Voltage rating of SMD capactor

[Pedro147]

I am building a Atmeg32u4 based quadcopter flight control board and the design calls for some 2.2nF 0603 ceramic capacitors rated at 16v ( http://www.digikey.com/product-search/en?keywords=490-11512-1-ND ) I cannot source these locally but I can source the same thing but rated at 50v  http://au.rs-online.com/web/p/ceramic-multilayer-capacitors/2040707/ )

If I were to substitute the higher voltage caps could this lead to any problems in the circuits operation please. My feeling is no but I prefer to ask someone much more knowledgeable.

Thanks Pedro

[Philfreeze]

It's fine to use the 50V cap, actually it is probably even better.
The voltage rating just says how much voltage it can withstand without taking damage.
You can imagine it a bit like a tank filled with something. The voltage rating would be kind of like the preassure rating of the tank. If you go over this rating the tank might expolde and you don't want that.

TL;DR buy it, it's ok


Edit: Here is a summary of capacitor characteristics: http://www.electronics-tutorials.ws/capacitor/cap_3.html
"2. Working Voltage" explains the voltage rating a bit more than I did.

[Pedro147]

Thanks Philfreeze

That is what I expected but it never hurts to ask the advice of someone more knowledgeable

Cheers Pedro   

Edit - Thanks for the additional info it is appreciated Philfreeze

[amyk]

Note that MLCC capacitors actually have a voltage-dependent capacitance, which decreases as voltage increases.

https://www.maximintegrated.com/en/app-notes/index.mvp/id/5527

[Pedro147]

Hello amyk

excuse my ignorance, but how might this affect my potential use of a higher voltage related capacitor in the circuit please

[bitshift]

In general capacitors have very loose tolerances, in the order of 10-20%. Your designs should try to avoid relying on a specific/stable capacitance value as it will vary with voltage, frequency, temperature and age.

Here is a nice introduction to different capacitor parameters.

[Pedro147]

Thanks for pointing that out amyk and bitshift.

I suppose that this is really more an important design consideration,than applicable to my specific situation here the way I read it, but I may be wrong  This circuit is not of my design and has been proven to be a sound performer but I am grateful that this has been pointed out to me so thank you both.

[T3sl4co1l]

A higher rated voltage is always better.

Regarding capacitance, the higher voltage part will maintain its capacitance to a higher voltage than the smaller part.

Neither of which I'm guessing matters, because the reduction occurs somewhere around rated voltage.  For smaller values like 2.2nF, I would expect capacitance is nearly constant up to rated voltage (for either part).

For larger values, like 0.1uF 50V or 1uF 16V, the drop-off point occurs at lower voltages (as little as 25% of rating).  But if the circuit is using it at 3.3V or 5V, out of a 16 or 50V rating, even that is fine.

Tim

[Ian.M]

Exception: totally weird circuits designed to exploit this behaviour, e.g. using two identical caps with a dielectric that strongly exhibits this effect, in series as a variable capacitor by applying a DC bias from a hi-z source to the center tap.

[Neilm]

The only possible issue I could think of would be if this cap was used for supply decoupling. The different capacitor would have a different characteristic and may not be as good for decoupling. However, this would be a very "way out" corner case that would probably be detected in an EMC facility but I doubt would actually cause real world issues unless you are incredibly unlucky.  Basically, just try it - I doubt you will see any issue.

[suicidaleggroll]

A higher rated voltage is always better.

Regarding capacitance, the higher voltage part will maintain its capacitance to a higher voltage than the smaller part.

Neither of which I'm guessing matters, because the reduction occurs somewhere around rated voltage.  For smaller values like 2.2nF, I would expect capacitance is nearly constant up to rated voltage (for either part).

For larger values, like 0.1uF 50V or 1uF 16V, the drop-off point occurs at lower voltages (as little as 25% of rating).  But if the circuit is using it at 3.3V or 5V, out of a 16 or 50V rating, even that is fine.

Tim

From what I've seen with MLCCs, the capacitance starts dropping immediately with increasing bias, not when you get near the rated voltage.  In fact the voltage rating of the cap has little or nothing to do with the drop in capacitance as bias voltage increases.  It's almost entirely a function of the dielectric and package size (0402 vs 0603 vs 0805, etc.).

https://www.maximintegrated.com/en/app-notes/index.mvp/id/5527

So if he sticks with an 0603 MLCC with the same dielectric type as the original, it should behave the same, regardless of the voltage rating.

[TimFox]

MLCC capacitors, like other types of ceramic capacitors, act differently depending on the choice of dielectric.
NP0/C0G capacitors do not have an important voltage dependence of capacitance.
X7R capacitors do have some dependence:  see data sheet, since the actual thickness of dielectric for a given capacitance and voltage vary between manufacturers.
Z5U capacitors have an extremely high dependence, so that the capacitance has fallen drastically by the time you apply full rated (safety) voltage.  Again, see the data sheet.
2.2 nF should be available in 0603 package in any of these dielectrics.

[Pedro147]

Who would have thought that such a seemingly simple situation could have so many various outcomes 

Seeing as this is being used as a decoupling capacitor maybe it is best if I just spend the extra money on the exorbitant freight charges and get the exact specification capacitor that is in the circuit design.

Thanks again to all of your for the comprehensive explanation of a seemingly simple situation. Never a dull moment   

[Ian.M]

Where in the circuit are these decoupling capacitors?  I strongly suspect that *ANY* SMD ceramic caps from 2n2 to 4n7 with an adequate voltage rating and that fit the footprint could be substituted with absolutely no ill effects.

[Pedro147]

Hi Ian.M

The capacitor in question is C8 attached to the MPU6050 gyro IC
Schematic attached

Thanks

[Ian.M]

In that case, you go to the MPU6050 Gyro product page, and see if there are any special requirements called out for the caps in the datasheet.  It turns out its *NOT* a decoupling cap and the line in the recommended BOM is: "Charge Pump Capacitor (Pin 20) C3 Ceramic, X7R, 2.2nF ±10%, 50V" which exactly matches your Murata cap from RS. 

It looks like your board designers were cutting corners slightly by specifying a cheaper cap!

[T3sl4co1l]

Yes, we've drifted fully away from the purpose, and are now arguing semantics.

From what I've seen with MLCCs, the capacitance starts dropping immediately with increasing bias, not when you get near the rated voltage.  In fact the voltage rating of the cap has little or nothing to do with the drop in capacitance as bias voltage increases.  It's almost entirely a function of the dielectric and package size (0402 vs 0603 vs 0805, etc.).

You're misinterpreting my statement -- I meant as a percentage of rating.  A 50V cap will roll off (taking a fixed capacitance reduction, say 30%, as the "rolloff point") at a higher [absolute] voltage than a 16V cap will.

As for actual subtleties, it's even worse than that: most ceramics show a slight hump near zero volts!  Slight as in ~5% of rated capacitance, and in the range of 1-5% of rated voltage.

This can actually be harnessed, I think by biasing the dielectric during manufacture:
http://www.digikey.com/product-detail/en/B58031I5105M002/495-6707-ND/5039884
Notice the energy capacity of these is much lower than a proper constant-C type, but when all that's needed is a low impedance at the rated voltage, they're magnificent (unfortunately... magnificently expensive, too!).

When it comes to physics of nonlinear materials, it's worth not getting your hopes up, because they can always do new and weird things...

Tim

[Pedro147]

in the recommended BOM is: "Charge Pump Capacitor (Pin 20) C3 Ceramic, X7R, 2.2nF ±10%, 50V" which exactly matches your Murata cap from RS. 

It looks like your board designers were cutting corners slightly by specifying a cheaper cap!

 

Thanks so much Ian.M.

That is what I like, some good solid detective work rather than getting bogged down in theoretical considerations however important and pertinent they may be

I am glad that I can now use those 50v rated caps seeing as that is what are specified for the MPU6050. Also thanks for educating me on how to investigate what the requirements are for a specific component by 'reading the data sheet'

Thanks again great work 

[LukeW]

If you're working with a small capacitance - say 2.2nF, or 100nF - then it's easy to get 50V capacitors cheaply, even in an 0603 package, and you can just specify 50V for all those capacitors without really thinking about it.

But if you wanted to go to a significantly larger capacitance - say 2.2uF, or 10uF - then you're much more limited in the tradeoffs you have to make. The capacitors are often much more expensive, the dielectrics are often a less well behaved class, and the voltage ratings are often significantly lower, say 10V or 16V, especially if you want to stay in the 0603 package. So then you have to start thinking about harder choices - being less generous with the voltage rating, how much margin you need, how the dielectric behaves under voltage and how much capacitance you really need, how much you're willing to pay for the capacitors, and whether you're able to move to a larger package footprint.

[Pedro147]

Thanks Luke I only just saw your reply now and thanks for your contribution to my question it is appreciated