TDA3XEVM: Questions about algorithms in the radarFft2.c file

w w3

Intellectual 520 points

Part Number: TDA3XEVM

Edition:PROCESSOR_SDK_RADAR_03_06_00_00

Hi,

1.Is numBytesPerSample = sizeof UInt64 because it has I and Q two parts?

2.Why is pObj-> intermediateBufSize divided by 2？

pObj = (AlgorithmFxn_RadarFftObj *) alg_handle;
pInputVideoBuffer = (System_VideoFrameBuffer *) in_buf-> payload;
inBufAddr = (UInt8 *) pInputVideoBuffer-> bufAddr [0];
outBufAddr = (UInt8 *) pOutputVideoBuffer-> bufAddr [0];

Is this a type cast?

4.How to explain this operation？

5. FFT_points is selected by nextpowerof2, why does buf point to coeffaddr?

But where is the specific FFT operation performed? This is just select the number of FFT points

waiting for the replys，thanks！

over 5 years ago

0 Anuj Gupta over 5 years ago

TI__Genius 15985 points

Hi,

I will answer few of your question

1.Is numBytesPerSample = sizeof UInt64 because it has I and Q two parts?

Yes

3 Is this a type cast?

yes

Regards,

Anuj

0 Piyali Goswami over 5 years ago

TI__Mastermind 30245 points

2.Why is pObj-> intermediateBufSize divided by 2？

This is only when the output container format is 32 bits. This is because the intermediate buffer 1D FFT output is 16 bit container (hence factor of 2 (16 bit versus 32 bit). The intermediate buffer size is calculated as per the final output buffer size.

4. This comes from Chirp Profile format. We put a constraint to find the right profile in advanced frame configuration one must enable the CP data in the AWR12xx for the profile information to be captured.

Refer: 8 Chirp Parameters (CP) and Chirp Quality (CQ) data in mmwave_dfp_01_02_00_00\docs\AWR1xxx_Radar_Interface_Control.pdf

5. This Buf is just the windowing coefficients. The actual FFT happens in AlgorithmFxn_RadarFftDoWorkProcess on the EVE.

Thanks and Regards

Piyali

Processors

Processors forum

TDA3XEVM: Questions about algorithms in the radarFft2.c file