Sensor to detect drops of liquid

whirlymark

I am looking for a sensor that can detect drops or a small stream·of liquid.· It is a scientific application where a vertical syringe squirts a liquid·into a small glass vial.· I need to detect if the liquid is squirted into the vial, because if not then the syringe is blocked.· The output of the sensor would be interfaced to the instrument to stop it running (using of course a Stamp).

Details:
The amount of liquid is no more than a drop (10 microliters), so it is not possible to use microswitch or weigh-based sensor.· The liquid may be water or a solvent so it is not opaque. A force sensor is not sensitive enough. The glass vial is 1cm in diameter so the sensor needs to of a simialr size.

So I am currently thinking about a Photo Interrupter type sensor or an optical sensor with an IR beam and detector.· But I am not sure if such sensors are sensitive enough to detect a short burst of liquid between the beam and detector. Does anyone know if they are that sensitive or have used them for a similar application?

Any other sensors I could look into?

Phil Pilgrim (PhiPi)

How much can the horizontal position of the falling droplet vary? Does it always fall directly into the center of the vial? Off to one side? Does it ever run down the side instead of falling through the air? How far above the sensing position is the dispensing point? What is the diameter of a typical droplet? These will help to determine an appropriate solution.

-Phil

ElectricAye

Just a crazy idea....

If the syringe can be electrically isolated from the vial, how about putting an electrostatic charge on the syringe/fluid and just below the outlet of the syringe have a conductive ring that is connected to something like an electronic electrometer circuit. As the charged drops pass through the ring, would it create a signal on the ring that could be amplified, etc.??? You might need some sort of electrostatic screen between the ring and the syringe to play with the electric field. If such a thing could work for drops, then I suppose a stream might be detected, too, so long as the vial was not grounded, etc. I have no idea if this would work, but just thought I'd toss it out there for the higher pay grades to shoot full of holes....

Just thinking out loud...

Beau Schwabe

whirlymark,

"a small glass vial...1cm in diameter...so it is not possible to use microswitch or weigh-based sensor"· - But you could measure the resonant frequency of the glass vial.

If you can hold the glass vial at one location in the exact center you can measure the fundamental frequency, adding liquid to the vial will change the resonant frequency.
If you can hold the glass vial at two locations 1/4 of the way in on either side you can measure the 2nd harmonic frequency, again adding liquid to the vial will change
the resonant frequency.

In both cases if you sweep the frequency and look for where the frequency peaks, you should be able to determine how much fluid is in the glass vial.


Reference: Here is something that touches on the concept using a metal rod... a glass vial would have similar characteristics.
http://online.physics.uiuc.edu/courses/phys498pom/Lecture_Notes/Vibrating_Rod/Longitudinally_Vibrating_Singing_Rod.pdf

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

ElectricAye

Whirlydude,
The mere fact that the liquid is possibly transparent should not preclude an optical design. Even the clearest fluid will refract light and disrupt an optical path. Also consider that some fluids, though they might be transparent in visible light, are totally opaque in, say, ultraviolet or infrared. So you might have some options using optical filters, various LED emitter/detectors, etc. You might look up the absorption spectra of your candidate fluids and see how they behave.


Beau,

I like your resonator idea, but I'm guessing the natural frequency of such a small vial would be pretty high. And a lot would depend on the dimensions of the vial vs. that of the fluid inside. (If the fluid barely wets the bottom, for example, then the sides of the vial would remain free to sing.) Also, material thickness and density of the vial would have to be considered. For this idea, I'm guessing the thinner the vial walls, the better. Shape, too, would be a factor. Most plastics would probably suck.

Something I saw referenced on the internet:

Rossing, T. D. “Acoustics of the glass harmonica” J Acoust. Soc. Am. 95(2) 1106-1111 (1994)

Phil Pilgrim (PhiPi)

ElectricEye said...
The mere fact that the liquid is possibly transparent should not preclude an optical design. Even the clearest fluid will refract light and disrupt an optical path.

My thoughts exactly, and what prompted my set of questions. With a properly-sized and -positioned opposing light source, the clearest liquids can be made to look opaque.

-Phil

Bean

Could you use a type of geophone/microphone to detect the drops hitting the bottom vial ?

Bean.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
There is a fine line between arrogance and confidence. Make sure you don't cross it...

·

kwinn

It is possible to use an optical (ir led/phototransistor) to detect transparent materials as pointed out previously. A drop or stream of liquid will refract the beam and produce a signal on the output. This type of detector has been used to detect clear plastic test tubes on a conveyor with very good results. It was done by having an op amp control the current to the emitter so that the output from the detector was at approximately 50% of the supply voltage, and using another op amp as a comparator to detect fluctuations when a tube refracted the beam.

If you are interested in using this approach let me know and I will see if I can find the schematics and post them for you.

parsko

www.keyence.com/products/sensors/fiberoptic/fu/fu_features_3.php

They may be a bit pricey, but I am pretty confident that you will find what you want there. Plus, they are wicked easy to use. Interfacing them to a Ucontroller would probably be pretty easy, though I have never looked into it or done it myself. I know my company uses (probably) hundreds of these things in each tool we sell. If all you are looking to do is detect a drop, then you should be fine with the light curtain.

The price might be in the area of a hundreds of dollars, but it's all figured out for you. If you are in a university, the cost might be easier to swallow. These would solve a lot of my problems in my projects, but they are cost prohibitive to the hobbiest, IMHO.

-parsko

whirlymark

Phil

Thanks for all the input and ideas.

To answer some questions..
The drop never flows down the side of vial, but the horizontal variation is probably 5mm.

Good news that the refraction of a liquid would be enough to allow me to use an ir led/phototransistor. I will attempt to go down this route.

Kwinn - I would like to try your approach, so if you could post a schematic that would be great.

To answer the other idea - a microphone would not work as the vial assembly sits on top of an instrument that contains a variety of fans, a turbo pump going on in the background, in fact the whole room is quite loud!

The light curtain idea is elegant and accessible but the cost is prohibitive for the current application (I am in a university, but ecoonically challenged!).

Phil Pilgrim (PhiPi)

How much physical clearance do you have on opposing sides of the vial? Is ambient light an issue, or can it be controlled?

-Phil

Post Edited (Phil Pilgrim (PhiPi)) : 3/27/2009 5:43:25 PM GMT

Tracy Allen

As a side note on this, I set up instrumentation using a Kenence laser curtain to measure the size of raindrops. The device measured both the size and velocity. Data processing was based on the height and width of the dip in transmitted light intensity. Secondary drops falling off leaves on high tree branches tend to be large and to have high terminal velocity. This was deployed in several sites in Japan and SE Asia in relation to effects on erosion. (E.g. Kazuki Nanko, Shigeru Mizugaki & Yuichi Onda, "Estimation of soil splash detachment rates on the forest floor of an unmanaged Japanese cypress plantation based on field measurements of throughfall drop sizes and velocities", Catena 72(2008) 348-361, available from www.sciencedirect.com).


For your situation, a 10 microliter droplet would be about 2mm in diameter, and you estimate 5mm horizontal variation. It might take a wide beam, but with a bit of optics it should be possible. The circuit would be analog with a threshold for detection of the dip in light level.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

whirlymark

Phil Pilgrim (PhiPi) said...
How much physical clearance do you have on opposing sides of the vial? Is ambient light an issue, or can it be controlled?

-Phil

On one side I have 1cm clearance and the other side I have about· 1 meter (so plenty).· Ambient light is an issue however I would be able to shield the sensor to a significant degree, but stray light would still impact the sensor.· I can't get the whole room to be dark during the operation.

Post Edited (whirlymark) : 3/27/2009 6:48:57 PM GMT

whirlymark

Tracy Allen said...
As a side note on this, I set up instrumentation using a Kenence laser curtain to measure the size of raindrops. The device measured both the size and velocity. Data processing was based on the height and width of the dip in transmitted light intensity. Secondary drops falling off leaves on high tree branches tend to be large and to have high terminal velocity. This was deployed in several sites in Japan and SE Asia in relation to effects on erosion. (E.g. Kazuki Nanko, Shigeru Mizugaki & Yuichi Onda, "Estimation of soil splash detachment rates on the forest floor of an unmanaged Japanese cypress plantation based on field measurements of throughfall drop sizes and velocities", Catena 72(2008) 348-361, available from www.sciencedirect.com).


For your situation, a 10 microliter droplet would be about 2mm in diameter, and you estimate 5mm horizontal variation. It might take a wide beam, but with a bit of optics it should be possible. The circuit would be analog with a threshold for detection of the dip in light level.

Tracy

Thanks for the paper - I like the technical blue bucket.· I can see that this could possibly work and will do a little reading around this topic.· Do you need a pdf of this paper?

kwinn

I will look through my obsolete manuals and dig out the schematic over the weekend.

Scratch the above comment. Miracle of miracles, the manual I was looking for was on my bookshelf, not packed away in a box so here is the schematic.

Please keep the following in mind:

This circuit detected three different tube sizes so you should only need one of the three sections.

This is a very old design so the 72720nop amp and MPS65xx transistors may not be available. I would try substituting something like the LM324 for the 72720 and 3904/3906 for the MPS6521/34. That would also allow operation from a single 5V supply, and Q1 may not be needed.

The LED and phototransistor may also be obsolete. Almost any LED/photo transistor used in an opto interrupter should work.

The circuit is fairly simple. The first op amp and transistors Q1, Q2 maintain the current through the LED (CR1) at a value that produces 2.5V across the resistor on the detectors (Q10) emitter. Capacitors C1 and C2 (250uF) act as a low pass filter.
The second op amp is used as a comparator and provides a positive pulse to the base of Q3 when the voltage on the emitter of Q10 drops below 2.25V.

Post Edited (kwinn) : 3/27/2009 11:31:47 PM GMT

whirlymark

Thanks for the schematic and updated components. I will let you know how it works out

Phil Pilgrim (PhiPi)

I hadn't wanted to say anything until I was reasonably certain this would work, but it's the reason for all my questions. In a couple or three weeks there will be a linescan sensor daughterboard available for the MoBoStamp-pe, that will give this board the gift of sight. (The details of the availability haven't been worked out yet, and pricing is yet to be determined. The reason is that the assembly house, who's currently building a batch of 300 units, is doing only the soldering. I'm doing the final assembly, testing, and packaging and will be timing these operations so I can get an accurate quote for them.) This card, based on the TAOS TSL1401R sensor chip, includes a lens and is able to "see" a field of view 128 pixels wide by 1 pixel high. That's the reason it's called a "linescan" sensor. The freely-downloadable coprocessor firmware will enable the BASIC Stamp on the MoBo to perform basic machine vision functions, such as counting pixels, locating edges, etc., with ease.

The water droplet detection problem seemed like an ideal application, and I can happily report a successful trial. Pictured here is the experimental setup:



The field of view is backlit by a miniature fluorescent tube driven by a high-frequency, 12VDC-powered ballast. The camera's field of view is horizontal and coincides with the narrow backlight. (It's also possible to use a diffused array of closely-spaced SMT LEDs for this. I just didn't happen to have one.) The droplets fall between the camera lens and the backlight, where they are silhouetted. In a nutshell, the PBASIC program instructs the firmware to take snapshots of the scene and report the level of the least-bright pixel it sees. If this level drops below a certain threshold, it means that a droplet fell during the exposure. Using an eye dropper, I was able to capture the drops that fell from it.

Admittedly, this is a pretty crude test and doesn't account for the presence of a vial wall. But the firmware includes other processes for handling that eventuality, so I'm pretty confident this will work.

-Phil

Note for those who are wondering: This isn't the PropCAM and doesn't need a Propeller to operate; although the Prop, too, would make an excellent host for it.

_

Post Edited (Phil Pilgrim (PhiPi)) : 3/29/2009 3:18:38 AM GMT

ElectricAye

Awesome!

whirlymark

Phil
That is amazing - I look forward with great anticipation to seeing this. It looks like it will be the perfect solution. Thanks for the preliminary test with the eye dropper and I will keep looking for the product.

CounterRotatingProps

Very cool Phil !

This is interesting for me too because I'd like to have a " rain-drops-keep-fallin'-on-my-head " detector. The basic rain gauge only accounts for total volume in time. And a mere "it's-raining" sensor - like those used to automatically close venting skylights, are only binary.

I've had a project brewing a while that would need to figure out how *many* raindrops are falling in a giving area in a given time sample. It would also need to check the size of the drops (which are usually about the same).

The question (and calculation) is how fast would a drop fall through the detection area.

cheers - and nice work!

-Howard in Florida
~~~~~~~~~~~

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Buzz Whirr Click Repeat

CounterRotatingProps

If I might ask a favor: can you post the schematic for the 12v hi freq balast and lamp set up? I've wondered how these work, but haven't had good luck searching on the web for one that's about the size of the one you've used. thanks - H

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Buzz Whirr Click Repeat

Phil Pilgrim (PhiPi)

I don't have a schematic. The board (more accurately called an "inverter", rather than a "ballast", I guess) is stamped "TDK" and "PCU-620A". I've had it for probably 15 years. I believe it came from JKL Components. This is probably its replacement. The 6mm x 50mm cold-cathode fluorescent lamp (CCFL) probably came from there, too, but they don't list it anymore. Found it: #BF650-20B.

BTW, DigiKey carries their lamps and a whole host of inverters from various manufactureres, including TDK.

-Phil

Post Edited (Phil Pilgrim (PhiPi)) : 3/29/2009 9:57:12 PM GMT

CounterRotatingProps

Thanks Phil - thats enough info for me to roll with it! - H

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Buzz Whirr Click Repeat