Maximum current for capacitors?

[uoficowboy]

Hi - I am working on a very high current system and am trying to evaluate and identify capacitors for it. Problem is - it seems that most ceramic capacitors do not have a current rating. For example, I have a couple sample kits for the TDK C series ( **broken link removed** ). I don't see any mention of maximum current in this system. How can this be evaluated?

If you're interested - I'm looking at putting maybe a +/- 20A sinusoid into a ~1-10nf capacitor at about 5MHz. I highly suspect I need C0G or similar for this, but I'm not sure if that is even enough.

Any suggestions?

Thanks!!

 

[KerimF]

First you may like calculating the capacitor voltage from the basic formula:

Vp = Ip / wC

where:
Ip = 20 A
w = 2 * PI * 5,000,000
C = 5 nF for example

Vp = 127 V

I think this exceeds the maximum rated voltage of this series which is 50V.

 

[uoficowboy]

First you may like calculating the capacitor voltage from the basic formula:

Vp = Ip / wC

where:
Ip = 20 A
w = 2 * PI * 5,000,000
C = 5 nF for example

Vp = 127 V

I think this exceeds the maximum rated voltage of this series which is 50V.

Oddly enough, I have some 250V rated C series TDK capacitors. I noticed that the datasheet doesn't call those out - but they're in the sample kit.

So - assuming I'm not blowing through the voltage rating - how do I tell if I'm OK?

 

[KerimF]

Do you mean 250V for 5 nF too?

 

[uoficowboy]

Do you mean 250V for 5 nF too?

The largest 250V C0G in this kit is a C4532C0G2E473J. That is a 250V 47nf C0G.

 

[RCinFLA]

As mentioned you will likely exceed the voltage rating first.

But if not then get the ESR number and evalute the heat generated by current profile. You have to have some intuitive judgement based the physical size and total power dissipation possible based on its ability to expell the heat. If the caps have a maximum operating temp spec that would be a good reference point for maximum power dissipation levels.

Many switching power supply using electrolytic caps ,which have a high ESR, suffer this.

 

[KerimF]

As you know, for a fixed current as 20 A, the voltage on the capacitor is inversely proportional with its capacitance.
So for 10 nF, Vp is about 64V

In theory If Vp doesn't exceed the maximum rated voltage (Vm), the capacitor can handle the current.
I personally prefer that Vp doesn't exceed Vm/4 (if real necessary Vm/2).

For instance, don't forget that we can use two capacitors in series to lower Vp. For example, two 10nF in series gives 5 nF. In this case Vp of each C is about Vm/4 only.


Note: As mentioned by RCinFLA, at high frequencies the capacitor dielectric may get warm or even hot at relatively high Vp and this gives another limitation.

 

[uoficowboy]

As mentioned you will likely exceed the voltage rating first.

But if not then get the ESR number and evalute the heat generated by current profile. You have to have some intuitive judgement based the physical size and total power dissipation possible based on its ability to expell the heat. If the caps have a maximum operating temp spec that would be a good reference point for maximum power dissipation levels.

Many switching power supply using electrolytic caps ,which have a high ESR, suffer this.

I am not worried about the voltage limit - I can choose parts that avoid that. Heat is my concern. I don't see it speced in any C0G datasheets. So I'm not sure how to handle that...

 

[BradtheRad]

I'm looking at putting maybe a +/- 20A sinusoid into a ~1-10nf capacitor at about 5MHz.

Even if ESR does not cause overheating...

Picture running 20A through the lead wires (steel, copper, whichever). Are they substantial enough to handle it?

Through the internal plates?

Through the internal connection (tack weld?)

Should you use a bank of capacitors?

 

[uoficowboy]

Even if ESR does not cause overheating...

Picture running 20A through the lead wires (steel, copper, whichever). Are they substantial enough to handle it?

Through the internal plates?

Through the internal connection (tack weld?)

Should you use a bank of capacitors?

Well that's sort of my question... Sure you can look at a component and say "that looks too small for 20A" - but that isn't a spec. That's a gut feeling. I want a spec. Or a formula. Or something!

I've asked a manufacturer the same question - I hope they get back to me with something useful!

 

[FvM]

Trivial solution, google for RF power capacitors. You'll find devices with current specification, e.g. **broken link removed**

 

[BradtheRad]

As to what size lead wires can carry 20A...

Here are links to charts of wire gauges showing safe amp-carrying capacity:

http://amasci.com/tesla/wire1.html



http://www.powerstream.com/Wire_Size.htm

Here's a link to another wire gauge table. Its ampacity values are much different than the other tables.

**broken link removed**

Possible reason for the disparity: One source may base ampacity on a higher temperature which makes the wire get too hot to hold, while the other source may use a lower temperature where copper's electrical characteristics start to change even though it is only warm.

 

[FvM]

Here are links to charts of wire gauges showing safe amp-carrying capacity.

Most of the tables are about to meaningless for 5 MHz operation, some mention at least skin effect. In a short, litz wire is the only chance to carry the intended current with small conductors, otherwise copper tubes or ribbons of sufficient size.