AVR Freaks

This is very similar to what we did. We used the
mpx5010gp.

Tom Pappano
Tulsa, Oklahoma

Good find, the capacitive sensor is a cool technique, didn't think of that one. The trick for you will be in building a double walled tube, so that the capacitive plates stay dry. It shouldn't be too hard.

Create the sensor tube, as they show in the article, except leave about a centemeter clearance at each end of the tube (this will be for mounting).

For the outer tube & support, you will need a larger diameter tube roughly 2cm shorter than the sensor tube, and 2 end-caps, for support. Drill holes in the center of the 2 end caps large enough to snugly accept the sensor tube.

Next you will need some epoxy to assemble, and seal this thing. Get the slow drying kind, so you have time to work. Start with the bottom cap flat on the table (opening up), and insert the sensor tube. next fill the end cap about half full of epoxy, to anchor, and seal the sensor tube in place. Next (before the epoxy hardens) insert the outer tube into the end cap, insert it in as far as possible. finally to keep the tubes in alignment, place the top cap on, to hold the sensor tube in place. Leave the assembly in the vertical position for 24hrs for the epoxy to fully cure.

Test the tube to make sure water does not leak into the inner chamber that we created. Test it by plunging it vertically down into water to a depth a few cm from the top of the tube, and holding it there for about 10 minutes (the longer the better).

If it does leak, try adding some silicone around the end cap joints at the bottom of the tube. You can also try adding more epoxy inside.

Note that if you're liucky and can find reducing couplings for the size of your sensor tube to your outer tube, use those, and the proper adhesive for the type of tubing you are using, instead of epoxy. This will give you a more reliable result than the drilled end-caps.

In AN27, the water is passed into the sensor tube from a flexible hose (look at the diagram, the water goes from the bucket into the sensor via a flexible hose, the water never contacts the exterior of the sensor tube), the sensor tube is not not submerged in the bucket. The technique works when the sensor and water container are at the same level. In AN27, they had the convenience for measuring the water that is in a separate small container. In your case, the sensor is inside the container (the cellar) so you need to place the sensor itself in another container to keep it dry.

The plates need to be kept dry, because impurities in the water, will create a short between the plates, thus killing the capacitive effect the sensor is based on.

Hi,

with wet electrodes: you can measure the change in capacitance with the level of water.
To avoid destroying of your electrodes i recommend to decouple the DC par with (much larger) capacitors.

One simple way could be to use a Port as an opencollector output to discahrge one capacitance and use the internal pullup to charge the capacitance. connect the capacitance to the analog comparator input (is it possible to simply use just the analog comparator input?).
Measure the time when the capacitor is charged to the certain voltage level. The time tells you the water level.

maybe you can find some CrNi18/8 (INOX) material for the electrodes.

Ken think about a capacitor. The plates are separated by an insulator, not a conductor. The capacitance of the plates is determined by their surface area, and the dielectric constant of the material that separates them. You get a change in capacitance because the water partially fills the void between the plates, while the rest is air. (the tubing remains constant over the entire surface, and is therefore not mentioned, but it does have an effect on the overall base capacitance) The two plates act as an infinite series of small capacitors in parallel, as the water moves up, the capacitance of the virtual capacitors separated by water changes from what it was with air, to what it is with water. while the top portion remains unchanged. The result is a change in the overall capacitance of the sensor, relative to the depth of the water.

Water does not need salt to be conductive. Any number of impurities in the water can increase it's conductivity. As it'll be ground water that is seeping in, it will be full of various metal minerals thus increasing it's conductivity, just as the sodium in salt water does. Obviously the higher the impurity content, the higher the conductivity. (actually pure water is a semi-conductor, otherwise you would never be able to break it down into hydrogen, and oxygen via electricity alone, so I'm not even convinced wet electrodes in pure water would work as desired)

If you put a conductive solution in contact between the plates, it will bleed off the charge, ruining the capacitive effect. At that point you need to measure resistance, not capacitance. So, as you can see, you need an insulator separating the plates for this to work. (note that you can still have wet plates, they just need to be conformally coated, or similar, to keep the water form contacting the conductive plate directly.

As you can see, using wet electrodes is not likely going to work as desired. For wet electrodes you'd be better off reading resistance, not capacitance, which, essentially, is what the original idea was to do. However, unless the resistance of the water at that time is known, you can only test for the presence of water, not it's depth (with only 2 electrodes)

If measuring for resistance, you have another problem the resistance measurement would die off, as the electrodes got coated with other non-conductive impurities from the water, reducing their effectiveness. Oxidization will also have a negative effect.

Again, in AN27 the electrodes are dry, they are taped onto the outside of a 1/4 inch diameter tube. The water runs on the inside of that tube.

[afterthought]
The dielectric constant of the water is going to change as well with the varying impurities. So this will have a negative effect on the accuracy of the measurement, unless you can calibrate the sensor to the dielectric constant of the water being measured. I have no idea how dramatically the dielectric constant will vary, and how dramatically it effects this type of sensor. Should make for an interesting experiment. Measure with distilled water, and then again with a high salt content.

There are many level controllers using ultrasonic measurement. You can try hacking some commercial ones, or develop your own. Google can give you many schematics that can be easily attached to an AVR instead of the usual PIC.

Ultrasonic level measurement for liquid works really nice, since I have one at home (AKO-53190)that controls the water level in my swimming pool from a height of about 4m with a precision about 1 cm.

Guillem.

You can do this remotely by using a bubbler made from small air pump and a length of plastic tubing. If one end of the tube is a the floor of the basement, the other end can be in the dry somewhere. Just pump air into the tube and measure the pressure of the air in the tube with a pressure sensor. The air pressure will vary linearly with the depth of the water. you can read it directly with a gauge of you like.

if you do the pressure method, make sure the setup is such that the column of air to be compressed will always be the same, no matter how fast the water runs in, so the bottom of the tube needs to be vertical, NOT lying on the floor.

Also you will need to take temperature into account, as pressure is greatly affected by temperature.

Each technique has it's own advantages and pitfalls.

Good luck with it! It should be a fun project.

Using the electrodes wet, you no longer have a capacitor. And since I was talking about measuring voltage drop with DC voltages, resistance was the correct term. Had I been talking about the drop with respect to a particular frequency, then yes, impedance would have been the correct term to use.

Now where's the fun in that? :D

Neil :P

True, Neil, very true !
Bingo, topics like these belong in the off-topic: they're too much fun :wink:

Now serious: I once did the capacitive one (with the rising water as dielectric between two plates) and found TWO effects: capacitance-change, but also: increase of dielectric losses. And yes, we did get it to work. But it's not a simple solution.

This one IS:

The idea with the PTC's is great, but the snag is always: how to keep the water out of the electronics.

When I was a student, I made a level-measuring system for the Big reservoir of a central heating system, and used CMOS and four electrodes (silver) on a PVC-pipe in the reservoir. The reservoir itself was grounded. It worked, but I had to go to the attick every month to rub-off the insulating oxide or whatever the stuff was ... it felt a bit gelly.
Yes, it was fun .... but it didn't work very well.
Students huh ?

Nard

This is getting very complicated. Why not just write a simple C program. Have a variable called
WaterDepth, and declare it as a float.