Measuring small volumes of water by mass?

Discussion in 'Analog & Mixed-Signal Design' started by blah2222, Aug 29, 2016.

  1. blah2222

    Thread Starter Well-Known Member

    May 3, 2010
    554
    33
    Hello,

    I am interested in continuously acquiring load cell measurements with a resolution of 0.01g or lower. I need to be able to identify individual droplets of water (~0.5g) vs. time when added to the liquid container on the scale.

    Naively, I bought a 1kg (0.1g resolution) pocket scale to see what kind of performance I could get. I opened it up, disconnected the four load cell leads (+/- excitation & +/- signal) from the PCB, and connected them to my amplifier/DAQ setup (excitation voltage = +/- 5V).

    Scale that I used (its bridge gauge resistors are 1kOhm each):
    http://www.usbalance.com/us-minibench-1000g-x-0.1g

    Playing around with the filters and gain, I settled on LPF @ 5 Hz and G = 1000. This gave me roughly 2 mV/g. Unfortunately the signal (left on its own) fluctuates quite a lot with temperature/noise and drifts around at magnitudes ~10x more than the droplet signals that I am trying to measure. No good.

    I am considering getting a 0.001g scale instead of this 0.1g scale, but I am wondering if these differ only in their signal processing hardware and not the actual load cell setup.

    Any suggestions on a plausible way to measure these small droplets without being squashed by ambient/temperature noise?


    Thank you!
     
  2. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,544
    1,251
    Schematic ... ?
     
  3. DGElder

    Member

    Apr 3, 2016
    345
    85
    How about amplifying the weight with mechanical advantage. I see a lever and spring. Or instead of measuring the change in weight, let the drop fall further and detect the impulse. If you are just counting drops there are a number of ways that are easier or cheaper than a precision scale.
     
  4. blah2222

    Thread Starter Well-Known Member

    May 3, 2010
    554
    33
    See attached (very crude).

    I need the change in weight because my end goal is to measure the volume in mL. I don't really care about counting drops, I need volume of each drop. Using the known density of the liquid I can back-calculate the volume.

    The use of a dropper is for explanation but this will be used to measure rodent urine which exits in droplets.

    Also, mechanical advantage doesn't get rid of the inherent temperature/drift of the load cell.
     
    Last edited: Aug 29, 2016
  5. DGElder

    Member

    Apr 3, 2016
    345
    85
    Still not clear. If you don't care about counting drops why do you need a resolution equal to weight of a drop.
    In any case your percent error will be a function of the scale resolution and total volume of urine. You need a more thorough specification of functional requirements before designing the apparatus.
     
  6. blah2222

    Thread Starter Well-Known Member

    May 3, 2010
    554
    33
    I stated that I needed a resolution of 0.01g or lower and each drop weighs ~0.5g. I am indirectly measuring volume.
     
  7. RichardO

    Well-Known Member

    May 4, 2013
    1,235
    384
    One way of measuring a very small mass is to use a force balance. This is done by creating a known force counteracting the unknown force of the mass. The known force is adjusted until the forces cancel. The value of the mass is now known because it equals the known force.

    A simple (conceptually) force balance is to lift the unknown mass using a solenoid and sensing the power in the solenoid needed to lift the mass.
     
    blah2222 likes this.
  8. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,544
    1,251
    Two thoughts with nothing to back them up.
    1. 50 Hz or 60 Hz high-Q notch filter in case the noise is power related.
    2. Second, identical load cell, unloaded, physically adjacent to the one collecting data. Its temperature-induced errors might the data cell, something you can subtract out in either the analog or data domain.

    ak
     
    blah2222 likes this.
  9. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,690
    2,752
     
    cmartinez and blah2222 like this.
  10. DGElder

    Member

    Apr 3, 2016
    345
    85
    You say you don't care about counting droplets, just weight so you can calculate volume, yet you keep talking about registering droplets, the weight of droplets..........

    Feels like an XY problem to me.

    http://xyproblem.info/
     
  11. blah2222

    Thread Starter Well-Known Member

    May 3, 2010
    554
    33
    I suppose "I need volume of each drop" was a tad misleading... I need the total volume after a single urination cycle (1+ drops).

    If the rodent urinates and ten droplets exit, I can measure that because the weight is substantial enough, but sometimes it only blesses me with a single droplet. The mass of a single droplet is difficult to measure if noise and drift are swamping it out.
     
  12. wayneh

    Expert

    Sep 9, 2010
    12,146
    3,057
    Measuring liquids with precision in the mg or lower range is a challenge, since the weight is changing as water evaporates from the droplet. If you know the water activity of the solution, you can set up a humidity chamber at roughly that water activity to minimize weight loss or gain from the ambient air.

    Measuring anything in the mg range requires shock and draft elimination. People use enclosed scales on heavy grant tables and such. Going below 0.1mg typically requires a special room.

    Only after these things are in place can you begin to work on the electronics.
     
    blah2222 likes this.
  13. blah2222

    Thread Starter Well-Known Member

    May 3, 2010
    554
    33
    Mains is not an issue because I have a 5 Hz LPF. I was also thinking of a second load cell, but I like the benefit of a commercial scale with nice a nice package. Thanks


    Thanks, I have seen this and I was thinking about purchasing this scale.
     
  14. KeepItSimpleStupid

    Well-Known Member

    Mar 4, 2014
    1,147
    204
  15. blah2222

    Thread Starter Well-Known Member

    May 3, 2010
    554
    33
    After hacking my first trial, I concur. The lack of windshielding was an issue for sure.

    I do not require anything below 10mg.
     
  16. dannyf

    Well-Known Member

    Sep 13, 2015
    1,825
    364
    Other than better sensors, better ADC/amplifier and better environment, no other solution fundamentally.

    For slow drift, you may be able to counter it in software but it is dicey - without calibration but is unlikely to work reliably.

    You really need a commercial solution or rethink your approach.
     
    blah2222 likes this.
  17. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,804
    1,105
    Are the drops significantly different in size? I would guess not; in which case counting the drops might be simpler than weighing them individually. The weight of a batch would be easier to determine.
     
    blah2222 likes this.
  18. wayneh

    Expert

    Sep 9, 2010
    12,146
    3,057
    Drop them thru oil and measure their size with a laser.
     
    blah2222 likes this.
  19. blah2222

    Thread Starter Well-Known Member

    May 3, 2010
    554
    33
    Thanks, I am trying to figure out if the load cells in 0.001g are that much better than 0.1g scales or if it comes down to better signal processing and ICs or both. I couldn't care less about the ICs, I just need a really accurate load cell.


    Thanks and good idea. I am collecting this data real-time along with bladder pressure simultaneously.

    Interesting idea. What is the purpose of the oil? Thanks.
     
  20. atferrari

    AAC Fanatic!

    Jan 6, 2004
    2,648
    764
    Wondering how the TS/OP will make the rodent to urinate precisely "there"...
     
Loading...