• State Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 1 answer
  • Subscribers 30 subscribers
  • Views 208 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.

AWRL6432BOOST: Reviewing Motion and presence demo

JINHYUN JEONG
Prodigy 200 points
Part Number: AWRL6432BOOST

Dear Expert TI team

Hello, 

i am reviewing motion and presence demo. (  MMWAVE_L_SDK_05_03_00_02 / using TrackingClassification_MidBw profile - Major&Minor both enable)

and i want to check whether what i understand is correct or not.

Q1. According to Timing Diagram in auto mode, Major Radar cube is updated for every pair of chirps (64 - ping / 65 -pong )

Thus, only the last pair of chirps will likely be reflected in the last radar cube(B). Is it correct information?

Even though only one frame is used, radar cube[0] size is params->numRangeBins * params->numVirtualAntennas * sizeof(cmplx16ReIm_t) * params->numDopplerChirpsPerProc; 

Actually, i found the code that only update ping or pong same buffer for every pair of chirps.

 (rangeprochwa.c "rangeProcHWA_ConfigEDMA_DataOut" )

Q2. contrary to major cube case ,Accumulates minor cube data for a certain number of frames(8) only in low power mode.

(Line 995 in rangeprochwa.c). Is it correct information?

Q2-1 . is there a reason why debugging is only in low power mode? , if i want to debug it in lower power mode , how can i do? 

Please share your opinion.

Best regards

JINHYUN JEONG

  • 0
    Prodigy 200 points

    To supplement Q-1, the document says that is is updated for each chirp, but in the code it seems that only Chirp 0/1 is updated.

  • 0 in reply to JINHYUN JEONG
    TI__Expert 7075 points

    Hello,

    I will need a day or two to confirm a few points with another engineer on your first two questions, but the reason why debugging is disabled in low power mode is because the JTAG connector disconnects when entering low power mode. If you would like to debug in low power mode, you can emulate low power mode - i.e., the application will wait for the length of time the device is normally in low power mode without switching modes - by changing the lowPowerCfg to 2. 

    Once again, give me some time to look into your first two questions, but hopefully my prior answer helps regarding low power mode debugging.

    Regards,

    Kristien

  • 0 in reply to Kristien Everett
    Prodigy 200 points

    Hello, Good morining :)

    Thank you for your quickly reply. 

    The question of Q2-1 is clear by your opioion.

    i will set lowPowerCfg value 2 to debug in lower power mode. 

    Thank you

    Best regards

    JINHYUN JEONG

  • 0 in reply to JINHYUN JEONG
    TI__Expert 7075 points

    Hey Jinhyun,

    Thank you for your patience. I will need another day as I'm waiting for a response from one of our DPC experts, but I will update you as soon as I can.

    Regards,

    Kristien

  • 0 in reply to Kristien Everett
    Prodigy 200 points

    i found some clue of my question.

    1. HardwardAcclerator perform FFT (2point fft and Range fft) during N_loop(32 " numChirpsperFrame ") before EDMA(out) transefer data form HWA to Radar cube.

    Therefore, EDMA(out) only need to consider the one destination address. 

    if it is correct information, then Q1 will be clear.

      

    2. Because the parameter set of 2 point FFT is set to HWA_TRIG_MODE_DMA, HWA(2-point fft) is performed every time data is transferred from ADCbuffer to HWA(M0 or M2)

    On the contrary, the parameter set of range FFT is set to HWA_TRIG_MODE_IMMEDIATE, HWA(range FFT) is performed immediately.

    if it is correct information, i want to ask 1 more follow-up question.

    i would like to see in the code what properties trigger the 2-point FFT inside EDMA,

    in other words, i want to know connection between EMDA-in and HWA 2-point FFT parameterset.

  • 0 in reply to JINHYUN JEONG
    TI__Expert 7075 points

    Hey Jinhyun,

    To clarify on the two radar cubes: in minor mode, the radar cube is accumulated across four or eight frames. This is determined by the third to last parameter of the sigProcChainCfg CLI command, numMinorMotionChirpsPerFrame, which is four frames for all demo .configuration files, expect for the tracking classification configuration files where its eight instead.

    Regarding the auto mode, I'll use an example to illustrate how the cubes are divided up. Let's assume there are 8 chirps per frame and 2 chirps per frame are used for minor motion detection - i.e., numMinorMotionChirpsPerFrame = 2. In that case, the minor radar cube will combine the first 2 chirps from the current frame and previous three frames to create an 8 chirp block whereas the major radar cube would be made up of all chirp samples from the current frame. Note that the first chirps are always used for sake of simplicity in the software architecture, but there would likely be very little difference in performance if using another set of chirps in the frame. You can also see the diagram below as a visual aid. 

    Hopefully this helps, but let me know if you have any other questions.

    Regards,

    Kristien

  • 0 in reply to Kristien Everett
    Prodigy 200 points

    Hello , Kristien

    i can understand detail of making two radar cube process. Thank for sharing your opioion.

     my original question is  " only the last pair of chirps will likely be reflected in the last radar cube(B) ": 

    however, i have found some information taht says the above original question is incorrect.

    "The FFT(2point , range FFT) has been completed all of chirps within one frame before EDMA transfers the data from HWA to Major Radar cbue ( N_loop )"

    is this correct?

    if yes, i have additional questions lkie the blow

    • Q1. i would like to know connection between EMDA-in and HWA 2-point FFT parameterset.

    When data is transfered from the ADC buffer to HWA by EDMA(in) , a trigger occurs and 2-point FFT performed

    i was able to find "HWA_TRIG_MODE_DMA" in the HWA parameterset, but i could not find a parameter in EDMA-in taht wolud trigger the HWA.

     

    Thank you for helping.

    Best regards

    JINHYUN JEONG

  • +1 in reply to JINHYUN JEONG
    TI__Expert 7075 points

    Hey Jinhyun,

    The FFT has been completed on all chirps for a given frame before the EDMA transfer is correct. The range processing HWA config function, DPU_RangeProcHWA_config, should configure the EDMA trigger through rangeProcHWA_ConfigEDMA_DataOut which is called by rangeProcHWA_ConifgureHwaEdma. This is mentioned around line 1023 of the rangeProcHWA_ConfigEDMA_DataOut function.

    Regards,

    Kristien

  • 0 in reply to Kristien Everett
    Prodigy 200 points

    Thank you for helping Slight smile

    Have a good day

    Best regards

    JINHYUN JEONG

  • 0 in reply to JINHYUN JEONG
    TI__Expert 7075 points

    Hey Jinhyun,

    I'm glad I was able to help! I will be closing this thread now.

    Regards,

    Kristien