• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 1 answer
  • Subscribers 32 subscribers
  • Views 590 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.

TIDEP-01012: TDA2 Cascade Radar RxAntenna Offset issue

Nishant G
Expert 1600 points
Part Number: TIDEP-01012
Hello,
Based on the documentation of TI cascade EVM (RF Board) and the TX Antenna offset values in the use-case 'cascade_radar_object_detect', for the 4 AWR1243s present on the EVM, I believe the order of the Tx Antenna on the board is
S3- M- S2- S1 (Slave 3, Master, Slave 2, Slave 1- From Left to Right)
(considering Tx3 of slave 3 as reference position)
 
This is defined in the following code:
Chanis_RadarAntOffset gCascadeRadarTxAntArray[CHAINS_RADAR_CASCADE_NUMSENSORS][CHAINS_RADAR_CASCADE_MAX_TX_ANTENNA_PERSENSOR] =
{
    /* Master */
    {   /* TX1, TX2, TX3 - {Azimuth offset, Elevation offset (in lambda/2)} */
        {11U, 6U}, {10U, 4U}, {9U, 1U}
    },
    /* Slave 1 */
    {   /* TX1, TX2, TX3 */
        {32U, 0U}, {28U, 0U}, {24U, 0U}
    },
    /* Slave 2 */
    {   /* TX1, TX2, TX3 */
        {20U, 0U}, {16U, 0U}, {12U, 0U}
    },
    /* Slave 1 */  // I think there is a typo here in the SDK. It should be Slave 3
    {   /* TX1, TX2, TX3 */
        {8U, 0U}, {4U, 0U}, {0U, 0U}
    }
};
 
However, in the same c file, the receiver Antenna offset seems to be different than what it is expected.
Considering for the table aoa_sensorRxOffset index 0 = Master, 1=Slave1, 2= Slave2 and 3=Slave3 and considering the Rx1 of Slave 3 as reference for the relative position of every Rx antennas,  I think that the value sensorAzimOffs configured is incorrect for every sensor.
It is defined as below in the SDK:
Void Chains_cascadeRadarFillRxAntennaOffsets (aoa_rxAntOffset_t *aoa_sensorRxOffset)
{
    aoa_sensorRxOffset[0].sensorAzimOffs = 0; 
    aoa_sensorRxOffset[1].sensorAzimOffs = 11;
    aoa_sensorRxOffset[2].sensorAzimOffs = 50;
    aoa_sensorRxOffset[3].sensorAzimOffs = 46;
}
 
 
Could you check if the rxAntOffset should be as below?
  aoa_sensorRxOffset[0].sensorAzimOffs = 11; 
  aoa_sensorRxOffset[1].sensorAzimOffs = 50;
  aoa_sensorRxOffset[2].sensorAzimOffs = 46;
  aoa_sensorRxOffset[3].sensorAzimOffs = 0;  
 
I look forward to your reply. Thank you!
 
Best Regards,
Nishant

 

  • 0
    TI__Guru 50045 points

    Hi Nishant,

    Thanks for catching the typo in the comment.

    For the RxAntennaOffsets, it was intended to set them that way as the algorithm expecting the order to be "Slave4", "Master1", "Slave2", and "Slave3".

    However, I need to check with my colleague to determine where in the data flow the mapping is being set since captured radar data is in the order of "Master1", "Slave2", "Slave3", "Slave4".

    Regards,
    Stanley

  • 0 in reply to Stanley Liu
    Expert 1600 points

    Hi Stanley,

    Thanks for your reply and for confirming that the order is 'Slave 4, Master 1, Slave 2, Slave 3'.

    Awaiting your reply where the data mapping is done for the captured radar data to set it in above order.

    Thank you!

    Best Regards,

    Nishant

  • 0 in reply to Nishant G
    Expert 1600 points

    Hi Stanley,

    Is there any update on this issue? Also, could you confirm with the team if all the use-cases for cascade have been tested with the Cascade EVM?

    Best Regards,

    Nishant

  • 0 in reply to Nishant G
    TI__Guru 50045 points

    Hi Nishant,

    Sorry for the delay. I am still waiting for the response from the team for both of your queries.

    Regards,
    Stanley

  • 0 in reply to Stanley Liu
    Expert 1600 points

    Hi Stanley,

    Thanks for your help. It will be great if I could know the tested maximum frames per sec for Network Tx for Cascade use-case 'b'- cascade_radar_beam_form. When we tested it with the default configs, the frame rate was found too low (around 2 fps). Maybe it's just some config issue but wanted to know your test results as well for this use-case as well as other Cascade use-cases.

    Thank you!

    Best Regards,

    Nishant

  • 0 in reply to Nishant G
    TI__Guru 50045 points

    Hi Nishant,

    Algorithm team indicates the expectation is data will come in with the device order of 4-1-2-3.

    For beam forming use case, are you using DSP or EVE for FFT?

    Regards,
    Stanley

  • 0 in reply to Stanley Liu
    Expert 1600 points

    Hi Stanley,

    I plan to use EVE for FFT. Does it matter which I use for FFT?

    Best Regards,

    Nishant

  • 0 in reply to Nishant G
    TI__Guru 50045 points

    EVE 2D-FFT performance was worse than DSP initially.

    But we found out that was due to EMIF configuration not set to duel EMIF interleave mode in bootloader by default.

    If you get different performance from DSP vs EVE, that means your bootloader doesn't have the optimal build option.