High Ripple Current Capacitors (and no elko)

[Magnethicc]

Hi everyone,
I am designing a high power high current DC-DC with low voltage at both sides. Simulation results shows high ripple current (and it is interleaved).
The costumer does not want any electrolytic capacitors in the system and the his input is ripple current sensitive thus I need a high input capacitance (to de-couple from his input) with high ripple current handling capabilities.

I first though of Film Capacitors but I can't get high capacitances (north of 1000uF) without compromising on a ton of space + Film capacitors don't have any information on ripple current at 60kHz.
My next though was multiple MLCC in parallel.
Does anyone have any recommendation on how to achieve high capacitance and high ripple current capabilities without using electrolytic capacitors (or oil capacitors)?

Thank you.

 

[Easy peasy]

lots of parallel X7R caps on both sides of a 1mm 3 Oz pcb, lots of these as plug ins / solder ins - to the main pcb

 

[cupoftea]

I once saw an automotive board where they had used loads of MLCC instead of El caps. They had to use the ones that are "kind of on stilts" as otherwise they got too much problems with cracking if ever the PCB was roughly handled. (which it often was when eg mech engs took the board to do measurements on, etc etc) Ive forgotten what you call them now, its like a ceramic cap but its on like a platform, so it is spared flexion stress.
Other way is to use radial ceramics of course. Or ceramics on "plug-in" platform PCBs, which may (or may not) have been what Easy Peasy was referring to?

Shame that you cant use supercaps. I dont think all are as bad as el caps.
With multiple interleaved phases you can very heavily reduce ripple as you know.

One alternative is to just have supercaps, but make the bank easily remove-and-replace-able, so when it fails, you can take it out...or just replace it anyway after a year or so.

Woudl have thought modern tantalum capacitors could be OK too, as they are not as bad as el caps, and the newer ones dont have surge current problems like the older types.

Wet electro , AYK, is the king of handling high one-off surge currents.

 

[Magnethicc]

And what about Organic Polymer Capacitors?

 

[cupoftea]

They have just the same lifetime problems as el caps....you can see it here...eg 2000hrs at 105degC.

https://www.mouser.co.uk/datasheet/2/445/875105344010-1727105.pdf

Maybe they are a little less likely to dry out than wet el caps.

 

[Magnethicc]

In that case, if it's not to much to ask an answer to:
How would you recommend doing a MLCC vertical board with a very low chance of the MLCC cracking?

 

[cupoftea]

You mean a vertical daughter board?......keep the board small enough and it should be ok.....stop anything causing flexion to it.....2mm thick to reduce flexion.
What did you mean by vertical....you mean the main board is vertical?
Beware depanelisation cracking of SMD ceramics though.

You do know MLCC looses capacitiance with voltage?....eg a 100n,50V ceramic is only about 10nF when its got 50v on it.

 

[FvM]

What's "low voltage" for you? 3V, 12V, 60V?

 

[Magnethicc]

Low voltage input is ~30V. Output is 48V @5.5kW

 

[Easy peasy]

Oooh, 192 amps input at 30VDC ....

 

[Calm down]

If the voltage is not high, try to use tantalum capacitor or solid polymer capacitor.

 

[mtwieg]

You haven't given hard specs on ripple current, but metallized film caps should be a good choice. Generally the ripple current handling is does not vary strongly with frequency. TDK has a good online tool which you should look at, it lets you customize a ripple current spectrum, and spits out estimated temperature rise and service life.

Only reason I would lean towards MLCCs is if space constraints make film caps a no-go. But you should also consider using a LC input filter to reduce input ripple, that could be much space efficient than simply piling on more capacitance.

 

[KlausST]

Hi,

Oooh, 192 amps input at 30VDC ....

Just to clarify (EasyPeasy surely knows) that 192A is the DC/average load current.
These 192A is not the ripple current.

Ripple current usually is measred in "A RMS" and additionally needs the frequency information to select a suitable capacitor.
The capacitor ripple current depends on how the 30V are generated ... and on load current ripple.

Klaus

 

[Easy peasy]

usually a current fed full bridge is used for low V inputs just because an L is good at providing low ripple ( input or output ) such that less uF are then needed to provide lower volt ripple ....