TMS320F28377S: VCU2 RFFT

Steve Mangette

Prodigy 20 points

Other Parts Discussed in Thread: TMS320F28379D, C2000WARE

With respect to the output format of the 2837x_vcu2_rfft_512 function,

Please describe the format of the output data.

DC, f1, f2, . . . fmax OR -fmax, -fmax+1, . . . -f1, DC, f1 . . . fmax-1, fmax OR ?

If single sided, have the positive and negative counterparts been combined, or 2x been performed on single ended?

Has the output been divided by N, the number of points?

Is output complex requiring (r^2 + i^2)^0.5?

Steve

over 4 years ago

0 Lori Heustess over 4 years ago

TI__Guru* 93110 points

Hi Steve,

I would like to caution using the VCU FFT - we are not supporting the FFT/Viterbi instructions on future devices. Only the CRC instructions are on our latest devices. The reason is it is only a 16-bit fixed point FFT and was developed for a specific application area. We see more applications requiring a 32-bit floating-point FFT and recommend that instead.

For your question, I did some searching and found some information that may help. In this thread Vishal mentions the output format and references the paper the algorithm comes from.

https://e2e.ti.com/support/microcontrollers/c2000/f/171/t/554522?FFT-example-on-VCU-II

Regards

Lori

0 Steve Mangette over 4 years ago in reply to Lori Heustess

Prodigy 20 points

Hello Lori,

Thank you for the reply. The actual TI part we are using is the TMS320F28379D. I am a systems engineer and so do not know much about the part. What is the appropriate FFT algorithm we should be using for this part?

Steve

0 Lori Heustess over 4 years ago in reply to Steve Mangette

TI__Guru* 93110 points

Steve,

I suggest using the 32-bit floating-point FFTs instead of the 16-bit fixed-point VCU.

C:\ti\c2000\C2000Ware_3_04_00_00\libraries\dsp\FPU

Regards

Lori

C2000™︎ microcontrollers

C2000 microcontrollers forum

TMS320F28377S: VCU2 RFFT