Post Go back to editing

ADIS16136 weird behaviour

Hi. I'm sampling the ADIS16136 gyroscope using an EVAL_ADIS with the following settings changed from the default ones:

SMPL_PRD 000Fh

AVG_CNT 0006h; 16sps

DEC_RATE 0007h; 10hz

I'm seeing a weird behaviour, the gyrsocope outputs some values with extremely more likeness than what one would expect. I'm attaching a .csv with raw values and a graph where straigh "railway like" lines are visible, indicating this behaviour. The same behaviour can be seen with the settings I used in this post: http://ez.analog.com/message/131287 but the values are different (I guess because of different filter and decimation settings).

This is not a conversion error on my part (my first guess) because when plotting the raw data I can also see this lines, that's why I'm attaching the .csv file. The raw value 171264 in the .csv corresponds with the top line from the graph.

Is this behaviour expected? Is it due to some error of mine in the configuration of the filter/decimator/sampe rate? Thanks!

attachments.zip
Parents
  • PeterSate,

    You and I have already connected on this, but for public closure of this discussion, I want to share what we have discovered.  In the course of investigating your claims, we have discovered a quantization error in the ADIS16136's signal processing, which explains your observation.  On a positive note, we had already addressed this root cause of this issue when we developed the ADIS16137, which shares the same internal signal processing solution as the ADIS16136.  Moving forward, we going to do the following in the next 2-3 weeks:

    1. Provide official documentation of this observation on the ADIS16136 product page, in the form of a "Silicon Anomaly" sheet.
    2. Send a Quality Advisory note to all of our distributors and register customers, with an offer to upgrade existing units with the latest version of the firmware.

    We sincerely appreciate you investing in using the ADIS16136 and for sharing this experience with us.  While this issue was already addressed, you provided us with an important perspective that would have been difficult for us to envision or observe through our normal gyroscope testing.  It is a great pleasure to work with you. We can't wait to hear more about your work!

    Very best,

    NevadaMark

Reply
  • PeterSate,

    You and I have already connected on this, but for public closure of this discussion, I want to share what we have discovered.  In the course of investigating your claims, we have discovered a quantization error in the ADIS16136's signal processing, which explains your observation.  On a positive note, we had already addressed this root cause of this issue when we developed the ADIS16137, which shares the same internal signal processing solution as the ADIS16136.  Moving forward, we going to do the following in the next 2-3 weeks:

    1. Provide official documentation of this observation on the ADIS16136 product page, in the form of a "Silicon Anomaly" sheet.
    2. Send a Quality Advisory note to all of our distributors and register customers, with an offer to upgrade existing units with the latest version of the firmware.

    We sincerely appreciate you investing in using the ADIS16136 and for sharing this experience with us.  While this issue was already addressed, you provided us with an important perspective that would have been difficult for us to envision or observe through our normal gyroscope testing.  It is a great pleasure to work with you. We can't wait to hear more about your work!

    Very best,

    NevadaMark

Children
No Data