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.

IWR6843AOPEVM: CFAR-CA or CFAR-CASO

Part Number: IWR6843AOPEVM
Other Parts Discussed in Thread: IWR6843AOP, , IWR6843ISK

Hello,

I have some questions about CFAR detection of IWR6843AOP.

1.
When I used the configuration1.cfg (range CFAR is CFAR-CASO), AOP EVM works well. But incase the configuration2.cfg (range CFAR is CFAR-CA, parameters are same as doppler direction),
the detection result is strange. Not peak range bin is detected but "valley range bins". Is configuration2.cfg not correct? (the picture below is example.)

2.
As written above, when I use configuration1.cfg, it works well but looking at the SNR value in detected object side info, the minimum value is about 11 in spite of CFAR threshold value is 15 in configuration2.cfg.
I think SNR in detected object side info is SNR of range CFAR. Why minimum value of the SNR is not equal to CFAR threshold?

Regards,
Kei

configuration1.cfg

configuration2.cfg

  • Hi Kei,

    When I used the configuration1.cfg (range CFAR is CFAR-CASO), AOP EVM works well. But incase the configuration2.cfg (range CFAR is CFAR-CA, parameters are same as doppler direction),
    the detection result is strange. Not peak range bin is detected but "valley range bins". Is configuration2.cfg not correct? (the picture below is example.)

    If the configuration is able to run, then it is not "incorrect" I do notice that the noise window is longer for configuration 1. Perhaps it's too long for your use case? THis would affect where points get detected.

    I think SNR in detected object side info is SNR of range CFAR. Why minimum value of the SNR is not equal to CFAR threshold?

    See the SDK User's guide for more info about this parameter at C:\ti\mmwave_sdk_03_05_00_04\docs

  • Hi Nathan,

    Regarding question 1, in case configuration2.cfg, please see the picture below.
    It is the snapshot of range profile when I use configuration2.cfg.
    Because guard length is 2, and noiseWin is 4, average of noise is higher than CUT obviously.
    I can't understand why the CUT is detected.
    What kind of things do you think could be the reason?

    Regarding question2, I can't find more info about detected object side info in SDK user's manual.
    Could you let me know what page should I see?

  • Hi Kei,

    Regarding question 1, in case configuration2.cfg, please see the picture below.
    It is the snapshot of range profile when I use configuration2.cfg.
    Because guard length is 2, and noiseWin is 4, average of noise is higher than CUT obviously.
    I can't understand why the CUT is detected.
    What kind of things do you think could be the reason?

    This does seem like problematic behavior. Are you using the most up-to-date code? Which binary and which visualizer are you using?

    Regarding question2, I can't find more info about detected object side info in SDK user's manual.
    Could you let me know what page should I see?

    See the section on cfarCfg for a description of that value.

    Best,

    Nate

  • Hi,

    Regarding question1, I am using IWR6843AOPEVM, xwr64xxAOP_mmw_demo.bin(SDK3.5, pre-built binary), demoVisualizer 3.6.
    By the way, does 6843AOP + 64xx binary support CFAR-CA?


    Regarding question2, I wasn't able to get the information that I wanted in the SDK user's manual.
    I have got additional result and let me explain.
    As I wrote above, when I use configuration1.cfg, since CFAR threshold is 15dB, detected object's minimum value of SNR should be 15dB.
    But the minimum value I got is about 11dB. Please see the figure below.
    But when I use some 2D detection setting, CFAR threshold and detected object's minimum SNR is the same(15dB).
    I would like to know the reason of this behavior.
    I found a thread that seems to be related to my question.
    https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/1109407/iwr6843-iwr6843--snr-level-drops-between-3d-and-2d-mode/4112164?tisearch=e2e-sitesearch&keymatch=6843%20side%20info%20snr#4112164

    Regards,
    Kei

  • Hi Kel,

    Regarding question1, I am using IWR6843AOPEVM, xwr64xxAOP_mmw_demo.bin(SDK3.5, pre-built binary), demoVisualizer 3.6.
    By the way, does 6843AOP + 64xx binary support CFAR-CA?

    I'm working on replicating this now. Unfortunately we're having some site issues that are preventing anyone from accessing the demo visualizer right now, but I will get back to you by the end of the day on Monday with results once the tool is back online. Thank you for your patience.

    For question 2, can you show a screenshot of a detected point with less than 15 dB of power? I see that you've made a plot, but I'd like to see it in the visualizer so we can ensure we're on the same page.

    Best,

    Nate

  • Hi Nathan,

    Regarding question1, I appreciate your cooperation.


    Question2, I think demoVisualizer does not have function to show SNR of detected object side info, So I used parser script example located at
    C:\ti\mmwave_sdk_03_05_00_04\packages\ti\demo\parser_scripts.
    When I tried to find root cause, I used 'IWR6843ISK', though the title of this post has 'IWR6843AOPEVM'. Because 6843AOP could not accept 2D setting somehow.
    Let me summarize condition again.

    board : IWR6843ISK
    bin : C:\ti\mmwave_sdk_03_05_00_04\packages\ti\demo\xwr68xx\mmw\xwr68xx_mmw_demo.bin

    1.
    configuration : configuration1_AddBpmCfg.cfg
    result : 1.dat
    -> minimum SNR of detected object is about 11.

    2.
    configuration : configuration1_2d_AddBpmCfg.cfg
    result : 2.dat
    -> minimum SNR of detected object is 15. It is equal to range CFAR threshold.

    Regards,
    Kei

    configuration1_AddBpmCfg.cfg     5707.dat

    configuration1_2d_AddBpmCfg.cfg     2146.dat

  • Hi Kei,

    I notice that the only difference between those two files is the number of antennas used. I will investigate internally further about this, but what happens if you set the threshold to say 19 in the faulty case, does this yield a minimum SNR of 15?

    Best,

    Nate

  • Hi Nathan,

    Yes, in the faulty case, the threshold 19 yields  minimum SNR of about 15.

    Regards,

    Kei

  • Hi Kei,

    What happens if you use 2 antennas instead of the 1 or 3 that you've shown? Do you see similar behavior?

    Best,

    Nate

  • Hi Nathan,

    I will out of office until next Wednesday and get back to you next Thirsday.

    Regars,

    Kei

  • Thank you Kei,

    I will investigate further on my end until then too.

    Best,

    Nate

  • Hi Nathan,

    When Tx1 and Tx3 was used, minimum value of SNR was 15.

    Regards,
    Kei

    configuration1_2d_Tx1Tx3_AddBpmCfg.cfg      Tx1Tx3.dat

  • Hi Kei,

    After conferring with our experts, we've found that in the main.c file found in 

    C:\ti\mmwave_sdk_03_05_00_04\packages\ti\demo\xwr64xx\mmw

    there is a function that converts the cfar threshold to a linear scale, but it does this based off the number of antennas:

    /* Convert CFAR threshold value */

    subFrameCfg->objDetDynCfg.dynCfg.cfarCfgRange.thresholdScale =

        MmwDemo_convertCfarToLinear(subFrameCfg->objDetDynCfg.dynCfg.cfarCfgRange.thresholdScale,

                                                    staticCfg->numVirtualAntennas);

    ... inside the MmwDemo_convertCfarToLinear() function...

    linVal = dbVal * (256.0 / 6.0) * ((float)numVirtualAntennas / (float)(1 << mathUtils_ceilLog2(numVirtualAntennas)));

    You can disable this by commenting it out and rebuilding.

    Best,

    Nate

  • Hi Nathan,

    Thank you for your confirmation. I understand.

    By the way, do you have any updates on my question1?
    I clicked 'This resolved my issue' button by mistake.
    Should I open a new thread?

    Regards,
    Kei

  • Hi Kei,

    Yes could you please open a new thread for that question while I investigate! Thank you for your patience.

    Best,

    Nate