AWR1843BOOST: MRR Beamsteering - HWA

Hi Cesar,

For "HWA input is Mapped "and "radar cube format" you mentioned, is it like the diagram shown as below? 

./ti/mmwave_sdk_03_06_02_00-LTS/packages/ti/datapath/dpu/rangeproc/docs/doxygen/html/dpu_rangehwa.html

 Or any other similar diagram  to show it?

Thanks

BRs

Bruce

Hi Cesar,

 Where I can find  HWA configuration(Mapped input mode) in MRR Beamsteering source codes?

Thanks

BRs

Bruce

Hi,

The HWA State Machine Register FFT1DEN controls the ADC buffer sharing mode.

For Range FFT it is enabled in dss_main.c in

Yes, this is similar

thank you

Cesar

Hi Cesar,

  Is it correct for below understanding?

    "HWA_FEATURE_BIT_ENABLE"  => Mapped

    " HWA_FEATURE_BIT_DISABLE" => Isolated 

Hi Cesar,

 For MRR in MmwDemo_interFrameProcessing() function, I am confused for Slow chirp .

For  radarCube[slowChirpOff] to obj->dstPingPong[0], I copy related codes and have the chart to show my understanding.

void MmwDemo_interFrameProcessing(MmwDemo_DSS_DataPathObj *obj, uint8_t subframeIndx){

 ...

 slowChirpOff = obj->numDopplerBins * obj->numVirtualAntennas;

...

 // setup the first virtual antenna's data to be transferred.
 / / This needs to start with chirp 128 - the "slow" chirps.
 Config_2d_and_Azm_Edma(obj, 0, (uint8_t *)&obj->radarCube[slowChirpOff]);

 ..

}

Config_2d_and_Azm_Edma() => EDMAutil_configType2b()

In EDMAutil_configType2b() function for this case, related EDMA config parameters are as below,

bCount = 128;
cCount = 4
sourceBindex = 16
destinationBindex = 4
sourceCindex = 4
destinationCindex=512 

 The Radar cube and dstPingPong are as the picture:

The value copied from Radar cube[512] to dstPingPong [0], is it correct? Or any misunderstanding?

Thanks

BRs

Bruce

Hi Cesar,

  Thanks for the document, one thing I'd to clarity for the source codes about MRR.

  Since  array size of  dstPingPong is 256, but cCount is 4 about below operation.

   Will it repeat 4 times automatically or just once ? 

   if it's just once, why set cCount as 4?

In EDMAutil_configType2b() function for this case, related EDMA config parameters are as below,

bCount = 128;
cCount = 4
sourceBindex = 16
destinationBindex = 4
sourceCindex = 4
destinationCindex=512 

Hi Cesar, 

Yes, I have read EDMA chapter in the TRM.

From the codes , CCount is the number of Rx Antennae, so it will repeat 4 times. 

Then My question is:

Since  array size of  dstPingPong is 256, but cCount is 4 .

/* 2D FFT. */
MMW_ALLOC_BUF(dstPingPong, cmplx16ReIm_t, heapL1start, MMWDEMO_MEMORY_ALLOC_DOUBLE_WORD_ALIGN, 2 * obj->numDopplerBins)

In EDMAutil_configType2b() function for this case, if it repeats 4 times, it will exceed 256. (128*4 = 512)

Any EDMA configuration I missed? 

Thanks

BRs

Bruce

Hi,

I understand your question.

I would need to review and run the code

Unfortunately I will only be able to do that next week

thank you

Cesar

Hi Cesar,

Thanks for your reply.

Another question is about EDMAutil_configHwaTranspose() function.

It is SyncAB type transfer as well, how does  it repeat 128 times(cCount = numChirpsPerFrame / 2) ?

It will transfer HWA data to radar Cube as below chart. 

Is any different setting between EDMAutil_configType2b() and EDMAutil_configHwaTranspose() ?

EDMAutil_configType2b()  performs the transfer of one frame.

EDMAutil_configHwaTranspose() performs 128 times frame by frame.

EDMAutil_configHwaTranspose() => 

config.paramSetConfig.sourceAddress = (uint32_t)pSrcAddress;
config.paramSetConfig.destinationAddress = (uint32_t)pDestAddress;

config.paramSetConfig.aCount = numAnt * 4;
config.paramSetConfig.bCount = numRangeBins;
config.paramSetConfig.cCount = numChirpsPerFrame / 2; => 128 times
config.paramSetConfig.bCountReload = 0; // config.paramSetConfig.bCount;

config.paramSetConfig.sourceBindex = numAnt * 4;
config.paramSetConfig.destinationBindex = numChirpsPerFrame * numAnt * 4;

config.paramSetConfig.sourceCindex = 0;
config.paramSetConfig.destinationCindex = numAnt * 4 * 2;

config.paramSetConfig.linkAddress = EDMA_NULL_LINK_ADDRESS;
config.paramSetConfig.transferType = (uint8_t)EDMA3_SYNC_AB;

Thanks

Bruce

Hi

You will notice that transpose transfer has intermediate chaining enabled. So, a new EDMA sync event is created after one frame teansfer has completed. This EDMA event is produced by the EDMA chained.

Thank you

Cesar