Practical Leakage Current Measurement on Caps

[luky315]

I was wondering how to quickly, yet accurately check the leakage current on ceramic capacitors. Some of them have a rather high isolation resistance (the manufacturer claims >10GOhm) and the insulation resistance meter that I know use quite high voltages, much higher than the rated voltage of the capacitors. Tek claims that the 2450 Sourcemeter is able to measure the leakage current on capacitors while applying a low voltage (for a price...), but is there another, cheaper solution for quick checks and comparisons of different capacitors?

[Whales]

How big are the ceramic caps?  What range of leakages do you want to measure (what ranges of leakages would cause you problems)?

You need a voltmeter with a mixture of fast measuring speed, low input capacitance and high input impedance.  Perhaps a vacuum tube or jfet voltage follower and then a normal oscilloscope (or multimeter)?  Charge the cap to voltage, disconnect it for x time, reconnect it to the jfet+scope and then use that to measure the voltage quickly (before it drops too much).  A jfet+multimeter might be enough perhaps, I'm not sure.

[Alex Nikitin]

What you need is an electrometer or a picoammeter + a stable adjustable DC supply for the biasing. The Keithley 617 I have in my lab can measure leakage down to single digit femtoamps (and some better ceramic caps, for example TDK NP0/CG0 will measure that low) for bias voltages up to +/-100V .

Here is a discussion on this very subject.

Cheers

Alex

[luky315]

So building a picoammeter would be the suggested solution? I'm expecting leakage currents in the range of a few pA to 100nA and would like to check with 3.3 and 5V bias voltage (battery powered design)

[Kosmic]

So building a picoammeter would be the suggested solution? I'm expecting leakage currents in the range of a few pA to 100nA and would like to check with 3.3 and 5V bias voltage (battery powered design)

You could try to build one of those: https://www.eevblog.com/forum/projects/picoammeter-design/msg790045/#msg790045

[bdunham7]

Are you looking to accurately characterize them or just check that the leakage is not excessive?  For the latter, the conductance range of handheld DMMs that have the feature should do reasonably well. 

[armandine2]

this method came up recently in my youtube feed

[jonpaul]

ceramic have nonlinear lkg vs voltage.  Depends on dielectric type eg NPO, X5R, etc.

Higher the dielectric contast, the high the leakage.

You cannot measure gigohms with 3..5V.

Lab testers use 100..500..1KV and up.

Use plastic not ceramic cap for very low lkg eg integrators.

Use HV cerami for HV use.

[iMo]

For example

[bdunham7]

You cannot measure gigohms with 3..5V.

Lab testers use 100..500..1KV and up.

DMM conductance ranges do just that, although not with great precision at the low end (high resistance) end of the scale.  However, you can't use a 1kV test voltage on a 16V MLCC.  A DMM conductance range using 3V or so open-circuit is likely as good or better than anything that can be thrown together easily on a breadboard.

[mawyatt]

Use plastic not ceramic cap for very low lkg eg integrators.

A study of modern designs show using C0G/NP0 not plastic capacitor for precision integration. Good examples are the modern 6 1/2 + digit DMMs which utilize C0G/NP0 capacitors in the core integrators.

Best,

[Kleinstein]

With an integrating ADC in a DMM the leakage is likely the smaller issue. The larger point is usually dielectric absorbtion. The 2 effects are however a bit tricky to separate - the longer the time scale the more it's leakage. The DA is reversible and looking at the current at zero voltage should eliminate the leakage part. The actual leakage is than abtained from the difference in the current at voltage and after a discharge.

When just looking at the voltage one has to take into account the temperature effect that can also change the voltage.

I have done some tests on capacitors for a high quality integrator and found the typical C0G capacitors about as good as good film types (PP or polystyrene). In my tests the best choice was a C0G type from TDK.  So it makes sense that modern DMMs tend to use C0G capacitors and avoid the PP and PS types that are not available as SMD because they are temperature sensitve.

For the leakage it can also make sense to look at the humidity. At least with a bit higher humidity and relatively small capacitors a good part of the leakage can be a surface effect from the water film that can form above a certain humitity level.