Sensor design - dielectric constant of water

Remember to use AC, otherwise you'll electrolyze the water and the measured impedance will go up.

Since normal water is not a very good insulator it will make a very leaky capacitor

All sensors I have seen use the resistance to detect presence or absence of water

Dan

-- Dan Hollands

This was done back in the early '70's with a simple metal collar around the upper radiator hose and measuring capacitance from the collar to ground.... no water means you've lost your lower capacitor plate.

I've also used a copper pipe inside PVC, then a wire cage about 0.5" outside of that, to measure (irrigation canal) water DEPTH.

...Jim Thompson

See my other post. Don't fret over the "polluted" water, consider it a conductor... make IT the "other electrode".

...Jim Thompson

"Klaus Kragelund" schreef in bericht news: snipped-for-privacy@f14g2000cwb.googlegroups.com...

Klaus,

As so often, it depends. Pure water is an insulator so the capacitive approach may work pretty well. But pure water is rare and you need only very little pollution to make a conductor. Measuring conductivity using an AC current may work better but you have to account for a wide range of conductivity values. (Unless the water has a more or less constant pollution of course but I understand from your question that this is not the case.) To use a capacitive sensor you can insulate the plate from the water. The effect can be compared with old wet electrolityc capacitors. With fluid high capacitance, without low capacitance.

petrus bitbyter

If you have to move enough water to require a pump why not a float sensor ie a magnetic float anda reed switch !.I can agree with "Exotic sensors for small volumes but for Gallons !

Yukio YANO

The current design is with a float sensor, but we are trying to optimize the costs and design of the product

Regards

Klaus

Thanks

Yes - the current sensor design is isolated from the water. So in effect the path is a capacitor with first a plastic distance before reaching the electrolytic - and then direct contact to the other electrode. We are measuring 4pF with no water and >20pF with water (1inch^2 area). But the capacitance measurement is I suspect heavily "poluted" by the conductance of the water.

Thanks

Klaus

Honeywell make some very nice liquid sensors based on total internal reflection of a IR LED. Single unit, single hole mount, no contact between the liquid and electronics.

I read in sci.electronics.design that Klaus Vestergaard Kragelund wrote (in ) about 'Sensor design - dielectric constant of water', on Tue, 20 Sep 2005:

It doesn't matter, or at least you can make it not matter. If you use a bridge to measure the capacitance, you simply balance out the resistive component. If you measure voltage and current, measure the current with a phase detector to isolate the quadrature component. Use as high a frequency as is convenient, so as to make the capacitive reactance as close to the same order of magnitude as the resistance as you can.

You don't need to reinvent this - its all been done before for low cost water level sensing.

Use a metal pipe standing vertically in the water.

Down the center of the pipe you run a teflon insulated wire which goes down the bottom, round a rod and back to the top, so that only insulation is under the water and you don't have to seal the ends. Between the wire and the pipe you have a nice capacitor with zero DC leakage. Capacitance is pretty well proportional to the water depth and is insensitive to water salinity and wire positioning. The pipe protects the whole thing nicely.

The secret is that the water acts mostly as an AC short and the teflon is the dielectric.

Don't use PVC wire - eventually the PVC leaks.

To measure the capacitance, you use a 7555 timer with the wet capacitor in the timing circuit. The pipe is the earthy side of the capacitor. The 7555 is stable with temperature. The logic output of the 7555 runs to your microprocessor.

Google will find you schematics, pictures of this sort of thing.

Roger Lascelles