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.

decrease in calculation speed during computation of fft in msp430f47177

Other Parts Discussed in Thread: MSP430F47177, MSP430F47127

Hello,

            I m using fft computation in my 3 phase power meter project. The MCU i am using is msp430f47177. The problem is when i use double values in the main fft loop..the calculation speed reduces. As of now the timer interrupts for calculation ever 1 sec.  i have restricted my code to float values but still its not working. Please help me on this.

Thanks in advance,

Varun

  • varun varghese said:
    i have restricted my code to float values but still its not working

    msp430 does not have floating point math unit so nobody can help you speeding-up floating point FFT calculations on it. It is not clear what precision you need. If 15bit fixed point math is enough, then you can try fixed point speed-optimized FFT routines (use internet search, they will pop-up).

    Other option: use floating-point DSP processor.

    Why you need FFT in power meter?!

  • Sir,  the fft does not do floating pint calculations. But some of the calculations I am performing to calculate THD values are in the fft routine. There is some delay that is slowing the calculations. Not able to solve the problem.  Sir i think using a DSP processor would not serve any purpose.

  • varun varghese said:
    Sir,  the fft does not do floating pint calculations.

    How's that? You said you use float values.

    You did not answer for what purpose you need FFT calculation in power meter.

    Please realize that msp430 is ultra low power microcontroller having no floating point math accelerator so FFT which needs lot of computing power will be painfully slow on ms430. If you are looking for some magic trick that gives msp430 math performance comparable to Intel Core I5, then answer is simple - there's no magic trick.

  • Sir,

            I am using fft routine for harmonics measurement for a 3 phase supply. And for that i use some calulations that require floating point values not the fft porgrams.

           I first developed this code on a msp430f47127 using CCE. I changed my MCU to msp430f47177 on CCS. While compiling on CCE it was working properly as needed. Problem occurs while using CCS.

  • varun varghese said:
    I first developed this code on a msp430f47127 using CCE. I changed my MCU to msp430f47177 on CCS. While compiling on CCE it was working properly as needed. Problem occurs while using CCS.

    If different compilers gives different performance, first thing you do is: check compiler optimization options, try various and see if this helps. If does not help then check generated code and see where difference is, think what you can do to achieve your goal.

    Other option: just don't switch compiler.

  • Thank you Sir.

**Attention** This is a public forum