- James Lewis
- Our guest this week may or may not be an Engineer or may or may not have Hair!
- James Lewis’ passion for teaching non-engineers has led to him to create the Bald Engineer Blog and the AddOhms video tutorial series
- With 15 years of experience in electronics, marketing, sales, and teaching James boils seemingly difficult concepts down to the core, so that anyone can learn what they need to finish that next great project
- Worked at Agilent as an FAE for oscilloscopes (for about 10 years), then as an FAE and marketing at KEMET for about 7 years
- Now working as oscilloscope product manager at Rohde and Schwarz
- MLCC – Multi – Layer Ceramic Capacitors
- Where did all the MLCCs go?
- Looking back, it was obvious we were headed towards a shortage but companies didn’t see it coming
- What are they?
- Common uses?
- 4 ways Ceramics lose capacitance
- DC Bias
- Aging effect
- What to consider when choosing an alternative dielectric/technology for capacitors
- Where did all the MLCCs go?
- Questions from the Slack Channel
- Insights into selecting bootstrap caps for high-frequency, high-voltage H-bridges
- What footprint should we be designing with, going forward? Presumably different sizes for decoupling vs bulk on the power inputs for getting different ESRs?
- Is MLCC cracking due to pcb flex something that still gets seen in 0402 and below?
- General rules of thumb on choosing appropriate packages based upon desired voltage rating and capacitance?
- What *is* the fundamental physical mechanism behind this voltage-dependent capacitance drop?
- Why does the Y5V dielectric exist?
Visit our Slack Channel and join the conversation in between episodes and please review us, wherever you listen (PodcastAddict, iTunes). It helps this show stay visible and helps new listeners find us.
Special thanks to whixr over at Tymkrs for the intro and outro!
About The Hosts
Parker Dillmann is MacroFab's Co-Founder, and Lead ECE with backgrounds in Embedded System Design, and Digital Signal Processing. He got his start in 2005 by hacking Nintendo consoles into portable gaming units. He also runs the blog, longhornengineer.com, where he posts his personal projects, technical guides, and appnotes about board layout design and components. Parker graduated with a BS in Electrical and Computer Engineering from the University of Texas.
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.
Host 1 00:11
Welcome to the macro fab engineering podcast. I'm your guest James, the bald engineer Louis.
Host 2 00:16
And we are your hosts Parker Coleman
Host 3 00:19
And Steven Craig.
Host 2 00:20
This is episode 141.
Host 3 00:24
So our guest this week may or may not be an engineer and may or may not have hair. James Lewis, his passion for teaching non Engineers has led him to create the bald engineer blog and the ad ohms video tutorial series. With 15 years of experience in electronics, marketing, sales and teaching James boils, seemingly difficult concepts down to the core, so that anyone can learn what they need to finish that next great project.
Host 2 00:53
So James, is there anything you would like to add to that? Extensive bio?
Host 1 01:00
Yeah, I guess let me talk a little bit about where my professional experience has taken me. So right out of school, I worked at Agilent Technologies, which is now Keysight. And I'm sure next week they'll have an another name as an fa e for oscilloscope and Logic analyzers. And I worked with computer customers. So I was doing like frontside bus, memory bus, those kinds of applications. And then after about 10 years of that, I switched to components and I worked at Kemet, which is a primarily a capacitor company, both as an FAA and marketing for about seven years. And just recently, I made the transition back to test and measurement and now I'm working at Rohde and Schwarz as a product manager for a telescopes.
Host 2 01:42
So I've actually never heard of Rohde and Schwarz. So is, is there they make something that I would recognize or
Host 1 01:50
Um, yes, so Rota, like Rohde, and Schwarz is a German company. And they are primarily known for RF test equipment, okay. And they are most famous for 4g test systems. Now they do. They've done a lot of other cellular stuff, too. But for the 4g generation, they, they really had a lot of product in that space. And so without question, the phones, you have, regardless of vendor have probably been tested on something from Rohde and Schwarz.
Host 3 02:17
Gotcha. That's cool. The first time I heard of them was on the Eevee blog, one of the one of the videos, he did a something on some test equipment from them. And Dave was absolutely drooling on the test equipment. His his ideas, were like you do not get better than that equipment.
Host 1 02:37
Yeah, it's so. So their whole entry into scopes is kind of interesting. And I know that's not what we're going to talk about today. But they've only been in the scope market for eight to 10 years. And so that's why I'm not kidding when I say they are primarily known as an RF company. And so it's interesting to look at a scope from the perspective of RF engineers, and there are things about it, where it's like, every time they tell me about something new that I haven't figured out yet, or haven't learned about yet. It's like, really, you guys figured that out? It's like, holy cow. Yeah, that's it. It really is droolworthy.
Host 2 03:13
So, James, what are we going to talk about today?
Host 1 03:16
So I thought we might talk about something called a capacitor. I'm not sure if you guys are familiar with this component?
Host 3 03:23
Nope. Never heard of it.
Host 2 03:25
I thought we were talking about resistors. But
Host 3 03:31
No, I'm actually pretty excited because we've talked about doing something like this for quite a while on the podcast. And I just kind of, I guess it kind of scared Parker and I a little bit so
Host 2 03:42
I decided to phone an expert.
Host 1 03:45
Oh, art, are they gonna join us later?
Host 3 03:51
Yes, yes, they are. That's that's the answer to that.
Host 1 03:55
Yeah, actually, you know, it's funny because I think a lot of us in school we, we sort of learned about capacitors and we were told things like, you know, they change with frequency. The only lesson I remembered was that there's polar and nonpolar and you don't put polarized in backwards. And then that was sort of it. And even when I went and joined Kemet, so I was interviewed interviewing for this job as fa e. And I asked at some point, why do you guys need FA ease mean, how hard is it to pick a capacitor? And as a challenge, the hiring manager said here, let me show you. And that led me into that stage of my career.
Host 3 04:34
So what's what's the thing that actually charges up and holds that storage for, you know, running these headphones?
Host 1 04:41
Well, that'd be a battery.
Host 2 04:46
So capacitors, what kind of capacitors are we talking about today? So not polarized and non polarized?
Host 1 04:53
Right. So there's all kinds of capacitor types out there. But I think the most common and almost anyone that's done anything electronics knows about ceramic capacitors.
Host 2 05:02
So what do we normally all, I guess some background normally called mlcc, which is multi layer ceramic capacitors. Correct. So why is that?
Host 1 05:13
Well, it's because they are actually literally designed with multiple layers. And so you can just picture a if you were to make a sandwich, and you had bread and cheese, you would just kind of stack up bread, cheese, bread, cheese, bread cheese. That's what a ceramic capacitor is your bread is typically Barium titanate. And then the electrode is something like nickel, and it's a really thin metalization on top of that, that that ceramic layer, and then those just get stacked up. And the more stacks there are, the more capacitance you get.
Host 3 05:45
So out of curiosity, this is a slight tangent, and we don't have to go deep into this. Are you aware of like the actual nitty gritties of how that's accomplished? Do they lay a layer down and then apply some metal to it and then just keep it
Host 2 05:59
Up? Take the nickel out of a jam jar and kind of just spread all over the platter? It's
Host 3 06:03
Pretty like peanut butter on some ceramic.
Host 1 06:05
Alright, so let's talk a minute about that. Because I have a trivia question. And I want everybody listening at home to try to guess along with this. So let's say we start with raw material. And we get we go all the way through production to capacitors in a real ready to ship somewhere. How much time do you guys think? How much time elapses between start from materials to parts are ready to ship?
Host 3 06:31
I'm guessing because this is like a this is a trivia question. I'm guessing it's an extreme of some sort. So it either takes a long time or a short time. And given given the cost of a capacitor, I bet you it's rippin. Fast.
Host 1 06:45
You guys got to take a guess. And then
Host 2 06:49
I'm gonna take this, I'm gonna take like the Carl Sagan, where like, if you're going to bake the cake, create the universe. So long time, like, you know, a couple months, maybe? Okay.
Host 3 07:00
A couple of months.
Host 1 07:02
That's one extreme.
Host 2 07:03
Yeah, that's like, you know, I'm talking like, you know, at least coming out of the ground somewhere. And then refining, going to production and transporting halfway across the world and then ending up in Melzer?
Host 1 07:17
Yeah, yeah, actually. Okay. So Steven, did you want to make a guess? Are you just going to go with really short?
Host 3 07:25
Yeah, let's hear let's, let's say, I'm just going to go the complete opposite extreme, let's say it takes 30 seconds, from nothing to real.
Host 1 07:35
Okay. So you both are wrong. And to the extremes. On average, it's about two to three weeks. And there's lots of it depends on it depends on a handful of things. But you're, it's on the order of two to three weeks, from when the manufacturer starts mixing powders to when parts are on a reel. And in that two week period, almost the entire time they're spent in an oven. And so it's it's yeah, they take a long time to make, but you also make like 10s of 1000s of them in a single batch. And so you get a lot of speed because of that. But yeah, it's it's about two weeks, which is when I was selling caps, I used to tell people, you know, keep that in mind when you ask for samples. Because, you know, if we say the lead time is four weeks, it's really two weeks, because it's going to take us two weeks just to just to bake the things.
Host 2 08:28
So is it bake because of the ceramic?
Host 1 08:31
Yeah, so so back to Stephens question about the kind of the nitty gritty, I'll give you a really high level view. And I think this is helpful, because I think this is one of the things that we sort of all take for granted as electrical engineers that we just buy this capacitor and it does what we want. So ceramic, and by the way, this is totally focused on ceramic capacitors, almost nothing I tell you applies to other dielectrics. Every dielectric has its own process. And so with with ceramics, you start out with powder. And basically the powders are effectively 4x Seven, ours is going to be barium, titanium, and a couple of binders to sort of just mix the powder up, that turns that gets some water gets added to that to turn into into a slurry. And then that slurry is poured onto or cast onto a plastic sheet. And so literally at this stage in the production, you have a flexible ceramic material that gets rolled up into a roll. The next step is you start to unroll that and then you take a silkscreen process and you silkscreen, an electrode pattern on top of it. And so if you actually look at pictures or get to see pictures of this, you can actually kind of look at the electrode pattern, and you can sort of see the individual layers of the capacitor get printed onto the material. From that point, the whole sheet is taken off the carrier, the plastic carrier, and then the ceramic layers are started to be stacked up. Next, they go through a cutting process and depending on the size of the cap, they're either sliced or die cut. And then around that point, they go into an oven. And so the oven is basically there to bake out the binders and basically center the the ceramic material. And it like I said, it takes a couple of weeks of process to do that it's they're not actually in like they're not at 2000 degrees C for two weeks straight, there's a whole process that has to has to happen to reliably bake them. The cool thing is when they come out of the oven, they're about 50% Smaller than when they went in. And that's just from all the organic shrinking, or the organics get burned out and then the material shrinks as centers.
Host 2 10:37
Well, that's cool. So I guess you you have to you would design your capacitors your Oh 805 would actually be a 1206 before it goes in.
Host 1 10:46
Well, yeah, yeah. Yeah, effectively. But you know, and this this actually, since you brought that up, Parker, you know, something that I'll I'll lead back into that. So you can edit. I just realized I just realized I was calling you Parker and Steven, Steven, without making sure that I knew what you guys look like. You picked
Host 2 11:08
Correctly. You're good. Yeah.
Host 1 11:10
It just occurred to me. I was like, I'm, I think cuz like, you guys sound different to me, since you're at normal speed instead of 2x. When I usually listen, almost always sound like
Host 2 11:19
Host 1 11:22
So since you asked that, Parker, you know, a question that comes up sometimes is or let me rephrase the question I had was, do we know before the process starts? What's going to come out on the other end? And what I was thinking of is, okay, do we just batch or do we just make a batch of stuff and say, Okay, well, these are 10. This one's 10. Microfiber, this one's 15. This one's 22. And the reality is every case voltage and capacitance has its own recipe. And so you know, ahead of time, everything that's going into place to make a specific part number. Now, tolerance gets measured and sorted at the end, but case voltage capacitance, you know what that's going to be when you go into the process.
Host 2 12:01
So tolerance is actually sometimes binned, then, all it's always been Okay, that's interesting. I thought that would be more on the designing. But yeah, that's, that's interesting.
Host 3 12:11
So would you have a different chemical mixture for different package sizes? Yes. Okay.
Host 1 12:18
Yeah, yeah. So. So when we look at ceramics, there's a couple different types. There's like x seven, our Caesar a G. And then well, let me rephrase that there's a couple different types there, c zero G. And then there's a whole class that are x Seven, r x five, our Wi Fi, B, that all have those letters. So that c zero G is class one. Everything else is basically class two and three, class two and three is mostly made out of barium titanate. And then for a specific recipe, you'll have different binders or what what one of the gentlemen I used to work with, he would just call magic fairy dust that makes differences in the makeup of the capacitor to make the whole process work.
Host 3 12:58
That sounds intense. It sounds like there's a lot that goes behind it and probably an enormous amount of trial and error.
Host 1 13:05
Yeah, yeah, I think I think the biggest eye opener for me was capacitors are almost as complex to design and manufacture as an integrated circuit. And I'm not talking like microcontrollers. But like, if you look at like a linear voltage regulator, or something relatively simple, and I'm not trying to disparage people that make regulators I'm just saying in terms of IC process, it's a relatively simple process. I think capacitors are easily as complex as that.
Host 2 13:33
Yeah. The what's interesting about that is it's, it's a lot sounds like a lot more like material science than electrical engineering is. Oh, yeah.
Host 1 13:42
Yeah. And I can so like, these were these were shared publicly so I can I can tell you these numbers. So when I was at Kemet, we had about 10,000 employees worldwide, roughly, let's just round. And of that we had something like over 1500 with an engineering degree. However, of those 1500, only about 100 had electrical engineering. Almost all the other engineering disciplines were either some form of science, chemical or material. That's cool.
Host 3 14:14
Aaron neat. So I guess one more quick question on that. A lot of times when you when you open up a electronics textbook, or you go look up a capacitor on Wikipedia, it'll show two plates next to each other and say this is a capacitor. That's not the case. In ceramic capacitors, correct? It's not two plates. There's a bunch, right.
Host 1 14:35
So okay, that's it's not the case with any capacitor. Ceramic is probably there are monolithic ceramic. So like, so there's four AC, AC mains, there are x and y capacitors. And so like if you look at an X or Y ceramic capacitor, it is literally just a chunk of ceramic material with silver electrodes on either side. So it's literally like the capacitor drawing But almost every other capacitor type outside of even ceramics are some other form. I mean, technically, the two plates are there, but it's not as clean as what the diagrams we like to look at are.
Host 3 15:13
Great. So one
Host 2 15:14
Of the big thing that gets brought up a lot in ceramic capacitors is this big part shortage that I'm having. And this is actually something that you wanted to bring up, James. And it's something that, you know, our listeners, and everyone on our Slack channel complains a lot about is, you know, they design with this one capacitor, and they go to make go order online, and they have to go switch the part out, sometimes multiple times.
Host 1 15:41
It's a question that I get asked quite a bit, and I've been outside of Kemet for about a year. And so I've had an opportunity to look back and sort of ask how did we get to this point? Because it sort of seems like the opposite of what you would think as a manufacturer that everybody should have known this was coming. And I don't think we did. So here's, here's what I noticed happened over the last couple of years. The first thing and let me see if you guys agree with this. Are there more or less electronics being packed into automobiles today? There are more electronics, more electronics? Right? So we're seeing tons of stuff happening there. In the aerospace and defense industry? Is that becoming more commercial or more governmental based? Commercial based? Yep. Right. So you know, we see this and it's happening, even with some of the stuff that the ESA is doing is more focused around private sector versus a more governmental approach. And that's important because traditional design and traditional requirements are sort of getting skirted. And so we're seeing a lot of aerospace and defense. So that means, you know, stuff that goes into space and stuff that only has to work one time, is starting to move away from military grade products. And they're saying, hey, why don't we just use automotive grade, because cars last a long time, so they must be good parts. And then even in industrial applications, we started to see traditional film and electrolytic designs transition to ceramics. And again, they want it to have automotive grade parts. And so what sort of happened is these three macro things happen. And it puts a big constraint on how many or puts a lot of constraint on how many automotive grade parts are available. And so if you combine that with the electronics industry, so Marotta a couple years ago, said they're getting out of the large case sizes, so that creates a credit disruption, I won't say did bad or good. It just disrupted the capacity of the industry. And second is to invest in a ceramic line is on the order of 10s of millions of dollars. It's not like it's a Oh, I get a million dollar machine to do this. It's it's a significant investment. And so all manufacturers have been reticent to sort of invest into that area. And then the third thing is, is that what's affected ceramics is because they keep getting thinner in layers, we technology allows for thinner layers, lower voltages, higher capacitances, you're starting to display some of these other capacitor technologies. And so you sort of take Okay, so the industry move people towards more ceramics and outside of the capacitor or component industry, everybody is moving towards ceramics. So it shouldn't be a big surprise that we got to a point where all of a sudden, there's less ceramics available.
Host 2 18:26
Yeah. So how did the industry just not think about that? Or it just snowball all at once? And hit?
Host 1 18:35
Yeah, I think I think it's more of a it was almost like a snowball is probably the closest I mean, I really feel like this all occurred over the last two to three years. I say even less than that. By Well, what I mean is, so I'm looking at like the macro trend, like okay, car content. The other the other one I meant to mention is in electronics, switching power supplies, their frequencies are going up. So you need smaller capacitors, so that draws even more like ceramic content into the design. And so I think, and then the final thing, and this is this is the one that I think people miss, because if you're not in the industry, you don't really see it. There's just been a huge constraint on automotive grade parts. Non automotive customers think that automotive grade parts are magical for some reason. And it turns out, we can or manufacturers can charge a premium for them. And so that's that's where they've put their focus is how do we get more for these premium parts? Without realizing,
Host 2 19:30
Yeah, that aq 2000 certification is just magic.
Host 1 19:35
Yeah, which is ironic because most tier one manufacturers test to q 200. Anyway, and you know, they might skirt it a little bit for their highest capacitance value, but most test plans are based on that. That's that reliability document anyway. And I think the thing that most people miss is a automotive grade ceramic capacitor is made with the same materials and same equipment as a commercial grade. So quality wise, they're not really that much different. It's they've been, they have higher reliability standards. But quality wise, it's like for two given parts if one is CERAM, or commercial one is automotive, there's more test done on the automotive. But from a failure point of view, they're almost the same.
Host 3 20:17
Hmm. So yeah, I guess you just maybe assume just a slight bit more risk. On the one that's not tested?
Host 1 20:26
Yeah, yeah. And so I
Host 2 20:28
Think what he's getting at is for your Arduino, you don't need a fancy dancy ceramic capacitor.
Host 1 20:34
Yeah, yeah. So you don't need an auto grade for most applications, I think. So for commercial applications, it's really, it's really overkill in my personal opinion. But where if you look at say, like the aerospace defense industry, they're used to buying a 10 cent part, along with $10 worth of test. And so they look at these automotive grade parts, which are only 15 cents, and they come with a packet of paperwork. So like, Hey, that sounds great. But like I said, they're really not all that much better. It's reliability wise, they're roughly the same. Now, again, the exception is if you look at the highest capacitance value of a ceramic Commercial Series, you won't find them in the automotive series, because they don't pass the the AEC q 200.
Host 3 21:19
Just Just out of curiosity, what what kind of range would that be in terms of value?
Host 1 21:25
So um, oh, that's a good question.
Host 2 21:28
Can I guess?
Host 1 21:30
I mean, I, yeah.
Host 2 21:32
Oh, 805 10, microfarad. That's pretty, that's pretty big.
Host 1 21:35
I mean, well, because I think you can get a 10 Mike 805 in x five, or x seven, or you could you if you can't, then you probably get a 4.7. And so let's just say that's the case, is that on an X seminar, which is slightly better than x five r 805 10. Bolt is the max you'd get is 4.7. Mike, so that you'd see that in commercial series in the automotive series, you may only see 2.2, Mike as the maximum, and it's because that design isn't quite pushing the limits of that process.
Host 3 22:07
Now, that makes sense. And I guess we touched on it just a little bit before but just to make sure for any of the listeners that are not aware the kind of the the phrases and terms which are acronyms, we call out x five RX seven are that that refers to the temperature coefficient, which that's about as far as I know what it actually means. I mean, I'm sure it's a chemical composition. And it's, it's a measure of how much capacitance change happens over temperature? Correct?
Host 1 22:38
Correct. Actually, I was I was getting a little bit nervous, because there's only two ways you can go usually it's the temperature coefficient, which is totally correct. Or a lot of people say it's the different dielectric type. And that's sort of correct. Even within I remember, when I was at Kemet, we would even call c zero g and x seven are different dielectrics. And technically, they are completely different material sets. But that's not what the three letter designator. It's exactly what you said seven, it's, it's the temperature coefficient. And so the three letters actually tell you the characteristic of the part. So x seven are the x is x is negative 55. Seven is 125. An R is plus or minus 15%. Which means across that temperature range, negative 55 to 125, the capacitance can change by as much as 15%. And just a quick pro tip is that it's almost never plus, it's almost always minus
Host 2 23:39
The plus there is a marketing thing
Host 1 23:41
Is, well, it actually covers there's some really strange dielectrics that depending on how the voltage applied to them, they'll actually go up a little bit and then come back down. And so sort of just as a catch all, but you're not going to go buy a ceramic capacitor from Digi key that goes up in capacitance across its temperature range, it's that they're just, that's not going to happen.
Host 3 24:02
This may be a bit of a question that we may not be able to answer, but I'm always curious, what is it that actually is changing with temperature?
Host 1 24:13
Host 2 24:15
I think it's how fast your electrons moving around.
Host 3 24:19
You there there? Can you convince them a little bit more with some heat.
Host 1 24:24
So that that's actually for other characteristics. That's true. On a ceramic. That's that's not what temperature does like if you look at a if you look at like a aluminum electrolytic it actually is a matter of the ions move slower in the dielectric when they get colder. Not in the dielectric. The ions move slower in the electrolyte when it gets colder, which causes the ESR to go up and Illumina electrolytics are basically an RC structure. So if you have more resistance, you have less capacitance honestly, I don't in Maybe Maybe it is the electron speed changes. I can't imagine that's what it is, though. But I offhand I'm I'm drawing a blank on what the what the temperature why the temperature affects it?
Host 3 25:08
Yeah, no, no worries. The that's always been, I've always been curious as to why and I'm sure it has a lot to do with the chemistry behind it.
Host 1 25:16
My, my guess is it's that kind of temperature range is actually changing the shape of the barium titanate molecule. And that would affect the permittivity, which gets into the DC bias effect, which I'm sure we'll talk about in a minute. But I'm not positive. That's why they shift with temperature. And shame on me for not knowing.
Host 2 25:36
Well, I guess we can just segue right into like, Why do ceramics lose capacitance, and one of them we just talked about was temperature. So what's the other reasons?
Host 1 25:46
So we mentioned the first one, which was tolerance, right? So right off the bat, you buy capacitors that are 510 or 20% of their nominal value, and just a hint as you go, so we so when you look at a chart for capacitor, across one axis will be case size, across another axis will be capacitance and then voltage gets worked in there somehow. And so the higher you go in, in capacitance, voltage or smaller in case size, the, the tolerance or as you get closer to the edges, the tolerance values tend to be closer to 20%. And not 5% or less, right? It's, it's you're, you're pushing the process to the limit. So it wasn't unusual for us to make a 200 mile or 220, micro farad capacitor. And then internally, we kind of laugh and say, Well, really, it's only to 200, because we're going to sell it as a 220 at minus 20%. But yeah, tolerance is probably and I think tolerance is the one that we all probably know the best, right? You do a search on a on a distributor, you can see tolerance easily in one of the columns. So that's the first one. The second one we thought we just talked about was temperature. And then the third one you guys know about DC bias effect? Yes. And is that a good thing or a bad thing? Very bad thing?
Host 3 27:03
Host 2 27:06
The funny thing is with DC bias is, we'll talk more about it. Yeah. But is it's one of those dirty little secrets of like, capacitor data sheets, because sometimes they don't even say what they are, you got to dig even farther. And I always find that interesting, as well as you just don't learn about this thing at all in school. And it's really the most I feel is like one of the most important specifications for capacitor.
Host 3 27:34
Yeah. Let me interject here. I think one of the things that kind of sucks about electronics is in general, it's not the easiest thing to learn the theory behind stuff. And then as soon as you're like, Oh, I finally understand the theory, everyone comes up. And it's like, everything you just learned is kind of wrong. There's a lot more there.
Host 1 27:56
So So DC bias for capacitor, sir, remind me of learning about the the, the, the voltage threshold for a MOSFET. It's like, you sort of get people to the first level of understanding of what is VGS. And then you can kind of, you can kind of see if they understand by switching it from an n type to a p type, if they can comprehend what a negative voltage means. And it's like, okay, I got it, or they've got it. And then the next step is okay, you understand that that is the that the maximum number is the minimum number to consider for it to start to turn on. And then you just get that totally glossed over look. Okay. What is this? What is this mythical voltage? Yeah, so DC bias, let's, let's talk a minute about that. I want to go a little bit deeper than I normally do. Just because,
Host 2 28:42
Well, you're in the right spot for it.
Host 1 28:45
It's fun. Yeah, you guys are I think the right people to hear about this. So DC bias, if listeners not aware of it, as we're talking about, with ceramic capacitors, and it's only certain ceramic capacitors. As you apply a DC voltage, their capacitance drops. So even though it's a 100, micro farad capacitor, and let's for make numbers easy, let's say it's a 10 volt 100 micro farad capacitor. When you apply five volts, it is not unreasonable for it to lose about 30% of its capacitance, meaning at five volts, it's only about a 70 micro farad capacitor. And at 10 volts, it may be as low as 25. It may lose 75% of its capacitance, meaning that this capacitor is actually only about 25 microphones, even though the part number clearly says it's 100 Micro ferrets and capacitor manufacturers just expect everybody to know that. And as Barbara was alluding to, they're really good about not telling you what's going on in the datasheet. What I think is even worse is some capacitor manufacturers will put a typical graph for their DC loss or DC by we call we always call the DC bias effect. I know other people use different terminology for it. But that's that's What I'm talking about when I say DC bias or DC loss, so you would see a typical graph, which personally always frustrated me, because there are a lot of factors that go into what is the DC bias of a particular capacitor. It's not as simple as capacitors from vendor a lose this much, and capacitor vendor B lose this much it. I mean, even within a case in voltage different or within, within within a, within a case size, different rated voltages will have different loss factors associated with them.
Host 2 30:32
Yeah, the ones I've seen the, when you actually start digging into it, a lot of times you have to find that specific part numbers. Mm, kind of specialized data sheet that has that in there.
Host 1 30:45
Yep. Yeah. Yeah. So, yeah. Yeah. And actually, it's sort of a it's sort of one of these catch 20 twos in a way because capacitor manufacturers aren't actually trying to hide this information. It's just it's really difficult to figure out how do you how do you convey it to somebody in a reasonable, concise manner? And so, okay, so like I mentioned before I used to work at Kemet at Kemet, we have a online simulation tool called casein which will actually plot the graph where specific parts because we actually measured it and then we're that's how we convey it. Now they don't work for Kevin, I can say, Marotta has a online tool called sim surfer, which provides similar information. The only caution I'll give you their is based on what I can see Marotta looks like it's they're doing everything based entirely on math. While at Kemet, we had a combination of math models that we did, we along with measured measured data just to make sure that our models seem correct. So I'm not saying that to say something negative about Murad. I'm just saying that's another thing you have to consider on top of everything else is, okay, what is this data really showing me? And so that's just something to keep in mind when you're looking at that.
Host 2 32:01
Yeah, I was gonna bring up is Kemet and merata wood, the two companies I found that like, it's very easy to find that information. I think AV is AVR AV X AV X. Sometimes you can find their datasheet for their part easily. It's sometimes somewhere in the Digi key page. Sometimes you have to go to their website and find it.
Host 1 32:21
Yeah, yeah. I think in terms of their online tools, I just don't know it as well. But they're actually they're getting better. You know, one thing like, and I'm not trying to plug him in a whole bunch of it's just what I know. So for distributors like Digi key or mouser, or Newark, we actually created part number specific data sheets. And the idea was, I as an engineer found it frustrating. And I'm not saying they did it because of me. But I like this about what we did is I hate it like going to a ceramic part and downloading a 45 page PDF that is like the entire family. It's the
Host 3 32:54
Family. Yes, yes. Yeah, that's right. Everyone does that.
Host 1 32:58
Right. And so we're like, Alright, so we, we had built an entire system so that we could actually, so kind of has, when I left, we had 6 million unique part numbers, and you could get a part number specific datasheet on all of them.
Host 3 33:10
Wow, that's, that's fantastic. Now,
Host 1 33:13
It didn't they didn't all have data, like the DC bias. But you know, that's a whole nother subject.
Host 3 33:18
So so this all kind of brings up an interesting question. If ceramic capacitors lose capacitance, when you apply voltage, which, that's what you do with a capacitor, why do people use them? Right?
Host 1 33:33
Um, so there's, you know, there's a couple ways you can mitigate it, the best way is to simply get a higher rated voltage, because it is a so it's a function, the DC bias effect is a function of the thickness of the dielectric. And so if you go to a higher rate of voltage, it means a thicker dielectric, which means the DC bias is less. And so if you can try to avoid that 50% range, you can actually you're you're only down to 80 to 10 to 20%. And so it's a little bit easier to design around. But the other side of it is even when you look at like there was this 1206 6.3 volt 470 microphones ceramic capacitor, I mean it the thing lost like 99% of its capacitance when you looked at it, yet, when you applied its full D rating. It was still cheaper. It was still cheaper than say a tantalum or Talon polymer. And so it's like, okay, well, I do need to have for these, but it's still cheaper than an alternative part of this kind of size and voltage. So to me, it's and this was sort of my philosophy when I was at, at the capacitor company was let's just give people the data because I think if you at least know what's going on, you can decide should I design around this or not?
Host 3 34:46
Sure. And, and it sounds like it's, it's, first of all, it's up to the designer to understand that this is a factor that will affect or could affect their design. So if you had a really I don't know time A sensitive charging circuit, you wouldn't necessarily want something that changes its value as it charges.
Host 1 35:08
Now you, I mean, if you're doing anything like a timing circuit or some kind of filter, you really want to use Caesar G, if you can. Caesar a G is a different class different material set. It has virtually no measurable DC bias or minimal temperature effect. But it comes with a 10th of the capacitance that you can get in the same case of voltage size. So ideally, you want to try to use something like that. That's ultra stable versus something that's going to change if you cough on it. Well
Host 3 35:39
NCC zero GX come with a little bit more of a price tag on
Host 1 35:43
Them. Yes, they do. Yeah, yeah. But they're Yeah. And that's totally a volume thing. It's, it's funny, because there's nothing exotic in their material set. It's just we don't we don't as an industry manufacturers, many of them.
Host 2 35:55
Hmm. So the the funny thing with this voltage bias is we were manufacturing some boards for for a company and they were had they had some 2024 microphone, I can't remember it's 20, something microfarad capacitors on the backside of the board. And they're like, oh, 805 or something like that. Or maybe 12 or six. Anyways, we were testing them. And they were testing like the machine was actually measuring the capacitance. And they were measuring 12 microfarad. Yep. And I'm like, Oh, I bet I know what this is I dug around, found the DC bias chart and met and then got the specifications for the tester and testers testing it at 1.5 volts. And the DC rating was 6.3. And at 1.5 volts, it was this capacitor is already SPECT to lose 40%. So I'm like, oh, yeah, that's something spec. And that was when that customer learned about DC bias capacitors is during production.
Host 1 36:56
So I have a follow up question to verify if this was the correct diagnosis. Are these measurements done before or after the reflow? Oven? Both? Okay. The suspicious thing about that measurement is that most capacitor testers use an AC waveform and not DC. And that's the Get rid of the DC bias. This side of DC I'm not even sure why I just I think it's just been. That might be why Oh, no, that makes sense. Usually, it's like a one volt. It's a one volt AC measurement. So okay, so, um, I'm glad to hear that story. But I totally thought this was gonna have a different ending, because the fourth way that ceramics lose capacitance is through aging. And you normally see that when somebody tries to measure the parts after they come out of a reflow oven. Over time, ceramics lose capacitance. And it's, it's in what we call decade hours, which occur at 110 101,000 10,000. so on. And I used to get calls from contract manufacturers all the time freaking out because their capacitors were not measuring correctly, or filters weren't working correctly, as soon as they came out of the reflow oven. And my internal joke was, well, I'd wait two days and call them back and then have them measured again, and they'd be fine. And the reason why is right after during reflow, the capacitors reach their curie temperature, which causes the barium titanate molecule to shift into a different or phase shift into a different shape. And that changes the permittivity of the part. And so over time, it relaxes and then we get the rate of capacitance back. But for that first, like 10 hours and 100 hours, it was totally normal for the capacitors to be way outside of their their tolerance band because they were basically too excited. And so that's that's where I thought you're gonna go with that one. Because I, I would see that all the time from people measuring stuff right off right off the oven. Now I've
Host 2 38:51
Never heard of that. But no, we they were it was tested by the the actual pick and place was picking up the parts and Okay, doing an E test on them. Okay, and the spec says it was a bias had a DC bias on at 1.5 volts and like, oh, yeah, so that was an easy one. But I've never heard of that, that reflow thing. That's interesting.
Host 1 39:12
Yeah, it's called an aging effect, which is a little bit of a misnomer, because it's not every time the part sees the Curie temperature, it follows the same curve again. So to me, aging sort of suggests that it degrades over time. And it's really just that the capacitance shifts over time. And so like I said, the magic time is up to about 1000 hours. And then after that, it changes like two to 3% over like years and 10s of years. But we I use used to I mean, when I was an FAA it was about once a month I'd get that call of a panic line assembly person that has boards failing and it's they just happen to have parts that that shifted during the reflow.
Host 2 39:55
So is that just a common thing then? Is that happen all the time? I guess
Host 1 39:59
All All Barium titanate have experienced that okay, and this is one of those. So we talked a little bit a few minutes ago about Caesar g and x seminar. So let me talk, you'll also see letters like x seven R and x five R. And they're both basically they're both Barium titanate, but x fiber is rated for a lower temperature range, it's only up to 85 instead of 125. And one of the tricks to sort of do that is you make the dielectric layers thinner, which is why x five Rs will tend to have slightly more capacitance than the next seminar. Another trick is that, because they're sort of not meant for automotive type applications, their referee time or measurement time is usually around 48 hours. And so that's just something else to watch for is if you use x five bar versus x seminar, when it comes out of the reflow oven, the x five bar is rated at 48 hours instead of 1000. Which means within the first couple of years, it could actually measure a few microphones lower than it should have, because of this whole aging effect. For most people, it's probably not that big of a deal to worry about the design until about 10 years. Like if you get something going into service for 10 years, then I'd worry about aging. But basically, it's when it comes down to reflow. Know that capacitance is going to be off. And about 10 years after life, it's going to be it's going to start drifting down outside of its tolerance range.
Host 3 41:24
So that being said, if, say you have a design that that needs measurement, testing or calibration, just having the device be cool out of the oven is not enough. Yeah, well, it would depend on the design. But but you if you have a design that utilizes these capacitors as part of a calibration circuit, it would almost be best to wait a day or two before doing calibration.
Host 1 41:50
Ideally. Now here, the good news is, like I said, this isn't a degradation mechanism. And so it's not that difficult to characterize it. And so I don't think it would be unreasonable to say, Okay, for this part number, we've tested it, we've characterized it, here's the calibration based on all of these factors, it's totally possible to carry it out. You know, it's not one of these things that has a whole lot of chaos involved in it. But it's all again, it's one of those if you don't know about it, then it can bite you.
Host 3 42:17
Once again, it seems more like one of those things where you'd like why do we buy these cats?
Host 2 42:25
That's a good segue to the next question. Is, what do you consider when choosing the alternative dielectric technology for capacitors? So let's say the say James here is TR pletely turned you off of ceramic capacitors, and you want to use something else?
Host 1 42:44
I have a I have an old boss at Kemet, that'd be really excited if that was the case. So let me Yeah, so I don't want to go into deep dive about all the other dielectrics. But I thought, let me at least touch on. What do you think about if you're trying to get away from a ceramic under design? Since we do have this shortage happening? And you know, there are cases where it's like, I got to get my I got to get my board in production? What can I do? So, you know, I think the obvious thing is, you look at the specs, and people say okay, well, if I look at capacitance, ESR, ESL, those are going to be the main things you got to watch out for. The disappointing thing is that usually, I mean, if it's a non critical circuit, or non critical use, okay? May just just get whatever ceramic you can get. But like say if you're doing like a PDN, or something, or a power delivery network, it's going to be really difficult to say swap that ceramic that you spent weeks designing in for something else, because you'll you can find the capacitance. And you might even find something that has similar ESR. But its inductance will be nothing like the other part. And so the thing that I wanted to stress is make sure you're watching all three of those parasitics or I guess, spec and parasitics. Because when you start going to say from ceramic to polymer tantalum, which is probably the most likely change, you can get close to ceramic ESR is with Polymer tantalum, but the inductance is much different between the two. And so you have to take that into account. Is that going to mess up my design or not? And then, yeah, so like I said, I just want to sort of mention that real quick, because I think a lot of times we think about capacitors as a monolithic component, and they're really not each dielectric is its own type.
Host 3 44:31
Well, and and sort of, it's both good and a little bit unfortunate. But that's the way we're taught, you know, a capacitor is just a single item. And even if you use like a simulation tool, there might be a difference in the symbol, but there's not gonna be a difference in the way the simulation handles it.
Host 1 44:47
Yeah, and how many simulation tools take into account dC by DC bias or temperature loss, right. Oh, and that reminds me of another point that I wanted to make is even if you're switching ceramics for from vendor to vendor, the thing that I would sort of watch out for there is, well, like I said, Watch capacitance, ESR, and inductance. Those are going to be these things compare against. But I will also tell you ceramic manufacturers are not the best about characterizing the inductance of their parts. And so it's really helpful if early on when you're setting up your sources that you get second and third sources for your parts and measure them yourself. Because I mean, an 805 has about the same inductance from vendor A, B, and C, but there's going to be some differences. If you just look at their datasheet they can be wildly different. And it's all about how they got measured. And so it's just another thing to take into account is make sure you've kind of taken due diligence to do your due diligence to take those into account.
Host 2 45:52
Cool, I've learned, I think my favorite is the the reflow. That's something that completely new I've never thought of before.
Host 1 46:01
I will find I have a I know of a datasheet or an application that covers that I think you'll enjoy looking through that. I think it'd be really good for you guys to know about that. Because I'm I'm shocked if it hasn't come up
Host 2 46:16
Yet. After five years.
Host 1 46:19
Yeah, yeah. If it hasn't come up yet. You're it's just a matter of yet. Yes.
Host 3 46:25
I know, in multiple multiple times, I have, you know, taken things right off the reflow oven, you know, given them 20 minutes to cool down and then you know, fire him up. Give him a go.
Host 1 46:37
Yeah, actually essential mentioned that man. Let me talk to hobbyists for a second because you know, I've, I've got a toaster oven, I convert it to a reflow oven. And I saw this happen in maker spaces, please do not open the oven door and just let a massive amount of cold air come rushing in as soon as the board has done. Give it a little bit of time for the I mean, even just a few seconds of holding the door open to slowly allow the heat to escape. Because manufac ceramic capacitor manufacturers are not joking about the cooldown rate of their caps. And what makes it even worse is if you thermally crack a ceramic, it could be a while before you know about it. And so one day your circuits just going to start acting funny and you'll have no clue what happened.
Host 2 47:22
Yeah, that's that's the big thing with our reflow oven profile is making sure that cooldown rate is set right. And we Yes, Beckett towards a ceramic capacitors and icees.
Host 3 47:34
Ceramics don't like to, you know heat up and cool down real fast.
Host 1 47:40
Like they're made out of ceramic.
Host 2 47:44
Another thing we haven't touched on on ceramic capacitors, though, is also as physical cracking through board flexing. And that's another big one that I actually work with our customers a lot when they do enclosure designs. I'm like, You need to move those ceramic capacitors away from those mounting holes because you will have fractures and failures and field because of that.
Host 1 48:07
Oh, absolutely. Yeah, the number one failure for ceramics is what we call flex cracks. And for for those that aren't familiar, it's basically after the capacitor has been mounted on the board. If the PCB flexes a certain amount, it will literally crack the capacitor. What makes it really insidious is sometimes depending on the the amount of fracture, when the part comes back together, the electrodes realign or maybe one or two electrodes don't can't connect. And so you don't notice right away, there's a problem. And so it's not until some moisture seeps in and causes a short, which then causes an open after some explosive results that you realize something happened. Can I tell you guys a story about flux cracks? Oh, yeah, totally. Oh, please. Alright, so I had a customer and they were experiencing failures. And there's a couple of telltale signs that I would ask questions about to find out. So I could, in my mind, say, Okay, is there a quality problem? Are they flexing them? And 90% of the time they're flexing them? So the first question was, is it the same part in the same position on the board? And the answer was, yes. And in fact, they had two copies or the part in two places, but only one was failing. So that's Red flag number one. Number two was is it near a mounting screw or mounting hole? And they said, Yes. Okay. flag number two. Is it your connector? Yes. Okay, well, there's number three is near the edge of the board. No, and that's how I know it's not a flex crack. Okay. Yeah, you got you got three out of four on that test. And that's, that's not a good, good test, or that's not a good grade for that test. And so, they went so far as they took capacitors directly off of a reel, sent them to a third party Test Lab and had them do cross sections to see how many of the classes had flex cracks in them from the factory, because they could not accept that the parts are getting cracked after they went on the board. Which is, from a manufacturer's perspective, flex cracks are my were my favorite failure because they're the only failure, pretty much the only failure where there's nothing we can do to cause it. It's all on the customer. So this was interesting. So the guy calls me up one day, he says, James, I got good and bad news. The bad news is I got back the test results, and I'm pretty disappointed across the board, he sent in our caps with four other manufacturers. And I know three of them would be what I would call tier one. So like a merata, or a TD case style company. And then one was like a Asian company I had not heard of. So you had for tier one and one questionable manufacturer. So great news. Only 15% of your parts have flex cracks in them. The worst case 50% of the parts had flex cracks in them. I was like, Okay, I want you to step back and think about this for a second. You're telling me that, on average, the number one capacitor manufacturers in the world ship 20% of their parts broken? Hmm, I guess that doesn't make a lot of sense, does it? And it turned out they had cranked up their microscopes. So high, they were actually seeing the grain boundaries between the the ceramic particles, and counting those as flex cracks when it's just like a perfectly normal structure. Right. So so at that point, I'm like, Look, let me come to your manufacturing site and see what you guys are doing. So we met at their, at their I think they were they were actually doing it in house at this point. And so he's walking me through the line. And I'm like, Yeah, I got to the last step. I'm like, Yeah, I don't see what's going on. Well, it was a device so that I don't accidentally identify them was about the size of a palm. And it only had a couple of icees on it plus the the passives. And then they took a plastic cover and snapped it onto the board. And I watched as this guy literally bent the board around his hand as he snapped the cover on. And I mean, I could actually watch the parts break. And it's like, Okay, I think I found the problem, guys. And it was like, it wasn't until then, because he had never actually come and looked at it. He's like, Oh, yeah, I guess they're bending the boards during manufacturing. I was like, Man, that sounds a lot like something I said about two weeks ago. So here's the here's the the the final like mic drop part. The size of the capacitors, he was cracking or Oh, four Oh, twos, which meant they were bending. Like I said they were bending them so far that they were able to crack one of the smaller components. Yeah, cuz that's
Host 2 52:45
Actually the thing is with Flex cracking is it's usually the larger packages are wrongly susceptible to them. And not the smaller ones.
Host 1 52:54
Yep. Yeah. Yeah, yeah. Usually it's like, oh, 603 is I would call questionable oh, five and bigger. Those are always susceptible. Same thing with hand soldering. Be really careful with anything bigger than an 805 or bigger, that you don't let your soldering iron touch the terminal. Because they're just big enough that you'll actually get a thermal gradient and it'll crack along the electrode pattern.
Host 2 53:17
I need to re solder Oh, my boards. I did my hand.
Host 3 53:21
Yeah, I didn't. I didn't know that one. That's interesting.
Host 1 53:25
Yeah, actually, I'll try to find a link for I'll see if, if the paper paper is still up, because this will be really fun for you guys, as well as the listeners. Kemet has a really old application note about hand soldering. And to give you an idea how old it is, it starts out with Surface Mount Technology is a new method for mounting parts to a board that may or may not take off.
Host 3 53:48
That's great. Well, so I think above and beyond that. There's there's another thing that is super common, that we all kind of gloss over and I'm sure we've all been there but panelized your boards up in an array and V scoring in between the boards and then just taking the whole panel and bending it over. I personally broken things that way. And V score was never really intended to just be held in your hand and snapped apart. It was it was meant to actually be cut on there. With it with a proper cutting tool.
Host 1 54:22
Yep, yep. Yeah, yeah. I mean, that's that's yeah, I would say outside of just absolute abuse, like the story told, that's probably the most common reason that they get cracked is V scoring. Now, there is good news, you can do a little bit about it to sort of mitigate the problem. There are ceramics with flexible terminations, and so flexible termination Flexi cap. There's probably a couple other words, but if you see something that talks about flexible termination, it's depending on the manufacturer, it's basically like a silver epoxy. that is in the in termination that actually gives the the solder point, a little bit of pliability. And so generally parts that have the flexible termination will be able to be bent, the board can be bent twice as much before the part cracks. And so that actually mitigates a huge amount of this problem if you just get flexible terminations. In fact, most automotive customers get flexible termination, just because they'd rather have the reliability, the added reliability versus dealing with the cost.
Host 3 55:32
Host 1 55:34
They're usually about 20%. More Okay, on if you just kind of pick a number, and like I said, it's, it's one of these things where, you know, yeah, sure, if you're building an Arduino, who cares? But, you know, if you've got a board, and you know, it's gonna be be scored, or you've done your pilot run, and 10% of the boards had random ceramic failures, it'd be worth built doing a build with the flexible terminations to see if that changes. Yeah,
Host 2 55:55
That improves your yield. Mm. Cool. So I want to go over a couple questions, because we're running like an hour. And sorry, notice, could we go? This is awesome. Um, so we're gonna run through some quick questions. So this would be kind of like the RFO section. It was just Steven nice. It's the our CEO rapid capacitor.
Host 3 56:18
Host 2 56:20
Host 3 56:22
It sounds like the capacitors have opinions about temperature and voltage dependence, their personal opinions about
Host 1 56:28
Them. I'll tell you how I personally feel about each of the different types I talked about.
Host 3 56:34
Yeah, so we started, we had some people on the Slack channel if they had questions, and it was just sort of like, earlier today, I was just like, hey, if anyone has any questions, and there was like, tons of questions, it just exploded into the channels like, okay, great. We got the cap guy, get you get your questions out.
Host 1 56:52
So hey, before, before you guys get to the questions, I just wanted to give you a, like a plug on your show for your Slack channel. Because, you know, I know a handful of people that would probably enjoy, like some of the conversations that happen, but they're like, oh, I don't want to install another app or whatever. But I just wanted to say it's totally worth it. I think it's it's a really good place to kind of hang out and ask questions. I don't talk a lot there. But I see a lot of really interesting things come by. So if listeners aren't already in the slack, they need to think about it.
Host 3 57:21
Look at that. We don't even have to plug our own stuff anymore.
Host 2 57:27
Alright, so the first question is, this person's asking insights into selecting bootstrap caps for high frequency, high voltage H bridges. So my question is, though, is what's a bootstrap cap? Because I've never heard that term before. Yeah,
Host 1 57:43
So that so this is this is uh, so normally, when I would do like seminars, I always tell people up front that when you ask a question, don't be surprised if I answer it depends. And this, this question actually makes me go immediately to it depends. So Bootstrap, right, that's like, when I read this, what I'm thinking of is probably a situation where they've got a MOSFET that they're trying to turn on. And so they're using a cap to basically act as a voltage doubler to make sure that the MOSFET turns all the way on. I don't know if that's actually what's going on. So I could be wrong about that. But I do know, when people were selecting caps for H bridge designs, the biggest issue we would run into is they would forget that at some point, the cat might see twice the applied voltage, and so they would not spectable to tie enough. And so just as a general Okay, one thing I do know to to look at is, make sure you're looking at a cap that has at least twice the rated voltage or the twice the applied voltage for its rating. If possible, whenever I hear high voltage, high frequency and if we're thinking talking about ceramic capacitors, if possible, you see zero G, since it doesn't have the voltage bias effect, it is less likely to break due to electro stiction
Host 2 59:03
Or what's electro stiction? That sounds it's pretty rad term.
Host 1 59:08
I might have actually set it slightly right. It's electro striction. It's the opposite of Pisa.
Host 2 59:13
Electricity makes electricity. What doesn't vibrates.
Host 1 59:17
Now I'm going to get them backwards. And you guys are getting all kinds of emails about how dumb I am. So PA, so is so one of them is you apply a voltage and the part moves? Yes. I want to say that's pa Oh, yes. And the other is if you move the party generates electricity. That's electric stiction. Okay. And so what actually happens is, let's say in a in a high voltage event, so the high voltage causes the capacitor to vibrate, which then then in turn the vibration as it comes back down, starts to generate a voltage and then you get like twice the event and it causes the whole thing to
Host 3 59:53
Crowd cascades and runs away. Sort of Yeah,
Host 1 59:57
Yeah. And I'll finish that by saying that's my explanation, and I wouldn't be surprised if I'm wrong. But that's how I that's like what I understand is going on in like in a, I say an events like an ESD event, that's, that's what's causing the cap to break down.
Host 2 1:00:11
So that's kind of similar to, I'm gonna say kind of similar because I could be completely wrong here too. It's like a inductor or coil. And when you shut off the, the voltage to it, it spikes. You have this flyback event.
Host 1 1:00:25
Mm hmm. Yeah, yeah, it's it's something along those lines. And my guess is there's like 500 events that are happening that add up to something else. But that's, that's my understanding is it's sort of like, like an inductor. Cool.
Host 2 1:00:38
So the double your radio on your voltage and pick, CZ oh, gee, a dielectric.
Host 3 1:00:46
C series on that actually,
Host 2 1:00:48
Host 3 1:00:49
Real quick, easy way to select bootstrap caps is a lot of times manufacturers will just give a bill of materials for an example design and use that calf.
Host 2 1:01:01
I like that answer.
Host 1 1:01:02
Use webpage. You know, on my side note on my YouTube channel, I did a couple videos on linear and switching regulators. And anytime someone says, Well, you show how to design a switching regulator. I just give a link to T eyes web bench because that's that's how I do it.
Host 2 1:01:18
That's how I do it too. I'm not ashamed.
Host 3 1:01:20
Yeah, I did it the other day.
Host 2 1:01:25
Okay, cool. Um, so next question just kind of goes towards the Where did the mlccs go? This is what footprint should we be designing for going forward? Presumably different sizes for decoupling verse bulk on power inputs for getting different ESRs.
Host 1 1:01:43
I'm gonna, here's what I'll take the answer that one, pick a case size. So for for bulk. I mean, in a lot of ways, it doesn't really matter, right, all you care about is you don't have a lot of capacitance. So get large case sizes so that you get a reduced voltage bias effect. And you get more effective capacitance. If you're designing the decoupling network, or you're trying to do filtering, like on a power distribution network. One recommendation I might suggest is pick a case size, like, Oh, 40206, or three or 805, whatever you're comfortable working with, or your manufacturers come comfortable placing and sort of stick with it. Because the inductance for individual case sizes is going to be roughly the same regardless of the capacitance. And so you can really get into a kind of a rhythm of okay, well, I know what the inductance of this package is going to be. So I get an you'll have a good feel for what capacitances should you be picking? If you try to say, Okay, do 805 This time and go forward to next time, it just gets to be a mess. So that that'd be my recommendation.
Host 2 1:02:47
And I have a kind of, at least for the first one first question. Part of that is, I've been doing a lot of research on what parts companies are like getting rid of and which ones are bringing, like new parts of bringing online. And there has been a reduction in like, oh, eight oh, fives. Yep, that seems to be the one that's signed kind of dying out. And then Oh, 603. And oh, four twos are going way up. But 1206 and bigger seem to be sticking around because they're kind of you need them for bolt capacitance or high voltage that Oh, fives are kind of dying. And there's been a big cut in like different voltages and different dielectrics, too. So like, instead of having this, like, you can get a 6.3 volt and in everything right there kind of cutting that down to, you know, more specific things, I guess, cutting their catalogue down? Well, I wouldn't say cutting trimming is probably a better term.
Host 1 1:03:47
No pun intended.
Host 3 1:03:47
Okay. Okay. Little tiny side note. And we I apologize if we could, you know, talk about this for a long time. And maybe I'm just looking for like a quick answer on this. But when we use the word bulk for bulk capacitance, you could pick a 10, micro farad, alumina electrolytic. Or you could pick a 10, micro farad 1206 ceramic. And let's say that's the input to your 3.3 volt regulator, or whatever you're messing with, which one do I go with? Which one should I pick?
Host 1 1:04:15
That's a great question. And the answer is it depends. So, the the the key, the key point of your question was that it's on the input side. And the thing is input side really doesn't care as much about the parasitics. Now, calm down for the people that have had issues where they pick too much ESR something in general, on the input side of a regulator, the ESR is not going to be a huge problem. It's at most going to cause a voltage drop, which you can probably account for, anyway. I mean, the whole idea is you're probably you're boosting or bucking it anyway. So what difference does it make? I think when you start getting into say again, on the input side, that's He's starting to get into this whole other mess of Okay, what about lifetime, because if you get a wet aluminum electrolytic, it's going to have a very defined operational lifetime. If you go with, say, an aluminum polymer, it will have nearly infinite lifetime and much lower ESR. But it's going to cost a little bit more. And so I think you just have to start to weigh in. Okay. Is there another secondary thing other than capacitance that I should take into account for, say, input side? And that that's where your sort of your judgment has to come in? I don't know if that that actually answers your question. No, that's,
Host 3 1:05:36
That's perfect. I love it. Next question.
Host 2 1:05:42
So we already talked about PCB flex. So I'm going to skip that. Let's see this next one is, generally general rules of thumb on choosing appropriate packages based on desired voltage rating and capacitance. I think you've covered that already. It's kind of like, pick the largest package, right? Yeah, so you can fit.
Host 1 1:06:04
It's been a while since I said this. So let me see if I get it right on the first try. So generally, and this sort of applies to all the technologies, first look for the capacitance and look for the capacitance you need. Next, pick the highest rate of voltage that you can, in the smallest case size possible. And the kind of the knob, you can twist there is if you don't, if you're not really size constrained, then get a larger, larger capacitor, because then you'll either get more capacitance or higher rated voltage, if you are size constrained, then pick the pick a cap that fits into your size constraint. And then within that get the highest rate of voltage that you can for whatever your capacitance need is.
Host 2 1:06:46
It's good answer. So then the next question is, I think we touched on this a little bit is what is the fundamental physical mechanism behind this voltage dependent capacitance drop or DC bias? And I like the answer that someone gave me a Slack channel, which was spooky quantum bullshit.
Host 1 1:07:07
Yeah, absolutely. Okay. So first note for listeners, I do have or Kemet has a video where a guy named James that is a bald guy with a Twitter name bald engineer talks about this with some really fantastically drawn graphics. I mean, they're just some of the best graphics ever made. So okay, now that that's over. So here's the short version of what's going on. It is not it's not actually Quantum. It's entirely chemical based. So like I said, barium titanate is made out of our most ceramics are made out of barium titanate. And in its natural form, it's a cubic molecule, which is basically the the points of the cube are all barium, with a titanium ion in the middle. And it turns out, the titanium doesn't fit inside the molecule. And so it sort of sticks to one of the corners, which gives the whole molecule a dipole, or its ability to hold a charge. And so remember, earlier I was talking about the crazy tests where they were, they were trying to find flex cracks, but what they actually measured were grain boundaries. Well, barium tiny is a crystalline material. And so it forms like these little like little chunks inside of the monolithic material. And so within those kind of grains, the molecules sort of self align some direction, and you know, there's, it's virtually random. So that's called self polarization, because all of the parts are sort of naturally line whatever way they fit. And that gives it gives the material really high permittivity, which equates to a high dielectric constant, which people which actually dielectric constant means relative is the permittivity for a capacitor, but it turns out, it's not constant. So we all stopped using dielectric constant. So well, nothing is applied. Everything's kind of sitting in its natural state. That's why we get this really high capacitance. As a electric field is applied, that titanium ion gets pulled in the direction of the electric field, which then causes these domains which were kind of which were random to start polarizing in the same direction. And as they polarize the permittivity drops, which means the capacitance of the material drops. Sea. So
Host 2 1:09:42
That is actually a material problem, not a unknown quantum bullshit problem.
Host 1 1:09:48
Yeah, yeah. And it's actually it's totally it's like it's it's a really well understood mechanism. So it's, you know, there's a couple other things that I've, I've learned about that. It's like well, To really intelligent, just brilliant, not intelligent. That's not the word I want to use two brilliant material scientists have given me fantastic explanations on why this phenomenon occurs. And they completely disagree with each other. But they both sound so correct. It's like how can they not be both? Right? That's great.
Host 2 1:10:19
So the second last question I have is why dielectric Wi Fi V.
Host 1 1:10:28
A long time ago, that was the best we could do for high capacitance and ceramics. But formulations have gotten so good layers have gotten so thin processes are so robust today, that it's, it's most of the time that you buy a Wi Fi ve, it's actually going to be an x five or it's basically around so that manufacturers can offer a low price point to whoever My recommendation is never ever design in Wi Fi B or Z five U or things like that stick to x 5x Seven something
Host 3 1:11:00
You know, maybe this doesn't apply to surface mount components. But I saw I saw one that I haven't seen before just the other day was SLS zero or something. Which I saw in through hole ceramic capacitors not not surface mount and and when I looked it up, I believe it was about as awful as you can pause. It basically is this thing is a capacitor when it feels like being a capacitor like it like it was the specs were all over the place. And I hadn't seen that.
Host 1 1:11:34
Yeah, I I've never heard that one. I'm just trying to think like this. This is all based on like a chart. And I just, those letters are not. But none of those sound like good letters.
Host 3 1:11:48
And they they're not.
Host 2 1:11:51
I found a datasheet for a vicia through hole that's got the S L zero. So I'll share that around with everyone after the podcast.
Host 1 1:12:01
Just real quick. Is there anything on it that says like class one, class two, class three or? Yes,
Host 2 1:12:05
Yes, it's class one, Class Two. class three. Says general it's a it's a family data sheet. Okay. Okay. So it's got class one, class two, class three, and then it's got a, they make this type in. SL zero s three n x seven ry five p x five FZ five U? Wi Fi V. It's all over. So you guys ever have like my tongue is just going a little crazy. You guys
Host 1 1:12:35
Ever have like leftovers night at your house where you just pull out everything in the fridge and make try to make me a lot of it was that? I think that's what these guys did. They just took whatever powders were leftover from the normal manufacturing and mix it up or like Well, that's a secret that
Host 2 1:12:48
They measured point one microfarad. Once Okay, so last question. James. Is what do you have any haircare tips?
Host 1 1:13:05
So, okay, growing up, my brother was almost 10 years older than I am. And he lost his hair, almost exactly as he turned 21. And so when I made it to about 25, without losing my hair, I thought for sure I've got this down. And then it's like, as soon as my next birthday happened, it was it was all turtles from there. And so I had a really close friend who told me wearing hats is the key to losing hair. And it turns out, no, it's all genetic. Because I never wore a hat. Or my brother always my brother never wore hats. I always wore hats. That hats had nothing to do with it. So so no, I have no haircare tips. The best place to find me is bald engineer.com. It's my blog where I kind of post everything that I'm doing as well as new tutorials. I have a YouTube channel called add ohms. I'm also on the element 14 presents YouTube channel on occasion. And if you happen to be in Minneapolis, the last week of October, I'll be at the embedded systems conference.
Host 2 1:14:08
Anything else you want to add James? Or do you want to sign us out?
Host 1 1:14:13
Just want to thank you guys for letting me come on. This was really fun. You know, I think one thing I'd like to say to everybody listening, don't be afraid to ask questions about capacitors. As you can probably tell from this conversation, even people that claim to be experts get kind of flustered around things. And what I noticed in my career was there was always an engineer who was afraid that they're going to sound stupid. And the thing is, is when it comes to capacitors, it's one of the few electrical engineering topics where there is I promise somebody else around you that has just as little clue as you do. So don't be afraid to ask questions about something as simple as a capacitor.
Host 2 1:14:50
And with that, you want to sell James
Host 1 1:14:52
Alright. That was the macro fab engineering podcast. I was your guest James Lewis
Host 2 1:14:57
And we are your hosts Parker, Dolman and Steven Greg later everyone take it easy
Host 2 1:15:09
Thank you yes you our listener for downloading our show if you have a cool idea project topic or sweet capacitor that you want Stephen and I to discuss, tweet us at macro fab or email us at podcast at macro calm. Also check out our Slack channel that James conveniently let us know about if you're not subscribed to the podcast yet to click that subscribe button that way you get the latest episode right when it releases and please review us wherever you listen as it helps the show stay visible and helps new listeners find us
Transcribed by https://otter.ai