This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADS8900BEVM-PDK: Unipolar Reference Selection is noisy when inputs are grounded

Part Number: ADS8900BEVM-PDK
Other Parts Discussed in Thread: ADS8900B, THS4551, OPA625

Hi,

I have the EVK up and running.  It looks like its setup with + and - inputs grounded and the common mode reference set to 0V.  I see a nice and low noise histogram, cool.

Questions:

1.  Why is there a 0V common mode option?  The common mode range of this part is 0V to VREF.  I can't think of an application where you'd want to operate at the minimum of 0V.  We have a biopotential signal that is bipolar, it will likely be close to ground within +-10mV or so, but because of the negative -10mV, we intend to analog level shift this signal to VREF/2 common mode.  this brings me to the second question.

2.  When I select VREF/2 common mode on the EVK (J1-closed, J2-open) and keep the + and - inputs grounded, I get LOTS of noise and the mean conversion value is near -36.7k.  Did I miss something?  I was expecting basically the same results as common mode 0V.  When I let + and - float things are happy again.  The OPA625s should be able to drive the common mode current when + and - are grounded, I don't understand what is going on?  I read 2.586V on both - inputs of the OPA625s so it does seem to be sourcing current to keep the - inputs tracking the + inputs at 2.5V.

Thanks,

Erik

  • Well, looking at the schematic a bit longer, it looks like the output of the OPA625s are probably at the rail when the inputs are grounded.  So, I guess I need a level translator to translate our +-10mV signal to 2.49-2.51V.  That's why I like the LTC6362, because it takes care of the level shifting, does TI have a similar amp?

    This is excerpt from LTC6362 is basically what I have designed into my schematic so far, only our swing is very small and we'll have a single-ended LNA before it.

    Notice you can ground the inputs here and keep a 2.5V output, because the LTC6362 level shifts to make the outputs referenced to 2.5V, which is nice.

  • Hi Erik,

    The ADS8900BEVM only supports fully differential inputs with the on-board amplifier configuration.  You will want to use an FDA (similar to your above post) to go from a single-ended input to a differential output suitable for driving the ADS8900B inputs.

    The recommended TI amplifier to convert a single ended input to differential suitable for driving the ADS8900B is the THS4551 amplifier.

    Take a look at this cookbook circuit that describes this approach in more detail:

    http://www.ti.com/lit/an/sbaa246a/sbaa246a.pdf

    There are also a few TI Reference Designs using the ADS89xx family and the THS4551.

    http://www.ti.com/tool/TIDA-01054

    http://www.ti.com/tool/TIDA-01351

    Answers to your specific questions:

    1.  The ADS8900BEVM Common mode setting refers to the common mode of the input signal.  If you jumper both J1 and J2, then the input range can be from +5V to -5V differential.  Each of the inputs would be +/-2.5V and opposite phase.  

      For example,

      Vs(-)=-2.5V, Vs(+)=+2.5V, Vindiff=+5V

      Vs(-)=+2.5V, Vs(+)=-2.5V, Vindiff=-5V

    2.  When setting the common mode to Vref/2, again, this refers to the common mode level of the input source.  The amplifiers are expecting an average voltage between the two inputs of 2.5V.

    A few examples for an input common mode voltage of 2.5V:

    Vs(-)=0V, Vs(+)=+5V, Vindiff=+5V, Vin-common=2.5V

    Vs(-)=+2.5V, Vs(+)=+2.5V, Vindiff=0V, Vin-common=2.5V

    Vs(-)=+5V, Vs(+)=0V, Vindiff=-5V, Vin-common=2.5V

    When you short the inputs to ground, the input common mode voltage is 0V, not 2.5V.  The amplifiers are saturated and not operating in the linear region, which can result in higher observed noise.

    Regards,

    Keith Nicholas
    Precision ADC Applications

  • Still confused on #1:

    Keith Nicholas said:
    If you jumper both J1 and J2, then the input range can be from +5V to -5V differential.

    And my understanding from the EVK user guide table 3 is that the EVK will then support a 0V common mode at the input.

    But then if I GND the inputs (common mode 0V) it doesn't work because as you point out the amplifiers become saturated.

    Thanks for pointing me to THS4551.  I will take a look. Thank you!

    Erik

  • Hi Erik,

    If you set the input common mode range to Vref/2 (J1 closed, J2 open), then connecting both inputs to 0V will result in saturated amplifier outputs and high noise.

    However, if you set the input common mode range to 0V (J1 closed, J2 closed), then connecting both inputs to 0V will not saturate the input amplifiers and you will get low noise.

    I verified this on an EVM board with the inputs connected to ground and input common mode set to Vref/2 and observed the high noise that you did.  However, setting the input common mode range to 0V with both inputs connected to ground, and the noise went back down to typical levels.

    Regards,
    Keith

  • Hi Keith,

    Thanks for checking at the bench.  I think I understand better after looking at the EVK schematic and probing the voltages.  The key is which common mode we are referring to, which is the input of EVK and NOT the input of the ADS8900B and NOT the input of the OPA625 (~1.3V when J1/J2 are closed).

    Thanks again!

    Erik