Capacitor ESR good but dissapation factor bad

[MathWizard]

I have a couple of 1000uF 25V caps from the input and output of a 2A 5V buck converter, their 100kHz ESR is 0.062 which I think is pretty good, but their 1kHz tan delta % is ~35%. In the datsheet it says 14% for 25V working voltage. IDK if they mean the max voltage there, but for 6.3V it lists 23% anyways.

So should I worry about that ?

[wasedadoc]

See https://3roam.com/capacitor-dissipation-factor-calculator/

DF can be calculated from ESR.  That tool gives 38% assuming that the ESR at 100kHz can be used to calculate the DF at 1KHz.  On that basis I think your capacitors are OK.

[elecdonia]

Do these capacitors get warm or hot when the converter is operated over a range of different load currents?
If they remain close to room temperature then there is nothing to worry about.

In my opinion the most important parameters for electrolytic capacitors in power supply circuits are uF and ESR. However I’ve never encountered an issue where the “dissipation factor” parameter itself was significant.

[Kleinstein]

The ESR is relevant when used with a relatively high frequency and thus high current, but still low AC votlage. The DF is relevant with lower frequencies (e.g. 120 Hz) with significant ripple voltage. The losses are possibly frequency dependent and there is no simple conversion from the ESR measured at 100 kHz and the DF at lower frequency (e.g. 1 kHz or 100 Hz).

14% sounds reasonable for an electrolytic capacitor, 35% is a bit on the high side. The DF can be higher when measured with a low DC bias or even without DC bias.
It depends on the ripple present if the loss is relevant and causes significant warming of the capacitor.

[T3sl4co1l]

Electrolytics normally have quite poor DF; this sounds reasonable to me.  On the high side perhaps, but not ridiculous.

Note that ESR, by its name (effective), varies with frequency; for electrolytics, it's normally quite a bit higher at low frequencies, thus the DF stays high over a wide frequency range.  DF approaches 1 of course near the resonant frequency (Xc = Xl, ESR alone remains).

If the capacitor is suspected of old age, I would try heating it up, see if that reduces ESR significantly.  Chilling will always cause ESR to increase (typically ESR rises significantly below 0°C or so), but when the temperature where this occurs shifts upward (i.e. heating "cures" its high ESR), that's sign of a failing capacitor.

Tim

[MathWizard]

Ok they're probably ok then, IDK where the 18V input came from, but the 5V output was from a 400kHz buck converter/inductor, but it seems to have failed. The caps are 105deg rated at least. I don't think the have to supply anywhere near the 2A rating of the buck chip.

[Vovk_Z]

I have a couple of 1000uF 25V caps from the input and output of a 2A 5V buck converter, their 100kHz ESR is 0.062 which I think is pretty good,

0.06 R for 1000 uF 25V doesn't seem to me good. It's rather somewhere on an edge or rather a bit more than is allowed.
An average LowESR 1000 uF 25V cap has to have <=0.04 R (look at a datesheet of any average LowESR series cap).
0.06 R can be fine for a general use series or wide temperature range series cap (but not LowESR).

[elecdonia]

If the capacitor is suspected of old age, I would try heating it up, see if that reduces ESR significantly.  Chilling will always cause ESR to increase (typically ESR rises significantly below 0°C or so), but when the temperature where this occurs shifts upward (i.e. heating "cures" its high ESR), that's sign of a failing capacitor.

I’ve observed that small switching power supplies which fail to turn on when cold will often function after being warmed up. I suspect this occurs because of this phenomena where ESR decreases as the capacitor gets warmer.
As a result I think a feasible way to ID a device which contains failing high-ESR capacitors is to place the device in a refrigerator overnight and then test to see if it will still power up and function correctly when cold.

[T3sl4co1l]

Yup, or with the use of freeze spray or hot air (in a open-chassis repair context).

Tim