F28027 - Toubles with a FFT

May be you should try with simple input(predefined) and check output to debug FFT by it self, and then go to experiments with ADC?

Also, probably You should double check your code step by step..looking to fft lib documentation and examples.

http://www.random-science-tools.com/maths/FFT.htm - was VERY useful for me.

I can't get any detailed info, because i never used F28027 and corresponding fft lib.

I think fft.init(&fft);  and fft.win(&fft);   looks strange(if FFT lib is coded in this way - then OK).  My FFT lib(for F28335) have dedicated finctions such as RFFT_f32u(&fft); to compute FFT for example....

Hey Vasily,

thanks for the answer.

I changed my program that i´m using the acquisiton module and then the fft module. The aquisition module works fine, but I have still some problem with the FFT module. I wrote my program like the examples fromt the ti fft library documentation. So should work.

Does anyone have a working example for me? I tried the examples from TI but ther is the stb.h include file missing, so I can´t compile it.

Thanks

Thomas

Yes I found some mistakes in my file and i defined my N in an include file.

I will try to debug my program now to find exactly the mistake where my program stops working correctly.

Here you can find the fft library: http://focus.ti.com/docs/toolsw/folders/print/sprc081.html

Hey Tom,

I believe I am working on a similar project as you. I am trying to read an input off of the ADC and perform an FFT on the data.  I am also having some trouble with the FFT library.When I enter this line of code "RFFT32 fft=RFFT32_128P_DEFAULTS;", it wont compile and says the symbols "CFFT32_calc, CFFT32_init, RFFT32_mag, RFFT32_split, RFFT32_win" are undefined.

Here is my code, it looks pretty similar to yours as I have just been going from the example code given in the document that came with the library.  I'm not sure if I'm even going about this in the right manner, and haven't been able to debug at all because the compiler stops after the RFFT32 fft=RFFT32_128P_DEFAULTS line.

#include "PeripheralHeaderIncludes.h"
#include "DSP280x_EPwm_defines.h"         // useful defines for initialization
#include "fft.h"   

int i;
#define N 128;
unsigned short inputsig[128];

#pragma DATA_SECTION(ipcb,"FFTipcb");
#pragma DATA_SECTION(mag,"FFTmag");

RFFT32 fft;
RFFT32 fft=RFFT32_128P_DEFAULTS;

void init(RFFT32_handle);
void calc(RFFT32_handle);
void split(RFFT32_handle);
void mag(RFFT32_handle);
void win(RFFT32_handle);

   
long ipcb[128+2];
long mag[128/2+1];
const long win[128/2]=HAMMING128;

void main(void)
{

.......

.......
    fft.ipcbptr=ipcb; // FFT computation buffer
    fft.magptr=mag; /* Store mag. square in separate buff*/
    fft.winptr=(long *)win; /* Window coefficient array */
    fft.init(&fft); /* Twiddle factor pointer initialization */
  
//=================================
//    Forever LOOP
//=================================

    for(;;)  //infinite loop
    {
            // reads the values from adc into inputsig[128] <-this is what I am looking to take the FFT of
        for(i=0; i<128; i++){
       
            adc_result = AdcResult.ADCRESULT0;
            inputsig[i]= adc_result;        // Store ADC measurement to adc_result buffer
           
        }   
                                        // level of the PWM1A_DAC output (the value

   
    fft.win(&fft); /* Window data*/
    fft.calc(&fft); /* Perform FFT*/
    fft.split(&fft); /* Split function computation*/
    fft.mag(&fft); /* Obtain the magnitude square*/
                                       
    }                                               

    
}

Any help you can give would be greatly appreciated.

Thanks,

Matt

Hey Matt,

try it like this and it will work. The most is the same than in the TI example. Did you download ist?

#include "fft.h"                            // FFT include file
#pragma DATA_SECTION(ipcb,"FFTipcb");

//--- Create an instance of RFFT module
RFFT32 fft=RFFT32_1024P_DEFAULTS;            // Instance the FFT module

//--- FFT initialization
    fft.ipcbptr=ipcb;                            // FFT computation buffer
    fft.magptr=ipcb;                            // Store mag. square in separate buffer
    fft.winptr=(long *)win;                 // Window coefficient array
    fft.init(&fft);                                     // Twiddle factor pointer initialization

//--- Perform the FFT
    RFFT32_brev(ipcb,ipcb,N);            // resort the ipcb to N/2-point complex data sequence in bit-reversed order
    fft.win(&fft);                         // Window the input data
    fft.calc(&fft);                        // Perform N/2 complex FFT
    fft.split(&fft);                        // Split function computation to obtain the computation of a real FFT
    fft.mag(&fft);                       // Obtain the magnitude square

Now you have the result in your ipcb. I performed the calculation of the fft and the magnitude in the ipcb.

Best regards

Tom

Hey Tom,

Thank you for the very quick response. It helped me take care of some of the errors I was getting. I'm sorry to bug you again, but I am a bit of a novice when it comes to programming.  I think the way I am reading from the ADC is causing errors.I am creating an array, and reading in 128 values from the ADC (to which I am currently feeding a simple 1Khz sine wave) like this,

for(i=0; i<128; i++){
       
            adc_result = AdcResult.ADCRESULT0;
            inputsig[i]= adc_result;        // Store ADC measurement to adc_result buffer
           
        }   

I then tried

RFFT32_brev(inputsig,ipcb,N);

which I thought would take the 128 values of 'inputsig' as the source and 'ipcb' as the destination, but I am just getting random numbers popping up on the ipcb variable in the watch window.

Is this similar to how you acquired your data from the ADC? I have tried using the RFFT32_ACQ function instead of the RFFT32_brev, but it was giving me some errors.

Again, thank you very much,

Matt