• State
  • Locked Locked
  • Replies 5 replies
  • Subscribers 45 subscribers
  • Views 678 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

ADS1220: Stability issues - too much noise with ADC counts

Lauren Mayfield
Expert 7875 points
Part Number: ADS1220
Other Parts Discussed in Thread: ADS1235

Hi team!

My customer is having some stability issues with this part - there seems to be too much noise with ADC counts.

The ADS1220 is configured with following parameters:

  1. Gain => 128
  2. 50Hz/60Hz filter => on
  3. 110 samples per second using a one shot conversion command
  4. Data rate => 175Hz

The ADC is attached to a analog load cell with 2mv/ V sensitivity. The excitation voltage is 5V.

 singleshot.xlsxAttached is the spread sheet of test results with 0 weight on loadcell and the schematic.

 

Following are the register settings:
R0  0x0E   ... GAIN 128
R1 0x60   ... DATARATE 175
R2 0x40   ... External reference using dedicated REFP0 and REFN0
R3 0x02  ...  DRDY is indicated on DOUT/DRDY

The voltages on signal path and AVDD are:
TP1=5.015V  => REFP0
TP2=2.508V => SIG+
TP3=2.508V => SIG-

TP5= 0.000V => REFN0
TP7=5.0174V => AVDD

Please reach out to me internally for the schematic and I can share that with you as well.

Thank you!
Lauren

  • 0
    TI__Mastermind 23275 points

    Hi Lauren,

    I did some calculations based on the provided measurement data in the Excel sheet. It does look like the noise performance is in the expected ballpark for the ADS1220 in this application.

    The LSB size of the ADS1220 in this setup is:
    1 LSB = (2 x 5 V) / (128 x 2^24) = 4.68 nV

    Looking at the data in the 'Raw' column:

    • Mean = 4559 codes (equals 21.2 uV)
    • Sigma = 77 codes (equals 0.36 uVrms)
    • Peak-Peak = 490 codes (equals 2.28 uVpp)

    The input-referred noise of ADS1220 using Gain=128, Data Rate = 175 SPS is 0.26 uVrms according to Table 1 in the datasheet. This however assumes shorted inputs. But as you can see we are not far off from the rated noise performance.

    As a sanity check you could short the inputs (MUX[3:0] = 1110b) and select the internal VREF (VREF[1:0] = 00b) and record a data set. We should get close to datasheet values in that case if there are no layout-related issues.

    If you want better noise performance you will either have to reduce the data rate or switch to a lower-noise ADC such as ADS1235.

    Regards,

  • 0 in reply to Joachim Wuerker
    TI__Expert 7875 points

    Hi Joachim,


    Thanks for your help!

    Attached is a graph of shorted inputs (MUX[3:0] = 1110b) and internal VREF (VREF[1:0] = 00b).

    Please let me know your feedback on this.

    Thanks!
    LaurenCopy of short2.xlsx

  • 0 in reply to Lauren Mayfield
    TI__Mastermind 23275 points

    Hi Lauren,

    Happy New Year and thanks a lot for the update.

    If I read the file correctly then the only difference between the data sets is that in one case the sensor in connected and in the other one not. But in both cases the internal Mux short (MUX[3:0] = 1110b) and the internal VREF was used. I calculated the input-referred noise in uVrms and uVpp to compare it with the noise table in the datasheet.

    I have no explanation as to why the noise performance is 4x better when disconnecting the sensor. There shouldn't be any difference in the noise performance as long as the internal Mux short of the ADS1220 is used.
    Also the absolute codes should be in the same ballpark in both cases. It seems like just connecting the sensor adds quite an offset. When the internal Mux short is used it shouldn't matter if a sensor is connected or not. In both cases the noise and offset should be very comparable.

    Also the noise performance with the disconnected sensor looks much better than what we specify in the datasheet. That is a quite odd to be honest. However our data is recorded using continuous-conversion mode and not single-shot conversion mode.

    I am not sure exactly what to make of this data. If you want you can record the same data using continuous-conversion mode, but I am not sure we will get much different results.
    In the end my initial assessment is still valid that if you want better noise performance then you might have to choose a lower-noise ADC.

    Regards,

  • 0 in reply to Joachim Wuerker
    TI__Mastermind 23275 points

    Hi Lauren,

    one question for clarification. Are the values provided in the Excel sheet raw ADC codes or are they manipulated in some way?
    I am asking because the codes between Gain=1 and Gain=128 are almost the same. This shouldn't be the case.

    Regards,

  • 0 in reply to Joachim Wuerker
    TI__Mastermind 23275 points

    Hi Lauren,

    sorry for the many posts, I think I figured things out.

    Could it be that all data is recorded at a data rate of 20 SPS (default setting) instead of 175 SPS?
    And also that all data on the 'with sensor' tab is with Gain=128, while all data on the 'without sensor' tab is with Gain=1? Maybe just a mix-up when the data was put together?!?
    If that is the case then we would be meeting the noise performance stated in the ADS1220 datasheet.

    Regards,