best capacitors for high ripple and large capacitance

A whole bunch of NPO ceramics in parallel is the "best" if you aren't worried about cost or space.

15A RMS at 200KHz is quite a lot of ripple to put through a single capacitor.

I've used CDE's 381[huh]R at about 0.5A per every 100uF with some success. They are worth a look.

[huh] My brain froze there is a character in there

Those figures appear incorrect. 15A rms ripple figure is a very large number. Are you sure that the number isn't 15mA?

Actually, to get high ripple, one uses NO caps. To ABATE high ripple, one uses a cap bank, and /or an RC filter, as long as it still provides the current requisite of the load circuit.

Where do you derive that "ripple figure" from? It is not on the page you gave a link to.

from digikey 381LX I think, wow you really drive those caps hard,

datasheet specs for part#: 381LX332M063H052

3300uF, 63V, 120Hz ripple 2.5Arms, 20kHz ripple 2.9Arms, both at 105Celcius.

Do you think that cap could be operated with 15Arms ripple at 200kHz, with ambient temperature below 50Celcius?

I guess it would be better to get two smaller ones so they have more surface area for cooling, maybe two 1800uF ones:

part#: 381LX182M063H022

1800uF, 63V, 120Hz ripple 1.7Arms, 20kHz ripple 1.96Arms, both at 105Celcius.

cheers, Jamie

The way I'd normally think to handle something like this is to NOT use an electrolytic directly, since they invariably have high enough ESR that 15A RMS will cause significant heating. Rather I'd handle the ripple with some very low ESR capacitance: ceramic or more likely polypropylene in this particular case. WIMA make some very good polyprops for applications like this. Use enough capacitance to get the ripple voltage down to a reasonable value, and then you don't need much inductance between that and bulk electrolytic capacitance to keep the ripple current in the electrolytic at a more reasonable level. Check out the WIMA MKP10 series. Panasonic ECWF and Vishay MKP379 series should also work well. Five 4.7uF 100V caps in parallel should very easily handle the current and get the ripple down to a couple volts p-p or less (depending on waveform). Even two or three would handle the current easily, but might not get the ripple voltage low enough--your design, your tradeoffs. Then even a fraction of a microhenry off to bulk capacitance limits the ripple in the bulk to a low value.

Cheers, Tom

381LR221M400A012 ESR=0.271@20KHz 2.03A@20KHz

It is a different part targeted as a "ultra high ripple capacitor"

No, they will heat themselves up too much. I don't like heat sinks on capacitors.