LAUNCHXL-F28379D: Understanding Fixed Point RFFT

Can anyone please help with this ? 

Hi,

Please give me a couple of days to look into this and revert back.

-Shantanu

HI Shantanu,

Thanks for your reply. Please try to help as earliest possible. 

Thank you.

-Kuldeep

Hi,

Anyone can help with this ?

Thank you.

-Kuldeep

Hi Shantanu and Others,

Can anyone help with this?

Thank you.

-kuldeep

Hi,

I am unable to replicate your problem. I ran with both N = 64 and 256. Please verify the hamming window and RFFt parameters that you are using. Also try to increase the value of N to see if the problem goes away. 

-Shantanu

Hi Shantanu,

I have the code as below - 

1. Declaration

//
// Included Files
//
#include "driverlib.h"
#include "device.h"
#include <time.h>
#include <stdlib.h>
#include <stdbool.h>
#include <stdint.h>
#include <string.h>
#include <stddef.h>
#include "fft.h"
#include "fft_hamming_Q31.h"
#include "math.h"

//
// Defines
//
#define FFT_SIZE            64
#define q31convfactor       1651910   // Range of the input - 0-1299 (0-1V) => to bring into range of 0-1 divide by 1300 and then to get Q31 format multiply with 2^31 = (approx.) 1651910

//
// Globals
//

// Create the FFT buffers, ipcb and ipcbsrc, in sections with the same name
#ifndef __cplusplus
#pragma DATA_SECTION(ipcb, "FFTipcb");
#else
#pragma DATA_SECTION("FFTipcb");
#endif
int32_t ipcb[FFT_SIZE+2];

#ifndef __cplusplus
#pragma DATA_SECTION(ipcbsrc, "FFTipcbsrc");
#else
#pragma DATA_SECTION("FFTipcbsrc");
#endif
int32_t ipcbsrc[FFT_SIZE];

// Declare and initialize the structure object.
// Use the RFFT32_<n>P_DEFUALTS in the FFT header file if
// unsure as to what values to program the object with.
RFFT32  rfft = RFFT32_64P_DEFAULTS;

// Define window Co-efficient Array
// Note: Windowing is not used in this example
const long win[FFT_SIZE/2]=HAMMING64;

int16_t test_input[FFT_SIZE] = {645,745,842,931,1011,1080,1138,1185,1222,1250,1270,1283,1290,1290,1283,1269,1247,1220,1188,1152,1114,1077,1041,1006,973,941,908,874,835,793,747,697,645,594,544,498,456,417,382,350,318,285,250,214,176,139,103,71,43,22,8,1,1,8,21,41,69,106,153,211,280,360,449,546};

2. execution code -

//Initialization 
     //Clean up input/output buffer
     for(i=0; i < (FFT_SIZE+2); i=i+2){
       ipcb[i]   = 0;
       ipcb[i+1] = 0;
     }



//In While Loop
       //input signal
       for(i=0; i < FFT_SIZE; i++){
          ipcbsrc[i]  = (long)  q31convfactor*test_input[i];  //Q31
       }

        RFFT32_brev(ipcbsrc, ipcb, FFT_SIZE); // real FFT bit reversing

        rfft.ipcbptr = ipcb;                  // FFT computation buffer
        rfft.magptr  = ipcbsrc;               // Magnitude output buffer

        rfft.winptr  = (long *)win;           // Window coefficient array
        rfft.init(&rfft);                     // Twiddle factor pointer initialization

        rfft.calc(&rfft);                     // Compute the FFT
        rfft.split(&rfft);                    // Post processing to get the correct spectrum
        rfft.mag(&rfft);                      // Q31 format (abs(ipcbsrc)/2^16).^2

Please have a look and let me know if I am doing something wrong in this?

Hi,

This looks correct. Have you aligned ipcb correctly in the linker cmd file? Also, can you run the stock examples in c2000ware and compare the input and output in your matlab model? We also provide a matlab script in c2000ware to help you verify. This is just to isolate the issue.

-Shantanu

Hi Shantanu,

Thank you for your input.

As per document the magnitude output is in the form of Q30 format but in the example code they have mentioned Q31 as below - 

After checking the document and scaling the values as per Q30 we are able to get the correct dc component but th fundamental and other harmonic components require a multiplication of a factor 2.

where q30convfactor is 1073741824 and adcrefconv is 1volt 

can you help with this if its due to RMS value or some other thing ? 

Thank you.

-Kuldeep