Measuring capacitor leakage

[vindoline]

I have a question about measuring capacitor leakage.

I want to build a 0.1-10 Hz low noise amplifier after Andreas's design. There has been quite a lot of discussion about the importance of selecting the input caps for low leakage, but little guidance on what techniques or conditions are appropriate. My understanding is that the leakage should be in the single digit nA range.

Is it as simple as charging the cap to 10V while monitoring the current? I have a Keithley 616 electrometer or a Keithley 196 DMM I can use to monitor current in the nA range.

What about self-discharge? Is this the same as (or a surrogate for) leakage? I can charge the caps to 10V, then remove from power and monitor the voltage drop over time with the electrometer (it's input impedance is > 2 x 10^14). The 616's resolution is poor, but I can use it's buffered output to feed my HP 3456A and log the voltage drop over time via GPIB.

Also, it looks like this may take quite a bit of time. Any ideas or tips on being efficient would be welcome. Thanks!

[doktor pyta]

Slightly off topic but there is another approach proposed by Mickle T.
He made large foil capacitor built from smaller ones.
https://www.radiokot.ru/forum/viewtopic.php?f=10&t=52829&start=2660

[DC1MC]

Personally for the elcos I charge them with a small current until they reach the nominal voltage then measure the discharge rate.
It's scary how miserable are the the wired ones, especially the axials, even bought new from reputable suppliers like Digikey and Mouser  .
The new tantalums and smd caps are really good.
One thing that I've observed, is that for the high voltage aluminum elcos, if I let them charged without any load their internal resistance do improve a bit (discharge rate decreases after 2-3 nominal voltage to to zero by self discharge), but I have no idea if under real load this helps in any way and it does take a while.
I will love to hear/read a reasonable explanation/paper about that.

 Maybe it will help you,
 DC1MC

[MiDi]

How I did, alternative is to put high resistance between low noise DC source and cap, let settle/soak and measure voltage across it.
Depending on how stable and noisy your PS is, the resistance has to be chosen accordingly as you do not want to measure your PS 

[Kleinstein]

Measuring the true leakage is a rather difficult task with capacitors that have a high dielectric absorption (DA).  The description with an equivalent circuit made from an ideal capacitor and parallel resistor for leakage just is not a good approximation. The DA is to a first approximation modeled with another capacitor and large resistor in series, parallel to the main cap. With electrolytic caps DA is often quite strong and thus the extra cap with series resistor quite large - up to about 20%. The time constants involved can be rather long -  more than hours and maybe days.

So the measured voltage change during self discharge is not only due to leakage, but also (and often to a large extend) DA. Also external effects like changing temperature and maybe air pressure or sound might effect the measured voltage.

To get an idea on the DA effects, one can charge a cap for maybe 10 minutes, discharge it for a few seconds with a low resistance and maybe dead short later and than measure the voltage that develops over the next minutes. With electrolytic caps this can be a significant  part of the initial voltage. One might get an idea about leakage (though at a lower voltage), when the voltage finally starts to drop again.

A similar effect but other direction also happens when watching a fresh charged cap: the voltage first drops quite fast and than after hours kind of settles to a much slower, about constant discharge rate that finally might correspond to the leakage rate. The initial part (like the first few hours to days) is usually dominated by DA.

Other ways to measure the leakage are effected in a similar way.

[Echo88]

For Andreas Design a single 3300µF-input-capacitor is used (or 2200µF + 1000µF).
The leakage of 105°C-rated caps appear to be higher compared to 85°C-rated caps, according to my limited sample observation of 10 105°C caps vs 10 85°C caps. Also there are low-leakage-specified caps like https://eu.mouser.com/datasheet/2/293/e-kl-14230.pdf
I tested about 10 3300µF/16V/105°C Panasonic-NHG-series capacitors and all of them were at about ~40nA after 48h at 10V, but i didnt form them at 16V before.
After that i bought 10 3300µF/50V/85°C Panasonic-M-Series (16V-capacitors were not available at that from moment from the seller), formed them at 50V for about a day, so the aluminium-oxide-layer could heal/stabilize.
Then i discharged them to 10V and let them stay at that voltage for 48h, i measured leakage-currents of <=5nA for 6 of them i think.
I used a calibrator to generate the stable 10V and 1Meg-resistors in series with each capacitor. My 6.5-Digit-DMM 34465A was used to measure the voltages over the 1Meg-resistors, 1nA -> 1mV over the resistor. But a stable voltage-supply might also be used to generate the test-voltage.

You can also use the discharge-current method you mentioned, but i guess it takes at least the same amount of time as the charge-current-measurement-method, since were basically waiting for the dielectric absorption in the cap to stabilize.

If youre using the PCBs made by mimmus78: the pcb fits tightly inside a TEKO 393, if you cut of a few mm on the width of the pcb with lever scissors. Also the mentioned 3300µF/50V/85°C are bigger than the 16V-types, but still fit on the pcb and inside the TEKO 393.

TL:DR:
I suggest you get 10 3300µF/>=16V/85°C-capacitors, form them at their rated voltage for a day and then let them stabilize at 10V for 48h. That should get you at least one very good capacitor with <=5nA leakage-current. Remember to cool the capacitor-leads during soldering, otherwise its leakage might jump up due to the temperature-shock...at least thats what ive heard.

I tried to measure the leakage-current first with my Keithley 2500 Picoampmeter, but it wouldnt show me reasonable readings (100Giga-Ohm-resistor and Keithley220-current-source was tested fine with it). I guess it didnt like the heavy capacitive component? Maybe your Keithley 196 might show this behavior also, thats why i used them 1Meg-resistor-method.

[e61_phil]

@echo88: I think every transimpedance amp will have problems with high capacity directly at the input. Did you try to put a high value resistor in series? That should limit the bandwidth for the noisegain and it shouldn't matter for DC leakage current.

[Echo88]

  Thanks for the link doktor pyta, i also thought about realizing the input-cap with parallel-connected 100µF-foil-caps but gave up on the idea, because of the resulting size and the expected emi-noise which will be received. Seems Mickle is one step ahead again.  Hmm, seems he just didnt find good electrolytic input caps and then resorted to his solution.

@e61_phil: Thats a good idea phil, i might try that.

[Andreas]

Hello,

I use a (stable) power supply with about 10V and a 100K resistor in series to the capacitor to measure the voltage across it with a DMM.
Alternatively I used the 200mV range of my cheap 3.5 digit DMM directly as series resistor (it has 1 Meg input resistance so 1 mV = 1nA).
Usually it needs 2 days forming before we get values in the nA range.

I go for a constant power supply because of the dielectric absorption which makes dynamic methods difficult.
(and a constant source is also that what I want to measure finally).

with best regards

Andreas

[vindoline]

Lots of great replies. Thank you everyone!

I go for a constant power supply because of the dielectric absorption which makes dynamic methods difficult. (and a constant source is also that what I want to measure finally).

Thanks Andreas, that makes perfect sense to me. I've got the caps forming now at 10v all in parallel. I'll add series resistors to all of them tomorrow and let them charge a few more days before screening them.

[BravoV]

I used bench PSU to measure the cap leakage in series with DMM that has 10M Ohm resistance, details -> Capacitor Leakage Measurement

Example of a polymer OSCON forming at rated Volt for about 1.5 hours, at the chart reading, Y axis = 1 volt equivalent to -> 0.1uA or -> Volt / 10M Ohm DMM internal resistance.

[David Hess]

I want to build a 0.1-10 Hz low noise amplifier after Andreas's design. There has been quite a lot of discussion about the importance of selecting the input caps for low leakage, but little guidance on what techniques or conditions are appropriate. My understanding is that the leakage should be in the single digit nA range.

Since this critical coupling capacitor is used as part of low frequency high pass filter, the leakage is limited by its error contribution across the following shunt resistance.  Lower shunt resistance produces less noise but require a larger capacitance to maintain the 0.1 Hz cutoff frequency.

Dielectric absorption in an electrolytic capacitor will require 10s of hours to days of conditioning at the test voltage before valid measurements can be made.

The noise in bad parts tends to be like popcorn noise but I do not know what mechanism is behind it.  Metalized film capacitors can produce their own noise from their self healing mechanism.

[MiDi]

My 0.1-10Hz LNA settles in max 10 minutes to steady state (0V->10V).
I cannot recognize any significant settling after that.
So DA for my unit does not have a great impact.

[Andreas]

Hello,

DA is filtered out by the 0.1 Hz high pass.
Thats why you dont see it

with best regards

Andreas

[Echo88]

@David Hess: I didnt see that effect myself in electrolytic capacitors. Which types did you test to see that effect / how do you measure the caps?

I was always wondering how the DA affects noise measurement.
How big is the impact in noise-error/how long is the settling-time when you connect a low noise reference, say an LTZ, while the input voltage before was 0V and the input slowly ramps up to LTZ-voltage due to DA?
I only know that the AN124 states that a settling time of 24h is required for meaningful results?

[wraper]

For Andreas Design a single 3300µF-input-capacitor is used (or 2200µF + 1000µF).
You should use a 85°C-type instead of a 105°C-type, as the leakage of the 105°C-types is higher.
I tested about 10 3300µF/16V/105°C Panasonic-NHG-series capacitors and all of them were at about ~40nA after 48h at 10V, but i didnt form them at 16V before.
After that i bought 10 3300µF/50V/85°C Panasonic-M-Series (16V-capacitors were not available at that from moment from the seller), formed them at 50V for about a day, so the aluminium-oxide-layer could heal/stabilize.
Then i discharged them to 10V and let them stay at that voltage for 48h, i measured leakage-currents of <=5nA for 6 of them i think.

You make extraordinary generalized claim based on sample of 2. There are a lot of different capacitors with different electrolyte and different purity of aluminium. Both 85^o and 105^o can vary a lot. It's just temperature rating, not "type". If you need guaranteed low leakage, buy low leakage series https://eu.mouser.com/datasheet/2/293/e-kl-14230.pdf

[Echo88]

I make claims based on my measurements of 10 caps with 105° specification compared to 10 measured caps with 85°C-spec, which is also mentioned by other people who tested 105°C-specified caps.
I changed my wording to "The leakage of 105°C-rated caps appear to be higher compared to 85°C-rated caps, according to my limited sample observation. Also there are low-leakage-specified caps like https://eu.mouser.com/datasheet/2/293/e-kl-14230.pdf".

Im not willing to spend a big amount of money/time to leakage-test various caps and as you state that there are big differences in seemingly same caps, please share your knowledge.

[MiDi]

Rule of thumb:
85°C types have lower leakage then 105°C types
lower voltage types have lower leakage then higher voltage types

Echo88 is not the only one who measured this, there is evidence of this effect.
All low leakage claimed types are 85°C rated, this should be good enough indicator.

Sure you cannot rely on small quantity measurements for generalized statements.

[wraper]

I make claims based on my measurements of 10 caps with 105° specification compared to 10 measured caps with 85°C-spec, which is also mentioned by other people who tested 105°C-specified caps.
I changed my wording to "The leakage of 105°C-rated caps appear to be higher compared to 85°C-rated caps, according to my limited sample observation. Also there are low-leakage-specified caps like https://eu.mouser.com/datasheet/2/293/e-kl-14230.pdf".

Im not willing to spend a big amount of money/time to leakage-test various caps and as you state that there are big differences in seemingly same caps, please share your knowledge.

You bought 2 series from one manufacturer. Isn't it wildly exaggerated to make conclusion about all capacitors in the wild based on that? Heck, you could get a different batch of those and receive opposite result. If you bought 20 different 105^oC capacitors and 20 different 85^oC then you could at least make a humble conclusion that results could extent to capacitors as a whole.

[wraper]

Echo88 is not the only one who measured this, there is evidence of this effect.
All low leakage claimed types are 85°C rated, this should be good enough indicator.

Nichicon datasheet says they can supply 105^oC variant as well.

[wraper]

Im not willing to spend a big amount of money/time to leakage-test various caps and as you state that there are big differences in seemingly same caps, please share your knowledge.

It would be much more meaningful if you bought 1 capacitor of 10 different series instead of 10 capacitors from single series which are from the same batch.

[Echo88]

It would be meaningful if you either state info about caps that you tested (did you?) or you send me a large amount of money to test different caps from all manufacturers, so i can make conclusions that benefit all members here.

[branadic]

I have tested different caps available at Reichelt for the low noise amplifier of Andreas too and found cheap Yageo 85°C caps to be low leackage. After forming them over a 24h time span on a 9V block an arrangement of 1.000µF/25 || 2.200µF/35V showed <5nA.

https://www.mikrocontroller.net/topic/207061#3164476

-branadic-

[Echo88]

The pdf-link in https://www.mikrocontroller.net/topic/207061#3164476 is not available anymore. Ive found the following document: https://tadiranbatteries.de/pdf/applications/leakage-current-properties-of-modern-electrolytic-capacitors.pdf is it the one you mentioned ?
Good read about the increased leakage after soldering and the healing-effect/tunneling.

[branadic]

The pdf-link in https://www.mikrocontroller.net/topic/207061#3164476 is not available anymore. Ive found the following document: https://tadiranbatteries.de/pdf/applications/leakage-current-properties-of-modern-electrolytic-capacitors.pdf is it the one you mentioned ?

Yes, seems like they renamed the file.

-branadic-