WOOT! I got me a flowmeter!

[Alchemy]

Originally posted by bigben2k
But I still need a cheap build guide for a manometer. Google time. The pressure gauges are starting to look better.

Few scientific instruments are easier to make. You could probably do well with an oil manometer. You could probably build one for less than $10. Hell, I could make full instructions on design and use if it was going to go on the site.

The only danger - and in fact, a major one - is blowing it. Cleaning out your entire system to remove indicator fluid = not fun.

Alchemy

[bigben2k]

Given the pressure range I'd like to have, mercury would be far better,so I think I'll drop it all in favor of this Gizmo (at Since87's suggestion).

All you do is
-amplify the output of the sensor.
-Run the amplified output into a square root circuit.
-Run the output of the square root circuit into a voltage to frequency converter.
-Run the output of the voltage to frequency converter into the RPM input for the CPU HSF.
-Make sure all of your scaling is right.

When Motherboard Monitor tells you that your fan RPM is 1422, you know that actually means that your flowrate is 1.422 gpm. (+/- whatever)

More planning to do.

[bigben2k]

Using a VCO (Voltage Control Oscillator, PDF, 24 pages, see page 1), is half the circuit.

Now to convert the input voltage to ^2...

[BillA]

ready for another lesson, eh ?
ok

the output is proportional to the excitation voltage, so you need a 'stiff' source
and the factory cal is at 10 VDC (I would guess 10.00)
. . . .

the DPs we're dealing with are pretty small (for an appropriatly sized unit),
are the pump pressure fluctuations going to be hopping around ?
if so, how will you 'smooth' them ?
. . . .

[bigben2k]

Hum... Maxim's MAX1452 might do it...

[bigben2k]

Originally posted by unregistered
ready for another lesson, eh ?
ok

the output is proportional to the excitation voltage, so you need a 'stiff' source
and the factory cal is at 10 VDC (I would guess 10.00)
. . . .

the DPs we're dealing with are pretty small (for an appropriatly sized unit),
are the pump pressure fluctuations going to be hopping around ?
if so, how will you 'smooth' them ?
. . . .

Since87's idea is to convert the signal into RPM, which would be translated into 1'000 rpm per gpm.

I think I ought to start with a straight pressure reading, where 1'000 rpm equals 1 psi.

Stabilizing a 10 Vdc supply isn't a problem, but I did make a note of it, thanks.

I fully expect fluctuations in the flow (and as a result, the pressure). I'm going over starting conditions (i.e. pressure bursts) specs/recommendations. The response time will be critical, but the Maxim chip seems pretty responsive, at 3.2 kHz.

[JSimmons]

I wonder if the water traveling through a copper tube would setup some sort of measurable resonance that changes in frequency depending of flow and/or pressure...

Hmmmm...

Or maybe sound waves transmitted through moving water are displaced in the direction of the flow, or take longer to flow upstream with an increase in downstream flow. Sound also travels faster through cold water than through hot water, so there may be something to look at there as well.

Or how about magnetic field deviation based on flow...

None of these methods would inhibit flow or pressure, and the hardware could be mounted on the outside of a clear vinyl tube. I'm not an engineer by any stretch of the imagination (but I did stay at a Holiday Inn Express last night), but I'm happy to throw a grenade like this in the room and then walk out.



P.S. If anyone does come up with something based on one of these ideas, I have to insist that you call it the "TurkeyMonkey Flow Meter", in honor of my avatar.

[bigben2k]

Keep walking... or identify the purpose of a "pressure snubber".

[Since87]

Originally posted by bigben2k
Using a VCO (Voltage Control Oscillator, PDF, 24 pages, see page 1), is half the circuit.

Now to convert the input voltage to ^2...

You are starting out on the wrong end. Forget about the VCO.

You should be looking at sensors at this point. There's no reason to think that the one I linked is anywhere close to the best choice. There's definitely not enough technical data on that Omega web page. (Omega doesn't want you to know how to use the sensor itself. They want you to buy a box to hook up to it.)

What accuracy are you going for?

How closely do you think your heatercore will match one that Bill has tested? Or, do you plan to calibrate it / have it calibrated?

How much of your error budget will be eaten up by the uncertainty of the heatercore calibration?

How much by the dp sensor? How much by the amplifier? Etc...

Things to look for in sensors:

Full scale that is only slightly above the maximum dp you expect to apply.

Availability of detailed specs and ideally applications notes.

Relatively high output voltage. The higher the sensor output voltage, the lower the signal to noise ratio coming out of your amplifier stage. Sensors with preamps built in are available and can save on additional circuitry. Extra cost for a sensor that has an internal reference and already amplified output is probably money well spent.

Good accuracy at an excitation voltage around 7V. (Because I can send you an LM399 voltage reference which has an output around 7V. We cherry pick them to tighter tolerance than the manufacturer, so we've got a thousand or so 'dropouts'.) Maintaining 10V +/- 5mV is not that easy. The LM399 is overkill though.

I'm sure there are other things that I haven't mentioned, but it's definitely worth searching through what's out there with these issues in mind.

WRT "square rooter":

http://www.analog.com/UploadedFiles/...471AD532_c.pdf

[bigben2k]

Actually, I think I'm going to concentrate on the Maxim chip for now: it's got a built-in temperature compensator. This means that I'm going for a pressure reading, not a flow reading. I can add flow later, in a permanent installation.

The Maxim chip has a number of DACs built-in for correction, and is programmable, to some extent. The VCO would follow the Maxim chip.

As for the pressure drop measured, if I don't exceed 1.5 gpm, +/- 1 psi would do fine (for heatercores only), but if I later make a permanent installation of it across my pump, even 5 psi won't be enough.

Hum... what to do, what to do...

At +/- 1 psi, the output will spread between 0 and 16.7 mV, which is actually more sensitive than the 50 mV spread (aka "span") on the 5 psi unit.

I could then later adapt this sensor for a flow reading, taking a reading across the HC... and if/when I remove the heatercore, I can add a simple little restriction, to have something to measure.

+/- 1 psi sounds good.

[Since87]

Originally posted by bigben2k

Actually, I think I'm going to concentrate on the Maxim chip for now:

Ok, good luck.

Effectively, what you are telling me is, "I want to build a car. Oooh, look a subwoofer. I'm going to concentrate on this subwoofer."

Have fun.

[bigben2k]

Hum... not sure what you're saying...

I understand what the sensor can do, what it's limits are, what type of output it has, and I've even spec'ed how it's to be mounted.

The next logical step, it seems, is to start interfacing it, and that's where the Maxim chip comes in.

If I want to ^2 anything, I can do it after the Maxim chip.

Then add the VCO, and patch it in the CPU RPM sensor.

No?:shrug:

[pHaestus]

Start by considering how to power it at 10.00VDC?

Start by finding detailed technical specs for sensor?

This is all going in a case still, right?

[bigben2k]

I found the specs: it's a Honeywell part.

The p/n for the 5 psi differential model is 26PCBFA6D. That's what I went over, this morning.

http://catalog.sensing.honeywell.com...D&FAM=Pressure

It's all there.

I can use a simple diode to drop the +12 PC supply down to 10. Easy.

[bigben2k]

(finally) got a reply from Blue-White, as I asked them about the pressure drop, and the required length of tubing before and after the meter.

Pressure Drop For Model F-44500LH-8:

.5 GPM = .12 lbs
5.0 GPM = .25 lbs.

Please access our tech support website www.blue-white.com to view detailed specifications for your model.

On variable area flowmeters certain before and after straight lengths of pipe run is not required to ensure accuracy.
The meter must be protected from "water hammer" and or fast opening valves. Pleas view the installation instructions
on the tech support site for more details.

so I asked them for for an actual graph. We'll see.

I also e-mailed the Sales dept, 'cause I figured that the tech dept was too busy:

Thank you for your inquiry. Please visit our product support web site at www.blue-white.com for additional specification details on the Blue-White F-440 series flow meter. The pressure drop at the low scale reading of 0.5 GPM is .12 PSI, and at the high scale reading of 5.0 GPM, the pressure drop is .25 PSI. These pressure drop ratings are approximate values obtained using bourdon tube type pressure gauge, 0-30 PSI.



For additional assistance in the Houston area, please contact our distributor:
Ryan Herco Products
Phone (713) 670-0300
Fax (713) 670-0308

A bit more helpful, but not much.


The good news is that the pressure drop at 5 gpm is only 0.25 psi, which is equal to 7 inches of water (@39.2oF, 4oC).

As I suspected, it's not a bad flowmeter.

Now if I could apply Since87's equation, I could draft up my own curve...

[bigben2k]

Originally posted by bigben2k
Now if I could apply Since87's equation, I could draft up my own curve...

Given the numbers, and using dP=Q^2 * Rf, where
dP = Pressure drop
Q=flow
Rf = {flow resistance}

and

dP = 0.12 psi @ 0.5 gpm
dP = 0.25 psi @ 5.0 gpm

I get 0.48 and 0.01 for Rf, respectively.:shrug:

[Since87]

Originally posted by bigben2k

I get 0.48 and 0.01 for Rf, respectively.:shrug:

Pretty obvious it wouldn't match that equation.

If you want it to match that equation, glue the 'slider' in place.

Wait, maybe not such a good idea...

[BillA]

http://cgi.ebay.com/ws/eBayISAPI.dll...category=11809
of its type, the best I have seen - and at a bargan price (though other methods are superior)

as is true of most things, one gets what is paid for
aka: you can't make a silk purse from a sow's ear

Since87
the numbers are the goal, not their value or relevance

[brucoman]

I apoligize for coming into this thread late, but...

That Blue White flowmeter is a poor example of what is available on the market. Just because it's cheap doesn't imply it is suitable for your use. As for history, I have personally removed ~30 Blue White (Orange is another name to stay away from) flowmeters in the last 4 years due to breakage and warpage. If you trully want accuracy as well as low/no pressure drop you need to spend/borrow/steal a magnetic flowmeter, with a nominal size equal to your tubing ID. THere are quite a few available, just be patient.

[bigben2k]

Thanks for the tip.

Going over all the related docs, I could tell that the housing was going to need special attention, and that it should be treated as fragile.

As for the accuracy, I have to take it for what it's claimed, but after Bill's constructive comments, and a massive lack of info from BW, I might as well treat it as having a 10% error margin, that increases over time.

However, I can still use it to give me some idea of actual flowrates, and it's an improvement from having nothing, or even thinking about a bucket test.

Homebrew/garage science?

[brucoman]

I do not believe in ballpark numbers as they can be very lisleading. A more accurate measurement then that Blue White would be a simple litre cal column on the pump intake. We do it as a backup on our industrial chemical pumping skids even with mag flowmeters.

Example: Put a T at the pump intake (preferrable with at least a foot between T & intake to reduce turblance at the pump) with a FULL port ball valve on each side with 1 going to cal column, other to original intake line. Since it's also a closed loop, most likely need another T setup on return side to collect liquid.

BTW, I do like your pump choice.... don't see enough Little Giant pumps on the forums.

[bigben2k]

I did a flow test on the heatercore yesterday, and here are my results.

First, the setup: I stacked a ball valve and couple of straight fittings on top of the pump, set the flowmeter, then the (new 3/4) airtrap, then a 3/4 barb. I used a 1/2 barb at the pump inlet. With the lines connected, I stuck a couple of tees, at the core inlet and outlet, and attached some extra 1/2" lines to measure the pressure drop.

Notes: There was still air going through the system, so the gpm readings are probably unreliable. Finding the right level for the whole contraption (pump vs core) proved difficult, and limited my tests up to 1.5 gpm, and I still managed to make a mess!

Accuracy: I wouldn't trust the pressure readings beyond +/- 1/4", which isn't too bad. The flow rate accuracy is somewhat mysterious, because of the air going through, but the flowmeter and pump manufacturer info might help define that margin.

Sources of error: air in the water (distilled), which appeared to come from somewhere within the loop.

Procedure: set the valve to obtain 0.5, 1.0 then 1.5 gpm, and measure the pressure drop. Lastly, remove the manometer, and run a "full pin" test.

Pic later.


Reference data: dP of flowmeter @ 0.5 gpm: 0.12 psi (3.3" water). dP@ 5.0 gpm: .25 psi (7" water).

Results:
dP of heatercore @ 0.5 gpm: 1.5" (0.054 psi)
dP of heatercore @ 1.0 gpm: 4.0" (0.14 psi)
dP of heatercore @ 1.5 gpm: 8.5" (0.31 psi)

"Full pin" test: 3.5 gpm

Interpretation: next post.

[bigben2k]

According to the pump's curve, the dP @ 3.5 gpm should be 11.2', or 134 inches (4.8 psi). Some of it will be within the loop, some from the flowmeter, and more from the airtrap.

Guesstimating, I'd put 0.2 psi at the flowmeter, 0.12 for the airtrap (calculated), leaving 4.48 psi for the core, but including tubing and fitting restrictions.

I believe that the results are close to what they should be.

[bigben2k]

...and here, finally, is the picture:

[bigben2k]

attached, this time: