In case you needed reminding that ceramic capacitor voltage is meaningless

[T3sl4co1l]

Exhibit: http://ds.yuden.co.jp/TYCOMPAS/ut/detail.do?productNo=UMK316B7225KD-T&fileName=UMK316B7225KD-T_SS&mode=specSheetDownload

It's being sold as a 50V, 2.2uF part, but it's -30% at only 6V!

On a complementary note, today I spec'd a "2.2kohm 300mA" ferrite bead, that is -30% at 150mA.  Not bad, most ferrite beads are as weak as, well, the above capacitor!

Tim

[exe]

This one is just particularly bad.

I don't even consider buying any MLCC if there is no comprehensive data available. For Murata I use this tool (simulator):  http://ds.murata.co.jp/simsurfing/mlcc.html?lcid=en-us . It shows capacitance, impedance, dissipation factor etc. Other reputable manufacturers also have similar tools.

But yeah, it's a bad surpise that voltage rating for MLCC is just voltage that doesn't kill the capacitor, not the "useful" voltage.

[james_s]

Isn't it only a DC voltage across it that results in the reduced capacitance? If used with AC I thought it would retain the rated value. It's a characteristic that I do think is not nearly as well known as it ought to be, it should be very prominently features on the datasheet since these are so often used for decoupling on DC power rails.

[chris_leyson]

Thanks Tim, that one is particularly bad for X7R, it's a good reminder to check the data sheets.
It's got me thinking about a simple relaxation oscillator circuit to measure capacitance under DC bias, found a neat app note from Maxim. https://www.maximintegrated.com/en/app-notes/index.mvp/id/6014

[T3sl4co1l]

It actually goes up with AC, then back down as it spends more time in saturation than not.

The B-H curve, er, well, the D-E curve (displacement charge density versus electric field), is not actually a single hysteresis loop, but a bit flattened in the middle.  When it's nearly flat in the middle, this is called a butterfly curve.

Tim

[Bud]

The Subject of this topic is misleading. Voltage rating is not for retaining capacitance value, it is for dielectric breakdown.

[T3sl4co1l]

Yeah, but do they say so in the datasheet?  I wonder why?

Hence serving as the reminder.

Tim

[james_s]

As far as I know, most capacitor types do retain their value all the way up to the voltage rating so it doesn't matter. These types of ceramic capacitors don't, which is a pretty serious issue that can bite those who are not aware of it. I only learned of it in the last year or two and frankly it's a little embarrassing that I didn't know it sooner. It is not widely publicized though.

[T3sl4co1l]

They're the only ones, really.

Hmm, anything else will have some small kind of voltage effect due to simple mechanical stresses and electrostriction.  Mostly very small, ppm range.

Electrolytics can, over a very long time scale where, at lower voltages, the oxide degrades and thins, so C goes up; and when maintained at nominal voltage, or reformed back to it, C goes back down (or down further if over-formed?).

Self-healing films will drop over time, but that's a persistent effect, and only after overvoltage.

Varactors obviously do, but arguably aren't capacitors, because they're diodes.  They're also explicitly made and sold for that purpose.

Tim

[james_s]

Your expertise never ceases to impress me.

I wonder if a ceramic capacitor could be used as a varactor? Not that I would expect it to work very well, but it's one of those things that could be interesting to try.

[RoGeorge]

Capacitance varies widely with polarization voltage, it's the normal behavior for X7R and X5R.
https://www.murata.com/support/faqs/products/capacitor/mlcc/char/0005

[rfeecs]

I wonder if a ceramic capacitor could be used as a varactor?

There are products out there:

https://www.murata.com/en-us/products/capacitor/variable/variable

https://www.st.com/en/emi-filtering-and-signal-conditioning/rf-tunable-capacitors.html?querycriteria=productId=SC1881

At least the ST part seems to be ceramic.  It's not clear if the Murata part is ceramic or something else like a semiconductor.