From: Bruce Murphy
Subject: Re: Water sensor
Date: 10 Jan 2003 21:16:17 +0800
Organization: The University of Western Australia
X-Newsreader: Gnus v5.5/XEmacs 20.4 - "Emerald"
> ah, i get it, so you want to know if the condensate pump has failed and the
> air con unit is leaking ?
That's the one, although I was also thinking of converting this for another
application that involved sensing various smallish water volumes simply, by
the positioning of point sensors at appropriate graduated levels up the
side of a pipe or vessel.
> If thats the case, i have a few tips for you.
> You can use a couple of probes to test for conductivity of water between
> them. However, you must use stainless steel, and the signal between them
> must be ac coupled, or else you will get electrolysis and the probes will
> An easy way is to generate a logic level square wave, and feed it into one
> end of a small capacitor (about 0.47uF), the other end of which is connected
> to one of the probes.
> Then all you have to do is look for an ac signal to be present on the other
> probe. If there is, there is water between the probes.
I'm not sure of the exact conductivity of water, particularly that
resulting from air-con semi-distillation, but I'd imagine it wouldn't
be very high. I can imagine a system like this being prone to picking
up stray signal from the other probe's feed wire. Ideally I'd like the
probes to be as simple as possible and on the end of relatively long
> Another way is to use a capacitive sensor, but these can be quite prone to
> problems with dirt buildup.
> You could also use a TDR system. Have an exposed, un-insulated pair of
> conductors running round the areas you wish to monitor, terminated at the
> end with the correct impedence for the cable used.
> At the other end of the cable, transmit pulses, and look for reflections.
> You can work out where the water is from the reflections.
This seems a very elegant solution apart from the small difficulty of
obtaining known-impedance wire lengths that contain uninsulated but
separated conductors without having to butcher something yourself, and
figure out impedance afterwards.
However, I think that the electronics required to implement something like
this lie a little beyond the simple input stage to microcontroller that
I had envisioned for this project.
Was there a fourth option? :)