Trouble finding high ripple current cap that isn't physically very large.

[Dmeads]

Hi all,

I am designing a boost converter for class.

I have attatched my ltspice file for anyone to look at. I have all my components picked out, except the output caps. the value needs to be about 10uF.

The trouble is, the boost converter outputs 10.6A @ 56.5V, so the ripple current is very high (depends on cap value and how many are in paralell, but its around 2Arms) I have looked on mouser and digikey for high ripple current caps but they are really large (like 5000uF) and very expensive. I can put a bunch of small ones in paralell to reduce the ripple current per cap, but I cant have like 7 or 8 electrolytic caps on my board to due space issues (the inductor is already pretty big and the MOSFETS need big heatsinks).

Should I try a different type of cap?
can film caps withstand high ripple current?
should i put some caps in series and others in paralell?

thanks,

Dom

[Weston]

You can find single electrolytic capacitors with a current rating of 2Arms or above. You want to look at the "ripple current at high frequency" rating. Aluminum polymer capacitors have a higher current rating than normal electrolytic capacitors. Here is a random example off digikey that works: https://www.digikey.com/product-detail/en/panasonic-electronic-components/80SXE56M/P124983-ND/10315800

Unless you have some weird control bandwidth requirement there is no harm in a larger output capacitor, it will just further attenuate the ripple. There is no need to put capacitors in series.

The current ratings for capacitors are based on some rather high ambient temperature and to reach some specified lifetime, something like a few thousand hours. If this is just for a class project you should be able to get away with exceeding the current rating by 1.5x or something like that for the relatively brief time you have this on the bench.

Both ceramic capacitors and film capacitors have higher ripple current per unit volume than electrolytic capacitors. Film capacitors are going to be optimized for higher voltages than you are running at and it is a bit difficult to get ceramic capacitors with high capacitance at these voltages. Ceramic capacitors also have lower capacitance with a DC voltage applied, which means you need a higher capacitance value.

Commercial designs often use interleaving, where they run multiple converters together with phasing such that some of the ripple current cancels.
 

[T3sl4co1l]

Who says you need ~10uF?

Also, you'll have a damn hard time using those transistors, at that frequency, at that current, with real parasitics (try 10nH in series with the drain and diode).

Is this simulation just to show switching dynamics, i.e., is there supposed to be a controller elsewhere in the real circuit?

Tim

[David Hess]

Should I try a different type of cap?
can film caps withstand high ripple current?
should i put some caps in series and others in paralell?

Film and ceramic capacitors have much higher ripple current ratings because of their much lower ESR.  If you use ceramic capacitors, then watch out for their voltage coefficient of capacitance.

Their low ESR however can cause problems with frequency compensation of the control loop if not accounted for.

[Siwastaja]

Look at the aluminium polymer electrolytic section. They are more expensive than the classical electrolytics, but you'll easily meet the spec. You'll likely still need to get "excess" capacitance to meet the ripple current requirement, but usually nothing wrong with that.

OTOH, 10uF at 56V is doable with ceramics, then the current rating won't be an issue. This would be around 40uF nameplate capacitance with 63V  rated X7R ceramic caps in 1206 or 1210 case, so you'll need to parallel many, for example 15-20pcs of 2.2uF nominal will likely do. Be very careful avoiding board flex and excessive soldering heat, though. And while using assumptions (like I showed) is OK for first order approximation, do look at the DC bias curves for the particular capacitor you are going to buy. Some are down to 15% at rated voltage, some are still at 30, even 40%, even if both have the same X7R dielectric rating.

[David Hess]

Look at the aluminium polymer electrolytic section. They are more expensive than the classical electrolytics, but you'll easily meet the spec. You'll likely still need to get "excess" capacitance to meet the ripple current requirement, but usually nothing wrong with that.

It is common with electrolytic capacitors in switching power supplies for the capacitor size to be completely determined by the ripple current requirement.

Polymer aluminum and tantalum electrolytics generally have pretty low voltage ratings but in this application, I would at least consider using 2 or 3 in series with voltage balancing resistors.  Maybe someone can say why this is a bad idea.

OTOH, 10uF at 56V is doable with ceramics, then the current rating won't be an issue. This would be around 40uF nameplate capacitance with 63V  rated X7R ceramic caps in 1206 or 1210 case, so you'll need to parallel many, for example 15-20pcs of 2.2uF nominal will likely do. Be very careful avoiding board flex and excessive soldering heat, though. And while using assumptions (like I showed) is OK for first order approximation, do look at the DC bias curves for the particular capacitor you are going to buy. Some are down to 15% at rated voltage, some are still at 30, even 40%, even if both have the same X7R dielectric rating.

They make special large ceramic capacitors for input and output applications which include flexible terminations but they are expensive.  I commonly see them used in brick type converter modules and applications where long operating life or high temperature operation is required.

[duak]

Panasonic has an OSCON polymer electrolytic series rated up to 100V: https://industrial.panasonic.com/tw/products/capacitors/polymer-capacitors/os-con/sxv?reset=1

It seems to me that paralleling capacitors is a better way to meet the ripple current spec because it increases the surface area and thus affords greater heat dissipation or lower temperatures.

Has anyone seen heat sinks or thermal management techniques used on capacitors anywhere?  I've seen heat sinks soldered to semiconductor or resistor leads to reduce spot temperatures.

[T3sl4co1l]

Only on the biggest ones, where you'll find metal-can, stud-mounted types (film and electrolytic), and the really big ones, conduction and water cooled (film and ceramic, induction heating, RF).

Seems there's a lot of room at the small end, without needing such measures.  Most importantly, avoiding situations where you generate extreme amounts of ripple in the first place -- e.g., don't use a single phase buck converter in DCM, use phase interleave in CCM.

Tim

[David Hess]

It seems to me that paralleling capacitors is a better way to meet the ripple current spec because it increases the surface area and thus affords greater heat dissipation or lower temperatures.

Paralleling capacitors avoids the voltage balancing issue but could result in the same ripple current rating improvement.  In series the ESRs add but the capacitors have to be proportionally larger in capacitance lowering their individual ESR.

Has anyone seen heat sinks or thermal management techniques used on capacitors anywhere?  I've seen heat sinks soldered to semiconductor or resistor leads to reduce spot temperatures.

I have not seen that but what I have seen is poor layouts where current sharing between parallel capacitors was very poor.