Fiber Optic very accurate water detection

[DaJMasta]

Could you coat the inside of the reservoir with a hydrophobic coating to narrow it and get more precise measurements of distance to the top?  Do you have control over the shape of the reservoir?  A capacitive sensor could be just fine if you had a flattened reservoir that was sufficiently thin, though I think that's going to be a challenge if you're really trying to get such minute changes noticed.  If you have to stick with the tube, you could string a wire through the middle (or better yet, a narrower tube) and if everything was covered in a non-conductive, hydrophobic coating, you could make a very nice little capacitive level sensor.  Have you considered floating something on top of the reservoir and using a sensor to measure the height of the float?  Could be easier to get accurate than trying to measure the water level itself.


My real question would be why you need such precise resolution - hundredths of a mL will be busy condensing on the side if you're just condensing from steam whenever it's in operation.  Would waiting for longer and measuring a larger change (tenths or individual mL) be sufficient if you interpolated and found the rate of change between viable samples, or do you need such precision in real time?  If you can take lots of samples and average them together, then some noise in your readings is fine.

If you are condensing into a vessel, then just going by weight may be entirely viable because you can physically decouple the reservoir on a load cell - it wouldn't necessarily have to be fixed in place.

[DTJ]

Maybe look at gravimetric flow measurement products made by a Jap company www.onosokki.co.jp/English/english.htm . They were very sensitive, accurate and the concept was straight forward.

I guess you've looked at the flow sensors offered by sensirion.com

[TimB100]

max_torque

Water/Condensate is flowing all the time. The flow rate may vary so I am looking to ascertain the flow rate but I need high accuracy. Its a VERY simple system to fill a tube and time how long it takes to fill the tube between 2 probes. It works, I have proven it. I see no reason to make it more complicated.

But the issue is that it's clean water with very low conductivity.

It is not impossible to detect very low conductivity water and I'm paying someone to do the R/D. He though is busy as well and not done anywork on it. And as I think its going to have issues I am looking at other systems where I can use the same proven probe system but without the conductivity issues. Hence fiber optic probes.

DaJMasta

I will probably have the inside of the tube coated in teflon but that is more to prevent drops clinging to the sided and messing with the volume.
Restricting the probes section is an obvious option and mentioned in the PDF. But I also would like some idea of real time flow rate and changes in volume at various sections will make using the pressure sensor difficult (but not impossible)
The Probe system works I need to be able to make it work with very low conductivity water. See above

"why you need such precise resolution" Without it the product does not work. I'm automating a manual system and that requires accuracy to 0.3ml which is relatively easy. but I have to have machine tolerances so aiming for <0.1ml. I can get 0.03 with probes.

Its not a condensor it is measuring condense. The condenser is separate.

DTJ
sensirion are £1000+ and would represent 200% of the aimed for BOM costs.

[Kalvin]

Although this capacitive method is not fiber optical method, this might be useful.

- The valve "out" is closed and the valve "in" is opened and the fluid will be pumped into the tube.
- The capacitive sensor will be used to measure the water level in the tube.
- When the predetermined fluid level is reached, the valve "in" will be closed and the valve "out" will be opened and the predetermined fluid volume will flow out.

The capacitive sensor is based on a precision capacitance sensor (for example, capacitance sensors from TI as LaserSteve suggested). The capacitance sensor is used as follows:

- When the tube is empty, the sensor's capacitance is very low.
- When the fluid level reaches the bottom of the sensor, the sensor's capacitance starts increasing. When the fluid level has reached this point, the pump rate can be decreased if needed. This can be use also to monitor the pump health as the fluid should reach this level in a certain time window.
- When the fluid level reaches the top of the capacitance sensor, its capacitance will stop changing: This is the detection level that the tube is full.**

Now, you just need to adjust the tube's diameter and the capacitive sensor so that the water level tolerance is sufficient for you application. You could form the tube so that it is fatter at the bottom, and it will much narrower at the top in order to get high measurement accuracy and resolution.

** This is quite different from determining the fluid level as a function of capacitance: This is not dependent of the actual sensor capacitance or absolute accuracy of the sensor capacitance, but only the change of the measured capacitance. It is very easy to measure the change of the sensor capacitance even if the fluid has different properties or the properties of the capacitance sensor changes in time.

[TimB100]

Kalvin

I have not used capacitive sensors in practice but these are all the issues I see.

1 The bore is ~13.6mm dia and I will probably have make it 2 bores in parallel at 11mm dia. It would be very difficult to place plates inside
2 There needs to be a top and bottom probe, Its hard enough with a single probe.
3 Probes that take up volume make my pressure based instantaneous flow very difficult due to varying volumes.
4 Capacitive probes only solve the detection issue but fail on the accuracy.
5 2 plates means that one may come in contact with the other. Unless its tilted then it will not and screw up any calibration.

Pumps? There are no pumps and introducing one would make the product not work.

My need is really really simple

1 A way to use a needle tipped probe to electrically detect pure water as soon as it touches the water (doubt it would work but paying someone to look into it)
2 A way to use a optical fiber based probe that would work in the same way as the probe above.

The later is what I need to test.

Thanks

Tim

[_Wim_]

Would it be possible to use a very hydrofobic air filter at the top of your tube (0.22µm pharma air filter). These filters let air pass though without any problems, but block the liquid. When the liquid hits the filter surface you get an instant pressure incease in your system that you can detect easily?

Edit: sorry, just realised water is probably only coming from the bottom in your calibration setup.

[TimB100]

Well that is a very novel and interesting approach. It could be made to work if you had a vacuum pump or sum such device and a pressure sensor.

The issue as I see it through is how you would managed the filter. The bore would be small and very difficult to attach a filter to. Also the water drop would be sucked against the tip so probably not notice the level drop.

Very neat idea though.

[_Wim_]

My original idea was a little different, but as I said, I was mislead by the picture in your pdf were I assumed the liquid was pumped in from below. If that was the case , then you could at a hydrofobic filter at the top of your tube, and when the tube is full, pressure would rise very quickly to the supply pressure. This would be the trigger to open your drain valve. The pressure sensor can be a simple low cost pressure swith, does not need to be accurate. But to work like this, you will need a sort of pump at the supply side to generate the pressure, which you would like to avoid...

[DTJ]

Good idea by Wim.


Check out Goretex filters. They do some circular disk types. Might be quite simple to install.


Vacuum pump==>=<== press sensor or switch
                             |
                             |
                    |--------------| Goretex disk
                    |                 |
                    |                 |
                    | known     |
                    | volume     |
                    |                 |
                    |                 |
                    |                 |
                    ==X===X==
                             ||  ||
                             ||   \===out valve
                             ||
                              \  === Inlet  valve
                         

• open inlet
  apply vacuum
  pressure will drop when cavity full
  close inlet valve
  open out valve
  pump air in vac port to expel volume of water

You could stick a thread in from the side to calibrate or change the set volume.

[TimB100]

So its a neat idea but for my implementation there would be a number of issues

Here are some of them

I need to time the fill so a bottom sensor is mandatory
Gortex of any filter that would be reasonably non water transmittance would not have enough airflow to allow gravity draining. I have to go to special measures to get it to empty as it is. (due to surface tension)
I have a 0-400mm h20 pressure sensor any pressure in the vessel above 400mm h20 would kill the sensor in a second.

Wim
It does fill from the bottom

[_Wim_]

"I need to time the fill so a bottom sensor is mandatory"  => I am not sure what you mean by this? The pressure swith, or some additional sensor you use ?

"Gortex of any filter that would be reasonably non water transmittance would not have enough airflow to allow gravity draining => I have to go to special measures to get it to empty as it is. (due to surface tension)" => This is for sure true. But if you use a reversable pump for the vacuum, you could use pressure to empty the container again.


"I have a 0-400mm h20 pressure sensor any pressure in the vessel above 400mm h20 would kill the sensor in a second." => Is this sensor also on the tube? Depending on the sensor type, 10 times the max range for overload is not very uncommen. To avoid very sharp pressure spikes, you could install the hydropohic filter between a soft gasket (or allow for some other type of deformation in the reservoir). This creates a little buffer/flex, making the rise time of the pressure peak slower and allowing you to detect the pressure rise, before it is to high to damage the second sensor. The flex could even be the pressure sensor itself (types with the stainless steel membrane).

[TimB100]

My main aim is to accurately get a flow value but the flow needs to be so accurate that conventional flow meters will not work. Hence I measure the time it takes to fill a fixed volume, a bit of maths and have a flow rate.

Going back to my main question on the use of fiber optics some issues I have since found out.

IR is attenuated by water but in my case the distance the IR beam has to travel is such a small distance not sure if it will be an issue
The angle of reflection to bounce IR is very low so in my tube example I doubt I would get much.

Some day soon I will get some time off to get in the shop to do some tests.

Cheers

Tim

[meeder]

What is the accuracy needed? Do you need temperature compensation as well since the volume is dependent on the temperature as well.

Edit: what is the estimated flow rate?

[meeder]

Tim, is 0,05% accurate enough? If so than you can use a flowmeter.

[TimB100]

meeder

Flow is around 8-16ml per min

I am aware of flow meters that work on doppler effect using ultrasonics but they start at £800 for the low end. I have gone up to £400 units but they failed to be reliable and bubbles in the flow cause issues.

I need ~0.1% accuracy.

[meeder]

Air inclusion is indeed a major problem for a flowmeter. The accuracy isn't an issue since Coriolis type massflow meters can reach 0,05% but they come at a cost. The air however will be a problem for any flowmeter no matter what the technology is.

[EmmanuelFaure]

Tim, by any chance is this an academic project?

[TimB100]

EmmanuelFaure

No its for a future product I want to develop. I have my own company and it will earn me money.

Hence why I pay people to do work for me.

Cheers

[bitslice]

At the level I'm working surface tension is a VERY big issue

Would adding a tiny amount of surfactant aid in achieving a more accurate measurement?

A concave surface sounds a bit more repeatable than a convex one, which I suspect is dependant slightly on the surface of the tube?
Just throwing it out there as you've looked at everything else.

[TimB100]

I cannot inject surfactants continually into the feed.If I could I would find a way to contaminate the feed to make it more conductive and solve my issues using my existing probes.

Surface tension is an issue but not insurmountable. Just have to remember its there.