Ripple current in electrolytic capacitor

[Faringdon]

Hi,
Customer has sent us an 8W Flyback (24vout, 132khz, 100-265vac). They want the flyback increased to 17W. We are limited to keep using the EFD15/8/5 Former and core set. The input electrolytics cannot change. They are two 10uF, 400V Electro's , each with ripple current rating of 90mA at 100Hz.

Capacitor 400AX10MEF:
https://www.rubycon.co.jp/wp-content/uploads/catalog-aluminum/AX.pdf

The new ripple current in each of them will be Approx 462mA total.....
(343mA at 100Hz; 305mA at switching frequency)

The device is only used for 20 minutes each day. But it will be at full power for all that time.
The caps will have reasonable blown air over them whenever the product is ON.
Air temperature around the caps will be around 40degC max, and as discussed, its blown air.
The caps are "2000hrs at 105degC"

I presume we will be ok to violate the ripple current rating?

What are the rules for "allowable violation" of electrolyic capacitor ripple current?

Presumbaly you just revert to the "2000hrs at 105degc" and Arrhenius equation?

[Berni]

If operated at reasonable temperatures and low duty cycle you should be fine.

Might certainly affect the life of the capacitor, they would heat up to operating temp in 20 minutes, but since you accumulate so little runtime per year it probably won't matter. More of a thing is how much ESR you can handle, the tiny capacitors with low ripple ratings can have a pretty lousy ESR even when new and this will get worse as the capacitor ages.

But one test is worth more than a 1000 expert opinions. So id say build 2 or 3 prototypes of the PSU, put them in an oven set to 100°C or however much you think the PSU will still survive, then leave them running continuously for a week or two. If it still works after that, then they will survive the field use too.

[Gyro]

Running small, low thermal mass capacitors at over 5 times their ripple current rating. Why are you even bothering to ask the question? The question you should be asking is whether they will vent in less than 20 minutes or not.

...
I presume we will be ok to violate the ripple current rating?

Why do you presume that?

What are the rules for "allowable violation" of electrolyic capacitor ripple current?

Datasheets don't have an "allowable violations" section, so there are none.

[Siwastaja]

Remember that losses go I^2*R. Exceeding current rating by 5x means 25x the power dissipation. It is just way too much even when taking into account that:
* ripple current is rated at 105degC ambient,
* ripple current is rated for 2000hrs

If the thing is well cooled and indeed driven only for 20 mins per 24 hrs, I would feel somewhat confident running them at +50, maybe +100% current and taking the risk. But what you suggest sounds too excessive.

[T3sl4co1l]

Why do you presume that?

What are the rules for "allowable violation" of electrolyic capacitor ripple current?

Datasheets don't have an "allowable violations" section, so there are none.

Actually, ripple ratings are at rated temperature inside the winding.  It's not exactly easy to measure the core temperature inside a commercial part (thermocouple'd specials are usually available from the manufacturer!), but a small part like this we can also assume temp rise over case will be small.

Given this detail, the proposed extrapolation is justified.

The proposed amount, however, will most likely be found impossible under the proposed extrapolation.

Tim

[mzzj]

I have pushed electrolytics to pretty wild overcurrents in some projects and cooked them in a matter of seconds. Like 15x rated ripple current. 

Cooling and ambient temperature makes huge difference.
Good guess for ESR could be 2 ohms, making about 0.5W losses in the capacitor. Not that much if you have forced air cooling.
For something stup...extreme like this only way to know is test it. Not that hard to do but of course you can also post yet another topic on eevblog. 

141 CFM 80mm fan seems suitable given the OP's history with sensible projects.
https://fi.farnell.com/sanyo-denki/9hvb0812p1g001/axial-fan-80mm-x-38mm-57-6w-12vdc/dp/3754937

[Faringdon]

The output of this flyback has two 100uf, 35v alu electrolytic caps in parallel, they are wuerth 865080545012...

https://www.mouser.co.uk/datasheet/2/445/865080545012-3100758.pdf

...We wish to violate the ripple current rating of them (300mA) but cant find the lifetime equation for this cap on their website.
Do you know where  we might find the info.?

As discussed, the product will only be used for 20 mins each day at full power, and there is blown air over it.
We dont really have room to fit more capacitance.

[mariush]

Just tell the to suck it up and pay a few cents more for a polymer capacitor ..  ex https://eu.mouser.com/ProductDetail/KEMET/A768KE107M1VLAS036?qs=sGAEpiMZZMs9%2FSpGH%2Fyc7Oi7G6YMoDvikdXleNAlI9shoZBAtX7SDA%3D%3D

bit bigger footprint, but not by much. may be able to bring that down to 82uF each or to a single 150uF capacitor instead of two, and save pennies there.
 

[wrangler]

If you drop in some series resistors to the bulk caps, you'll limit input peak current, and increase the conduction time of the input rectification. This has the added benefit of limiting in-rush since you imply that the device will be energized from a cold and dark state once per day. If you're adding in series resistors, you may also want to explore NTC thermistors on the input. Of course, you've got to keep that added heat away from the electrolytics, or else you're probably worse off, and accept the efficiency hit.

As far as the transformer primary current, you may be able to make some subtle changes to the magnetics to drop peak current, and explore going into CCM, but that probably won't be as efficacious as improving the bulk storage.

Can you find similar physical size electrolytics with better ratings?

[dmills]

If you are running parts off label like this, then some combination of the specific vendors FAE and testing is the only way to play. If you are doing sufficient volume you can likely get the vendor to do an investigation for you and provide a special datasheet for your weird use case, but you need to be ordering a million units on two year lead time for this to really work.

I would note that you need to be very leery of extrapolating from results with one vendor to a different one when running off label. While all make fine capacitors, I would insist on redoing the lifetime tests switching from CDW to Panasonic to Nichicon, and electrolytic caps are something that purchasing seem to often think are far more interchangeable then they are (Unusually as long as both parts are operated within datasheet limits they are fine, off label is where most of the danger lies).