Electrolytic capacitor leakage current – what is it and how to measure it

[momaka]

Okay, I know this is likely covered much better by many other places online. However, I have been playing with a home-built electrolytic capacitor reformer recently, so I thought I'd include a brief mention of what leakage current is, based on the research I have done so far. In particular, what I am trying to do here is focus more into the practical side of things – i.e. how to measure leakage current of an electrolytic capacitor in order to determine if the cap is (going) bad or not.

Basically, all electrolytic caps self-discharge, much like a battery does when you leave it in storage for a long time. This discharge rate is known as the leakage current. High leakage currents are undesirable because they produce heat inside the capacitor. Essentially, this wastes energy and makes a capacitor behave… well, less like an ideal capacitor and more like a resistor (especially if the leakage current is relatively high). In such cases, the cap is said to be “leaky” (which, by the way, has nothing to do with a bad capacitor that is physically leaking electrolyte.)

Therefore, cap datasheets specify a maximum leakage current for each cap. It is usually given as:
I_leakage ≤ 0.03 • C • V
or
I_leakage ≤ 0.01 • C • V
and
I_leakage ≥ 3 μA

...^ where I_leakage is the cap's leakage current measured in micro amps (μA), C is the rated capacitance of the cap in microfarads (μF), and V is the rated working voltage of the cap in Volts (V).

Let's say you have a 16 V, 1500 uF cap (something typical from a PC PSU or motherboard). First determine from the datasheet how the leakage current is calculated. For this example, I will assume it is the second formula from above. Then the maximum allowable leakage current for the cap is:
I_leakage ≤ 0.01 • C • V = 0.01 • 1500 • 16 = 240 μA = 0. 240 mA

The next step would be to find out the cap's actual leakage current and compare with the above value. There are two ways to do this:
(1) Put an ammeter in series with the cap and connect to a voltage source that has a voltage output as close as possible to the capacitor's maximum rated working voltage (but not higher). Let the capacitor sit connected to the source for a few minutes (datasheets usually specify 2-3 minutes minimum). Then note down the current (should be 3 μA to several thousand μA, depending on cap's rated voltage and capacitance, so set your multimeter scale accordingly if it is a manual one.)
(2) Measure the voltage across the cap in a cap reformer circuit (after the cap has been reforming for at least 30 minutes). Then, take the source voltage of the cap reformer and subtract the voltage across the cap from that. Finally, divide by the resistance of the current-limiting series resistor.
Example: let's say you have 15.5 Volts as the source voltage for that same 16 V 1500 μF cap above, and the cap has 14.9 Volts across it. If the series resistor is 10 KOhms (10000 Ohms), then the cap leakage current, I_leakage is:
I_leakage = (15.5 – 14.9) / 10000 = 0.00006 Amps = 0.06 mA = 60 μA

Thus, for the above example, comparing the actual leakage current (60 μA, at least from part 2 above) with the calculated from the datasheet (240 μA), we can conclude that the capacitor is likely okay and not leaky. I actually based those numbers above on some experiments I did with my cap reformer (therefore they are not too fictitious or arbitrary.)

So what if a capacitor has higher leakage current that stated maximum in datasheet?
Well, this is obviously not desirable, as stated previously. Moreover, if you are trying to measure the capacitance with a capacitance meter, chances are your meter will probably show higher capacitance than what the cap actually has. Also, there have been numerous instances on the forums here, where a cap with high leakage current would show excellent low ESR (0.02 or less) on an ESR meter, despite the cap actually being popped (i.e. bulged). I've experienced this myself too with some clearly bad Nichicon HM and HN caps from the faulty batches produced between years 2001 and 2005. Of course, for those of you who have ESR meters with capacitance measurement capability, you might have noticed that the cap probably also showed ridiculously high capacitance – and if that was the case, then you definitely had a bad cap in your hands.

Anyways, what I am really trying to get to here is show how to measure cap leakage current. This can come handy if you have some new-old stock (NOS) caps. In particular, you might want to know if they are still good after you reform them. Measuring the leakage current won't tell you if the caps are still good, but if you do have an ESR and/or capacitance meter, at least you will know that it is not showing you some bogus measurements. And of course, if your caps do have an abnormally high leakage current, then perhaps they are just too tired (too old?) and it is likely time to discard them.

[stj]

we need some reference-numbers.

i'v noticed that aging poly's have higher leakage but the other figures look o.k.

[japlytic]

On a related subject: how to measure electrolytic capacitor ripple current

[stj]

i'v seen expensive benchtop testers that do that.
i suspect they charge the cap, then dump it into a load through a shunt and capture the current-flow.

[momaka]

Originally Posted by stj

i'v noticed that aging poly's have higher leakage but the other figures look o.k.

Polys, in general, do tend to have about three to ten times higher leakage current - at least the ones I've worked with. That would be Chemicon PSA and PSC, Nichicon FPCAP R7 and R5, Sanyo OS-CON SEPC, as well as Fujitsu FPCAP RE. This is why you should always check what's in the datasheet for your specific cap.

Originally Posted by japlytic

On a related subject: how to measure electrolytic capacitor ripple current

I don't think there is an easy way to do that.
First off, the ripple current spec for a capacitor is not a "solid number". The cap will be able to handle different amounts of ripple current at different frequencies. Second, you can push any amount of ripple current through the cap that you want. However, the ripple current produces heat inside the cap. Therefore, it must be kept to a level that doesn't make the capacitor overheat internally. And to measure that, I think you would need some seriously sophisticated equipment.
So no simple answer here.

[stj]

the ripple current can be considered the same as the total holding current if you charge it and then do a load dump for most purposes.

that's how i think the bench testers do it.

[socketa]

That was very well written and explained!

Okay, I know this is likely covered much better by many other places online

i seriously doubt that - at least not in terms of simplicity and ease of understanding.

[CuloumbGuy]

Thanks for a really informative post! You saved me from buying an expensive leakage tester

[stj]

leakage can be metered with a $7 component tester.

[vhr]

Yes,Vloss.

[momaka]

Originally Posted by stj

leakage can be metered with a $7 component tester.

They're not always accurate., as Dave from eevlog noted himself. ESR and capacitance? -Yes. Vloss (leakage)... flip a coin and maybe pray a little. I get crappy Sacon FZ caps fooling my meter all the time. Same with some leaky Nichicon HN/HZs.

[sam_sam_sam]

Originally Posted by momaka

They're not always accurate., as Dave from eevlog noted himself. ESR and capacitance? -Yes. Vloss (leakage)... flip a coin and maybe pray a little. I get crappy Sacon FZ caps fooling my meter all the time. Same with some leaky Nichicon HN/HZs.

So what do you do then

how do what is good

or

do you just replace them

[momaka]

^ If capacitance is over 20% of stated value on the sleeve, the cap almost always has high leakage current and is likely on its way out. For those cases - yes, replace the cap for continued reliability.

If capacitance and ESR are normal, I sometimes also charge the cap to full rated voltage and leave it like that for some time to see how much the voltage falls down. For big caps over 1000 uF with a fairly low voltage rating (less than 50V), the cap should hold a significant portion of the voltage for at least 10-30 minutes - that is, only after you have reformed the cap, of course. Unreformed used capacitors will tend to drop down to whatever voltage they had across them while used in a circuit (i.e. for motherboard with DDR2 RAM, you will likely see the caps filtering the DDR2 Vdd voltage to drop down to 1.9-2.0V when left with a charge over a very long time).

This method with leaving the cap charged is not very precise, but will accurately show you if you are dealing with a cap that has severe high leakage current.

Alternatively, you can measure the drop across a 1-10 KOhm resistor while reforming a capacitor (but only after the cap has finished reforming for a good 10 minutes). Then divide that voltage drop by the resistance, and you get the leakage current.