capacitor ripple current

[V2]

Hey guys. I'm hoping someone can tell me what an ecap's ripple current spec means and whether that value should be high or low. I assume it is the amount of internal AC current the cap can handle, that it varies inversely with the cap's ESR (and other things like temperature), and that it should be sufficiently high (although I don't know how high; or even whether I am on the right track).

[R.G.]

Ripple current is the spec that the cap maker puts on the AC currents that flow in the cap. This current squared times the ESR is the amount of heat generated inside the cap by the current flowing. Caps can only get rid of so much heat, and their internal temperature has to rise to get rid of it. Temperature rise speeds the degradation of the innards, especially of electrolytics.

So you want **high** ripple current specs and low ESR. Generally, the lower the ESR, the less heat generated, so the higher the ripple current can be. You're on the right track.

Now for the bad news. It's very, very difficult to calculate how high a ripple current you need to specify in a cap, and even harder to calculate a priori what ripple current will be caused by a given capacitance, transformer winding resistance, diode equivalent forward resistance, and ESR. Those are generally in the range of imponderables that are hard to calculate.

The pros almost all put in specs by guesses based on experience or ignorance, and if they look, they build one, test it for RMS current in the caps, then calculate to see if the caps they specified will live long and prosper or vent in a year or so.

It's a good thing to worry about, but really hard to get a handle on, even for me, and I used to earn my living doing things like this.

[stretch2011]

a priori...

someone took a philosophy class XD

Great answer R.G.

[trobbins]

PSUD2 can give a good estimate of your average cap ripple current at the operating frequency of the capacitor ripple. Sometimes that frequency can modify the spec level - which is different for bridg/full rectifier compared to common doubler.

Be mindful that it depends on ambient temp, and if you operate close to the temp rating (eg. 85C) because of valves close by, and close to the ripple current level, then your cap won't last long.

[Stevem]

The unknown fact is why all filter caps should be rated for atleast 50 volts more that what is called for, no R G?

[R.G.]

The unknown fact is why all filter caps should be rated for atleast 50 volts more that what is called for, no R G?

Good question. The answer is that (a) the voltage coming out of the wall varies and that (b) capacitor voltage ratings are sloppy things.

The wall voltage is supposed to be a nominal voltage ±10%. What it really does is more complicated, and sometimes it runs high. Sometimes quite a bit high, like maybe 15%. The power transformers you use have to be designed to work OK at -10% and also at about +15% even if you have the luxury to specify one that's the ideal DC output voltage at nominal line voltage. On a 450V supply, +15% is 517.5V. With some luck the "surge" rating on the electro caps will cover that, but you can see that adding 50V is not a bad idea.

And then there's the actual voltage rating. Electro caps do not have an all-or-nothing voltage rating, bursting like a bubble when it's exceeded. Instead, they start leaking a little, then more, then massively before they die. At each micro leakage point, there is an electrochemical fight between the heat generated at that pinpoint space tearing down the insulating barrier, and the electrochemistry building up new insulator.

Letting a low current flow through the aluminum plates/foils is how the insulator is grown in the first place. The manufacturers set up a process where a current flows at just the right rate to maximize insulating layer growth and minimize heat-degeneration. This same fight goes on throughout the capacitor's life, with leakage both tearing down and building up the insulator when the cap's powered. When the cap is not powered, there is another purely chemical degradation that eats up the insulator in a few decades so it leaks and dies when you suddenly power it. This whole business of growing insulation is called "forming" and that's what you hear about when you hear of "re-forming" caps.

The voltage rating of a cap is where the manufacturer thinks high-90s of its caps from that process will last a long time, usually about 10 years of power-on-hours, before they get too leaky and start dying.

Specifying an additional X of voltage rating lets the cap withstand high line conditions more gracefully and survive more of them, and also gives some additional buffer against the day when the chemical degradation makes it die.

I say "X", not 50V because the same thing happens at lower voltages, where the additional amount you spec may be 5, 10, or 25V.

[David Root]

Somewhere a while ago I read a lengthy mathematical discourse on this subject from the application side, and the writer's conclusion was that a safe e-cap DC voltage rating is sq.rt of 2 x the quiescent DC voltage at the cap.
I do not recall the derivation of this number, but as a constant it pops up everywhere in what we do, so maybe has some credibility?

I have tried to follow this advice and have had no failures so far.

[V2]

Thanks R.G.! Thanks guys!

[gingertube]

PSUD2 does give the ripple current, it appears in the capacitor RMS current column. Ignore the 1st second where inrush currents are large and look at the value for the next second or so and choose caps to suit that ripple current. Be conservative. Due to the current squared x ESR heating power internal to the cap you get a power thing happening with respect to cap lifetime too. A cap rated at X2 the ripple current will last 4 times as long (actually due to some other things it will last a bit longer than 4 times).

EPCOS and other cap manufacturers have Application Notes on their websites which show you how to calculate the ripple current and from that the expected lifetime with lots of tedious algebra.

The root 2 quiescent voltage thing is to do with the peak voltage the cap will see before tubes start to conduct. Once the tubes start to conduct current the power supply voltage will drop from the peak value towards an average value.

Cheers,
Ian