Connect with us

how to read value into ADC from variable resistor?

Discussion in 'Electronic Design' started by Michael Noone, Jul 22, 2005.

Scroll to continue with content
  1. Hi - I'm going to be using a CUI pressure sensor
    (http://www.cui.com/pdffiles/SF-X.pdf). It acts as a variable resistor. I
    want to connect it to an ADC (fairly high impedance - I think about 10M).
    How does one do this? My only idea was to make a sort of voltage divider -
    with one pin of the pressure sensor held high, the other connected to the
    ADC. Then have a resistor between the ADC and ground. I suppose I'd have to
    have a large resistor in front of the ADC to raise the impedance as the
    pressure sensor's resistance goes up to 100M or so. Is this the best way to
    do it? How does one choose a value for the resistor going to ground?

    Thanks!

    -MJ Noone
     
  2. Andrew Holme

    Andrew Holme Guest

    What pressure range do you want to measure? The recommended load for
    that sensor is 1.5kgf, at which its resistance is in the region of 2k
    ohm. The specification section on the datasheet quotes maximum
    allowable voltages and currents. It doesn't matter how you use it, as
    long as you stick within the limits. You could either feed it with a
    constant current, or make it part of a voltage divider. The datasheet
    suggests an operating current of 5mA which implies the former.
     
  3. With a bridge, the best results are achieved
    when both resistors are about the same. Then
    use a high impedance buffer amplifier in front
    of the ADC.

    Rene
     
  4. Yes, you can do this. The only problem is that the relationship between the
    unknown (pressure sensor) resistance and the voltage applied to the ADC is
    not linear. That means that you need to do some math to get the resistance
    from the ADC reading. It also means that the resolution is not constant with
    resistance: for some pressure regions, a small change in resistance will
    cause a large change in the ADC reading, but it in others it will take a
    large change in resistance to produce any change in the ADC reading.

    This might actually be good though, because your sensor is non-linear, so
    you might be able to use this non-linearity to make the sensor a bit more
    linear in the regions where you care about that.
    I don't know anything about pressure sensors, but look at the figure at the
    bottom left of the first page. If I am interpreting it correctly then it
    shows data for four different samples of the same model of sensor. The
    response is all over the place below 250 gf: a 50 gf weight produced a
    little less than a 1meg resistance on one sensor, but more than 10meg on two
    others. I don't think that the sensor is intended for use in those regions.

    Also, impedance does not work like that. If you put a 90meg resistor in
    front of an ADC with an input resistance of 10meg, then you are right that
    the input resistance increases to 100meg. However, you have also thrown away
    nine tenths of your signal; more voltage will be dropped across the 90meg
    resistor than across the ADC's input resistance.

    In any case, are you sure that your ADC input looks like a resistance? In
    many types of A/D, the input looks like the sample and hold capacitor.

    Jonathan
     
  5. Andrew Holme

    Andrew Holme Guest

    Andrew Holme wrote:
    [snip]
    Or you could design your potential divider for 5mA.
     
  6. Now go back to the drawingboard and calculate the needed voltage to 5mA
    through a sensor wit 1 to 10MOhm resistor.....

    My suggestion: use a decent pressure sensor instead of this crap.
    Decent pressure sensors use strain gauges which can be prefectly mearured
    using a wheatstone brigde.

    Meindert
     
  7. Can you suggest a different pressure sensor of similar dimensions? (it
    needs to be about 3mm square/round) The only other similar sensors I've
    been able to find are made by IEE (http://www.iee.lu/nav04/lusense.htm) who
    don't have any distributors in the USA. Their parts do look quite nice
    though - take a look at Conrad (http://www.conrad.de) 182519-62, datasheet
    here: http://www.produktinfo.conrad.com/datenblaetter/175000-199999/182519-
    an-01-de-Drucksensor_FSR-149AS.pdf.

    -Michael
     
  8. Ideally I'd like to be able to measure about 0-1kgf, which unfortunately
    is in the least accurate region of the sensor. These are for a one off
    application and a total of 6 will be used - so I was planning on
    individually calibrating each to compensate for this inaccuracy.

    -MJ Noone
     
  9. No I can't Have you searched google for "miniature load cells" ?

    Meindert
     
  10. Tim Wescott

    Tim Wescott Guest

    You're asking to be able to read a tremendously large resistance span
    there. I would investigate the possibility of preloading the thing with
    a few Newtons (what serious engineer uses "kgf" for crying out loud!).
    This preloading has the potential to wreck your DC accuracy if you can't
    do it with a weight, so be ready to jettison the idea fast.

    If you can't preload and you need anything resembling accuracy then
    you're stuck with what you've got. While the resistance of the thing
    varies wildly with force, it appears that the conductance vs. force
    relationship may be more well behaved. If you could design a little
    circuit that would hold the sensor voltage constant and monitor the
    current you may do well:


    |\
    2mV <-------|+\
    | >---------o----------->
    .--|-/ | To ADC
    | |/ .-. .--->
    | | | |
    | | | GND
    | '-'
    | ___ |
    '-----|___|-----o----->
    To Sensor
    .----->
    |
    |
    GND
    created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

    Note that this circuit, with it's 2mV input to the opamp, requires
    either an ultra-low offset opamp or an AC coupled system. Since the DC
    coupled circuit would be incredibly sensitive to ground loops I'd
    probably do it AC coupled: I'd add appropriate blocking caps and drive a
    square wave from a microprocessor into the opamp. I'd use the same
    microprocessor to read the ADC at both the high and low point of the
    square wave, subtract the two values and average the hell out of the
    result -- and I'd still expect to have difficulties.
     
  11. gf and kgf threw me off too. I had never heard of that term before - so I
    had to look it up. Preloading would probabaly be possible, though it'll be
    hard to do accurately. Today I have been in communication with a couple
    other FSR (force sensing resistor) manufacturers, and they look to be more
    accurate in the low pressure region, so I'll probabaly end up going with
    one of those.

    If I do end up using these CUI sensors your idea definitely sounds
    interesting. Complicated, but interesting!

    -Michael
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-