if the whole board equally goes up or down, or however so it gets kind of funky.
Yeah, we we actually had one guy who listens to our podcast on NRC. Ask about if we can get a thermal imaging camera and pointed at the single when we turn it on, and then start actually running power through it and see how basically, how does the board heat up? Yeah. And I was thinking about that. And most thermal imaging cameras don't have really high resolution. Yeah. And I mean, that would be kind of a problem, because you have 40, you know, 40,000 resistors. So what's the grid by?
Is it? What 200 by 200 100? By two and so we need at least well, I guess 200 by 200 resolution camera, give or take some we could be able to see enough detail and see where stuff seated up in then well, actually, it's actually not as bad as I thought it would be. Well, you know, it would be cool, actually. So if you if you are able to hit this with a ton of power. And then you see how it heats up across the board, you can see which resistors are potentially lower or higher, based off of like, what's the path that electrons are trying to take? Yep, through it. And then you'd be able to see like the lowest, so
those are lower, and as a heat up, starts to redistribute? Right. Right.
That'd be kind of cool. But yeah, it seems pretty hard.
No, if we get, um, you know, a decent thermal imaging camera, we should see that.
Let's see what we can do. Yeah. So yeah, I'm gonna get this going to get this board on order here soon. Now that I got all the files, but had a thought that is interesting. So the design that I have puts all of the components on the top side of this board, and I was thinking, there's a whole backside of this board that is just unpopulated doesn't have anything on it. And if you think about it, you know, board this big going through a reflow oven, is it going to warp? Is there going to be any issues? And what can I do with the backside? And I was thinking, why don't we repeat this experiment with capacitors on the backside. So do a 200 by 200 array of one micro farad caps on the other side, because series in parallel with cats, even though it's the opposite of resistors, it, the equations still should come out the same way. So you have a calibration resistor on one side and a calibration cap on the other.
And, you know, I was thinking is when you charge up that big capacitor array, I wonder if it takes any longer to actually, you know, ramp up versus normal. No, one micro farad cap. Yeah, you have a lot more plate surface to fill up.
Well, and one thing that would be interesting, typically, when you put a bunch of caps in, in series, you put dropping resistors across all of those legs so that you equally balanced the load of voltage across each cap. It would be cool to put those balancing resistors charge it up. See how that charges, knock off all those resistors and do it again and see how it turns out changes different right? Because at that point, you're just relying on the leakage between the caps you have to balance to balance Yeah, is it gonna get all super funky that would be a fun experiment. I'd have to give it a go. Yeah. Cool. Yeah. So we'll get that on order. And, and we'll give you updates as it as it all trickles in. Yeah,
I think we should have it. Probably in three weeks done. Yeah. Yeah.
And we're talking about potentially doing some videos of the my 200 pick and place actually placing all of these because it's going to take a long time.
Well, I'm in the mind 500 pasting this entire board. Yeah, you think that's 880 1000 pads.
I wonder how fast it will do it. Because that that machine is super fast.
It is very fast. But it's still at its best
yet. Right? Right. We might have to buy our own vial of paste, because we're gonna go through a ton of it
go through a whole whole thing.
It's it's ridiculous. We actually told Chris, the CEO about this project today. And he kind of rolled his eyes. It was it was great. It was fantastic.
And then a lot last week, you're working on a bass amplifier, right? Yeah, friends.
Yeah. So a buddy of mine. He actually is a music teacher at a school here in Houston. I do I do some repairs for him occasionally. And so he dropped off a bass amp that I've actually seen in in my shop a couple times. It's had some really interesting issues that many times. No, but yes, at the same time, first time he brought it in, it was just toast. I mean, the thing was just absolute crap. And so I spent, gosh, a long time trying to fix it. I couldn't find any schematics. I couldn't find anything to go along with it. And eventually, I this is like a desperation thing. When I'm when I'm doing repairs, I end up calling the manufacturer. And the reason why I say it's desperation is because most of the time they won't help you. They're just like, oh, well, you know, good luck. buy another one. Yeah, exactly. Well, especially because this thing is probably like 400 bucks. They're not gonna give you a bunch of service time on it and stuff. Regardless, I called them up and they're like, oh, yeah, that's the old revision of the main power amp board that you have in there. And they're like, Would you like a new one? And I was like, yes. So fairly works. Yeah. And they shipped it from Italy, which is I don't It's random.
Is it made in the same factories? The Arduino? I doubt a real Arduino I should say, right? That
the Gen Duino. Know, where are they fixed all that? Well, okay, you're right. Yeah, that's all that's all done as of like, two months ago or whatever. Yeah, but But yeah, so So I get this, this main amp board. And it's what's really nice is the old amp board had trim pots all over it. And I've got no information on what to do with those and and I don't want to go replacing like power transistors, because you got to buy some and blah, blah, blah, regardless, well, this new board comes in. And it's it's, there's no trim pots, it's just a plug in Go. I literally just plopped it in the case, turned it on. And the bass amp works great. I played it for probably two hours. And it's like this thing sounds great. So I call my customer. He's all happy he comes, picks it up, calls me back not long after he's like, yeah, the first gig I took it, I put it on stage turned it on, and it stopped working. Like oh, great. Like this is this is exactly how it's Yeah, I mean, this is brand new off the factory floor. So he brings it back and I start digging through it. And the there's there's a preamp board and a power amp board. And on the preamp board, I'm starting, thou shalt check voltages. That's the first thing I do. And I look at it, it's got negative 15 volts, and I checked the positive rail and it's zero should be positive 15. So okay, great. So I've got the smoking gun already. That's that's the good carry the fuse, right? Yeah. So I'm looking all over the board. And this one's funky. It's got regulators all over it. And I'm testing everything's like there's 15 volts on the main board. And there's a ribbon cable that jumps across, I'm checking the ribbon cable, everything's great. And I'm sitting here just like, frustrated out in my mind, because I've got 15 volts where it should be and I don't have it, where it should be at on the other board. Yeah. So I ended up taking taking the board and looking all over the place. And then I lifted it up to such that I could see a light behind it. And lo and behold, there's traces on the inside of the board. And literally from the VO Kia blasphemy. Yeah, exactly. From the it's a it's a Joe Schmo. 70 7805 Regular sir, sorry, 70 815 regulator. It's got 15 volts right on its output pin, and a trace that goes all the way over to the preamp board. And there's a broken internal trace on the new board on a brand new board from the factory. And I'm like, Ah, really. So it had worked when I put it in and then the guy threw it in his car and a car probably got jostled around. I mean, you take it to get his roadies broken. Yeah, exactly. So I really like brand new from the factory in an internal trace go so I routed the trace external and everything works great. Perfect. So don't trust internal traces. They will break somewhere aboard. Yeah. But that was a giant pain in the butt
is speaking at trim pots. We got we're pretty close to finishing up that space echo. That's right. And you were actually going over? I don't think we've talked about this on the podcast, I don't you're we're going over on calibration, because we actually, like fired it up and sorry, you know, first thing we do is to start messing with the knobs, you know, see if it worked. And the reverb worked pretty well. Yeah, and the only thing that didn't really work was the echo. And it's a space echo. So we're like, Okay, we gotta figure out why that's not working. But it would if you hit because we basically plugged the guitar into it. And if you hit the guitar just right, it would work. Right? And why? Okay, so it does work. And then, and then you start digging into the service manual.
Yeah. So okay, something to note, we changed every cap on this board. I mean, we got it every single cap, and then we cleaned all the grime off. Oh, yeah, no, this thing, this thing looks great now and all the slime out of the pots and everything. It's, it's, it's back up and running. Well, but But you got to think nobody had done any service on this for the last, I don't know, however many, um, three decades probably. So all the components in this device kind of drifted together? Well, we just revitalized all the caps. So now all of our bias points, all of our calibration points are just skewed everywhere. So the thing needs recalibrated. And luckily, the service manuals are readily available for download. So I'm like a great, you just get service manual, and it tells you exactly what to do. Yeah, but
it's weird, because it's like, input a signal at, you know, was it negative 50 DB or something? Yeah. Which is not even how you like would measure that.
So it uses an old ish nomenclature for for audio terminology that, frankly, we don't really use anymore. And it's a pain in the ass because it's telling you to input signals in, but it's using DB U, which isn't even a term we use anymore. So I'm sitting here scratching my head, because it's saying like input this whatever number dBu and measure this voltage on this pin. I was like, why are you telling me to measure a voltage on a pin? Why couldn't you tell me to input a voltage at the at the input? Because that just makes sense. My signal generator doesn't output dB, you
know, its peak peak voltage at a frequency, right?
So, so it's talking about a certain level of line level voltage for audio applications. So there's a whole standard that we use, but not not really, and sort of, it's goofy. So we're having
to use it. Yeah, exactly. So we need a gray beard to calibrate this thing. We're
the thing is, I'm having to spend way more time interpreting the crappy service manual than I am actually doing the calibration, the calibration is going to take 10 minutes, but I have to figure out what all these numbers are, hey, if
I think Siemens actually figured it out. But if you're an old audio gray beard out there that knows what that is. Yeah, definitely let us know. And then hey, maybe come by and fix it for
us. So actually, okay, the two numbers that are the most annoying to calculate are negative 50 and negative 48 DB you if you know what that means? Let us let us know that we've got to figure it out. But
because I mean, it doesn't have a frequency what that's at, right. It's just a what it's whatever level it's, they
call out a couple numbers. For frequency. It's one kilohertz is the standard for testing audio gear. So yeah, it's a it's a sine wave. So a negative 48 and negative 50 dBu signal at one kilohertz sine wave.
Okay, so we'll put the the manual up on the podcast notes. Yeah. And if no one tells us anything, we'll just try it anyways. And see what happens. We got it. We were pretty good at not letting smoke out of electronics. Yeah. There's not much smoke in that machine. Either way. No,
no, no. With all the shiny new fine gold caps. Yeah, fine. Go caps.
Do if we, if I put those caps in backwards, I would have cried. That was a lot of work. RFO Yeah, our first section. Um, this is a little bit different than we normally do. It's, um, I think C is ces going on right now. Computer Electronic, whatever it is that yeah, looks like it is yeah. Um, because I had some friends like just posting links of really cool like laptops and stuff and I'm like, Oh, I guess CES is going on. But this one is really cool. I guess it's a it's razors project. Valcke, Valkyrie, something like that. Anyways,
you wrote you wrote Valerie, Valerie. Valerie, Valerie, I copy that paste it so that must be right. Okay. Yeah, Valkyrie is a lot cooler and probably should be the This yeah should be
a three screen laptop that uses robot arms to extend the screen now. That's incredible. Yeah. So it's got some cool specs, you know, three screens actually has a mechanical keyboard instead of a, you know, stupid Chiclet. Apple keyboard. It's got an actual proper mechanical keyboards perky and has a GTX 1080 GPU in it, which is pretty hot. Well, no mention of what power consumption is because this thing probably takes efficient battery to power it. And the fact that if you tried to use this on like an airplane, you had to get the middle seats, then open it up, and the screens would fall out into your other people's tray area.
It's the way how much is one of these cars.
It's not for sale yet. But the platform it's based off of, because razor has a couple of laptops. Those start at like $4,000. No GS, which is a order of magnitude more than my laptop.
See, I think that what would be great as to take this to like a hoity toity coffee shop. And then sit on one of their like tiny tables, get yourself a cappuccino and then unfold your monster computers.
As the robot arms are extending the screens, like knocks other people's crap off.
That's amazing. Yeah. So
the first person who uses this laptop on the airplane, take a picture of it cuz I want to see how much a douchebag er
It's fairly obnoxious. But it's cool.
Very cool. I guess I want this just so like, you could do like, you know, EDA tool stuff on the go, CAD, all that good stuff. You know,
it's not for that. No,
it's not with a GTX 1080. Right. Like, I need a laptop that's got at least two screens with like a wimpy CPU and GPU in it. Yeah, nice keyboard. That's all I want. Someone make it so next is Tesla turned on their giga factory, which is their ginormous lithium battery factory. That's in Reno, Nevada.
Yep. That was supposed to well, they were talking about bringing it here to Texas, right.
Texas was one spot California was that they were just all over. Okay, where they were going? I thought it was a split between Texas and yeah, it was it was up in the air. But it ended up being Nevada just for the land was way cheaper. And the power is way cheaper to make sure there.
And if you have a name like Gigafactory you you need a lot of power clearly.
Well, it's I think it's all gonna be solar. Really? Yeah. I think it Yeah. Just turned on to like this week. It's like 3000 people working there now. Wow, that's pretty big factory for 4.9 million square feet. Geez, I'm just for their batteries. Just to build batteries. Wow. Yeah. If they didn't have any stats on like, how many batteries they're producing right now? Or like, one out? Yeah, one one rolled out today? Because I want to know, like, how much power this factory consumes? How much water? How much lithium do they have to put in to get batteries out? Yeah, yeah. It's kind of cool. Well, you
visited one of the factories, right?
Tesla's motor factory in Southern California. Yeah. How was that? That's really cool. It's the first time I've ever been to a automotive factory, though. So all that machines and arms and robots are really cool. That's cool. The really cool thing they have are the six axis robots. Yeah. Where they actually like machined out the rear like axle. Out of like a solid block of aluminum. All that's in like one piece. Wow. Yeah, that's really cool.
There's a lot of money pumped into that. Oh, yeah. Well, it's
a luxury car. Yeah. So yeah, check out that. The the Gigafactory. Maybe, someday we'll be able to look at that thing. That'd be a tour that I guess would be in Reno, Nevada, though.
I've been out there before. It's pretty.
Is it cool, though? It's cool. That's cool. I guess we just go to Las Vegas right afterwards,
a couple hours away.
That's just right down the road in Texas terms. That's true. And the last RFO is the other kind of bypass capacitors. So this is an article talking about different kinds of bypass articles. And I picked this article, because you might know something about it. I
hope. I hope I wrote an article about bypass
catch. Well, you wrote about bypass caps in terms of power supplies. Yeah. And no, not really called bypass caps, immoral decoupling. Okay, masters, at least in this guy's terminology. What he's talking about is bypass capacitors for feedback circuits. Hmm. So for unlike option amps and stuff. Okay. And basically he's talking about all these different feedback circuits and how op amps actually have internal feed these internal feedback capacitors that basically bypass and help it stabilize lower frequency signals. Yeah, by using the higher frequencies stuff in the signal. Hmm. So you feed back more the high frequency stuff to help it stabilize faster. So you get less basically less a transients in your in your in your feedback. That's cool.
Yeah. Because Okay, so phase margin is a big thing when you're talking about feedback. Yeah. Because if you if, obviously, with when you have a capacitor on there, in any part, you're going to get a phase shift. And if you say shift enough, eventually you'll be so far out of phase that the amplifier will start to be in a positive gain situation. And in goes from negative feedback to positive feedback. Actually, I
actually misspoke. It's actually the opposite. It reduces high frequency gain and your feedback loop.
Okay. Yeah, that actually, yeah, that makes that makes a bit more sense. Yeah. So So what's interesting is the choice of verbiage there or calling them decoupling caps.
Well, that's the power. That's what you wrote an article on.
Yeah. Right. So when you talk about feedback,
he's called he calls these bypass caps, because how they work is win win when you first power on. It's actually bypassing some of your your feedback filtering. I gotcha. And letting that letting your op amp stabilize faster that way, with this through this capacitor. And then when you hit steady state, nothing flows through that capacitor, and your normal feedback loop works.
That's interesting. So it was all about turn on transients. Yes. Cool, or turn on to getting to steady state. Right? Because the game probably does some kind of funky spike, exactly what's going on, they can latch and get stuck in a deposition.
That's exactly what is. So it's nice. It's on the embedded related comm. website, we have a link in really cool. And it's actually really well written. And I probably would not have if I actually read the whole thing. I would not make it to the podcast today. Because, you know, I'd still be reading that article. It's very interesting.
So actually, interesting story, you whenever, whenever you have any chip on a board, on the power rails,
what do you normally put right next to the chip point one micro farad capacitor in the right voltage range?
Exactly. Except not always. So what would you
not use it for? Well, here's
the thing that's interesting. We had a circuit a while back, where we were actually producing a one megahertz output signal, and it was going down to a an eddy current probe. So effectively, this probe produced a one megahertz, em wav. And you if you put that in close proximity to a chunk of metal, mainly, it was just steel, and you would wiggle that steel, you could actually read the vibration back, you could literally read the position based off of how the load was on this probe. Regardless, the front end op amp that was actually aiding in producing that one megahertz signal, we were having a lot of trouble with it. And on both the positive and negative power rails on that we had point one Mike faired caps right next to the chip. And we pulled those caps off, and that system worked perfectly.
So Did y'all figure out why or is it just one of those, like gray beard scratched? You know, scratches chin and was like, let's remove those things. Well,
okay. So it was probably obviously it was due to the caps, but it was probably due to their value. So the value of those caps were causing the power rail to phase,
which is probably oscillating because of the one megahertz writing. Right,
exactly. So the op amp was trying to draw current at one megahertz. And that was causing some funkiness from the cat.
Yeah, I bet you if you looked, I don't know if you would know the part number those capacitors Oh, shoot that because, yeah, yeah, another life. But because if you look at your this is things that should be careful with bypass capacitors, as yet the look at the datasheet that you're using, you can't just go to the cheapest one you got to figure out is what is your power supply going to be? Basically was your ripple oscillating that? Yeah. And you have to make sure that your bypass capacitors don't have roll off before they're or a roll off in there. Oh, yeah. Yeah. And so because you want your reject, you want your capacitor to have a good rejection of ripple at that frequency. And so you have to pay attention to that and actually ran into an issue with the prop once is I was using, you know, we we basically we overclocked a parallel propeller from 80 megahertz to 106 megahertz. And the some of the bypass caps we're using, they had, they were great at 80 megahertz. But when you got up to 100, they were starting to drop off, right there that was on the drop off. And so we got crazy ripple on the power supply cuz we're running the clock at 100 megahertz or 106. Now, which is close enough to 100. Right? And so we just change it to basically raised just went up in package with the same brand. And it costs the same. Yeah, and but that, but basically the add in material shifted that that drop off higher up the frequency band, right and got rid of
it. Well, and that drop off is due to the parasitic inductance inside the cap. So just changing the physical size, we'll certainly do
that. And that used to be the case, but it's not the case anymore. How so? Capacitors are actually that that was actually due to just like, they made it bigger. And so there's just the different manufacturing techniques. Yeah, but now they're so good at making capacitors, you actually have to go in to the datasheet. And make sure that's the case. Oh, okay. I see. Right. Because usually used to be the old trick was to basically pick point one. microfarad. Yeah. And like, oh, 603 and then appoint one microphone at the same Brandon. Oh, 805. Yeah, just put them right next to each other. And then you wouldn't have any drop off? In words, way further out. Oh, it's wait for that. But your your notches don't overlap. Yeah, yeah. But now the most cases they do? Well,
and what a lot of guys do is they just do the low value, and then that same value times 10. So point one, in parallel, and that allows you to sweep from way low, all the way, way high. And if you're doing something super high frequency, you can even throw a 10 pico farad or something in parallel with that on top of that, and then you get you get crazy it low impedance across the whole frequency band. Yep.
And we got way more into that than I thought we would. Cool yeah, you know what, we should probably type what we just set into an article and add that as like bypass capacitors part two. Every time
we have a podcast we just come up with like more work for the podcast. I love it.
But you know, that's that's a
good thing. So I don't think you
actually talked about frequency analysis for bypass capacitors. Oh, in
the article that I wrote it No, I just was basically appraising just or just saying do it. Do it? Yeah. But but but not like the intricacies of what happens when you do it wrong.
Do it wrong. Maybe we should try to build a board that does it wrong. Yeah, that might actually be harder to actually build one that does it right.
Actually an oscillator High Frequency Oscillator. Yeah. Telling you that you can do it wrong. Easy.
Okay. Cool. Yeah, that's, that's the last RFO and this was episode 49. Of the macro fab engineering podcast. We were your hosts Parker Dolan and Steven Gregg letter, guys, take it easy. Oh, yes. And have a great new year because this is the first podcast of the 2017 Cheers. Cheers.
What separates good documentation from bad and which kind of application notes do you like? Also, Ancient Chinese Semiconductor 7-segment drivers!