cdr

I am trying to make a liquid conductivity sensor. Basically I tried to use a simple two electrode device to measure the resistance/conductivity but I have to much stray current as well as a ground probe in the tank of liquid. Both of which cause erroneous readings in my sensor.

I thought an induction based sensor would work better. Can anyone tell me how to make one, or suggest a better method?

go

Russlk

Why do you want to know the resistance of the liquid? That may have some impact on the way you measure it. The resistance of the liquid will no doubt vary with frequency, so that is a parameter you will have to evaluate. Do you have a liquid of known resistance as a standard? How do you know your measurments are in error? I think you are best advised to find out why there are stray currents and eliminate or compensate for them. Is the ground a third electrode? There should be only two electrodes, one of which can be ground, and the liquid has to be in an insulated container.

__________________
see my website: www.geocities.com/russlk

cdr

I am actually using a 555 timer circuit to generate a frequency based on the resistance of the liquid. I count the frequency and convert it to conductivity with an equation and a temperature compensation factor.

The stray currents are mainly caused by leakage from submerged pumps and other devices in the tank. Because of this we had to install a ground probe on the tank, the third probe. This probe is not associated with the conductivity probe, but seems to be causing the erroneous readings in the conductivity probe. If I take readings with the sensor in the tank I get one value, then if I remove a sample from the tank and put it in a seperate isoleted container I get a valid reading.

Roff

The one web site I saw on induction based sensors led me to believe that conductivity needs to be much higher (lower resistance) than the range you are expecting.
The diagram below is an idea which might solve your problem. It might still have problems - depends on how the ground shield is interacting with your circuit.
The idea is to prevent current flow between ground and the timing capacitor. The shield is kept at the same voltage as the cap. Current will still flow between the shield and pin 3 of the 555, and possibly from the shield to tank GND, but this should not cause a problem if your lowest resistance is indeed 10k, or possibly as low as 2k.

Edit:
It dawned on me after I posted this that you might get away with just having one probe down the center, from the capacitor, and driving the shield with pin 3.
Of course, you might not be able to put any sort of shield in your fish tank. Maybe it could be made of stainless steel fine mesh screen?

Attached Images

conductivity_probe.gif (10.5 KB, 265 views)

cdr

That sound like it will work.

I can make a shield out of stainless steel and slide it between my probe housing and the electrodes. Here's what I can do to my existing probe prototype, do you think this will work correctly....

Attached Images

schem_212.jpg (44.3 KB, 244 views)

Roff

I described two ways of shielding the probe(s). Your drawing implements neither of them. I don't think it will work to tie the shield to +5V. Current will flow from +5v, through the liquid, and to the capacitor, screwing up the bias.

cdr

I see now. I forgot all about the second option of using the capacitor line as a probe and pin 3 to the shield. I'll give bot of your suggestions a try.

Thanks

cdr

I tried the shielding ideas, I still get different readings with the probe in the tank versus the probe outside the tank in a sample of the tank water.

I made two shields, one was a thin sheet of stainless steel with a stainless steel mesh soldered to one end. The other was just a a cylinder of the mesh. Both were connected to pin 3 of the 555.

Also, when I dipped the probe in the tank it saturated my pH probe reading to the max.

Any other suggestion? Did I do something incorrect?

go

Roff

Did you actually try the version with the op amp? It was the one I had the most hope for.

I have one last gasp - see below. This doesn't require a shield. It relies on the fact that one probe is at virtual ground, so no current should flow to the tank ground. If your resistance from either probe to tank gnd is less than about 10k, this probably won't work - nor will the other two.

It is very important that circuit GND and tank GND be at exactly the same potential - two "grounds" are not acceptable.

Attached Images

conductivity_tester1_sch.gif (8.6 KB, 205 views)