series instrumentation amps for capactior/cell monitoring

Discussion in 'Analog & Mixed-Signal Design' started by jamesmaci, Sep 23, 2016.

  1. jamesmaci

    Thread Starter New Member

    Sep 23, 2016
    5
    0
    Im hoping to use 3 instrumentation amplifiers to measure a differential voltages over each cell/capacitor of a total of 3 in series. The first cell is 0 to 4.2V, the 2nd would be 4.2V to 8.4V, but the differential over it would measure 4.2V. The third cell is 4.2Vx3, in voltage with reference to the lowest cell, but it again should read 4.2V via differential. Ive attached an image of the format of the 3 cells. In real application, there are 12 cells putting the common mode up at 50V so a VDR has to be used to scale it down.

    I made a model of the application in spice and also modeled the behavior in a spice engine. The behavior is the same in both simulation and on the board. The lowest cell in the stack always measures about 0.07Volts lower than the cells in series above it. However, on the PCB, it is more like .2V. It seems to be an issue related to the lowest rail being a single sided and grounded. Im at a loss on how to make this work. Ive tried many configurations on the lowest - input

    The specific IC Im using is INA826 and some models were done in tiny TI spice.

    Im trying to get an understanding of what is causing this and the best way to deal with it. Is there an internal diode or level shift occurring? I understand from the datasheet that the DC level on the Vref pin can alter the output but this dosent seem to be the case in the model.

    Vref=0V (spice) Vref=1V (spice) Measured VRef=0
    Cell1 1.91V 2.91V 1.81V
    Cell2 1.98V 2.98V 2.06V
    Cell3 1.98V 2.98V 2.06V

    thanks
    Jim
     
  2. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    I am not sure what causes your issues, but as far as I can see the INA is only good up to 40v common mode voltage.
    What we recently used to cheaply measure 4 lead acid 12V batteries was as follows:
    On each battery is an opamp, powered by the battery and is used as a voltage controlled current source. On the low side is a resistor that converts the current back to voltage and is then multiplexed and measured.
     
  3. #12

    Expert

    Nov 30, 2010
    16,305
    6,814
    I see the datasheet says the INA826 accepts inputs clear to the ground voltage, but...try giving the chip a negative supply voltage and see if this suddenly clears up.
     
  4. OBW0549

    Well-Known Member

    Mar 2, 2015
    1,319
    888
    I haven't used the INA826, but on the instrumentation amplifiers I have used, the Reference terminal MUST be connected to some voltage (ground, or whatever voltage you want the output offset by). You CANNOT leave it floating; if you do, the amplifier will not work correctly.
     
  5. jamesmaci

    Thread Starter New Member

    Sep 23, 2016
    5
    0
    OBW - we do have the reference terminal grounded. Sorry, its not present in that schematic as I tried to simply it for presentation and must have removed it. The table I presented earlier shows what we get with different levels of Vref.

    #12 - We can try giving it a negative rail, but this is not single sided. Im not sure what it will tell us.

    kubeek - we have a voltage divider that reduces the common mode. It steps 50V to 5V.

    thank you all

    Its very odd behavior that Id really like to understand. Ive asked 4 EEs at work and no one seems to understand it.
    It can even be reproduced in TI tiny spice under the device model INA826 if you make a similar series stack but the real life measurement is much larger.
     
  6. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    I am guessing the input current of the amp might be the culprit, could you change your divider and use say 100k and 3k92 to verify?
     
    jamesmaci likes this.
  7. jamesmaci

    Thread Starter New Member

    Sep 23, 2016
    5
    0
    kubeek - we already tried that, exact same numbers, good thinking.
     
  8. crutschow

    Expert

    Mar 14, 2008
    13,022
    3,235
    What is the tolerance of the voltage divider resistors?
     
  9. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    Yes, the precision of the divider can get you quite unreliable results even with modest common mode voltage.
     
  10. jamesmaci

    Thread Starter New Member

    Sep 23, 2016
    5
    0
    we have the voltage dividers accounted for.
    Its also shows up in the model of tiny ti spice though too, don't forgte.
    I was going to try to upload it but was prohibited file type
     
  11. jamesmaci

    Thread Starter New Member

    Sep 23, 2016
    5
    0
    kubeek pretty much had it.
    we had gone to 1M to reduce current
    never thought we had to go to 100k
    awesome man
     
Loading...