• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 81 subscribers
  • Views 909 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Modifying RFFT-ADC Example on F28335

Jonatan Vanegas
Prodigy 150 points
Other Parts Discussed in Thread: ADS1278

Hi,

I am interfacing ADS1278 EVM-PDK  with Piccolo f28335 via MCBSP. When I get samples I want to compute FFT by using  the example from FPU library ("2833x_RFFT_ADC_RT"). this example use the ADC 12bits experimenter kit F28335 and as I know, It scale samples input int 16 from ADC by 1/4096 to float 32 and it compute FFT. As ADS1278 EVM-PDK has 24 bits resolution, I need to change the source code written in assemby and modify variable type to int 32 and to scale by 1/16777216 (2^24). Unfortunately, I dont know asembly lenguage and I dont know how do it. Could anyone help me out? 

This is the soude code.

RFFT_adc_f32.asm

  • 0
    TI__Mastermind 20850 points

    Hi, 

    You need to change the inverse scale factor in the assembly. Change this line

    MOVI32    R4H, 0x39800801

    to

    MOVI32    R4H, 0x33800001


    What you are doing is changing the 12-bit adc scaling factor (1/4095) to the 24-bit factor of (1/(2^24-1))

  • 0 in reply to Vishal_Coelho
    Prodigy 150 points
    Hi Vishal

    Thanks for you quickly reply,

    I did what did you say but the resuls don't agree, I think the variable type could be affecting because I get into unsigned int 32 input variables and the example works with 16 unsigned int input variables. and again, I don't know change it in the assembly code. Could this have to do?.

    Thanks,

    Jonatan.
  • 0 in reply to Jonatan Vanegas
    TI__Mastermind 20850 points

    Oh, i forgot about the change in data size, that will complicate things a bit. I guess we need a few more changes to the _rfft_adc_f32_Stages1and2and3andBitReverse() within the assembly file.

            MOV      AR0,#0Ah
        MOV      AH,*+XAR4[AR0]         ; FFT SIZE
;<---------------------------------------commented the line below
;       LSR      AH,1                   ; FFT SIZE/2 - for 16-bit input data
        MOV      AR0,AH
;       LSR      AH,3
        LSR      AH,2                   ; for 16-bit input data
        SUBB     AH,#1                  ; (Size / 8) - 1
        MOVL     XAR1,#0000h            ; index if memory is not aligned

        RPTB    _rfft_32_Last, AH
;------------------------------------------------------------------------------
;  Input buffer must be aligned for this code
;------------------------------------------------------------------------------
        NOP     *,ARP2
        MOVI32    R4H, 0x33800001       ;<------------(1/(2^24-1))
        UI32TOF32 R0H, *BR0++           ;I1 load
        UI32TOF32 R1H, *BR0++           ;I2 load
        MPYF32    R0H, R0H, R4H         ;I1 scale 
        UI32TOF32 R2H, *BR0++           ;I3 load
        MPYF32    R1H, R1H, R4H         ;I2 scale
     || MOV32     *-SP[I1], R0H         ;I1 save
        UI32TOF32 R0H, *BR0++           ;I4 load
        MPYF32    R2H, R2H, R4H         ;I3 scale
     || MOV32    *-SP[I2], R1H          ;I2 save           
        UI32TOF32 R1H, *BR0++           ;I5 load       
        MPYF32    R0H, R0H, R4H         ;I4 scale
     || MOV32    *-SP[I3], R2H          ;I3 save
        UI32TOF32 R2H, *BR0++           ;I6 load 
        MPYF32    R1H, R1H, R4H         ;I5 scale
     || MOV32    *-SP[I4], R0H          ;I4 save
        UI32TOF32 R0H, *BR0++           ;I7 load 
        MPYF32    R2H, R2H, R4H         ;I6 scale
     || MOV32    *-SP[I5], R1H          ;I5 save
        UI32TOF32 R1H, *BR0++           ;I8 load 
        MPYF32    R0H, R0H, R4H         ;I7 scale
     || MOV32    *-SP[I6], R2H          ;I6 save
        MPYF32    R1H, R1H, R4H         ;I8 scale            
        MOV32    *-SP[I7], R0H          ;I7 save                             
        MOV32    *-SP[I8], R1H          ;I8 save

    So,

    1. the first change is to comment out the LSR AH,1 so that AR0 = FFT_SIZE (and not FFT_SIZE/2)- this will ensure that each time you use the bit-reverse addressing - the instructions with BR0++ - its getting the right location for a 32-bit sized data.
    2. Change the scaling factor to 1/(2^24 - 1)
    3. change all the UI16TOF32 to UI32TOF32