Maker Pro
Maker Pro

I thought I'd measure the leakage current of a few caps

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
After measuring the leakage in some diodes, I thought I'd measure the leakage in some caps.

I have some nice (we'll find out how nice) 330pF polystyrene capacitors, and for comparison an X rated 3n3 capacitor.

The 330pF capacitor measures 1.95x10-13A, that's 195fA with about 12.8V across it. This is the equivalent of a 65TΩ shunt resistance.

The 3n3 cap, on the other hand, measures 0.9 x 10-11A, that's 9pA with about 12.8V across it. The equivalent shunt resistance is about 1.4TΩ

(edit that's TΩ, not just GΩ)

In the circuit I have, the capacitor shunts a 1GΩ resistor. So either cap might be OK. I'm expecting 1mV per pA of input current across the resistor, so assuming the leakage is linear with voltage, either cap would probably be OK. However the 3n3 cap has much higher dielectric absorption, so (even ignoring the greater capacitance) it's not a good choice.

At this stage none of the caps have been cleaned (just carefully handled). Whilst I can clean the 3n3 cap with isopropyl, I'm not supposed to clean the polystyrene cap with this (apparently distilled water is the right solvent). I'll give both a blast with isopropyl and see what happens...
 
Last edited:

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
OK, so a quick clean of both caps by spraying some isopropyl alcohol on them...

Let's look a the 3n3 cap first. The leakage current halved to 4.5pA (reflecting an increase in resistance to about 2.8TΩ

So how much better is the polystyrene cap?

It now measures 0.2x10-7A. That's a massive 20nA, or a reduction in resistance to only 640MΩ

After waiting a while the current dropped to 0.65x10-10A, a little better at 65pA or about 200GΩ

I knew that isopropyl does this to polystyrene caps. Apparently the resistance will continue to rise for a couple of days now. It may even get better than it was originally. Should I heat the capacitor to drive out the solvent? No -- the dielectric is subject to some sort of phase change at 70C which permanently alters the capacitance and leakage.
 
I knew that isopropyl does this to polystyrene caps.
Isopropanol (and most other alcohols, I suspect) does a good job of dissolving polystyrene, probably not a good thing for your capacitor. If I remember correctly, it's also hydrophilic and hygroscopic, and whatever water it picks up will most certainly increase the leakage current.

Should I heat the capacitor to drive out the solvent? No -- the dielectric is subject to some sort of phase change at 70C which permanently alters the capacitance and leakage.
I'm not aware of anything terrible happening at 70C, but I believe polystyrene melts at around 85C. I should think that sticking to 50C-60C, you'd be safe.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
My plan is to see what happens over a couple of days (or however long it takes to stabilize).

Some people have done this inadvertently and claim it had no detrimental long term effects.
 
Interesting--thanks for sharing.

Care to also share what circuit (or equipment) you used to measure such a small current?
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
I'm using a Keithley 610B electrometer.

There circuit is very simple, it's just the battery, a diode, a resistor, and the cap under test in series with the electrometer in its current measurement mode.
 
Thank you. I haven't had to take a measurement at that low of current. I see the Keithley 610B electrometer has some pretty good specs in the current measurement mode--I figured the meter noise would be larger than that (+/-3 fAmps).

Does your test circuit pick up much noise when taking these measurements? (i.e. from nearby electronics) I am curious because I could see your test circuit acting as a single loop antennae and throwing off your measurement rather easily.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Does your test circuit pick up much noise when taking these measurements?

Lots.

Moving my arms caused the meter to go wild :)

And at the more sensitive end of the measurements there was an offset that I had to subtract, with significant low frequency noise on top of that.
 
Top