• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 82 subscribers
  • Views 343 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.

Fcing problem in using FPU FFT library for 2048 point

Arpan Sadhankar
Intellectual 335 points

Hi,

 

I am using the DSP FFT library (FPU v120) for C2000 series (tms320c28335).

My intention is to compute FFT of an input sine wave 0f 50 Hz over sampling rate 1000Hz. When I try to compute FFT for 512, 1024 and 2048 points i am getting strange outputs.

Please follow for input sine wave 0f 50 Hz over sampling rate 1000Hz -

FFT Points          - Highest peak

512                       - 24/25

1024                     - 49/50

2048                     - 101/102

Yes the output is shifting to either sides, when using lesser or higher FFT points than 1024.

1. Is the limitation with the library, that it can be used only for 1024 points..??

2. If yes, how can i use the same library for computing the FFT for 2048 point resolution..?

3. If 2 is not possible, is there any other alternative solution, or any other library to compute the FFT for 2048 FFT points..?

 

Thanks in advance, please respond with any or all possible solutions..

 

- arpan

  • 0
    Intellectual 335 points

    Just to inform you, i have got the library from Control Suite.

  • 0 in reply to Arpan Sadhankar
    TI__Expert 3520 points

    Hello Arpan,

    I could be misinterpreting your question, but from what I can tell the various results you gave for varying FFT point values appear to be correct.  When computing the FFT of a signal keep in mind when you change the number of points of the FFT you are changing the number of values in your result.  So instead of an absolute point location for a given frequency, the position for any given frequency in an FFT result is related to both N, the number of points in the FFT, and your sampling frequency.

    Each result you have given is Peak/N ~= 0.05 (rounded).  When comparing your sampling frequency to your signal frequency you get fsig/fsamp ~= 0.05.  This means that in your specific case the point associated with 50Hz should always be at Peak/N ~= 0.05 regardless of what value N you are using.

    Hopefully this makes sense and that I have answered your question, if not let me know and I will try to help you further!

    Regards,

    Mark-