Practical or quick n dirty method to measure capacitor leakage ?

[BravoV]

Edit : Tested and results using Hero999's idea -> Post #10

The title should be clear enough, and since its for hobbyist/enthusiasts audience, I guess an electrometer is not an option, or at least not a norm cause even its used, most have crazy prize tag on it. 

Of course there must be compromise, say for the accuracy, but how much ?

TIA

[poot36]

If you lookup vintage capacitor tester you will find vacuum tube based capacitor testers that use a bridge circuit along with a adjustable high voltage supply and usually a tuning eye indicator that will change if there is too much leakage through the capacitor.

[kripton2035]

there is one project that measures esr, capacity and leakage, but is limited to low voltage capacitors :
http://kripton2035.free.fr/digital%20esr/esr-sprut-de.html

also siliconchip had a project that does it for higher voltages :
http://archive.siliconchip.com.au/cms/gallery/article.html?a=111776&i=8

and there is a previous thread about this in the forum :
https://www.eevblog.com/forum/projects/capacitance-leakage-tester/

[tautech]

We presume you're wanting to measure used caps? 

For new Electrolytics, selection is based on study of datasheet specs to fit YOUR application.

[Marco]

Deadbug LMC662 follower with bootstrapped power supply and a couple 100K series resistance (to protect the input)?

[Zero999]

What size are the capacitors?

If it's a large capcitor (over 100µF or so) I just use an ordinary multimeter.

Charge the capacitor.

Disconnect from the power supply.

Leave for a known period of time, say 5 minutes.

Very briefly measure with a multimeter. It's important to only connect the meter for long enough to get a reading, otherwise the meter's input impedance will significantly affect the result. To speed things up, manually select the desired range, zero and use the maximum reading function.

Do the maths.

[LA7SJA]

When testing low voltage caps i use a currentlimited powersuply and ucurrent gold or another sensitive ampmeter. Testing of high voltage caps i use my isolation tester at 125V, 250V, 500V, 1000V, 2500V or 5000V.
Practical? Yes I think so. Qucik? Yes I think so. Dirty? I realy don't know.

Johan-Fredrik

[Zero999]

Haven't tried it but you could charge the capacitor, then connect the DVM, set to measure voltage, in series with the capacitor and the same power supply you used to charge the capacitor. If the meter has a 10M impedance (most do) then it will read 1mV per 100pA of leakage current.

[BravoV]

@kripton2035 , thanks, I'm aware of those circuits, and your circuits repository too.  Its just it will be one off project, sort of ghetto setup as probably it will not be used oftenly. Building those circuit will be too much.

Deadbug LMC662 follower with bootstrapped power supply and a couple 100K series resistance (to protect the input)?

Out of curiousity, I downloaded that opamp datasheet, wow, 2 pA input bias current. Please, mind elaborate further how ? I know whats a follower is, what do you mean by bootstapped ? A quick circuit explaining it will be appreciated. Thanks for pointing that opamp, pretty affordable, I might get one or two just for curiosity sake.


@Hero999, caps to be measured varies, and majority are < 100 uF.

[BravoV]

Haven't tried it but you could charge the capacitor, then connect the DVM, set to measure voltage, in series with the capacitor and the same power supply you used to charge the capacitor. If the meter has a 10M impedance (most do) then it will read 1mV per 100pA of leakage current.

Ok, interesting, quick n dirty enough  , I will do that and report back, thanks for the idea.

[BravoV]

As promised, especially thanks to Hero999's idea, here are the results so far.

Method and test details :

• The setup was using adj. bench power supply and Fluke 287 that has 10 Mega Ohm input resistance as in the specification.
• Using this arrangement, so at every beginning of each test, the CUT was connected directly to the PS with the switch closed, and left for about 5 minutes, and then opened the switch to start the measurement.
  
  

The CUTs :



Top left to right :
Sanyo Oscon 150 uF 16V , Nichicon 1000uF 100V (PR series) , Nippon ChemiCon (LXZ Series) 1000uF 16V

Bottom left to right :
Polypropylene 560 nF 400V and polypropylene 330 nF 2000V

PS : Not sure about bottom red plastic caps brand or the type as well, so CMIIW.

Datasheet for the electrolytic caps can be found here (stolen from BadCaps forum) -> Electrolytic Datasheet Repository (suggesting you to bookmark it if you haven't  )


The results :

Sanyo (OSCON Series) 150 uF 16 V (light blue colored/top left)
Test voltage = 16 V
Measured voltage drop = 3.7 volt (more detail below especially for this cap)
Equivalent leak
= Voltage measured / DMM Internal Resistance
= 3.7 Volt / 10 Mega Ohm
= 0.37 uA leakage

Nichicon (PR Series) 1000 uF 100 V (brown/top middle)
Test voltage = 37 V (NOTE : My adj PSU maxed out at 37 Volt, so this is the highest test voltage)
Measured voltage drop = 7.8V -> 0.78 uA leakage

Nippon ChemiCon (LXZ Series) 1000uF 16 V (dark blue/top right)
Test voltage = 16 V
Measured voltage drop =  0.7V -> 0.07 uA leakage

Polypropylene 560 nF 400 V (bright red/bottom left) PS : Seen as photo below.
Test voltage = 37 V
Measured voltage drop = 0.3 mili Volt -> 0.03 uA leakage

Polypropylene  330 nF 2000 V (dark red/bottom right)
Test voltage = 37 V
Measured voltage drop = 0.4 mili Volt -> 0.04 uA leakage


Here is how the "quick n dirty" test jig looks like when measuring that plastic low leakage cap.




Found an interesting situation when measuring the polymer cap Sanyo OSCON and other two electrolytic caps as well. The leakage seems to have "curve", I just let the chart to tell the detail here.

The logged result (DMM measured voltage) once the switch was opened (after it was closed for 5 minutes for the cap to get charged and "settled").

Log interval is one second for > 1.5 hours.




Do these measurements valid (good enough) as in "quick n dirty" way to measure cap leakage ? Any other comments are welcome.


Again, thanks to Hero999.

[SeanB]

The curve is the capacitor reforming to the applied voltage, probably they have been in storage for a while. That is why you find worst case leakage is so high in a datasheet, while the typical is very low. Leave it powered for a month and it will be very low indeed.

I tried a 20000uF 6V capacitor, and it eventually levelled out at a very low self discharge current after a 2 week open circuit test.

[Marco]

wow, 2 pA input bias current.

Actually typical 2 fA at room temperature.

I know whats a follower is, what do you mean by bootstapped ?

A follower is simply a non inverting opamp with output connected to inverting input and signal to non inverting. The output follows the input, without loading the source. This is bootstrapping. The follower prevents the multimeter from loading the capacitor under test.

[BravoV]

The curve is the capacitor reforming to the applied voltage, probably they have been in storage for a while. That is why you find worst case leakage is so high in a datasheet, while the typical is very low. Leave it powered for a month and it will be very low indeed.

Yes, there are quite old, but never used.

I tried a 20000uF 6V capacitor, and it eventually levelled out at a very low self discharge current after a 2 week open circuit test.

How did you test it ?

wow, 2 pA input bias current.

Actually typical 2 fA at room temperature.

Femto Amp ? Yeah, I know its stated at the 1st page of the datasheet, isn't that "marketing" words ?

When on the detail DC Electrical Characteristics page, taking the "worst case" as Dave usually said, its 2 "Pico" Amp.

Which is which ?

I know whats a follower is, what do you mean by bootstapped ?

A follower is simply a non inverting opamp with output connected to inverting input and signal to non inverting. The output follows the input, without loading the source. This is bootstrapping. The follower prevents the multimeter from loading the capacitor under test.

Thanks, that may take a while to digest.

Mind elaborate the follower prevents the DMM from loading the cap ? Whats wrong with just plain follower ? Please be gentle, I'm learning here. Please, if you could post a quick scratch of the suggested circuit. TIA.

[Marco]

Few things are overstated in datasheets, some things are understated and some things are just right.

Nothing wrong with a plain follower, but it has less supply range.

[JackP]

Fempto Amp (spelling?) is the unit below pico Amp (i.e. 1 x 10^-15, or whatever is next from a trillionth)

[SeanB]

I tried a 20000uF 6V capacitor, and it eventually levelled out at a very low self discharge current after a 2 week open circuit test.

How did you test it ?

I measured ESR and capacitance after pulling it out of the box, then charged it via a resistor overnight to 6V5, then the next morning measured it at 6V5, then disconnected it and left it alone for a week, then checked again. Was around 2V after 2 weeks, having dropped fast initially but tapering off after the first week to a more stable voltage with lower drop. Leakage current was higher at higher voltage, If I reformed it with 9V it probably would be lower leakage at 6V, and originally it was intended as a decoupling capacitor on a 5V TTL rail in a minicomputer. After the test I checked ESR and capacitance again, and they were essentially unchanged from the initial reading. Not bad for a 40 year old Sangamo capacitor.

[BravoV]

Fempto Amp (spelling?) is the unit below pico Amp (i.e. 1 x 10^-15, or whatever is next from a trillionth)

Its Femto (10^-15), details here -> List of SI prefixes

I measured ESR and capacitance after pulling it out of the box, then charged it via a resistor overnight to 6V5, then the next morning measured it at 6V5, then disconnected it and left it alone for a week, then checked again. Was around 2V after 2 weeks, having dropped fast initially but tapering off after the first week to a more stable voltage with lower drop. Leakage current was higher at higher voltage, If I reformed it with 9V it probably would be lower leakage at 6V, and originally it was intended as a decoupling capacitor on a 5V TTL rail in a minicomputer. After the test I checked ESR and capacitance again, and they were essentially unchanged from the initial reading. Not bad for a 40 year old Sangamo capacitor.

Whats that cap's voltage rating ?