• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 31 subscribers
  • Views 817 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.

AWR1843BOOST: AWR1843 : AOA is unstable as processed by HWA

Jacki Hsu
Intellectual 330 points
Part Number: AWR1843BOOST
Other Parts Discussed in Thread: AWR1642, AWR1843

Hi TI teams,

We've used mmWave SDK Out-Of-Box Demo 18xx on AWR1843BOOST EVM.

We've tested the detection function with our own radar simulator. The object detection of DPCs seem okay (correct range and doppler bin).

However, the object AoA seems a little bit of unstable, as shown in the video.

Also, we used another lab0007_medium_range_radar to check this phenomenon, and the object AoA is quite stable.

We've found out that as the object possess a relative velocity the AoA seems more unstable than stationary (zero velocity).

  • 0
    TI__Genius 15046 points

    Hi,

    There is no such known issue with the SDK datapath or demo; I will have to discuss this with the wider team. Before I do so, could you clarify what the radar simulator is for me? Could you also give me details of your test setup?

    Regards,

    Aayush

  • 0 in reply to Aniket Mahajan
    Intellectual 330 points

    Hi Aayush,

    The radar simulator we used is ITS-9510C and the test is done in our chamber. The distance between radar and simulator is 2.5 m.

    Today, we compared the .bin file of AWR1642 and AWR1843 under the mmwave_sdk_03_05_00_04.

    The datapath of each them is below,

    mmwave_sdk_03_05_00_04\packages\ti\demo\xwr16xx\mmw\xwr16xx_mmw_demo.bin

    mmwave_sdk_03_05_00_04\packages\ti\demo\xwr18xx\mmw\xwr18xx_mmw_demo.bin

    After that, we applied the same config to both demo but the xwr18xx processes more unstable result.

    The setting and results of xwr16xx is shown as below.

    The setting and results of xwr18xx is shown as below.

    The radar simulator setting is shown as below.

    BTW, we also done some experiment on demo code of xwr18xx

    1. Compare with a 32-bit FFT processing calculation is implemented on DSP, and a 24-bit FFT processing calculation is implemented on HWA, is there any overflow or not-enough precision issue to make the AoA unstable?

    2. We have tried to modify “fftMode.butterflyScaling”, “source.srcScale” and “dest.dstScale” of dopplerprochwa and  aoaprochwa DPUs, but the AoA unstable is still occur.

     

    Do you have any comment about AoA unstable issue?

     

    Thanks

  • 0 in reply to Jacki Hsu
    TI__Genius 15046 points

    Hi,

    Thanks for all the details. I'll bring this up to the wider team tomorrow and let you know.

    Regards,

    Aayush

  • 0 in reply to Jacki Hsu
    TI__Genius 15046 points

    Hi,

    I checked with the team, we have not observed something like this with our tests with target simulators. We have a few theories. Could you do the following to help us understand the issue better?:

    1. Attach the configuration and a few seconds of captured output data capturing the instability (record start and stop in the demo visualizer). Could you also do the same for the 1642 device in the same setup without the instability? Please enable range and noise profile before capturing the data.

    2. I can see that the peak in the range profile for the 1642 device is more prominent as compared to the 1843 device. Could you increase the RCS and test with the 1843 device to see if a stronger signal eliminates this issue? 

    3. Could you see if the same behavior is observed while using a strong reflector? This is to rule out the radar simulator as a potential cause.

    Thanks,

    Aayush