Volumetric measurement of liquid in transparent container

[speechsupply]

I'm working on a way to measure the amount of water in a container.
The container is transparent, has no lid, and is made of glass or plastic. It holds 3 liters = 101 oz and stands on a tabletop mostly.

But, due to various challenges, I would prefer to measure the volume from beneath. ( To avoid dipping a MPXx2051 or float into the water)
Using a strain gauge is def. a viable option but I would prefer to have a sealed enclosure with no external moving parts.
Requirements for accuracy and precision are quite low. If I have to present actually numbers, I would say that a resolution of around 100 ml. with a +- 10 ml accuracy would suffice.


Possible solutions are: ** Updated **
A. Use a Sharp GP2Y0A41 or similar and hopefully manage to detect the water level. Will try this within the next couple of weeks.
B. Capactive sensing on the outside of the container. This solutions violates the desire to only monitor from beneath, but could be a preferable solution to weight cells. // Thanks to: forrestc, and larsdenmark for the suggestion.
C. Container resonance? Is this viable, also in plastic containers?
D. ideas?

Let me hear what you think

[EmmanuelFaure]

What are your requirements in term of accuracy and cost?

[jpanhalt]

Several years ago a participant on another forum tried to measure the surface distance to water with that device.   She had trouble getting good reflection without adding something to the water.  Maybe your experience will be different.

I did a very crude experiment with a TOF laser distance measuring device and bounced off the bottom of the container.  Since the speed of light is considerably different in air versus water, given a fixed distance above the bottom, I could estimate the depth of water.

[larsdenmark]

Knowing the required accuracy and response time would help a lot to direct the idea generation. Is the container connected to anything while the measurement takes place?

Weighing the container would be the obvious thing to do.

You could also shine a laser on the surface of the liquid at an angle and detect where the beam hits a detector on the bottom.

Try to measure the capacitance between two electrodes that are connected to the side of the container and see if you can correlate that to the water level.

You could use the resonance of the container (that will depend on the level of water), but you'll probably need some special (crystal) glass to get a good resonance. Also if the resonance is within the audible spectrum then this will be rather annoying to listen to.

[DTJ]

Maybe looks as the products made by Onno Sokki who make some accurate fuel flow instruments and see how they do it.
At least some of them use a strain gauge.

https://www.onosokki.co.jp/English/english.htm

[JS]

You don't nees moving parts, you don't need to measure the weight of the tank, just the preassure at the bottom, a pressure sensor connected to the bottom is enough, if you can poke a hole in the bottom great, if not a hose comming to the top with some gizmo to keep it full with air or water, if filled with watter or a hole on the bottom the sensor must be at the bottom level or you will get an offset wich you could correct for later, but the hose should have a soft diaphragm on the opening so it's allways filled with the same thing. Or sumerge the sensor, depending on reliability would be easier to pot the sensor and let it lived sunmerged.

JS

[DTJ]

If you could read from above the sharp distance sensor could read the distance to a float on the surface.

[speechsupply]

What are your requirements in term of accuracy and cost?

Requirements for accuracy and precision are quite low. If I have to present actually numbers, I would say that a resolution of around 100 ml. with a +- 10 ml accuracy would suffice.

I did a very crude experiment with a TOF laser distance measuring device and bounced off the bottom of the container.  Since the speed of light is considerably different in air versus water, given a fixed distance above the bottom, I could estimate the depth of water.

I was hoping to "reverse" your setup, shining from the bottom up, relying on the water-to-air density difference.. Not sure it will work. Waiting for parts 

Try to measure the capacitance between two electrodes that are connected to the side of the container and see if you can correlate that to the water level.

Yes! that could be a viable way - this sensor looks quite interesting:  http://www.ti.com/lit/ug/tidu736a/tidu736a.pdf 
 Only problem is, that I'll be moving up the side of the container.

You could use the resonance of the container (that will depend on the level of water), but you'll probably need some special (crystal) glass to get a good resonance. Also if the resonance is within the audible spectrum then this will be rather annoying to listen to.

 
Like a wine glass  - changing its pitch with the amount of volume. Perhaps a https://www.sparkfun.com/products/10917 surface transducer and a piezo pick-up could be used?

You don't nees moving parts, you don't need to measure the weight of the tank, just the preassure at the bottom, a pressure sensor connected to the bottom is enough, if you can poke a hole in the bottom great, if not a hose comming to the top with some gizmo to keep it full with air or water, if filled with watter or a hole on the bottom the sensor must be at the bottom level or you will get an offset wich you could correct for later, but the hose should have a soft diaphragm on the opening so it's allways filled with the same thing. Or sumerge the sensor, depending on reliability would be easier to pot the sensor and let it lived sunmerged.

JS

That is a way, most definitely. Unfortunately I would have to submerge or break through the container which is not an option ATM.

[forrestc]

I'm working on a way to measure the amount of water in a container.
The container is transparent, has no lid, and is made of glass or plastic. It holds 3 liters = 101 oz and stands on a tabletop mostly.

But, due to various challenge, I would prefer to measure the volume from beneath. ( To avoid dipping a MPXx2051 or float into the water)
Using a strain gauge is def. a viable option but I would prefer to have a sealed enclosure with no external moving parts.

I don't see anyone who has suggested this (sorry if I missed it):

The best way to accomplish what you are looking for is with a capacitive sensor.   It can be applied to the outside of the container and will reliably detect the height of the fluid inside the container.   There are numerous appnotes out there about how to do this.  I know in particular microchip had one. 

Searching, I found:  https://www.hackster.io/team-protocentral/non-contact-capacitive-liquid-level-sensing-using-fdc1004-9333c7  which is a good description of one process.      Microchip AN1375 describes another very briefly in section 6.

The summary of all of these methods is to arrange some wire/traces such that the level of the liquid changes the capacitance between the traces.   Measuring the capacitance between the wires will then give you the height of the liquid.

[speechsupply]

The best way to accomplish what you are looking for is with a capacitive sensor.   It can be applied to the outside of the container and will reliably detect the height of the fluid inside the container.   There are numerous appnotes out there about how to do this.  I know in particular microchip had one. 

Searching, I found:  https://www.hackster.io/team-protocentral/non-contact-capacitive-liquid-level-sensing-using-fdc1004-9333c7  which is a good description of one process.      Microchip AN1375 describes another very briefly in section 6.

The summary of all of these methods is to arrange some wire/traces such that the level of the liquid changes the capacitance between the traces.   Measuring the capacitance between the wires will then give you the height of the liquid.

I did link to the TI application note in an earlier post, but I hadn't found that from hackster.io.  Really great resource, thanks! I'll make sure to update the original post

[JS]

Usually specs are the other way around, so you have a few less significant digits of uncertainty. That being precision being smaller than accuaracy.

JS

[jaredwolff]

I've used capacitive in the past. Spaced out the pads per units of volume so I knew, approximately, how many mL were removed. Capacitive though becomes a pain when any other elements are involved (like your hands!) (This was, by the way, using a flexible PCB not something completely at the bottom of the cup)

I'd also suggest (if it hasn't been already), especially if you're working from the bottom, using ultrasonic. If you have a homogeneous mixture (water or water like) ultrasonic works like a charm.

Jared

[Berni]

Since you are aiming for pretty low resolution of 100ml on a 3 liter container it would make sense to glue a flex PCB with capacitive pads on the side of the container. One could then use a MCU with plenty pins to measure the capacitance on 30 such pads and determine at what pad the capacitance suddenly jumps up. Additional sloped pads between the existing pads could be used to narrow i down the resolution further if needed.

But usually the easy way of sensing level in a open tank is to simply screw a pressure sensor into the bottom (Provided you know the density of the liquid will always be the same)

[tpowell1830]

Some clothes washing machines have an open tube that goes to the bottom of the washing tank so that when you rotate the load size switch to different positions, the machine can sense how much water is in the tank. I have not examined one of these circuits or setups very closely, but there would be varying air pressure in the tube as the water rises, so I am assuming it has a pressure sensor at the top of the tube. Makes sense...

Hope this helps...

[DTJ]

Some clothes washing machines have an open tube that goes to the bottom of the washing tank so that when you rotate the load size switch to different positions, the machine can sense how much water is in the tank. I have not examined one of these circuits or setups very closely, but there would be varying air pressure in the tube as the water rises, so I am assuming it has a pressure sensor at the top of the tube. Makes sense...

Hope this helps...

I've seen industrial washing machines use this arrangement with an MPX style pressure sensor.

[Berni]

Some clothes washing machines have an open tube that goes to the bottom of the washing tank so that when you rotate the load size switch to different positions, the machine can sense how much water is in the tank. I have not examined one of these circuits or setups very closely, but there would be varying air pressure in the tube as the water rises, so I am assuming it has a pressure sensor at the top of the tube. Makes sense...

Hope this helps...

Actually yeah i remember washing machines from the old days already using pressure to determine water level. It was just a round pancake shaped device with a small tube running from the bottom of it into the bottom of the drum. When the water level got to certain levels the pressure caused switches inside to close and that signaled the mechanical program controller to advance to the next step in the program.

[max_torque]

Why are you not just weighing the tank??  Simple, robust, sealed, repeatable.  You have to be mad (or have a VERY good reason) to do any other method!

[pwlps]

I have recently built a capacitance probe to measure the level of liquid nitrogen in a dewar. But I have tested it with water too and it works fine.

I did not try the FDC1004 but I see two potential problems you should consider:

1) The capacitance sensor described in the documentation of FDC1004 uses bulky electrodes an the calibration seems to be extremely sensitive to any stray fields or small changes in geometry. I wanted to have a simple probe that I can insert through a small hole in the dewar and also a portable system i.e. a probe that I can withdraw and insert again without loosing the calibration. Therefore I'm using a simple coaxial capacitor made of two concentric stainless tubes (6mm/3mm diameters), with the inner tube held with teflon spacers. The coaxial geometry makes the capacitor completely shielded from the outside environment (the outer tube is connected to the circuit ground). There are a few small holes in the outer tube so that the liquid level inside the capacitor can adjust quickly to the level in the dewar.  The calibration is very easy: the change of capacitance is proportional to the liquid volume fraction and the dielectric constant of the liquid.


2) Beware of simple capacitance meters which are sensitive to the resistive part of the impedance : if there is a leak in the capacitance then you would need a true LCR meter to extract the capacitive part   (of course the constraints depend on the capacitance range, for my application it is of order of tens of pF).  The  commercial capacitive level meters I found for nitrogen are using simple circuits with either low frequency excitation or  time-of-discharge measurement and are rather sensitive to leaks, they usually specify that the leak should not be bigger than  10 MOhms or so. This is not always possible to achieve in an environment where the electrodes can be polluted by  moisture near the top of the dewar, some of commercial probes fail to measure anything if there are are water droplets making a leak of order of hundreds kOhms, this is why I designed my own circuit. 
The documentation of FDC1004 is not very clear about this point, as it only says:

"The FDC1004 can be used to measure liquid level in non-conductive containers."  (sec. 9.1.1)

What about the liquid itself? A pure non-conducting liquid does not exist, water can be considered as conducting or non conducting dependiong on the resistance range you are concerned with. 

My circuit (I can send you the schematics if you are interested) is a capacitance half-bridge working at 2MHz and using a phase-sensitive detection to discard most of the resistive part, it is insensitive to leaks down to a few kOhms. The circuit is in a small box mechanically attached to the top of the coaxial capacitor to suppress any stray capacitances.

Actually for water you might not need such precautions because its huge dielectric constant will give you much higher capacitances, so that simpler circuits might do the job (but I didn't do any numerical estimations).

[speechsupply]

Why are you not just weighing the tank??  Simple, robust, sealed, repeatable.  You have to be mad (or have a VERY good reason) to do any other method!

Yeah it's absolutely a solution. The reason why I want to avoid weighing is primarily due to the "mechanical" aspects of building a sealed enclosure with weight cells. Tried it before, not impossible, but once someone other than me starts throwing this stuff around...

Some clothes washing machines have an open tube that goes to the bottom of the washing tank so that when you rotate the load size switch to different positions, the machine can sense how much water is in the tank. I have not examined one of these circuits or setups very closely, but there would be varying air pressure in the tube as the water rises, so I am assuming it has a pressure sensor at the top of the tube. Makes sense...
Hope this helps...

I've seen industrial washing machines use this arrangement with an MPX style pressure sensor.

Absolutely, but due to contamination I would prefer to avoid direct contact with liquids within the container.

I have recently built a capacitance probe to measure the level of liquid nitrogen in a dewar. But I have tested it with water too and it works fine.

Have you shared this build on any website? Would love to see some pictures of it

[jpanhalt]

Absolutely, but due to contamination I would prefer to avoid direct contact with liquids within the container.

Sorry, but that doesn't make sense.   The "container" is made of something.  You can make your probe out of the same material.   How will a probe made of the came material as the container add contamination that the container doesn't add?

How did your experiments with the Sharp detector go?

[speechsupply]

Absolutely, but due to contamination I would prefer to avoid direct contact with liquids within the container.

Sorry, but that doesn't make sense.   The "container" is made of something.  You can make your probe out of the same material.   How will a probe made of the came material as the container add contamination that the container doesn't add?

How did your experiments with the Sharp detector go?

Sorry for the radio silence, vacation and various other projects got in the way.

- In essence you're right. However, the liquid in the tank could be water, orange juice, gatorade, etc.. So everyday, it has to be cleaned after use.
I'm thinking, let me know if you disagree, that having a float, tube or similar internal measuring device will be difficult to clean properly.

The Sharp detector is actually looking promising. BUT only with water... cloudy substances like milk, juice, etc. is obviously different. I expected this, but seeing that water somewhat works, is actually quite nice.

[Marco]

Yeah it's absolutely a solution. The reason why I want to avoid weighing is primarily due to the "mechanical" aspects of building a sealed enclosure with weight cells. Tried it before, not impossible, but once someone other than me starts throwing this stuff around...

You can buy presoldered, potted strain gauges ... what kind of abuse are you expecting? Do they randomly hit stuff with a hammer? Throw fuming nitric acid onto it for a laugh?

[larsdenmark]

I think capacitive sensing wins:

Non-contact capacitive liquid level sensing using FDC1004 (ProtoCentral)


Ask an Expert - How can a capacitive sensor be used in a liquid application?


Liquid-level sensing using the FDC1004 EVM



Using a pressure sensor is also possible:
Prototype Board - Pressure Sensor as Liquid Level Meter

but it is invasive, which is not what you want.


I think any acoustical or mechanical exploitation of resonanses or time of flight will be fairly complicated and expensive (unless you only need to know if the container is full or not). If the properties of the liquids differ a lot you probably need to make a lot of experimentation (or modelling) in order to make things work.

[Marco]

I don't see how the capacitance sensor can be accurate for arbitrary liquids at low levels (not that weight measurement works for arbitrary liquids of course). The reference sensor needs to be fully covered to be able to compensate the signal from the level sensor AFAICS.

[speechsupply]

I agree, it could change with the varying liquids. However, I have a pretty strict selection of liquids, that will go into the container. Mainly consumable (by humans) liquids.
I have the FDC1004 EVM kit now, I'll share results as I go along.