Questions on real and imaginary array for FFT implementation

[invexed]

Hi all !!

I'm tring to make a litte FFt application

I'm tring with visual C++ first and later i'll try with my C microcontroller compiler

I have some questions...

First of all, it seems to be necessary to have a complex array with real and imaginary values

I have an array with a full sampled signal and i don't know exactly what to do with his...
I need to convert this array to a new array with real and imaginary part?
How can i do this?

FFT transform have a real and imaginary part too

The i need to convert this to a new array to make a graphical representation

I have found some FFT sources but not completes i only have fft functions with something like this:


fft(fftLength,RealInput, ImaginInput, RealOutput, ImaginOutput);

I need to know exactly the steps to convert a real sampled signal into a Real an Imaginary array to make an FFT

Someone can help me???

Thanks

 

[viswa]

Re: FFT implementation

hi guy,
if u need to ustand fft implementation then try first with matlab.if u need i can send u my sample code for finding 16 point fft for a complex signal.

in fft implementation, u need to work with twiddle factors raised to suitable powers in each stage.so while using these twiddle factors convert them to r*e^i(theta) form rather than (a+i*b) form since the first form would be easier in raising to higher powers.

u cant view ur complex values.u can only view their magnitude in c(i hope so).

 

[invexed]

Re: FFT implementation

Thanks viswa!!! if you can post your matlab code i'll try it....

I have some FFT examples in matlab but i cant see how to work this:

----------------------------------------------------------------------------------
t=1:200;
sin1=sin(2*pi*30/1000*t);

sin1c=abs(fft(sin1));

x=[0.005:.005:1];
subplot(2,1,1),plot(x,sin1c,'.'),title('DFT de la señal SIN1');
----------------------------------------------------------------------------------

Here only need a sampled signal and and obtain FFT without problems but when i use a real C function i need real and imaginary parts

 

[echo47]

FFT implementation

To feed real data into a complex FFT, simply copy your real data into the real part of a complex array, and set the imaginary part to zero.

One way to display the complex FFT result is to compute the magnitude of each complex value, mag=sqrt(real*real+imag*imag), and then display those magnitues. You may also want to convert the magnitudes to decibels, dB=20*log10(mag).

 

[redsk_y]

Re: FFT implementation

To best view the results try to use a number of points that is a power of two (64,128,...).

Ex.
fs=1; %sampling freq
fc=fs/16; %input freq
s=sqrt(2)*sin(2*pi/16*(0:1023);
plot(20*log10(abs(fft(s))/1024))

Regards,

--rs

 

[invexed]

Re: FFT implementation

Thanks Echo47 !!!

I have try your "mag=sqrt(real*real+imag*imag)" and it works
making ABS(real) it works too but i'm not sure if this is ok

I'm trying with a C function and it works ok with and without imaginary part!!

Making Imaginary part=0 y have a duplicate spectrum and 1/2 amplitude

Is this right ??

 

[echo47]

FFT implementation

Yes, if you zero the imag part, then there's no way to distinguish positive frequencies from negative frequencies, so the output from a complex FFT appears duplicated and mirrored and half the amplitude of a full complex signal. Simply ignore the upper half of the output array (that's the negative part, somewhat counter-intuitive).

If you feed a complex number into MATLAB's abs() function, it computes sqrt(real*real+imag*imag). Very convenient.

 

[amithsanu]

Re: FFT implementation

hello viswa..
i needed the sample code for 16pt fft implementation usng vhdl.. Can u plz send it to ma e-mail id- amithsanu@gmail.com...??? we want to implement 1024 pt fft on fpga usng vhdl... Can u help me out in this regard..?
Thank you..