• State TI Thinks Resolved
  • Locked Locked
  • Replies 14 replies
  • Answers 8 answers
  • Subscribers 33 subscribers
  • Views 3514 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

IWR1443BOOST: Phase discontinuity between RX channels from each TX antenna

Santi Ortega
Prodigy 10 points
Part Number: IWR1443BOOST
Other Parts Discussed in Thread: IWR1443, IWR1642

PROBLEM

Platform: IWR1443BOOST

We are seeing a phase discontinuity between the RX channels from each TX antenna.

•The target is stationary, so this isn’t caused by Doppler shifts
•The target is at z=0, so it isn’t additional elevation phase
These phase jumps are causing additional peaks in the FFT and, as a result, errors in position estimation.
POINT TARGET EXAMPLE
 
VIRTUAL RX CHANNEL PHASE
AZIMUTH/ELEVATION FFT
  • Using only TX1 or TX3 gives the correct FFT peak, corresponding to 33 deg (wx = 1.75)
  • Combining TX1 and TX3 virtual channels gives the wrong peak, which corrupts the azimuth estimate
  • Incorrect wx and constant phase offset of TX2 phase gives incorrect value for z

 

NOTES

We have tried the calibration procedure described in: demo/xwr14xx/mmw/docs/doxygen/html/index.html -> "Range Bias and Rx Channel Gain/Offset Measurement and Compensation"

The phase offsets between antenna channels seem to not be constant, so calibration doesn't solve the problem.

Do you have any suggestions about what might be happening and how can we address the problem?

Thanks.

  • 0
    TI__Mastermind 25795 points
    Santi Ortega,

    Could you post Raw data and post processing scripts?

    Thanks and regards,
    CHETHAN KUMAR Y.B.
  • 0 in reply to CHETHAN KUMAR Y.B.
    TI__Mastermind 37500 points

    Hello,

    If you have fixed radial distance, and fixed angle, to a reference point on the antenna.    We would normally expect with MIMO processing with separate Tx1, and Tx3 Chirps,  That Rx1-4 with Tx1, would provide for Rx5-8 with Tx3.  

    - if your phase discontinuity does not show up with this style of two chirps, one with Tx1 and one with Tx3, then the physical antenna locations would be expected to be OK    Tx1 -> Tx3 azimuth separation of 2*lambda, Rx1->Rx2-> Rx3-> Rx4 azimuth separation of lambda/2.

    - we would expect the L3 data collected over CCS, or the use of DevPack&TSW1400 or DCA1000 over two chirps for the Rx receiver to provide for the 4 actual Rx, and Tx3 pseudo Rx antennas.    

    - in the case where you have activated Tx1 and Tx3 at the same time,  did you apply BPM to one of these Transmitters so you can separate  the phase difference?   The Tx1 and Tx3 with the same Tx-phase will have a different radial distance and have two frequency peaks at each receiver.   Using the BPM for one of the Tx outputs, and separately processing the Rx1.4

    - over the range of the chirp, the antenna spacing in phase changes.  If you cover a smaller chirp range does the behavior change.  

    - I have included the antenna designer simulation of the antenna, I don't have the IWr1443 antenna characterization report for the EVM, I will ask the other engineers.

    - as another application engineer has asked, if you can send the config file, and FMCW Rx data over the separate Tx1, separate Tx3, combined Tx1, Tx3 Rx data we can look at those items.

    Regards,

    Joe Quintal

    swrs554 - MIMO app note

    IWR1642 & IWR1443 (later slides) EVM antenna simulation

    swra554.pdf

    1643.AWR1642_AWR1443_antennas_simulation2.pdf

  • 0 in reply to Radio-Joe
    Prodigy 190 points
    Hi Joe Quintal,

    I also encounted this problem in Platform: IWR1443BOOST.

    When I get the tx1-rx4 and tx3-rx4 data from radio studio, phase discontinuity between RX channels from this twoTX. But I found, for the same data, the post processing of radio studio can get the same result compared 1tx-4rx with 2tx-4rx, while my fft processing code can not.

    I think it need to do the calibration for each Tx---maybe some deviation between two tx antenna, while it may have compensation in your post processing code, And how can we do?

    Thanks,
    Weihua
  • 0 in reply to WEIHUA JIANG
    TI__Mastermind 37500 points

    Hello,

    There is an update from radar studio to mmwave studio.

    For the radar studio, and had written some  instructions for a 1642 experiment.  It did not deal with phase processing.  There are some discussions of the  special post processing in matlab.  The  PDF is attached.

    In the mmwave SDK user guide, and ICD (radar studio docs folder) document.  There are special discussion of the newer mmwave SDK calibration functions:

     -    mmwave_sdk_userguide, measureRangeBiasAndRxChanPhase

     -    mmwave_sdk_userguide, compRangeBiasAndRxChanPhase

     -    mmwave_sdk_userguide, compRangeBiasAndRxChanPhase

     -     mmwave_sdk_userguide, bpmCfg - allows the extension of Rx through BPM and TDM MIMO

    The other concern with phase processing, is the reference point.  Tx,1 to Rx1-2, Tx1 to Rx2-3, Tx1 to Rx3-4 .  Eventually as you get close to Tx1, there is less Tx-Rx isolation.  You can adjust the Tx power backoff to a few db to see if that helps.

    You can capture the radar studio output from TSW1400 or mmwave studio DCA1000 , to make sure the same mmwave radar output is used in your matlab as is used by the Post Processing in Radar studio.

    I will ask some other application engineer.

    Regards,

    Joe Quintal

    notes_RadarStudioPostProcess.pdf

  • 0 in reply to Radio-Joe
    Prodigy 190 points

    Hi Joe Quintal,

    The Tx power backoff is all set as default value in 0dB.

    I do not have DCA1000 board, but I have get the TDM-MIMO raw data from TSW1400 using radar studio, and compared the phase processing result between my matlab code and radar studio post processing.

    1) According to my maltab processing, for 1 transmitter and 4 receiver, TX0-RX(1-4) and TX2-RX(1-4) can get the same result. But when I couple this two data and get virtual 8 receiver antenna, it always has 6 degeree offset compare to the 1tx-4rx system. I checked the phase of this 8 receiver, and found phase discontinuity between RX channels from each TX antenna just as shown by in this post.

    2) For radar studio post processing, 1tx-4rx and 2tx-4rx system can get the same result while there is phase discontinuity between RX channels from each TX. So I think it may have the phase compensation in Tx antenna with your  radar studio post processing code. And how can I do this calibration?

    I also have posted the question in there( ) at few days ago, but still do not get response. 

    Thanks,

    Weihua

  • 0 in reply to WEIHUA JIANG
    TI__Mastermind 37500 points
    Hello,
    I have forwarded your question to a systems engineer.

    My understanding is that the antenna spacing of lambda/2 for Rx-Rx, and lambda for Tx,
    result in the 2*lambda spacing for Tx1, Tx3. Using the azimuth angular resolution = (azimuth FOV) / (number Rx antenna * number MIMO Tx antenna). 120/8 = 15 degrees. In MIMO we can have TDM (one Tx on) or dynamic BPM. (two Tx on zero phase, two Tx on Tx3 -180 phase).

    I have asked for the PostProcessing Matlab source code so I can help answer, I do not have this currently.

    In the mmwave SDK there is a phase calibration API, you could look this up in the doxygen documentation, the source code is provided with the mmwave SDK Out of Box demo for both 1443(HWA & ARM) and 1642 (DSP & ARM)
    (in the docs folder\mmwave_SDK_user guide)
    - compRangeBiasAndRxChanPhase API
    - measureRangeBiasAndRxChanPhase

    Regards,
    Joe Quintal
  • 0 in reply to Radio-Joe
    Prodigy 190 points
    Hi Joe Quintal,

    Thanks a lot.

    1) As my prior post, I think it is not a problem of angle reslution.

    2) I have checked the mmwave_SDK_user guide document, that part as you mentioned introduce the Rx phase calibration, but in our case, I think we need to calibration Tx antenna.

    Thanks,
    WEIHUA
  • 0 in reply to WEIHUA JIANG
    TI__Mastermind 37500 points
    Hello,
    We are trying to track down the Post Processing radar studio source file, to see if there are any special processing steps for MIMO.

    Regards,
    Joe Quintal
  • 0 in reply to Radio-Joe
    Prodigy 190 points

    Hi Joe Quintal,

    Thanks. Please let me know if you have any progress.

    Sincerely,

    Weihua

  • 0 in reply to WEIHUA JIANG
    TI__Mastermind 37500 points
    Hello we, have contacted the Radar Studio software engineer to get the matlab script. We will post it when its received.
    REgards,
    Joe
  • 0 in reply to Radio-Joe
    Prodigy 190 points
    Hi Joe,

    Do you have any news about the radar studio post processing from your engineer?

    Thanks,
    Weihua
  • 0 in reply to Radio-Joe
    TI__Mastermind 37500 points
    Hello,
    I pinged the engineer again to get the file. I don't have it yet.
    Regards,
    Joe Quintal
  • 0 in reply to Radio-Joe
    Prodigy 190 points
    Hi Joe,

    Do you have detail reference document about the radar studio post processing ?

    Thanks,
    Weihua
  • 0 in reply to WEIHUA JIANG
    TI__Mastermind 37500 points
    Hello if you download the dfp mmwave (radar) studio package.
    " www.ti.com/.../mmwave-dfp "
    In the docs folder is a user guide.
    I am still waiting for the matlab script.
    Regards.
    Joe Quintal