• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 30 subscribers
  • Views 4923 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.

AWR1642: Zooming into range FFT

Greg Ratzel
Prodigy 30 points
Part Number: AWR1642

Our application requires computing range to a single object with high precision, similar to what is shown in the following demo. 

We do not need micrometer precision as shown above, however 5mm would be desirable.  Can this be demonstrated with the current SDK?  If so, what parameters should be used?

  • 0
    TI__Guru 58035 points
    Hello Greg,
    Thank you for your interest in AWR1642 device. Can you share more details like what is the max range, max velocity and velocity resolution you are targeting ?
    In the present SDK demo we can get upto 4 cms resolution. Also, in the demo, the accuracy would be limited to this resolution. For getting better accuracy further techniques would have to be used. In your use case would be sensor be seeing a single point-object only or could there be multiple objects adjacent to each other?

    Regards,
    Vivek
  • 0 in reply to Vivek Dham
    Prodigy 30 points
    Thanks, Vivek.

    Our desired maximum range is 5m. Unfortunately 4cm is not sufficient. Our understanding was that the 1.0 SDK release would support higher precision measurements < 1 cm.

    What is the best way for us to proceed? Do you have source code available for the demo pictured above, perhaps? Can the 'mmw' demo be simply modified to add zero padding?

    Regards,
    Greg
  • 0 in reply to Greg Ratzel
    TI__Expert 4720 points
    Hi Greg,

    To improve the accuracy to less than 4cm the best way forward is to modify the OOB demo's code to perform interpolation around the peak corresponding to the target, and thereby get a better accuracy in range estimation.

    For example, you could capture the range-bins at the range of interest, and then perform a quadratic interpolation on those samples to extract the true peak. This has to be done after the 2D FFT, during detection. Use the 2D FFT output for interpolation, not the 'detection matrix' output.

    The micro-meter demo requires that the target be perfectly stationary for a couple of seconds while the measurement is being taken. In case you need velocity information when using the radar, the micro-meter demo may not be suitable.

    Regards.
    Anil