Water Water Everywhere or It Makes No Sense.

[itsmoked]

I need to sense water level in a small tank with tap water.

I set out to detect it via conduction. I capacitively excite a probe with 1kHz 24V pulses. I look at the same probe by charging a capacitor via a simple diode rectifier. If the probe touches the water in the grounded metal tank the DC pulse train is supposed to diminish dramatically. The loss should appear on the sensing cap and the comparitor should signal the loss.

It's not working.. I have reduced the caps from 2.2uF clear down to 1000pf. At 1000pf the comparator trips every pulse for about 1/3 the period. It still doesn't definitively trip.

By "caps" I'm referring to both C25 and C27 at the same time.
The comparator's trip point is 4V.
The comparator is running the LED you see.
TP28 Has the 1kHz pulse train on it. This is cycling the drive comparator at the top.

http://www.box.net/shared/static/bopujnd0xu.png

Suggestions?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

Uh oh. Remember the 'World's Longest Eng-Tips Thread' about measuring the depth of water in a well? Here we go again... ;-)

Are your probe wires insulated from the water? I seem to recall a circuit that used capacitance to measure water height (similar) and the wires were (IIRC) fully insulated from the water (including the tips).

 

[benta]

Tap water using conduction?
My take on this is that tap water conduction varies immensely from location to location and also over time.

Wouldn't it be better to measure reactance, given that water has a dielectricity constant of around 80?

Just my 2 cents.

Benta.

 

[MacGyverS2000]

I was thinking the same thing, VE ;-)

Time to call up Polaroid and get one of their ultrasonic sensors

Dan - Owner

http://www.Hi-TecDesigns.com

 

[MikeHalloran]

GEMS level sensors; floats with magnets, reed switches.
Inexpensive, and reliable as mechanical devices go.

Retroreflective sensor looking at the flat end of a pencil-shaped piece of plexi will sense the difference btw air and water at the point. "Delco Eye".

Honeywell makes something commercially that sort of does the same thing through a penetration in the bottom of the tank using a transparent hemisphere. It's British, so follow the app notes exactly or it won't work.

If the tank is transparent, you can do the critical angle thing through the wall with no actual penetrations. It's simple enough to do with discretes; no one is doing it commercially.




Mike Halloran
Pembroke Pines, FL, USA

 

[btrueblood]

Not quite seeing the reason for back-to-back diodes to ground (D4)...won't those swamp anything the probe is doing?

 

[btrueblood]

Sorry, smoked, sent that last too soon.

What are the metals used for the probe and tank, and can you measure an emf from probe to ground with probe decoupled from circuit, and/or with power off to circuit. Idea being dissimilar metals will produce a battery-like emf, and it can fubar low voltage signals. 24 v should be enough, but those diodes in the path may not play nice, dunno.

 

[itsmoked]

Yes I'm bit concerned about wild conductivity variations.

The dielectric change method is probably the better way.. Dang. What do dielectric type probes look like? Just bare rods near each other? Or are they insulated to exclude water conductivity?

This is a pressure vessel. In fact a steam generator too. So it runs above atmosphere. So floats are out. The really cool total-internal-reflection eyes (Delco like) are out. Reed switches, etc., etc., are all out! No ultrasonics either.

btrue; These are standard probes that are stainless rods that are slightly bent. By twisting the rod you adjust the water level since it will touch the water at different points depending on its position. The tanks are stainless.

I am able to use a DMM to measure a resistance when I hook it from the probe to the tank.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[geekEE]

With pulsed DC, won't you end up with electrolysis occurring? Do you see any bubbles forming on the electrodes that could affect the current flow?

Glenn

 

[ScottyUK]

Floats are definitely an option, they are used in plenty of LP boiler applications up to about 20 Bar or so. How much room have you got though - some of them are kinda chunky.

What about a DP cell arranged top to bottom? If you're careful with the impulse lines that would work fine, it's done often enough with things like deaerators.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[itsmoked]

geekEE; It was thought that capacitively coupling the probe to the excitation would provide an average. But actually it's not. The pulses are all unidirectional. The cap charges up..


Scotty. The tank is six inches diameter and eight inches long, laying on its side. Not a lot of room for floats and if I could get them in there there wouldn't be a lot of room for the water!

Keith Cress
kcress -

http://www.flaminsystems.com

 

[ScottyUK]

Keith,


Google [blue]fossil aquarian[/blue] for some examples of probes used in industrial boilers. They're not enormous. Given the high purity of industrial boiler feedwater I believe these are capacitive rather than resistive although someone else might refute this.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[itsmoked]

Thanks Scotty.

Expensive stuff that!

I just got it working! I switched C27 and R38 to 10M Ohm resistors. C25 went to 1000pF.

This creates a voltage divider that gives about 12V to the input of the sense comparator. Water across the input drops that voltage well below the trigger 4V. That includes reverse osmosis water.

To head off electrolysis I excite to check, for only 3ms a second.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[btrueblood]

"To head off electrolysis I excite to check, for only 3ms a second."

Yup, it will help. You may still get some plating over time, and that may affect the overall path resistance, but it's slow and manageable, and descaling the boiler periodically should return things to the design condition.

We have a 15 foot water tank inside the building now, with 1 foot clearance between top and ceiling. Conductance depth probes, level checked for 10 seconds, once per hour. Home-built probes (1/8" brass wire fed thru 1/4" o.d. poly or PFA tubing, and the pair fed thru a Swagelok compression fitting. Works like a charm.

 

[zzzjim]

Looking at possible failure modes
Quoting Btrueblood
" descaling the boiler periodically should return things to the design condition. "

Are there any consequences for detecting FAKE Emptyness ?

The solution looks good ie Few moving parts to Fur up !

 

[itsmoked]

Well like the usual boiler there's a thermal/pressure relief valve. So if everything goes pear shaped the water starts flowing down the drain.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[btrueblood]

...assuming you remembered to hook up the drain pipe, otherwise it's a big puddle on the floor.

 

[zzzjim]

A wet and dry hoover can help with these teething problems ,

and sometimes help defy gravity !

 

[waross]

There are commercial well level probes on the market that use a transformer putting out about 300 volts and a relay.
The combination of low current and AC avoids electrolysis. Cheap easy and dependable in suitable applications.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[itsmoked]

I got it going. I took a DMM and cranked up the Ohms range. Then I "probed" tap water, filtered water, and reverse osmosis water. All of them were below 1.2M ohms. So I altered the above frontend slightly to be an ohmmeter feeding a comparator. I ditched the whole AC angle as a waste of time because I only turn on the excitation for 1ms every once and a while.

Let it try to plate out... Mwahahahaha!!

Keith Cress
kcress -

http://www.flaminsystems.com