• State TI Thinks Resolved
  • Locked Locked
  • Replies 11 replies
  • Answers 2 answers
  • Subscribers 46 subscribers
  • Views 1588 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.

ADS127L01EVM - shorted noise

Zerg
Prodigy 70 points
Other Parts Discussed in Thread: ADS127L01EVM, ADS127L01

Hi

Got my hands on ADS127L01EVM. Captured some  sinewaves, getting around +/-8M excursion for almost full scale signal, so it is about right for 24 bit space.  

Next thing I'm capturing a shorted ADC input noise. There are some tones are visible, which I suspect can be idle tones or some sort of interference.

ADC configuration  is WB1 filter, high resolution mode, OSR=32, 16MHz clock.

I checked AVDD and VREF noise and voltage, but didn't find anything criminal over there. To avoid any sort of ground loops setup consists only EVM and laptop running on a battery.

The question is, if anyone else observes what I do or I'm just missing something in the EVM configuration etc.

Spectrum and captured data are attached.

shorted_9_11_1_1.txt

 

  • 0
    Prodigy 70 points

    Really looks like some sort of interference.

  • 0 in reply to Zerg
    TI__Guru 66446 points

    Hi Zerg,

    Thanks for your post. This is actually a known issue with this first revision of the ADS127L01EVM. The interference you are seeing is intermodulation of two crystal oscillator frequencies and is not introduced by the ADS127L01 itself. We will be addressing this with an update very shortly. For now, there is a quick workaround detailed in this other post: e2e.ti.com/.../547727

    Best Regards,

  • 0 in reply to Ryan Andrews
    Prodigy 70 points

    Thanks!

    Perhaps it worthwhile of having a warning on the EVM page printed in capital red letters. Somewhere on the top preferably :)

    That looks better now:

  • 0 in reply to Zerg
    TI__Guru 66446 points
    Hi Zeng - I'm glad it worked for you. I think next time we'll try to avoid this issue altogether. :)

    Please let us know if you have any other questions about using this device.

    Best Regards,
  • 0 in reply to Ryan Andrews
    Prodigy 70 points

    Hi Ryan

    Was looking at the noise data and noticed that histogram has some noticeable small scale pattern on top of nice looking larger scale Gaussian shape. Any ideas what could be a cause of this?

    Best regards

    Zerg

  • 0 in reply to Zerg
    TI__Guru 66446 points

    Hi Zerg,

    To me, this looks like a matter of using a non-integer bin resolution, where the number of codes/bin is not an integer multiple of the LSB size of the ADS127L01. If you were size the bins to contain say 10 codes/bin, this smaller distribution pattern should disappear.

    Our new GUI has a built-in histogram tool that allows you to customize the plot by specifying either an integer number of bins (from which codes/bin is calculated) or an integer number of codes/bin. You can experiment with different bin sizes to see the effect on the same set of data.

    Best Regards,

  • 0 in reply to Ryan Andrews
    TI__Guru 66446 points

    For example:

    6.57 codes/bin

    7 codes/bin


  • 0 in reply to Ryan Andrews
    Prodigy 70 points

    Hi Ryan

    To avoid any truncation issue I aligned bins to get integers exactly at the bins center. So, I would say there should not be problems with math I'm doing. Going to double check it to be sure, but I'm quite confident on this stage.

    I agree that making bins wide enough would wipe those peaks, and it is unlikely  to be an issue in our application. Although it is good to have a good understanding of hat is going on.

    I would suspect that this is a result of truncation on  on of the decimation filter stages.  I'm a bit surprised that it is still visible with such amount of analogue noise present. 

    Regards

    Zerg

  • 0 in reply to Zerg
    Prodigy 70 points

    Here are my bin edges. Look quite half integer too me. Also TI software turns to agree with my results:

  • 0 in reply to Zerg
    TI__Guru 66446 points

    Hmm. In that case, I really am not sure what the cause may be. I agree that the way you've set up your bins looks correct to me and the fact that the GUI shows exactly the same result should confirm that's not the issue.

    I'm honestly not sure at the moment what this may be. Would you mind saving the raw data in the Standard format and attaching it below?

    Best Regards,

  • 0 in reply to Ryan Andrews
    Prodigy 70 points

    No problems. Here it is. 

    I'm wondering if this effect can be achieved by postfiltering and truncation of a multi-level quantizer output, in case if number of levels is not 2^n.

    Will try to model that if I have a spare time.

    Zerg

    15_11_2016_ads127l01_data.txt