Flow Logger

[kollinsb]

Does anyone have suggestions on a liquid flow sensor? I have a watering system in my yard and would like to monitor this. I had though a reed switch may do the trick but would like expert opinions. All I am looking for her is the sensor and will worry about the interface/logger later. I would love to be able to use I2C or better yet, the 1-wire interface (I already have a 1-Wire weather station)

Any suggestions?

 

[Andrew Cowan]

Do you want a digital reading (flow/no flow), or an analogue reading (32 litres/second)?

A

 

[QuIcK]

from my knowledge flow rate is something extremely difficult to measure.
oil companies spend million on R&D for this technolegy.
solutions range from accurate temperature sensors that detect the friction cause by the fluid, propellors that are turned by the fluid, pressure caused by a neck (known as the venturi method), to introducing microscopic amounts of tracable fluid, and very sensitive detectors to monitor their flow allong pipes.

for a fluid, the most obvious is the venturi method. the only problem is it constricts the flow in order to generate 2 pressures. if thats a problem, then you may be able to use the propellor method

http://en.wikipedia.org/wiki/Flowmeter

 

[papaof2]

The least expensive calibrated flowmeter I've found is the one used in the Whirlpool Duet clother washer. It delivers 815 pulses/gallon. They can sometimes be found on Ebay for under $20US. The flowmeter consists of an impeller with a magnet in it and a reed switch on the outside of the plastic housing.

I've measured flow (from a 20 gallons/minute pump) with the PICAXE interrupt capability and sending the data via serout, using an 08M running at 8MHz.

Test code from a year ago (because it's on the PC I'm using today).

'FlowMeter test code
'Uses flowmeter sensor from Whirlpool Duet washer
'Sensor gives 815 pulses/gallon
'Pulse speed depends on water pressure.  For my applications, 
'pressure will typically be less than 3psi, so count speed should 
'not be a problem..
'Copyright 2008 John E. Carter
'Initial code 6 March 2008


#picaxe 08m
#simspeed 10

symbol pulsecount = w4	'will be reset after each 815 pulses
symbol gallons = w5			'max tank size is 250 gal, but counting in 0.1 gallons
symbol doupdate = b2		'update display if set
symbol waiting = w6						'if pulse input stuck low

'interrupt for count
setint %00000000, %00001000	'interrupt when input 3 low

pulsecount = 0
gallons = 0
doupdate = 0


setfreq M8						'faster to catch pulses from flowmeter

		
main:

if pin4 = 0 then
	sertxd(" ", #pulsecount)
end if



	if doupdate = 1 then
'		'debug
	sertxd(" ", #gallons)
		doupdate = 0
		low 2
	end if
	goto main
interrupt:	
if waiting > 0 then
	waiting = 0
	setint %00000000, %00001000	'input 3 low
	return
end if
pulsecount = pulsecount + 1
do while pin3 = 0			'check for pulsecounter running slow or stuck low
	waiting = waiting +1
	if waiting > 0 then
		setint %00001000, %00001000	'input 3 high
	end if
loop
	if pulsecount = 81 then		'815 pulses for 1 gallon through meter - use 1 for simulation testing
	gallons = gallons + 1		'this is a count of gallons
'														'display functions handle decimal point
		doupdate = 1		
		high 2
		pulsecount = 0					'clear count
	'sertxd (" ",#gallons)
	end if
	setint %00000000, %00001000	'input 3 low
	return

John

 

[Jeremy Leach]

This might sound crazy - but I expect you could have a small feed from your watering system directed into a tipping-bucket rain gauge. Make it really small bore, so only a little water is fed into the bucket. The rate of filling the bucket should be pretty much proportional to the flow rate to your sprinklers etc I'd think?? Might need to calibrate and produce graph. Just an idea. A tipping bucket rain gauge is quite easy to make at minimal cost and could just have an 08M counting tips. At least you don't have to get some device actually in the main flow too.

 

[eclectic]

I love the problems on this Forum.

Papa John's “Whirlpool washer sensor” idea looks great for continuous flow.
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&Item=280341078264&Category=20714&_trkparms=algo=LVI&its=I&otn=2

Jeremy's “emptying” idea look perfect for total usage.

Might I add one more, based on Jeremy's idea?

Build an “Autosiphon” cistern, with a floating magnet, in a tube.
When it's empty, the magnet closes a reed switch.
The Picaxe counts the ”clicks”

e

 

[boriz]

Idea: Use a DS18B20 one wire temperature sensor with a coil of nickel-chromium resistance wire (Nichrome wire) wrapped around it as a heater. Drive the coil using a MOSFET and PWM. Use a closed loop feedback program that measures the temperature and if it is below a certain ‘key’ value, it increases the duty, and if it’s above a certain value it decreases the duty.

Immerse this sensor in the water flow. The faster the water moves, the higher the duty will need to be in order to maintain the key temperature. The water takes heat away from the sensor at a rate proportional to its flow speed. Though it’s unlikely to be a linear relationship, a little calibration would give you a good repeatable system.

For greater accuracy, use another unheated sensor also immersed in the flow (upstream of the heated sensor) in order to subtract the effects of variations in ambient/water temperature.

 

[Andrew Cowan]

This might sound crazy - but I expect you could have a small feed from your watering system directed into a tipping-bucket rain gauge. Make it really small bore, so only a little water is fed into the bucket. The rate of filling the bucket should be pretty much proportional to the flow rate to your sprinklers etc I'd think??

I think it would be proportional to the pressure in the hose.

No attachment on hoze = low pressure (but high flow)
Hoze blocked = high pressure (but no flow).

However, it could be calibrated for a certain device, eg a sprinkler.

A

 

[eclectic]

I think we're in Theory-land at the moment.

We need some more input from kollinsb,
to refine the requirements.

e

 

[QuIcK]

There is an industrial method like the one that boriz described.
its 2 sensors surrounding a heater:
---sensor ----- heater ----- sensor --
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
-------------------------------------

the flow "drags" the heat downstream. the temperature between the downsream sensor and upstream sensor is proportional to the flow rate.
the heater is kept constant, and you also have a control temperature of the water going in.


and if i read his whole post... i would see he mentioned this

 

[demonicpicaxeguy]

depending on your manufacturing capabilities the easiest way i've ever built a flow meter

a little imagination and you should be able to see how it works
*to adjust the measuring range you adjust the lenth and amount of rubber bands as well as the dismeter of the polymorph cyclinder
*also the red square inside the polymorph block is a rareearth magnet proferably a cylinder shaped one

 

[lanternfish]

This may be a good starting point - http://www.jaycar.co.nz/productView.asp?ID=ZD1200&keywords=flow&form=KEYWORD. There is a pdf giving more info. You may find something similar where you live. Cheers

 

[boriz]

A simpler approach: Two thermistors, one with a NiChrome coil, arranged as a divider so that any temperature changes that are the same for both just cancel out. Circuit:

As you can see, any increase or decrease in the resistance of BOTH thermistors will have no effect on the ADC level, which should always read roughly half of Vcc (2.5v) with no heater current. Now if you raise the heater PWM duty sufficiently to raise the ADC reading say 1v (3.5v) and hold it there, any flow will take away heat and require higher duty. Just keep the ADC at 3.5v and the flow rate is proportional to the duty.

 

[boriz]

LOL. I forgot the thermistor symbol. Rough scetch, but you get the idea.

 

[chipset]

i love this... you guys have figured out multple ways to measure fluid and not one of you know exactly what this person actually needs to measure lol. Its funny but also goes to show how creative minds start rolling when given a problem to solve. Im the same way and if this had to do with air flow and/or pressure from either air or fluid ID have quite a few suggestions myself. Without knowing it (or maybe you do) quite a few of the suggestions if lined up in order of occurance chronicles automotive mass air meters used for calculated airflow into an engine. Starting with the wheel flowmeter just like early 80s toyotas that used a spring loaded flapper door that moved in proportion to air flow all the way up to the idea of the heated resistor that calculated based on how hot the wire got. That is what would be called a heated film mass air meter used widely in fords and most bosch meters as well.

 

[boriz]

“...you guys have figured out multple ways to measure fluid and not one of you know exactly what this person actually needs to measure lol…” ???

Post #1 … Flow sensor … Water … Yard watering system …

 

[chipset]

total flow or gal/hr or is it hr/lt? It doesnt really say... A simple level sensor would tell you how much had moved through the system however if you wanted to know how much flow was occurring at any one time that requires something completely different. Think cars. A fuel gauge tells you that the tank is half full and you know your tank hold 14 gallons, then total flow was 7 gallons. If you wanted gal/hr though it would take a bit more to do it. Even then a clock and a level sensor would do it. Are they watering from city water and need a meter to know how much has gone through or is it from a tank?

What Im getting at is the variable to be measured is very important in regards to what meter you use. Maybe I missed it but I didnt see anything other than flow and garden.

 

[Jeremy Leach]

Yep the requirements are lacking but it doesn't stop us evolving ideas I feel like getting some thermistors already! and maybe the requirements should be made to fit our solutions !

 

[demonicpicaxeguy]

kollinsb, if by any chance you're in Australia , i have a domestic pulse output water meter thats free to a good home, the turbine still works but the counter has long since had it, however all it needs is a reed switch or hall effect sensor and you've got a fairly accurate flow meter pm me for details if interested

 

[kollinsb]

Wow.... Thanks for all the input. This is my application:

I have a sprinkler system with a backflow. I would like to measure the gallons used by the sprinkler system (gallons per hour, day, month, and year). Also, I'd though about using a reed switch to count the pulses from the water meter but not confident that will be accurate.

There are a lot of interesting ideas here. I'm not opposed to manufacturing anything within reason. For me it's all about doing it "myself" but I am not opposed to buying something prefab.

I don't want to re-invent the wheel when there's one that fits my application that is priced less than $100 or so. I would need to log the data to store it somehow (csv file or database)

Thank you for all the input! Gotta love this forum!

I'm near Nashvilel Tennessee by the way. Wouldn't be opposed to paying for the shipping if that's doable.

 

[AlC]

I'd go for "re-inventing the wheel" in form of some propeller with a little magnet inside the pipe ("flow indicators" that indicate flow in general are quite cheap for example) and a reed switch on the outside.

Then count the rpm.

or you could use a couple of fixed flow rate valves together with several solenoid valves connected in a "Y" way, to have "digital flow control" (can even "charlieplex" many solenoids on few signal lines if desired), for example ten steps with each step = additional 20ml/sec = 0 to 200ml/sec...

(solenoid valves are expensive, sadly)

 

[demonicpicaxeguy]

http://www.builditsolar.com/References/Measurements/DIYFlowMeter.htm

here's some food for thought, it's just a case of getting an analog or digital output

 

[Jeremy Leach]

Yes, that is interesting - but the big problem is 'how' to get the digital output at reasonable cost. Considering the pressure difference isn't much at all, whatever you use would need to be very sensitive. Hmmm.

 

[demonicpicaxeguy]

i'd maybe go for a floating magnet inside a tube and a hall effect sensor kind of setup,

 

[Dippy]

Some great ideas.
I was also after a flow sensor for central heating fluid.
Obv it would have to be magnetic as optical may not be very happy with water+Fernox.

I looked at pukka turbine magnetic ones - these seem popular at the lower/medium end of price range.
Wouldn't be able to use a magnet on elastic band - can't see that lasting too long.
Degradation/temp changes would cause the elastic properties to change and eventually fail, but its a neat principle nonetheless.

In my app I couldn't use the thermal method with any confidence, though I see that it is used quite commonly.

I like the propellor inside pipe with Hall sensors. Though I suspect by the time you've made one that can withstand the long-term rigors then you may as well spend £70 on a proper one which is made from tough materials with good bearings.

Wasn't there a sensor that used good old Bernoulli or something similar?

 

[boriz]

Another idea. Completely theoretical and untested.

A piezo disk with an ordinary (dressmakers) pin glued to the centre, protruding 90 degrees from the plane of the disk.

A piece of string/cord/nylon, a few CM long glued to the tip.

The string is suspended in the water flow and ‘flaps around’.

The amplitude and frequency of the flapping, sensed by the disk, can be used to derive the flow rate.

Indeed, the string might be omitted. The pin itself might produce sufficient turbulence for approximate flow metering.

 

[boriz]

Another...

A pinch in the pipe with a microphone to pick up the hiss of the expanding water just downstream of the pinch. Audio filtering at specific frequencies might produce useful results.

 

[boriz]

Another...

Generate a precise ultrasonic frequency and use a transducer to vibrate the water.

Downstream, pick up the sound, amplify and compare with the original signal.

Doppler shift caused by the adding of the speed of the water to the speed of the sound could be detected and used to provide a, possibly quite linear, flow rate reading.

This might even be made to work externally, just by contact with the pipe.

 

[boriz]

Another...

Two ultrasonic transducers, say 1 meter apart. Ping transducer 1, time the arrival of the pulse at transducer 2. Ping transducer 2, time the arrival of the pulse at transducer 1. Compare. Repeat.

With no movement of the water, the signals will take exactly the same time to travel both ways. With movement however, the signal will move faster downstream that it will upstream. One travels at speed of sound PLUS speed of water, the other at speed of sound MINUS speed of water.

 

[boriz]

Another...

A pinhole in the pipe would produce a drip rate proportional to the pressure. Count the drips.

 

[demonicpicaxeguy]

this is quite doable with the ultrasonic transducers, the resolution might be a little problematic i'd want to get the flow rate up nice and fast, although i'd be tempted to look at modifying a "pic ultrasonic range finder" project since most of the hard work has been done

 

[Dippy]

Some more great ideas.

Count the drips ... no comment.

How will you couple the U/sonic transducers?
You can't just strap a standard module to the pipe as the largest signal will probably come from the pipe itself.

I guess Boriz's idea (#29) could use transducers which are within the pipe or pentrate the pipe but accoustically isolated from the pipe - unless you have some DSP for looking at secondary pulses?

There are some nice ideas here:
http://www.omega.com/techref/flowcontrol.html

I'm sure there are many other sources of information.

I like the Doppler idea myself - gives me an excuse to dig out my FFT on dsPIC

 

[kollinsb]

Some great ideas here. I'll do some research and testing on the ones within my skillset.

I've subscribed to the thread. Any breakthroughs or light bulbs above anyones head is still appreciated.

 

[boriz]

Nice link Dippy. Particularly like the Coriolis meter idea.

 

[demonicpicaxeguy]

kollinsb,
here the picture of the water meter i mentioned

just finished cleaning and testing it , it was removed from a property about 2 years ago the reason listed for this one was "periodic replacement", and i got hold of it for prototyping purposes, i'm going to take it down the shop later tonight and get a proper weight

the specs http://www.relianceworldwide.com.au/files/spec_sheets/endurance.pdf
claim 1 pulse /100L but i think it's different for this particular one

the downside is that it's all in litres, and i can't remember the pulse output timing which is another problem, i think it's every 100ml or 1litre, they are all different the weight is aprox 1.5kg to 2kg which is around $40AU odd probably less to send it to to states through the snail mail

 

[Jeremy Leach]

Ok, here's one I've had, see attached pic.

Tube mounted vertically. Cone kept in place by gravity. Water flows upwards and raises cone proportionally(ish) to flow rate. Ferrite rod attached to cone. Rod moves within coil (on outside of tube). Inductance of coil varies and could be part of an oscillator circuit and freq varies with flow rate.

?

 

[AlC]

kinda like a reversed solenoid - clever!

 

[Dippy]

Look halfway down that link I gave Jer.
The Big Boys use a taper - obviously someone has done some work on this.

Looks feasible.

Anyone going to make one and have a go?

 

[Jeremy Leach]

Ok, I think I might build a DIY flowmeter myself, so I'm putting more thought into how to make a practical prototype! Cones aren't easy, so here's an alternative attatched. It might look complex but I don't think it is really and should be buildable ...

I'm thinking of using the pressure difference principle in the manometer example above, and having two connections off the main pipe (and the main pipe has a variation in bore diameter to create the pressure difference). This pressure difference creates a flow in a much smaller tube. This is Tube A in the diagram, which I'm even thinking could be a plastic drinking straw. Just thinking through how to make it ('Blue Peter' style!) ...

Construction
1. Get plastic straw and cut small slits as shown. The 'escape' slit should be nice and uniform if possible. The 'return' slit isn't critical and could be a hole, but must be positioned above maximum travel of rod.

2. At the bottom of the 'escape' slit insert and glue a suitable 'collar' into the straw. This just acts as a rest for the vertical rod when there isn't any flow so as long as it works it isn't critical.

3. Hopefully find a ferrite rod that can fit into the straw with minimal clearance. This might not be easy !! Maybe it just needs to be metal?? Might need to waterproof rod somehow to stop rust.

4. Wind coil. The whole priniciple of varying inductor is a bit of a guess. Makes sense that it will work best when the gap between rod and coil is small, so thin-walled TubeA and close fitting rod is best. Maybe magnet and hall effect sensor better ?? Cost??

5. Get another larger diameter TubeB that fits over TubeA. Not critical. The whole purpose of this is just to allow flow from the 'escape' to 'return' slits. Maybe just seal TubeB to TubeA with glue. The wires from the coil will have to come out somewhere, maybe through this glue.

6. Job done.

Operation - theory!
As flow rate increases, the rod lifts in the tube and changes the inductance of the coil. The water entering TubeA escapes through the 'escape' slit into TubeB, then returns through the 'return' slit back into TubeA.

As mentioned before, this variable inductor is part of an oscillator circuit and varies the frequency. The picaxe uses the Count command to measure the number of pulses per second. This method has natural inbuilt averaging.


@ DPG: you seem to know a bit about Hall effect sensors ...do you think it would be better to measure linear displacement using one? Are they fairly cheap??

 

[boriz]

What a great thread. Lovin it. You put my quick-n-dirty sketches to shame

One more idea:

A QTC pill squashed by the cone-like device indicated above.