• Resolved

IWR6843ISK-ODS: My Elevation angle FFT result is upside down.

Prodigy 80 points

Replies: 8

Views: 125

Part Number: IWR6843ISK-ODS

Hello. I’m trying to estimate the 3D location of the object using IWR6843ODS + DCA1000 data, but my results seem upside down.

 

There is no problem with Azimuth-angle FFT.

When an object is located on the left side (location with positive azimuth angle) of the sensor ( the position is specified in the above figure), the peak appears at (+vres), after 4-point FFT.

 

However, the result of elevation-angle fft is upside down.

When an object is located on the top (location with a positive elevation angle), the peak appears at (-vres).

Also, when it is located on the bottom (location with a negative elevation angle) the peak appears at (+vres).

 

Do you know what caused this error?

I don’t think it is code error, because the result is accurate for azimuth angle direction.

Is the TX/RX number above incorrect?

  • Hi Taehyeong,

    I've asked an expert to look into this and we should have an answer for you in the next few days.

     

    Cheers,

    Akash

  • For Azimuth angle, you need to look through the back of the antenna, because that is the TX direction. 

    So your antenna order for Azimuth needs to be reversed.   

    For elevation angle, your antenna order is correct.  

    Best,

    Zigang

  • In reply to zigang Yang:

    I still don’t understand why my range-elevation map is upside down.

    From what I understand, when the object is above the sensor the phase must have increased through x[0]~x[3] (the observation from TX0-RX1 /TX1-RX1/TX0-RX2/TX1-RX2), therefore peak must appear at X[1].

    However as I mentioned at the above, the peak appears at X[3].

    Likewise when the object is below the sensor, the beak appears at X[1], though the phase must have decreased.

  • In reply to Taehyeong Ha:

    You can look through the training video on angle of arrival estimation:

    https://training.ti.com/intro-mmwave-sensing-fmcw-radars-module-5-angle-estimation?context%5B0%5D=1128486&context%5B1%5D=1139153&context%5B2%5D=1128546

    Then regarding your elevation vector, yes, if the object is above the sensor, the phase should increased through x[0] ~ x[3]. And the angle estimation should be positive.  However, the FFT output should be FFT-shifted.   

    But regarding the azimuth angle,  there is a confusion when you say the object is on the left.  depend on where are you facing.  If you are facing the sensor facing direction and the object is on the left, then the antenna from left to right is actually x[3], x[2],x[1], x[0] instead of the other way.   

    By the way, when you do FFT, you should really use a 64 or 128 point FFT instead of 4 point FFT.  And you still need to interpret the final angle as theta = asin(w_est/pi), where w_est is the peak location in your FFT output. 

    Best,

    Zigang

  • In reply to zigang Yang:

    Thank you for your answer, I’ll try the 64 point FFT

    However back to my original question, whether I use FFT-FFT shift or Capon-Beamforming,

    peak appears at a positive elevation angle, even though the object was below the sensor.

    And this does not seem to match your answer or the training video.

    Do you have any more suggestions about this issue please?..

  • In reply to Taehyeong Ha:

    Can you check the chirp configuration?  

    chirpCfg 0 0 0 0 0 0 0 1

    chirpCfg 1 1 0 0 0 0 0 2

    chirpCfg 2 2 0 0 0 0 0 4

    Are you sending the TX order right?  By the way, you have named the TX as RX and RX as TX in your original post.  Please double check whether you are taking x[0] ~ x[3]  right from the captured data?

    Best,

    Zigang

  • In reply to zigang Yang:

    Hi, 

    Just realize that the elevation has a build-in phase difference for different antenna.  

    There is 180 degree of phase jump for RX2/RX3 comparing to RX1/RX4.   In summary, the phase shifting  of all 12 visual antenna is: 

    % ch-1 ch-4 ch-9   ch-12
    % ch-2 ch-3 ch-10 ch-11
    %                 ch-5   ch-8
    %                 ch-6   ch-7
    phaseRotRx = [1, -1, -1, 1, 1, -1, -1, 1, 1, -1, -1, 1];

    Please apply the phase rotation before angle spectrum calculation.

    Best,

    Zigang

  • In reply to zigang Yang:

    It works.

    Now I have an upright range-elevation map!

    Thank you very much!!