Capacitive water tank level determination

[GTBecker]

I'm considering building capacitive water (both clean and black water) level indicators. I'm attracted to the notion of applying two vertical strips of self-adhering aluminum tape to the outside of the plastic tanks and using them, and the water inside, as capacitors in the resonant circuit of an oscillator - probably running at a few hundred kHz. The device would either divide down to a measured period or count up during a prescribed period to determine a value that represents the current tank level.

Anyone done anything similar?

[GTBecker]

Here's a first experiment using a pair of aluminum tape strips on the outside of a plastic bucket that's about 9" diameter at the bottom, 11" at the top and 10" deep, ~4.5 gallons.

The multimeter shows 95pf with a ~100pf cap on it. Clip leads alone shows 2pf, clipped to the strips with a dry bucket, ~11pf. Slowing adding fresh water (dyed for visibility) indicates 14pf when the level just reaches the bottom of the strips, then 36pf, 64pf, 86pf and 107pf at ~25%, ~50%, ~75% and full, respectively. Then, filled to 30pf, I dissolved a tablespoon of table salt and saw a 10% capacitance increase.

An oscillator located close to the strips should produce a pretty linear period change that follows the depth, I expect. That'll be the next step.

[GTBecker]

More:

[GTBecker]

More:

[GTBecker]

Image attachments here seem to not be ordered well, and some display while others want to become downloads. Odd.

While doing those measurements the small capacitance values became a concern. 10pf is small, indeed, and an oscillator that uses it is probably going to yield a pretty short period, perhaps an unusable one unless it's well divided down. A digital solution might not be easiest.

That leads to the notion of passing a fixed AC signal through the capacitance, instead, and measuring a resulting smoothed voltage. A processor could do that easily, I think.

How else might a 10-100pf range be evaluated?

[twesthoff]

The capacitance values are so small that most anything that gets within a few inches of the strips will affect the value.  How you measure it won't really matter unless you can control the physical environment around the metal strips.  I would try to see if you can devise a way to increase the capacitance, possibly by using multiple strips in parallel, and see what that does.  I think the width of the tape is not as important since the capacitance is primarily between the edges of the strips.  How about trying this.  Use a knife and remove several 3/8" strips vertically on the tape.  This should make each strip turn into about four smaller strips.  Now connect them in parallel, then measure again.  If you end up with 8 pairs of strips the capacitance may be eight times higher and easier to deal with.

Another idea would be to use two brass brazing rods parallel and close to each other in the liquid. Then measure the conduction of an ac signal between them.

Tom W.

On 8/16/2011 10:39 PM, General wrote:

Image attachments here seem to not be ordered well, and some display while others want to become downloads. Odd.

While doing those measurements the small capacitance values became a concern. 10pf is small, indeed, and an oscillator that uses it is probably going to yield a pretty short period, perhaps an unusable one unless it's well divided down. A digital solution might not be easiest.

That leads to the notion of passing a fixed AC signal through the capacitance, instead, and measuring a resulting smoothed voltage. A processor could do that easily, I think.

How else might a 10-100pf range be evaluated?



Tom
http://www.ustream.tv/channel/bowcam
http://www.ustream.tv/channel/cape-coral-marine-radio VHF
http://67.207.143.181/vlf9.m3u Lightning, spherics

[GTBecker]

... unless you can control the physical environment...

The applications I intend are well isolated and should remain constant, I think.

... the capacitance is primarily between the edges of the strips...

Hmmm. Not the surface-to-water area? Well then, what about gluing down a strip of 40- or 80-conductor ribbon cable and wiring it as alternating combs?

Wet probes are not an option.

[twesthoff]

The ribbon cable idea may work, but the wires may be too close together.  The thickness of the plastic bucket would affect the optimum spacing of the wires.  I am making educated guesses on most of this based on other things I have done over the years.

I also wonder how the salt content of the water may affect the readings. It appeared to on your earlier experiments.  Having something that could be auto-calibrated, or making a sensor that would measure the difference between two sensors would be more precise.  Are you looking to achieve high resolution, or just High, Mediun, and Low indications?

On 8/17/2011 7:44 AM, General wrote:

twesthoff wrote: ... unless you can control the physical environment...
The applications I intend are well isolated and should remain constant, I think.
Quote: ... the capacitance is primarily between the edges of the strips...
Hmmm. Not the surface-to-water area? Well then, what about gluing down a strip of 40- or 80-conductor ribbon cable and wiring it as alternating combs?



Tom
http://www.ustream.tv/channel/bowcam
http://www.ustream.tv/channel/cape-coral-marine-radio VHF
http://67.207.143.181/vlf9.m3u Lightning, spherics

[GTBecker]

... measur[ing] the difference between two sensors would be more precise...

How?

I'm still puzzled by your suggestion that the active area is only between edges and will test that as you suggest, Tom, by using several pairs of narrower strips in parallel.

[twesthoff]

Capacitance depends on the surface area of the conductors, the distance (spacing) between them, and the dielectric.
Normally capacitors are made of parallel plates or sheets with as much surface area as possible parallel to each other.  The strips you have only have the edges of the tape in parallel, so not very much surface area (capacitance) is obtained.  I am proposing that multiple edges in parallel will generate more capacitance.  The spacing of the strips could be made very close to each other too, for instance 1/32" or 1/16" spacing, that should help.

The dielectric primarily is the plastic bucket material, that is affected somewhat by the water.  When you added salt to the water, the water became more like a conductor and increased the effective parallel surface area of the strips, with the water acting like a third strip that covered the two tape strips.  Now that I think about it more in this way, the width of the strips would have more of an effect than I first thought.

As far as the "how" I am still thinking about that.  I have seen other differential sensors that I thought might be adaptable, but can't figure out how they might work in this situation.

I'm sorry if I don't have it all figured out.  I haven't had much time to think about it fully since I am working at my day job today.
Tom W

On 8/17/2011 11:46 AM, General wrote:

twesthoff wrote: ... measur[ing] the difference between two sensors would be more precise...
How?

I'm still puzzled by your suggestion that the active area is only between edges and will test that as you suggest, Tony, by using several pairs of narrower strips in parallel.



Tom
http://www.ustream.tv/channel/bowcam
http://www.ustream.tv/channel/cape-coral-marine-radio VHF
http://67.207.143.181/vlf9.m3u Lightning, spherics

[GTBecker]

Well, I cut out strips of foil to form eight (thus seven edge-to-edge gaps roughly the same width as that between the original two wide strips). The dry capacitance was 38pf, about four times the single gap. Adding water to the bucket raised the capacitance to 56pF just below half-full, where it suddenly skyrocketed to ~400pf. I suspected, and found, a crack at that level; it didn't leak but capillaried water between a few strips. I'll let it dry overnight, tape it and try again.

The first 50% of depth, though, remains valid and showed only an 18pf increase from empty to that depth - while the single gap (two wide strips) showed a 50pf increase over the same span. This is going the wrong way and I don't expect tomorrow's test will change that.

I'll then try doubling the widths.

[twesthoff]

Yes, I believe you will find that to be true.  After thinking about it while writing that last e-mail, I think the water is acting like a third plate, effectively putting two capacitors in series.  So surface area is the main contributor rather than dielectric as I first thought.  Will you be calibrating this at some point?  Will the salinity of the water change?  I would expect that to affect the reading.

On 8/17/2011 9:34 PM, General wrote:

Well, I cut out strips of foil to form eight (thus seven edge-to-edge gaps roughly the same width as that between the original two wide strips). The dry capacitance was 38pf, about four times the single gap. Adding water to the bucket raised the capacitance to 56pF just below half-full, where it suddenly skyrocketed to ~400pf. I suspected, and found, a crack at that level; it didn't leak but capillaried water between a few strips. I'll let it dry overnight, tape it and try again.

The first 50% of depth, though, remains valid and showed only an 18pf increase from empty to that depth - while the single gap (two wide strips) showed a 50pf increase over the same span. This is going the wrong way and I don't expect tomorrow's test will change that.

I'll then try doubling the widths.



Tom
http://www.ustream.tv/channel/bowcam
http://www.ustream.tv/channel/cape-coral-marine-radio VHF
http://67.207.143.181/vlf9.m3u Lightning, spherics





Attachments:
DSCN1539a.JPG

[GTBecker]

The crack was still a problem above ~50%, but I essentially duplicated yesterday's values: 38pf, 43pf and 54pf (empty, 25%, 50%), using eight strips with relatively narrow spacing, which extrapolates to a 32pf change, empty-to-full.

Two broader strips, well-spaced, showed much greater change - yielding 6pf, 35pf, 68pf, 92pf and 125pf for empty, 25%, 50%, 75% and full, essentially linear.

Dissolving a good deal of salt had no effect on the capacitance; I suspect that is true because the effective series resistance of the plastic dielectric is very high, making the water conductivity comparatively less significant.

Suspending a damp paper towel near the strips outside of the filled bucket had little effect. In fact, even with the bucket empty and dry, I needed to get the damp sheet within 1/2" or so to see even a few pf increase. If I adhered the damp sheet to the inside surface, though, I saw the same capacitance, ~125pf, as when the bucket was filled. That suggests that the capacitance falls off exponentially with distance or dielectric thickness, making it pretty independent of surroundings.

So, two widely-spaced 4"-wide electrodes produces a ~120pf change, empty-to-full, with this plastic material, thickness, and depth of water (about 10") - and the immediate environment doesn't affect it much.

Unlike a bucket, the two tanks I want to measure are broad or long, shallower, and the walls are thicker. To expose more tape surface to the contents, I intend to mount the strips diagonally instead of vertically as in this experiment. A cylindrical tank might even have the strips wound in a helix to maximize the response.

Now, to some electronics to quantify it.

[spamiam]

As usual, a great idea and a great experiment!

I have a couple of questions.

Will RFI be an issue? It would seem that the foil "plates" would be a great transmitter antenna! I would think that shielding would be impossible without rendering the sensor useless.

How will you be doing the measurement? At first I though you would use a bridge circuit, but then I realized that this is probably more complicated than necessary. Probably you would only need a square wave at about 100Khz with a low pass filter using the capacitor under test as the filter cap. A 100K or 200K series resistor would probably be in the right ball park. Is this generally what you had planned?

-Tony

I'm considering building capacitive water (both clean and black water) level indicators. I'm attracted to the notion of applying two vertical strips of self-adhering aluminum tape to the outside of the plastic tanks and using them, and the water inside, as capacitors in the resonant circuit of an oscillator - probably running at a few hundred kHz. The device would either divide down to a measured period or count up during a prescribed period to determine a value that represents the current tank level.

Anyone done anything similar?

[GTBecker]

... Will RFI be an issue?

It might be. I won't be surprised if I see interference from VHF transmissions (this will be implemented on the boat; the VHF antenna is about 10 feet from either tank, and can radiate ~20Watts). It might be necessary to put a few small RF chokes in series with the electrode connections; we'll see.

My first stab at reading the capacitance will be simple, a '555 astable oscillator, placed right at the electrodes. Although 10pf is below the graph on the datasheet, 10pf (~empty) and 1meg extrapolates to ~50kHz; 100pf (~full) and 1meg should be ~5kHz. Measuring the period (or counting pulses in a fixed period and inverting) should produce a value that closely corresponds to depth.