Plotting Frequency Spectrum using Matlab 

Hey! I just found this is the latest reply for this video. I am doing a vibration analysis project, Would you have any tips or discussions to share with me... thank you

You probably dont give a damn but does anyone know a way to log back into an instagram account? I was stupid lost the account password. I love any tips you can give me!

@Tristan Jaiden Instablaster =)

@Donovan Atticus thanks for your reply. I got to the site thru google and Im in the hacking process atm. I see it takes quite some time so I will reply here later with my results.

@Donovan Atticus It did the trick and I finally got access to my account again. I'm so happy:D Thank you so much, you saved my account !

@Tristan Jaiden Happy to help =)

Actually, dividing it by N_2 to see the amplitude spectrum ==> each peak is amplitudes of each mixed signal

Thanks for your clarification.

@@toiyeuhoa Thanks so much!

Thanks for pointing that out. And btw, it's "plot(fax_Hz(1:N_2), X_mags(1:N_2)./N)" instead of "plot(fax_Hz(1:N_2), X_mags(1:N_2)/fs)"

+David Godfrey You should check with your tutor. Typical features that can be extracted more easily in the frequency domain are fundamental frequencies of a periodic signal (and the strength of its harmonics); the presences of 50-60Hz interference from a mains signal; the presence of noise and the frequency range the noise occupies. This list can go on but hopefully this might give you some ideas.

There are built-in functions to do this. periodogram is one that I have used in the past.

@@ddorran Yes I need how we could do PSD, value using Mat-Lab. Thanks for your videos they have been very helpful, please.

hold on

The stem function in place of the plot function should do the job.

@@ddorran Hi, David. Thanks for your prompt response. Yes: I substituted plot() by stem() and I guess that I have the graphic that I need. Thanks so much! I fully copied your code and I think that was enough to complete my task. :) However, I will try to show information about the phase. Can i show you my code? Thanks again for everything...

@@ddorran By the way, David: In your opinion, how we can give some use to Nyquist frecuency in the program?

x = rand (1,1000); X = fft(x); X_positive = X(1:length(X)/2); X_positive_sorted = fliplr(sort(abs(X_positive)))

David Dorran Thanks for the help ..

David Dorran Could you help me how to compute the Fourier transform for each data stream in xdm. and plot magnitude spectrum for each data stream? thanks in advance

There are no units in general - they are the magnitudes of the sinusoidal components which when combined will reproduce a signal. Its the same when looking at a time domain signal - in general a plot will have its y-axis labelled "Amplitude" but if you know what the signal represents then you can change how you label the axis. For example for audio signals the amplitude will correspond to pressure measurements. In the frequency domain the y-axis would then correspond to a measure of pressure at a particular frequency - the units would still be related to pressure.

David Dorran I'm doing a frequency spectra for acceleration, where the input file is in terms of gravity (g). So does this make the units for amplitude g.secs?

For a completely unknown waveform you can't know an appropriate sampling rate. However, you can take some knowledge about the waveform and what you want to detect into consideration. The basic rule is sample at twice the highest frequency present in the signal you're interested in (or twice the bandwidth). If you don't know the highest frequency present, you could filter out any frequencies above a certain value that may not be of interest. For example, with speech signals you could sample an audio signal at 8 KHz and get an perceptually intelligible representation of the signal. While there will typically be frequencies above 4 KHz in speech we can use our knowledge of how audio signals are perceived to select an appropriate sampling frequency.

@@ddorran Thanks!

You likely have the answer by now, but for anyone else curious; tan inverse theta of the values gained from fft will yield phase. Remember to NEVER sum phases to find ITPC, instead, find avg of the vectors then tan inverse theta of resulting vector

+Abhishek PRATAP SINGH You would get a corresponding increase in the number of samples over the same signal duration (in seconds).

+David Dorran Sir in frequency spectrum entire range is equal to fs. So if fs doubles(becomes 2000) the three sinosoids will now become 30 ,40, and 50 hz which is not true. How do we take care of varying sampling frequency. Also tell how to select sampling frequency.

pulse = [zeros(1,1000) ones(1,100) zeros(1,1000)]; plot(abs(fft(pulse)))