generating a noise signal

[laleh_mokhtarpour]

Hi
I want to design a narrow band singnal comprising frequencies of 100, 200, 300 and 400 Hz. it's variance is adjusted to 2, and a white noise with SNR of 30 db is added.
at the end I want to plot it's PSD, and my main problem is that I dont know how to use fft or how to set it's value
I wrot my matlab codes.
I am wondering if anybody could help me !?

Fs=2000;
t=0:1/Fs:1;
v=(cos(2*pi*100*t)+cos(2*pi*200*t)+cos(2*pi*300*t)+cos(2*pi*400*t));
snr=30;
k=awgn(v,snr);
Hp=spectrum.welch; Hp.segmentlength=600;
Hp.FFTlength='userdefined';
Pd=psd(Hp,k,'NFFT',3000,'Fs',Fs);

How should I find the correct value for nfft?
tnx

 

[laleh_mokhtarpour]

Hi
Tnx for your attention.
So in your opinion my code dosent work properly?
I wrote them from a thesis, and just changed the amount of nfft and segment length by trace and error.
my noise figure is similar to original but there is some variations

 

[JoannesPaulus]

your signal is 2001 samples abut you are calculating the fft for 3000 samples. You are zero padding your data and therefore it might look slightly different.

It is good practice, when using the fft, to have a number of samples that is a power of 2.

 

[JoannesPaulus]

It is good practice, when using the fft, to have a number of samples that is a power of 2.

As he says the NFFT length must be a power of 2 and should be close enough to the number of sample points in the data.

padding too much of zeros will make the spectrum different from the expected one.

Thank you for your support but your statement is not accurate:
NFFT length IS the number of samples and should be a power of 2, to avoid zero padding.