Is there a simple way to test a cap for leakage- BEFORE it's soldered in place?

[MikeH]

I'm really sick of finishing a build and having it pop because some crappy electro cap is leaking all kinds of DC on the input or output, and then having to replace the cap- sometimes twice.  How can I test caps for leakage before using them?

[alanlan]

All electrolytics have some leakage, so it may be that you're using the wrong part for the job.

But if you want to measure leakage simply connect a 100K resistor in series with the cap and a 9V battery and measure the volt drop across the resistor.  0.1V for each 1uA of leakage.

Also, make sure you're connecting them the right way around with regard to polarity.  Electrolytics do not like being plugged in backwards.

[frank_p]

Type "ESR meter" in the search function Mike.

[John Lyons]

ESR isn't leakage is it??


John

[frank_p]

ESR isn't leakage is it??


John

I think you are right John,  I will get an other coffee.  I mistaken series and parallel.
Then you have to test for the two problems when they are old ?

Sorry for the confusion Mike.

http://en.wikipedia.org/wiki/Capacitors


This can not leak. 



But case b) can.

[MikeH]

Thanks for the replies.

All electrolytics have some leakage, so it may be that you're using the wrong part for the job.

It's not an issue from something I'm designing myself, or I'm using an electro instead of another cap, or anything like that.  I'm using an electro where a design calls for one.  I'd say one in every ten electros is probably crap (could just be my sources).  I just want a way to test them before I use one.

[R.G.]

Connect the + side of a 9V battery to the + side of the cap. Connect the + lead of your DMM to the (-) side of the cap. Connect the (-) side of your meter to the (-) side of the battery, set the meter to read *Volts* and watch the voltage. It will start at 9V and decline to the ohm's law value of the cap leakage resistance and the meter internal resistance by the voltage divider equation. This will be slow, as the cap will be large (presumably 1uF to 100uF) and the resistance of your meter will too - 1M to 10M most likely. You're looking for 100M or so across the cap, so you want the meter voltage reading to get under 1M/(101M)*9V or 0.09V; better yet, 0.009V. Any more than that is excessive leakage.

It would be massively better to know ahead of time what the resistance of your meter is. You can do that by constructing a 1M/1M voltage divider and measuring across 1 of the 1Ms. If it changes to 2.3V instead of 4.5V, you have a 1M input meter. If it stays about 4.4V, you have a 10M input meter. Once you know that, you will be able to calculate the measured leakage resistance from above.

Take note: if you are measuring a 100uF cap with a 10M meter, the time constant is 100E-6*10E6 = 1000 seconds, and it will take 5 time constants (5000 seconds or about an hour and 20 minutes to stabilize just from the R-C time constant. The cap may be fine, but you won't know that for over an hour. Proportionately shorter for smaller caps and resistors.

[MikeH]

Take note: if you are measuring a 100uF cap with a 10M meter, the time constant is 100E-6*10E6 = 1000 seconds, and it will take 5 time constants (5000 seconds or about an hour and 20 minutes to stabilize just from the R-C time constant. The cap may be fine, but you won't know that for over an hour. Proportionately shorter for smaller caps and resistors.

Thank you for putting this in perspective for me- it would seem to me it's far less time wasted to simply replace shoddy caps as they come, rather than test them ahead of time. 

Oh well, none the less- good info to know.

[frank_p]

Thanks for the tips Mr Keen.