I'm doing a 12V->48V converter at 330W. Even with a four phase converter, 
the ripple current (according to LTspice) is about 7A rms. I need about 
1,000uF (63V will be okay) and the max ambient will be 60C.

The caps I've looked at from Panasonic and Nichicon can handle about 2A max 
of ripple current (at 100KHz).

What worries me is if I parallel a bunch of them, and especially if I allow 
multiple suppliers for the caps, I'll never be sure how the ripple current 
is shared between them. So, do I put ten in parallel to be sure 
(rhetorical)?

What would be nice is to find ONE cap that can do all the ripple current. 
Anyone know of such a beast?

Thanks.

Bob
-- 
== All google group posts are automatically deleted due to spam == 

On Thu, 24 Sep 2009 21:01:44 -0700, "BobW"
<n...@roadrunner.com> wrote:

>
>I'm doing a 12V->48V converter at 330W. Even with a four phase converter, 
>the ripple current (according to LTspice) is about 7A rms. I need about 
>1,000uF (63V will be okay) and the max ambient will be 60C.
>
>The caps I've looked at from Panasonic and Nichicon can handle about 2A max 
>of ripple current (at 100KHz).

330 W @ 48 V is about 7 A and the capacitor is only required to supply
this current when none of the four phases are supplying charge. At
1000 uF, the voltage drop is just 7 mV/us. In a multiphase design,
these times might last perhaps 10 us at a time, so the peak-to-peak
ripple voltage would be 70 mV, which is quite low for a 48 V supply.

Do you really need that large capacitor ?

A small parallel non-electrolytic capacitor could handle the steepest
transients. Some small series inductance (sufficient at such high
frequencies) would also reduce the electrolytic capacitor ripple
current. 
 
Paul

"Paul Keinanen" <k...@sci.fi> wrote in message 
news:1...@4ax.com...
> On Thu, 24 Sep 2009 21:01:44 -0700, "BobW"
> <n...@roadrunner.com> wrote:
>
>>
>>I'm doing a 12V->48V converter at 330W. Even with a four phase converter,
>>the ripple current (according to LTspice) is about 7A rms. I need about
>>1,000uF (63V will be okay) and the max ambient will be 60C.
>>
>>The caps I've looked at from Panasonic and Nichicon can handle about 2A 
>>max
>>of ripple current (at 100KHz).
>
> 330 W @ 48 V is about 7 A and the capacitor is only required to supply
> this current when none of the four phases are supplying charge. At
> 1000 uF, the voltage drop is just 7 mV/us. In a multiphase design,
> these times might last perhaps 10 us at a time, so the peak-to-peak
> ripple voltage would be 70 mV, which is quite low for a 48 V supply.
>
> Do you really need that large capacitor ?
>
> A small parallel non-electrolytic capacitor could handle the steepest
> transients. Some small series inductance (sufficient at such high
> frequencies) would also reduce the electrolytic capacitor ripple
> current.
>
> Paul
>

Paul,

No, I don't really need that big of a cap. I think that I can live with more 
voltage ripple. However, even with smaller total capacitance, the total 
ripple current is roughly the same.

This supply will be driving a 190mm radial fan (630cfm in open air). I'm not 
sure how a supply with a lot of ripple voltage will affect its performance. 
It'll be a few weeks before I'm able to get the whole thing breadboarded. 
Then, I'll know for sure.

I can use a smaller cap with lower (guaranteed) esr to take the larger 
component of ripple current. 4.7uF / 100V ceramics are tough to find, but 
they are available (AVX and TDK have them in 2220 packages). Film caps in 
this size are an option, too. Then, as you suggest, it would be wise (and 
required) to put some type of current-limiting impedance in series with the 
electrolytic caps to insure that their ripple current is minimized. The 
downside of this approach is its complexity, cost, and real estate usage.

If the esr of the electrolytic caps were specified as a min and max, then I 
would feel comfortable paralleling a given number of them so as to not 
violate any one of their max ripple current specs. This, however, is not the 
case. So, I'd really like to find one cap that can do it all.

There used to be caps that were specified as "extended foil". This technique 
assured that they could handle really big ripple currents. I don't see this 
verbage used at all, anymore. Maybe they were too expensive to build.

Bob
-- 
== All google group posts are automatically deleted due to spam == 

On Thu, 24 Sep 2009 23:23:21 -0700, "BobW"
<n...@roadrunner.com> wrote:

>This supply will be driving a 190mm radial fan (630cfm in open air). I'm not 
>sure how a supply with a lot of ripple voltage will affect its performance. 
>It'll be a few weeks before I'm able to get the whole thing breadboarded. 
>Then, I'll know for sure.


If the switching frequency is in the order of tens of kHz, this should
not have much effect, unless the commutator is causing some harm,
causing beat frequencies etc.

However, any EMC radiation issues might be relevant, so you may have
to use a shielded cable to the motor. 

If you have to use some EMC filtering, it will usually require some
series inductance, so placing some of it in front of the main
capacitor may also help the ripple current issue.

Paul

On Sep 24, 9:01=A0pm, "BobW" <nimby_GIMME_SOME_S...@roadrunner.com>
wrote:
> I'm doing a 12V->48V converter at 330W. Even with a four phase converter,
> the ripple current (according to LTspice) is about 7A rms. I need about
> 1,000uF (63V will be okay) and the max ambient will be 60C.
>
> The caps I've looked at from Panasonic and Nichicon can handle about 2A m=
ax
> of ripple current (at 100KHz).
>
> What worries me is if I parallel a bunch of them, and especially if I all=
ow
> multiple suppliers for the caps, I'll never be sure how the ripple curren=
t
> is shared between them. So, do I put ten in parallel to be sure
> (rhetorical)?
>
> What would be nice is to find ONE cap that can do all the ripple current.
> Anyone know of such a beast?
>

I never design in "panic sonic" capacitors any more.  They are forever
changing their mind about which ones that want to make.

Is there some reason you can't just put in several smaller capacitors?