Really confused about capacitor meter fluctuations

[LoveLaika]

I need some advice. I am doing measurements using a capacitance meter (BK Precision 890) to measure the capacitance between two points. I was using standard multimeter probes, but it seems that even when there is no load between them, the meter fluctuates in the 20-30 pF range. This is an issue, as I'm measuring capacitances in the pF range. I can zero it out with my meter, but the fluctuations still cause inaccurate measurements when accounting for the capacitances along the probes.

After a lot of research, I found that twisting the probes seems to make it better somewhat. The manual recommends that I use short, twisted lead for better shielding, so perhaps the fluctuations are caused by noise. Would it be fine to just twist the probe wires for measuring? I also have a short pair attached to a BNC plug (and a BNC adapter for such measurements) as well as a long BNC cable with tiny alligator clips at the end, but would that work without it being twisted? Also, going this route, if I had a low-pass filter that could be attached via BNC, would that also help filter out noise?

[TimFox]

Are you holding the unshielded probes and/or unshielded wires in your hands while observing the fluctuations?
To measure capacitances in the low pF range, you need to avoid hand capacitance and stop all motion of the probes and wires.

[David Hess]

It might depend on how the measurement is made, but the charge/discharge method of my Tektronix 916 DMM gets picofarad stability only if I am careful about dressing the leads and making sure that their position does not change during the measurement.

My DE-5000 produces stable readings but the leads are short and shielded.  I have not completed a long set of leads for it yet and they will be shielded anyway.

[LoveLaika]

Are you holding the unshielded probes and/or unshielded wires in your hands while observing the fluctuations?
To measure capacitances in the low pF range, you need to avoid hand capacitance and stop all motion of the probes and wires.

Thanks for replying. Well, yes, I have to hold the unshielded probes in my hand. They're probes, not clips; also, the space at the pins are not the easiest to access without holding them. Cramped spacing and all also make it hard to hold them steady. If I should avoid low hand capacitance, is it better to use clips/grabbers so I don't have to hold on to them?

[LoveLaika]

It might depend on how the measurement is made, but the charge/discharge method of my Tektronix 916 DMM gets picofarad stability only if I am careful about dressing the leads and making sure that their position does not change during the measurement.

My DE-5000 produces stable readings but the leads are short and shielded.  I have not completed a long set of leads for it yet and they will be shielded anyway.

My meter doesn't have a port for shielding, unlike the newer BK Precision Model 890C. At best, I can just wrap my leads with some foil, but I'm not sure if that works properly.

[bob91343]

You are using the wrong instrument to make such measurements.  I have been through this many times.  I have an old Boonton RX meter that is good for such stuff, and a GR 1658 bridge.

The problem with long leads is that the stray capacitance is unstable and can't be compensated for.  Yes a shielded bridge is a good answer but you don't have one of those.

If this is an important measurement, you need to acquire a suitable bridge.  Measuring capacitance between widely separated points is a problem in itself, due to (as you have found out) the difficult means of making connection.

My GR 1658 is a shielded bridge and I have made many satisfactory measurements in the sub-pF range.  Its resolution is 0.01 pF.  The Boonton meters, Q meters and RX bridge, have good resolution as well.

Some specialized capacitance bridges have been built with resolution down to 0.0001 pF or so, at least according to my measurements textbook.  I never saw one.  They were intended to measure interelectrode capacitances of tubes.

Tektronix model 130 is good too; it has a low range of 3 pF full scale and residuals can be zeroed out.  But only if they are stable.

[Kleinstein]

There are 2 possible problems: one is the unwanted capacitance of the test leads, that can change when they move or the hand near by moves. So the actual capacitance changes. In addition there is capacitance to the meters "ground".  Shorter leads with clips are a good idea

The other possible problem is capacitive coupling of mains hum and similar that can effect the measurement circuit. Some methods (e.g. slow charge discharge) are quite sensitive, while others (reasonablle fast AC excitation and phase sensitive detection) are quite tolerant.

It can help to do the test in a quiet area, away from mains wiring.

[LoveLaika]

Thank you for your suggestions. If I have a chance, I'll have to look into them. For the time being, all I can do is make do with what I have.

[LoveLaika]

There are 2 possible problems: one is the unwanted capacitance of the test leads, that can change when they move or the hand near by moves. So the actual capacitance changes. In addition there is capacitance to the meters "ground".  Shorter leads with clips are a good idea

The other possible problem is capacitive coupling of mains hum and similar that can effect the measurement circuit. Some methods (e.g. slow charge discharge) are quite sensitive, while others (reasonablle fast AC excitation and phase sensitive detection) are quite tolerant.

It can help to do the test in a quiet area, away from mains wiring.

Thanks for replying. I'll be sure to keep your considerations in mind.If I may ask, my own short leads are attached to a BNC head. I have an adapter for such a thing that I can put into my meter, but aside from that, would it harm anything if I attach a low pass filter in series in between my adapter and my short leads with the BNC head? I think doing so would help filter out noise but I worry about how it would affect measurements.

[bob91343]

A low pass filter would have capacitors in it.  Did you have a circuit in mind?  A filter is not a component; it is an assembly of components.

[TimFox]

A low-pass filter inserted into a sine-wave capacitance measurement circuit would stop the measurement:  you are seeing fluctuations in the low-Hz range due to hand capacitance and moving leads, but the meter uses sine waves between 100 Hz and 10 kHz to perform its measurement.