Ceramic Capacitor Behavior

[ArdWar]

For non price/performance constrained design it's usually easier to simply design with enough margin that any typical capacitor is sufficient (aka the jellybean method). Personally I also rarely stray too far from using my selection of proven good design unless necessary. In this case characterization is more like verification, especially for EMI and other unintended interactions.

[Siwastaja]

the bigger capacitor manufacturers do have their own online detailed parametric charts for each P/N. Use that instead if you have to.

Yeah. It's understandable they use a (dynamic) web page instead of a PDF document, but why the f*** can't they just say they do. It took me years to realize that the data is available, just not on the datasheet. Why not print "for more data, go to our website at qwerty.com <insert QR code here>" on the document? Or stop producing the half-assed PDF datasheets completely and ask the distributor to link to their website instead? Nowadays reading this forum it seems many people still don't realize the data is available on the manufacturer's websites.

[tggzzz]

...It took me years to realize that the data is available, just not on the datasheet. Why not print "for more data, go to our website at qwerty.com <insert QR code here>" on the document? Or stop producing the half-assed PDF datasheets completely and ask the distributor to link to their website instead? Nowadays reading this forum it seems many people still don't realize the data is available on the manufacturer's websites.

“But the plans datasheets were on display…”
“On display? I eventually had to go down to the cellar to find them.”
“That’s the display department.”
“With a flashlight.”
“Ah, well, the lights had probably gone.”
“So had the stairs.”
“But look, you found the notice, didn’t you?”
“Yes,” said Arthur, “yes I did. It was on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying ‘Beware of the Leopard.”
― Douglas Adams, The Hitchhiker’s Guide to the Galaxy, 1978

Yes, almost half a century ago. Plus ça change, plus c'est la même chose

[mawyatt]

These are some of the very reasons we've investing in a Hioki IM3536 LCR Meter as well as a Tonghui TH2830 and a bunch of LCR Meter fixtures, some OEM, some modified, some custom!

Capacitors are highly complex with all sorts of issues, parameters, features, surprises one must pay close attention too, ceramics seem to lead in all these categories

In our case awhile back we had a new project which involved 128 up to 256 (actually could go to 1024) independent AWG channels capable of producing ~ +-100V peak signals with 16 bit resolution 15 bit accuracy, and have direct Bi-Phase waveform modulation with less than 5mv waveform average offset between Bi-Phase polarities with 200Vpp signals. We were recommend for this project by an Advanced Research USG agency based upon our prior career and were not allowed to attend the actual system application testing (no need to know basis), which involved directly controlling a new type Advanced Phased Array Radar based upon Liquid Crystal technology.

This project started off right when component availability became scarce due to the Pandemic, and we couldn't find anything in stock anywhere, and had to change the design often to accommodate what was available rather than what we wanted. Frustrating as hell, and scary, as the system had to work first time without our participation with this unknown system defined only by the interface!!

With this in mind, and not wanting egg all over our face, we decided to thoroughly test every critical component and not rely on vendors data sheets, especially unfamiliar ones we had to use because of the component situation. So we got a number of KS34465A DMMs, IM3536 and TH2830, all new with Calibration Certifications and set out to check each critical component.

Anyway, we were told the overall system was setup inside a large Anechoic Chamber, Target Identified with Coordinates, and Phased Array Beam nailed the Target Spot On, First Time!! So guess this effort paid off as the overall system worked as expected, and we didn't get egg all over our face

This effort alerted us to the various issues with unknown vendor ceramic capacitors (film types also) and precision resistors. Also created a more conscious review of capacitor data sheets, even from well known vendors, and we began to do more testing with these LCR Meters just to verify device parameters, even from our usual "Go To" vendors.

As stated by others even the "Quality" Sources for ceramic caps have details buried in various data sheets, on-line resources and such, which makes it difficult to "extract" what's important for the designer!! Here is where the quality LCR meters can prove worthwhile, allowing one to make measurements with devices "in hand" to establish component confidence, something we've had to do often lately

Best,   

[T3sl4co1l]

This brings up an interesting question....

For all you DC-DC converter designers, what is your process for selecting specific part numbers for input and output capacitors? 
.. How much work do you do up front scrutinizing datasheets ..
.. vs buying some caps and characterizing them yourself on their own ..
.. vs putting them in circuit and seeing how the converter works and going from there?

If the voltage swing is large, on a DC input or output, something's gone terribly wrong with my design -- or it's a wide swing application like a class D amplifier, where linearity has already made the requirement clear.  So for most cases, a linearized model suffices.  Here's a recent example:



The capacitor parameters go into a series RLC model (see https://www.seventransistorlabs.com/Modeling/index.html#spiceasim RLCSER.ckt), for convenient representation.  The "[value] (xN)" is display only, and the parameters are ESR, ESL and C in order.  The inductor is a Coilcraft model.  Modeled are 12 and 2 x 1210s, and 3 x electrolytics, for a 12V 800W converter output.  Load assumed, wirewound resistor.  Resonant, so the rectifier output is reasonably sinusoidal and RMS current is given, and thus output voltage amplitude at/around 2Fsw will be representative, and other frequencies can be eyeballed lower as needed (of course, it's got enough lowpass filtering that harmonics are quickly irrelevant).  AC and transient analysis are used, with input and output sources active depending on wiring (I3 is connected to input or output (dangling wire) as needed).  Only doing step response for output (step load, see rectifier response).  Also, note this is the worst-case assumption, with current driving the rectifier; a resonant supply may well have some impedance there, I hadn't tested that as of this model.

In any case, the notes below (and others) survey candidate parts in various ratings, and I put in the total value as needed.

Tim

[Howardlong]

Something else to be aware of that I wasn't until recently is that class II MLCC ceramic caps (e.g. X7R, Y5V, Z5U) degrade over time. Class I (C0G/NP0) aren't affected by this phenomenon.

This aging varies significantly not only by dielectric, but also by manufacturer, apparently due to the may the molecular dipoles arrange within crystals in the dielectric.

Thankfully most of these caps are used in non-stringent applications such as AC coupling and DC decoupling, but there may be latent failures in waiting in marginal designs. If you're using them for filters or timing with fairly stringent characteristics, you might want to re-think your design.

https://youtu.be/yLuWcTo2cb4?si=A9nzO6_W9xLe6vnY

[floobydust]

For MLCC ceramics, I've mentioned in thread here Good practice for MLCC power by pass capacitor before that they are full of game.
MLCC Class II aging in the first few minutes of use is an elephant in the room. Note that soldering heat also resets the crystalline structure wrt aging apparently. Pulling the part to measure it changes its value.
The lobbying MLCC cap manufacturers to stop the fake specs, after an automotive product of Continental's failed (LDO oscillating). I have no idea if anything came from that.
Even finding voltage-coefficient is difficult, some manufacturers really hide the bad news on their website. I don't use those brands.

For electrolytics in the sim, I would include tolerance and any low temp ESR derating if necessary. The ZLJ is -20% tolerance and triple ESR at -10°C.