Help w Water Flow Sensor

[Jack Van Olst]

Help w Water Flow Sensor?

I need help with a design problem. I need to get an analog signal
from 200 water pipes that I can use to estimate flow rate. I have a
great analog-digital system from B&B Electronics that will give 12 bit
accuracy from a 0-5 volt signal for only about $6 per channel, but I
cannot decide what type of sensor to use. Forgive the length of this
question, but I am trying to get ideas from anyone that has experience
in this field. My biggest problem is COST - I cannot use traditional
technology, because we don1t have the $$$.

The flows are in 82 diam horizontal PVC pipes. The most common way of
doing this is to read water velocity in a full pipe with aSignet style
paddle wheel sensor, and convert the frequency of the pulses from the
prop to water velocity, and convert that to flow. This solution is
too expensive.

All the other methods we have thought of depend on using a rectangular
weir cap on the end of the pipe, and calculating flow from a simple
non-linear formula relating water depth behind the weir to flow rate.
This method works great visually - we can engrave a 0-8 inch depth
scale next to the weir opening, and read depth and GPM within 5%
accuracy - fine for our needs.

But computerizing this depth measurement has proved difficult.
Commercial solutions are usually sonic transducers mounted above the
water surface - too expensive for us to buy 200 of them. One
possibility is to buy consumer style SONIN sonic distance meters for
$36 each and try to gut them to get a useful signal - anyone tried
this?

A second commercial method is to read conductivity, but our water is
too variable to allow this.

Another commercial solution is a 3bubbler2 style water pressure gauge,
in which air is bubbled through a small vertical tube and the back
pressure that develops is converted to depth. The commercial versions
are too expensive. I built a system with a $60 Data Instruments 0-1
psi pressure transducer. It works great, except it is hard to set the
bubble rate so that it won1t be too rapid for shallow depths, or too
low for deeper water. Perhaps there is a cheap constant-flow valve
for air systems? Like a pinhole?

All other ideas involve some method of directly measuring the 0-8 inch
depth by using a float on the water surface. Here are some ideas:

1) Linear position potentiometer (82) - $200
2) VDO Marine fuel tank sender - $18
3) VDO tube type fuel sender - $46, but the 3pot2 is just a thin
strand of wire - very delicate
4) VDO Boat trim plate sensor - $60, but the pot is too stiff
5) Wirewound adjustable power resistor with a sliding tap - $18, but
the linkage to get a tap to slide along the 82 coil is awkward

All these methods would give a variable resistance via a pot linked to
the float - the electronics would be simple. But they all mean I
would have to build a lot of mechanical linkage gear to link the float
to the pot with shafts, etc. I don1t have a machine shop, and it would
be expensive to have the linkage fabricated commercially. Also,
each apparatus would need a large dustproof cover box to protect the
sensitive pot and linkage from the environment.

6) Muti-turn pot & retracting cable. UniMeasure in Oregon makes a
cute $88 5-turn pot with a retracting cable I could hook to the float.
This is a tidy solution, but more money than we would like to spend.
An alternative is to try to gut a Starrett consumer electronic tape
measure ($30) to get a usable signal - anyone tried this?

7) Buy high-resistance wire and build a linear pot from scratch.
Anyone have a source for high resistance wire? Comments?

Then come the 3high-tech2 concepts:

8) Build a metal detector circuit, put some metal on the float, and
try to get a proportional signal with distance.
9) Use a Hall Effect transducer , put a magnet on the float - but
will Hall Sensors work 8-102 away?
10) Move a magnet in an 82 fixed coil, and measure the change in
voltage due to disturbing the coil field - possible?

11) Strain gauge - One last idea is to have an external chamber near
the pipe and let water fill it through a flexible tube. Then weigh
the chamber with a strain gauge. Comments?

Again, sorry for the length of this letter. Any comments would be
appreciated.

Jack Van Olst

[Loren Wilton]

1) How about a mass flow sensor? Two thermosensors, one with a heater
element. Measure the water temp upstream with the first, and maintain the
second sensor 20 degrees or so above the temp of the first. The energy
required for this should be proportional to flow. Should be able to do the
whole thing for about 10 bucks a point with a PIC 16C74 microprocessor, a
resistor, and a pair of thermistors. Steal the entire circuit and control
logic from the Microchip FuzzyTech Explorer demo board.

2) A standard restricted-oriface pressure differential flow detector. Mark's
handbook can give you the basic equations for the pressure drop across a
restrictor vs flow rate, viscosity, etc. You need a pair of pressure
detectors or one differential detector.

3) A presure sensor on the bottom of the wier measuring the weight of the
water passing over the weir. This will be directly proportional to the height
of the water at the weir, so proportional to flow. The drawback may be
accuracy, since you will be working with a fractional pound of pressure.
Still, this might be a good choice if you can find a cheap 0..1 psig sensor.
I'm not quite sure what effect flow will have on the measured pressure, since
it would probably depend on the shape of the measuring orifice. Probably
easiest to build one and calibrate it emperically. May also have to
compensate for water density for best accuracy.

4) A velocity sensor using a paddle wheel and one of the sensors from a $4
mouse. This will give you an RS-232 output rather than analog, but that might
be even cheaper to process. Some mice can operate in a polled mode, which
could simplify the datacomm connections. If the pipes happen to be reasonably
close together, or even in pairs, you might be able to get two signals from a
single hunk of mouse electronics. Depending on how continuously you need
data, you might be able to do this with a single RS-232 channel and a 200
output 5V 10ma digital I/O board. Power up a mouse, read a couple of reports,
subtract the difference for flow rate, power the mouse down, go on to the next
pair of pipes. About half a second a pair.

5) The mouse encoder tied to the pivot of your tank float. The problem here
is that the mouse encoder is relative rather than absolute, so to do a
measurement you have to cause the float to move from a known position to the
current water level. This probably requires a small dc motor or a solinoid or
some such.