stft plot doesn't match FFT results

Hello

I have a signal sampled at 250e6 samples/s. The FFT result of the signal looks correct with fundamental at 1.1 MHz and some harmonics. But when I do short-term fourier transform on the signal using the command:

figure,stft(T_mean,250e6,'FFTLength',1024);

I am not sure if the result is correct or not. Can anyone please help?

I did detrending to remove the DC noise part.

Answer 1

[Star Strider]

If the frequencies are the same in both plots, the only difference is that the STFT plot spans -125 Hz to +125 Hz. The peaks are clustered between 0 Hz and 5 Hz, explaining the bands on either side of the 0 Hz line.

Include:

ylim([-5 5])

and the plots should look similar (except that the STFT plot will continue to be symmetric about 0).

.

[NASRIN AKTER]

@Star Strider

Thanks for your answer. The source signal here is actually supposed to be a mixture of 2 signals. Based on the stft plot here, do you think there is any matlab technique that can separate those 2 signals? I already tried ICA but didn't seem to work.

[Star Strider]

My pleasure!

ICA or a filtering approach are the only options I’m aware of.

I have no idea what ‘didn’t seem to work’ means.

.