• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 46 subscribers
  • Views 199 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

Original question:

ADS8900B: Noise Density

ADS8900B: 1/f noise and reference question

Erik Anderson41
Intellectual 915 points
Part Number: ADS8900B

Hi,

I've narrowed our part selection down to ADS8900B and AD4020.  The AD4020 specifies the 0.1Hz to 10Hz noise at ~6uV-pp.  Do you have any 1/f noise data to share for the ADS8900B?  Or, suggest a method I can measure it on the evaluation kit?

In addition, if we use 2.5V reference instead of 5V, can we expect to see 2x lower noise floor overall?  I understand if we are trying to run full scale swing, the SNR would not improve, but our signal is so small I'm thinking to go with a low reference to reduce noise further.

Thank you!

Erik

  • 0
    TI__Mastermind 36240 points

    Hello Erik,

    It looks like the plot was generated by sampling at 50ksps for 10s (500k samples total) and then averaging in 2500 sample blocks.

    The ADS8900EVM can be configured to take similar data, and then export to Excel to do the averaging.

    1.  Configure the ADS8900BEVM for 0V External Signal common mode by closing (shorting) J1 and J2.

    2.  Short the inputs to ground.

    3.  Using the ADS8900BEVM software GUI, select the Time Domain Display to collect the data.

    4.  Set the Target Sampling rate to 50k and the number of samples to 524288.

    5.  Press Capture and wait for the data to collect (10s).

    6.  Once the data is collected, you can then right click on the data in the display and select export to Excel.

    Please keep in mind that the total results will be the EVM noise, including the ADC, reference, and input amplifiers.

    Regards,
    Keith Nicholas
    Precision ADC Applications

  • 0 in reply to Keith Nicholas
    Intellectual 915 points

    Hi Keith,

    Finally getting around to this.  Using the procedure you provided I'm measuring 1.8uV-RMS noise from 0.1Hz to 10Hz.  I calculated ~0.3uV-RMS 1/f noise due to the amplifiers/reference (crest factor = 6).  This means that the ADS8900B dominates the 1/f noise for this EVK, which is calculated ~1.77uV-RMS.

    I'm not sure if Analog Devices accounted for external amplifiers or not.  They spec 0.1Hz to 10Hz noise at 6uV-pp. Using crest factor of 6 this would be 1uV-RMS.

    Anyways, the 1/f noise for the ADS8900B is still quite low, we can add a little more gain at the front end to effectively knock down the 1/f if needed. This of course assumes we select an amplifier that has much less 1/f noise than the 1.77uV-RMS.

    EDIT #1:  The only thing I'm not sure about is that the 0.1Hz to 10Hz might actually be lower that this calculation, because any white noise that was left over after averaging will fold back down to the 0.1Hz 10Hz band, right?  A more accurate method might be to keep the sampling rate high and average more.  I will think on this a bit more.

    Erik

  • 0 in reply to Erik Anderson41
    TI__Mastermind 36240 points

    Hi Erik,

    The sampling rate was done at 50ksps, so all of the white noise is spread over 0->25kHz.  Since you are effectively creating a low pass filter around 0.1-10Hz by averaging, most of the white noise will not be located in this frequency range, and the noise will be dominated by the 1/f region.

    Regards,
    Keith