Good brand capacitors with good ESR but reduced capacitance?

[Agent24]

Replacing capacitors in a PSU with a mix of good and bad brands, I checked even the good brand ones since it was old, found Nichicon PW 3300uF measured as 2400uF and UCC KY 1000uF measured as 820uF. ESR was fine though.
Replaced them all with brand new capacitors that measured in spec.

I have heard of failed capacitors losing capacitance but with good ESR, but has anyone else seen this behaviour consistently? I would have expected ESR to increase first, but perhaps the better brands just fall in capacitance instead?
Funny enough, the rest of the capacitors were LTec and all tested OK except one which had marginally high ESR.

[stj]

Re: Good brand capacitors with good ESR but reduced capacitance?

just different failure modes.

the foil can decay, the paper can dry up, the liquid can leak out or crystalise.
caps are mechanical more than electrical.

[Wester547]

Re: Good brand capacitors with good ESR but reduced capacitance?

They could have developed high internal leakage (in an open circuit manner), possibly fooling the meter, hence the low capacitance but in spec ESR. NCC's KY datasheet specifies a 25% tolerance after the endurance test so the KY is arguably in spec. Anything with a rubber seal and a liquid solution (no matter how high or low the concentration of H2O is) can dry out over time, so I think their bungs might be slightly compromised. Even wet electrolytics with better seals can dry out over time, the same way high voltage electrolytics with snap-in terminals lose capacitance, even from the good brands.

I'd say capacitors are chemical more than anything. The LTECs measuring okay is possibly why some people think them "mediocre", but I've seen enough of the small, general purpose ones dry up fast enough to think ill of them (but to be candid, I certainly consider them better than Teapo). I'd be curious to know what kind of thermal duress the KY and PW were under in the PSU and how old the PSU actually is. With the choice of brands, I'd guess it's a Delta or Newton Power.

[Agent24]

Re: Good brand capacitors with good ESR but reduced capacitance?

Good point, I didn't check leakage. I may do that later.

I agree the KY may still be considered OK by the datasheet, but this particular PSU I had partially recapped before (replaced one failed 3300uF 16v CapXon) and left the rest as they had tested OK at the time. With some spare time I decided to go ahead and replace all the LTec and check the rest. The KY did measure 909uF the first time I checked it, and 820uF now, so I suspect it is failing.

I don't know about thermal stress but you are correct, it is a Delta Electronics DPS-370AB A Rev 01F in fact.
It powers an HP ML310 G3 I bought 2nd hand. An older model so it probably has quite some years on it. I don't know the history.

The KY has "6(T) 2X" written on it aside from the usual markings. I don't know if that's a datecode, if it is, I don't understand it.

[Wester547]

Re: Good brand capacitors with good ESR but reduced capacitance?

Yes, I remember reading about the failed 3300uF 16V CapXon GL in your unit, about two years ago now. Of course the sole capacitor that failed catastrophically in the unit had to be a CapXon... 560uF and 1.5 ohms for a 3300uF capacitor is certainly worse than 820uF and fine ESR for a 1000uF capacitor...

So, the 1000uF 16V 10x20 NCC KY that was in parallel with the CapXon is starting to go bad. I wonder if the failing CapXon "encouraged" it to when it failed, stressing the capacitor more. I don't know why the 3300uF Nichicon PW is starting to fail, but it sort of reminds me of this thread. Capacitors testing fine for ESR but not for capacitance (too low for capacitance in this case). I know the KY in the thread was obtained through an unauthorized Chinese distributor but the capacitor in the thread certainly looks genuine, so I wonder if it's a manufacturing flub up. When you're producing millions, sometimes erroneous sleeving can happen (your KY could really be an 820uF 25V capacitor but the only way to know for sure would be to test its leakage current, and the PW might really be an erroneously sleeved 2200uF capacitor). That's just a theory, of course.

No, the DPS-370AB looks like an overspec'd and well ventilated unit to me. I don't think it was an exothermic failure. That datecode translates to February 24th, 2006, factory "T", according to UCC's datecode system. "T" means one of the factories in Japan, IIRC. Odd. I have seen good capacitors much older than that measure in spec.

[Agent24]

Re: Good brand capacitors with good ESR but reduced capacitance?

That long ago? Huh. It sure doesn't feel so!

Since the 12v rail had only the CapXon and the NCC KY, I'd imagine the KY did get significantly higher ripple to deal with after the CapXon blew, and I have no idea how long it was in that state.
So it would not surprise me if it has degraded due to those circumstances.

As for the Nichicons though, I don't know. They measured ~2400uF the first time I checked them, so maybe they came like that from the factory, and they were not on a rail along with any failed capacitors either, so that theory doesn't seem to fit there. I decided to replace them anyway though, since the full 3300uF would surely not hurt.