Just assign the signal contents to the variable x and the call the FFT routine as mentioned in the post
X=fft(x,NFFT); %x is your signal,

you can use any of the different representations to plot the FFT output.

I need to do fft to obtain the frequency content of my signal and calculate the energy to obtain the transmission coefficient, but I have a lot of problems since I don’t know exatly what is the correct way to obtain everything with a correct scale and units.

I have an output signal (voltage vs time : size(u)=(100000,2)) obtained from excitation: pulse (pulse width = 100 µs); on the oscilloscope : duration is 10 ms and 10 MSa/s of sampling rate.

for the frequencies, I expect to get a fundamental at around 8 KHz and the harmonics (2nd and third).

this is a very smal part of my signal :
t(s) t(volts)
3.99000000E-02 1.3574E-02
3.99001000E-02 1.4299E-02
3.99002000E-02 1.3969E-02
3.99003000E-02 1.3508E-02
3.99004000E-02 1.3706E-02

My questions are:

I need the values of the frequencies pics (fundamental, 2nd and 3rd) and calculate the energy of power in order to get the transmission coefficient of the signal.

1 – how should I proceed to get the frequencies in a correct scale? (plot of all the Spectrum in linear scale – Spectrum in dB scale)

2 – is it necessary to include a window in the fft ? which is the better one in that case?

P.S: I tried following the method explained in this website; here is my matlab code!!!!
u=load(‘u10.tsv’);
tau = 0.000000099 ;

Nfft=length(u);
w=blackman(Nfft);
u_windowed_temp=u;
u_windowed=u.*w;

Y1 = fft(u_windowed,Nfft);
%freq1(1)=10^-10;
freq1(2:Nfft/2)=(1:Nfft/2-1)/(Nfft*tau);
plot(freq1/1000,abs(Y1(1:length(freq1))))
thank’s a lot!

Thanks for pointing it out

I was plotting the NON-normalized magnitude spectrum. The amplitude is influenced the number of samples taken to compute DFT. For normalized magnitude spectrum please check the following post.
https://www.gaussianwaves.com/2015/11/interpreting-fft-results-obtaining-magnitude-and-phase-information

In the above example when I gave the number of cycles as 10, then it is giving a spectrum which has a higher amplitude. This seems so strange to me because the number of cycles is just a visualisation parameter and that should not chance the appearance of the spectrum.

nCyl = 10; %to generate ten cycles of sine wave

Kindly explain why this is happening.

I would ask why he does not divide only by L, as this is the usual scaling factor when calculating FFT

There’s an error on point 3. It should read
X=fftshift(x,NFFT); %compute DFT using FFT

Rgds

you divide by (NFFT*L) and not L^2?