Homemade capacitive sensor

[akaliuseheal]

TL;DR Summary

Capacitance of sensor is not stable, why?

Hello,

I have made a capacitive water level sensor. It is a parallel plate capacitor.
While measuring the capacitance of my sensor, I measure 53 pF. I then leave it for couple of minutes, still conected to the meter, capacitance then rises to 54 pF. Capacitance is slowly increasing. Why is that and how do i prevent it?

 

[hutchphd]

Is this in water?

 

[hutchphd]

Exactly how are you measuring the capacitance?

 

[akaliuseheal]

Is this in water?

It is in air.

Exactly how are you measuring the capacitance?

I have the capacitor connected to the multimeter set to measure capacitance. I have it conected and turned on for couple of minutes.

 

[hutchphd]

Can you make it move by breathing on it? Also this sensor geometry (this is the parallel trace device?) will be affected by the presence of nearby conductors...see if you can make it change by threatening it with a wire or your hand.

 

[sophiecentaur]

The secret of dealing with this sort of problem is to play about with absolutely everything and see what has an effect. Try to make the effect worse by changing temperature and humidity (boil a kettle nearby for instance). Try turning off the DMM between measurements. Calibrate the depth - measure C vs height of plate above water. Min and max values as water level is varied. Are you sure that Capacitance is actually that of the plate to water? etc. etc. Measure the C of a pair of plates (not water) does the value agree (near enough) with what the sums tell you?

Most of those things may not give you a clue ( you may have done some of them) but one of them may give the crucial clue. There is a great temptation, when you make something and it doesn't behave as you expect, to put it in a cupboard and not chase the problem. Zero learning experience. A half day fiddling with it as above and you will learn so much and it may even end up working!

 

[Tom.G]

Read the accuracy specs of your meter, they will likely be of the form xx% ±1 digit. A 1 digit jitter in any digital readout is quite common and expected for several reasons; i.e. battery voltage dropping, the actual input signal being very close to the transition value between steps (with random electrical noise occassionally forcing a transition), internal thermal drift in the meter... well, you get the idea.

Also that's only a 2%-of-reading, not bad at all in the real world.

EDIT: that accuracy of xx% may be either % of Reading or % of Full Scale

Cheers,
Tom