Hi all,
while taking FFT of a signal, we decide sampling frequency, time increment and time step as shown
Fs = 256
dt = 1/Fs = 0.0039
t = 0:dt:1-dt (for 1 second)
My question is:
when i take above values, i am getting accurate fft magnitude and phase information for my signal, but when i change 'dt' value (say 0.004) or 't' value (say 0:dt:255*dt), the results are not accurate ?
I think, irrespective of dt and t value, fft magnitude and phase should be the same always. Is there any relation between time vector, dt and fft results.
please help...
Thanks in advance
vivek
The dwell time defines the scale for your spectrum, the bandwidth. The FFT really just gives the portion of the signal that experiences a particular phase shift during each dwell time. That is converted to frequency by dividing that phase shift by the dwell time.
In MRI, for example, magnetic field gradients are used to impose specific relationships between frequency and spatial position. If you reduce the slope of the magnetic field gradient by a factor of two and double the dwell time, you will get identical appearing images( spectra) when the fft of the acquired signal is done, but the bandwidths will differ by a factor of two.
Hi Vevek,
I am not sure how you are calculating the FFT of your signal, but try this:
%suppose 'a' is your signal, 'Fs' is the sampling frequency and 'L' is signal length, 'NFFT' is the number of FFT point.
----------------------------------------
Fs=1000;
L=2200;
NFFT= 2^nextpow2(L);
Y= fft(a, NFFT)/ L;
f=Fs/2*linspace (0,1, NFFT/2+1);
% now folding: keeping positive magnitude:
ft= 2*abs(Y(1:NFFT/2+1));
%now you can plot 'ft'.
I hope this is helpful and you will always get accurate results.
Hello qassim sir,
yes! i am calculating fft in the similar way as u suggest, but i was asking for the step time for the signal.
For ex. if my signal is Y = 5*sin(2*pi*10*t+pi/6)+3*sin(2*pi*20*t+pi/4)
what should be the time step "dt" for it.
generally we take dt = 1/Fs, but what if we change dt value. should dt always be 1/Fs ? If not then why the results are not accurate with different dt ?
Thanks
Vivek,
The dwell time is 1/bandwidth, by definition. The edges of a spectrum correspond to the frequencies that will result in a positive or negative phase shift of 180° during dt. Higher or lower frequencies will alias. By choosing your dwell time, you are defining your bandwidth.
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators