I want to upsample by 5 a signal in frequency domain, and then plot (stem) it. I figured how to upsample,
Fk=(1/5)*upsample(ak_new,5);
now this creates a vector that is 5 times bigger than the original one, and I need to take the inverse Fourier series of this signal
Fn=(Fk*(exp((1i*2*pi/N*n'*n))));
where n is a sample vector (-1000:1000), as you can see, I can't make the transformation since n is not the same size as Fk anymore. How can I solve this?
You need to plot your upsampled signal in frequency domain, on a similarly "upsampled" frequency vector.
If your initial frequency vector was -1000, -999, -998 ..., now it should be -1000, -999.8, -999.6.
Here is a simple example:
Fs = 2000; % Sampling frequency
T = 1/Fs; % Sampling period
L = 2000; % Length of signal
t = (0:L-1)*T; % Time vector
S = sin(2*pi*400*t); % Signal
Y = fft(S);
ak_new = fftshift(abs(Y/L)); % Initial signal in frequency domain
Fk = upsample(ak_new,5); % Upsampled signal
f = (Fs/L)*(-L/2: 1 :L/2-1); % Initial frequency vector
fup = (Fs/L)*(-L/2:(1/5):L/2-1/5); % Upsampled frequency vector
subplot(2, 1, 1);
stem(f, ak_new);
title('Before upsampling');
subplot(2, 1, 2);
stem(fup, Fk);
title('After upsampling');
Related
I am trying to get the output of a Gaussian pulse going through a coax cable. I made a vector that represents a coax cable; I got attenuation and phase delay information online and used Euler's equation to create a complex array.
I FFTed my Gaussian vector and convoluted it with my cable. The issue is, I can't figure out how to properly iFFT the convolution. I read about iFFt in MathWorks and looked at other people's questions. Someone had a similar problem and in the answers, someone suggested to remove n = 2^nextpow2(L) and FFT over length(t) instead. I was able to get more reasonable plot from that and it made sense to why that is the case. I am confused about whether or not I should be using the symmetry option in iFFt. It is making a big difference in my plots. The main reason I added the symmetry it is because I was getting complex numbers in the iFFTed convolution (timeHF). I would truly appreciate some help, thanks!
clc, clear
Fs = 14E12; %1 sample per pico seconds
tlim = 4000E-12;
t = -tlim:1/Fs:tlim; %in pico seconds
ag = 0.5; %peak of guassian
bg = 0; %peak location
wg = 50E-12; %FWHM
x = ag.*exp(-4 .* log(2) .* (t-bg).^2 / (wg).^2); %Gauss. in terms of FWHM
Ly = x;
L = length(t);
%n = 2^nextpow2(L); %test output in time domain with and without as suggested online
fNum = fft(Ly,L);
frange = Fs/L*(0:(L/2)); %half of the spectrum
fNumMag = abs(fNum/L); %divide by n to normalize
% COAX modulation ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
%phase data
mu = 4*pi*1E-7;
sigma_a = 2.9*1E7;
sigma_b = 5.8*1E6;
a = 0.42E-3;
b = 1.75E-3;
er = 1.508;
vf = 0.66;
c = 3E8;
l = 1;
Lso = sqrt(mu) /(4*pi^3/2) * (1/(sqrt(sigma_a)*a) + 1/(b*sqrt(sigma_b)));
Lo = mu/(2*pi) * log(b/a);
%to = l/(vf*c);
to = 12E-9; %measured
phase = -pi*to*(frange + 1/2 * Lso/Lo * sqrt(frange));
%attenuation Data
k1 = 0.34190;
k2 = 0.00377;
len = 1;
mldb = (k1 .* sqrt(frange) + k2 .* frange) ./ 100 .* len ./1E6;
mldb1 = mldb ./ 0.3048; %original eqaution is in inch
tfMag = 10.^(mldb1./-10);
% combine to make in complex form
tfC = [];
for ii = 1: L/2 + 1
tfC(ii) = tfMag(ii) * (cosd(phase(ii)) + 1j*sind(phase(ii)));
end
%END ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
%convolute both h and signal
fNum = fNum(1:L/2+1);
convHF = tfC.*fNum;
convHFMag = abs(convHF/L);
timeHF = ifft(convHF, length(t), 'symmetric'); %this is the part im confused about
% Ignore,
% tfC(numel(fNum)) = 0;
% convHF = tfC.*fNum;
% convHFMag = abs(convHF/n);
% timeHF = ifft(convHF);
%% plotting
% subplot(2, 2, 1);
% plot(t, Ly)
% title('Gaussian input');
% xlabel('time in seconds')
% ylabel('V')
% grid
subplot(2, 2, 1)
plot(frange, abs(tfC(1: L/2 + 1)));
set(gca, 'Xscale', 'log')
title('coax cable model')
xlabel('Hz')
ylabel('|H(s)|V/V')
grid
ylim([0 1.1])
subplot(2, 2, 2);
plot(frange, convHFMag(1:L/2+1), '.-', frange, fNumMag(1:L/2+1)) %make both range and function the same lenght
title('The input signal Vs its convolution with coax');
xlabel('Hz')
ylabel('V')
legend('Convolution','Lorentzian in frequecuency domain');
xlim([0, 5E10])
grid
subplot(2, 2, [3, 4]);
plot(t, Ly, t, timeHF)
% plot(t, real(timeHF(1:length(t)))) %make both range and function the same lenght
legend('Input', 'Output')
title('Signal at the output')
xlabel('time in seconds')
ylabel('V')
grid
It's important to understand deeply the principles of the FFT to use it correctly.
When you apply Fourier transform to a real signal, the coefficients at negative frequencies are the conjugate of the ones at positive frequencies. When you apply FFT to a real numerical signal, you can show mathematically that the conjugates of the coefficients that should be at negative frequencies (-f) will now appear at (Fsampling-f) where Fsampling=1/dt is the sampling frequency and dt the sampling period. This behavior is called aliasing and is present when you apply fft to a discrete time signal and the sampling period should be chosen small enaough for those two spectra not to overlap Shannon criteria.
When you want to apply a frequency filter to a signal, we say that we keep the first half of the spectrum because the high frequencies (>Fsampling/2) are due to aliasing and are not characteristics of the original signal. To do so, we put zeros on the second half of the spectra before multiplying by the filter. However, by doing so you also lose half of the amplitude of the original signal that you will not recover with ifft. The option 'symmetric' enable to recover it by adding in high frequencis (>Fsampling/2) the conjugate of the coefficients at lower ones (<Fsampling/2).
I simplified the code to explain briefly what's happening and implemented for you at line 20 a hand-made symmetrisation. Note that I reduced the sampling period from one to 100 picoseconds for the spectrum to display correctly:
close all
clc, clear
Fs = 14E10; %1 sample per pico seconds % CHANGED to 100ps
tlim = 4000E-12;
t = -tlim:1/Fs:tlim; %in pico seconds
ag = 0.5; %peak of guassian
bg = 0; %peak location
wg = 50E-12; %FWHM
NT = length(t);
x_i = ag.*exp(-4 .* log(2) .* (t-bg).^2 / (wg).^2); %Gauss. in terms of FWHM
fftx_i = fft(x_i);
f = 1/(2*tlim)*(0:NT-1);
fftx_r = fftx_i;
fftx_r(floor(NT/2):end) = 0; % The removal of high frequencies due to aliasing leads to losing half the amplitude
% HER YOU APPLY FILTER
x_r1 = ifft(fftx_r); % without symmetrisation (half the amplitude lost)
x_r2 = ifft(fftx_r, 'symmetric'); % with symmetrisation
x_r3 = ifft(fftx_r+[0, conj(fftx_r(end:-1:2))]); % hand-made symmetrisation
figure();
subplot(211)
hold on
plot(t, x_i, 'r')
plot(t, x_r2, 'r-+')
plot(t, x_r3, 'r-o')
plot(t, x_r1, 'k--')
hold off
legend('Initial', 'Matlab sym', 'Hand made sym', 'No sym')
title('Time signals')
xlabel('time in seconds')
ylabel('V')
grid
subplot(212)
hold on
plot(f, abs(fft(x_i)), 'r')
plot(f, abs(fft(x_r2)), 'r-+')
plot(f, abs(fft(x_r3)), 'r-o')
plot(f, abs(fft(x_r1)), 'k--')
hold off
legend('Initial', 'Matlab sym', 'Hand made sym', 'No sym')
title('Power spectra')
xlabel('frequency in hertz')
ylabel('V')
grid
Plots the result:
Do not hesitate if you have further questions. Good luck!
---------- EDIT ----------
The amplitude of discrete Fourier transform is not the same as the continuous one. If you are interested in showing signal in frequency domain, you will need to apply a normalization based on the convention you have chosen. In general, you use the convention that the amplitude of the Fourier transform of a Dirac delta function has amplitude one everywhere.
A numerical Dirac delta function has an amplitude of one at an index and zeros elsewhere and leads to a power spectrum equal to one everywhere. However in your case, the time axis has sample period dt, the integral over time of a numerical Dirac in that case is not 1 but dt. You must normalize your frequency domain signal by multiplying it by a factor dt (=1picoseceond in your case) to respect the convention. You can also note that this makes the frequency domain signal homogeneous to [unit of the original multiplied by a time] which is the correct unit of a Fourier transform.
I just read the example of Mablab tutorial, trying to studying the FFT function.
Can anyone tell me that for the final step, why P1(2:end-1) = 2*P1(2:end-1). In my opinion, it is not necessary to multiply by 2.
Fs = 1000; % Sampling frequency
T = 1/Fs; % Sampling period
L = 1000; % Length of signal
t = (0:L-1)*T; % Time vector
%--------
S = 0.7*sin(2*pi*50*t) + sin(2*pi*120*t);
%---------
X = S + 2*randn(size(t));
%---------
plot(1000*t(1:50),X(1:50))
title('Signal Corrupted with Zero-Mean Random Noise')
xlabel('t (milliseconds)')
ylabel('X(t)')
Y = fft(X);
P2 = abs(Y/L);
P1 = P2(1:L/2+1);
P1(2:end-1) = 2*P1(2:end-1);
f = Fs*(0:(L/2))/L;
plot(f,P1)
title('Single-Sided Amplitude Spectrum of X(t)')
xlabel('f (Hz)')
ylabel('|P1(f)|')
Y = fft(S);
P2 = abs(Y/L);
P1 = P2(1:L/2+1);
P1(2:end-1) = 2*P1(2:end-1);
plot(f,P1)
title('Single-Sided Amplitude Spectrum of S(t)')
xlabel('f (Hz)')
ylabel('|P1(f)|')
Matlab sample
The reason for the multiplication by 2 is that the spectrum returned by fft is symmetric about the DC component. Since they are showing the single-sided amplitude spectrum, the amplitude of each point is going to be doubled to account for the contributions of data on the other side of the spectrum. For example, the single-sided amplitude of pi/4 is the amplitude at pi/4 plus the amplitude at -pi/4.
The first sample is skipped since it is the DC point and therefore shared between the two sides of the spectrum.
So for example, if we look at the fft of their example signal with a 50Hz sinusoid of amplitude 0.7 and a 120Hz sinusoid of amplitude 1.
Fs = 1000; % Sampling frequency
T = 1/Fs; % Sampling period
L = 1000; % Length of signal
t = (0:L-1)*T; % Time vector
S = 0.7*sin(2*pi*50*t) + sin(2*pi*120*t);
% Compute the FFT
Y = fft(S);
% Compute the amplitudes
amplitude = abs(Y / L);
% Figure out the corresponding frequencies
f = Fs/L*[0:(L/2-1),-L/2:-1]
% Plot the result
plot(f, amplitude)
When we plot this, you'll see that it's symmetric and the original input amplitude is only realized by combining the amplitudes from both sides of the spectrum.
A slightly more explicit version of what they have done would be to be the following which sums the two halves of the spectrum
P1(2:end-1) = P1(2:end-1) + P2((L/2+2):end);
But since by definition the spectrum is symmetric, the opt to simply multiply by 2.
I am trying to find the inverse Fourier transform of a simple filter in Matlab. In the first case (sinc filter / "brick wall"), I use the ifft function to find the time-domain function, which is a sinc, centered at t = 0.
I now want to now find the time-domain function for a simple Chebyshev filter. However, for some reason, the code below seems to give me the impulse response, where the time axis is incorrect. Should I not expect a similar looking sinc function centered at t = 0?
fo = 10; %frequency of the sine wave
Fs = 100; %sampling rate
Ts = 1/Fs; %sampling time interval
t = -1+Ts:Ts:1-Ts; %sampling period
freq = -Fs/2:(Fs/length(t)):Fs/2;
%% Sinc with bandwidth = fo. This works!
y = 0.5*sinc(2*fo*t);
YfreqDomain1 = fft(y);
figure('Name','Brick wall sinc filter (freq)');
plot(freq(1:length(y)),2/length(y)*fftshift(abs(YfreqDomain1)))
y_ret1=ifft(YfreqDomain1,'nonsymmetric');
figure('Name','Brick wall sinc filter (time)');
plot(t,y_ret1);
%% Chebyshev with bandwidth fo. This gives me a strange result.
[b,a] = cheby1(6,0.1,2*fo/Fs); % 6th order, 0.1dB ripple
[YfreqDomain2 w] = freqz(b,a,length(t),'whole');
figure('Name','Chebyshev Filter (freq)');
plot(freq(1:length(YfreqDomain2)), 2/length(y)*fftshift(abs(YfreqDomain2)));
figure('Name','Chebyshev Filter (time)');
y_ret2=ifft(YfreqDomain2,'nonsymmetric');
plot(t,y_ret2);
There are two problems with your computation:
First, you evaluate the time domain filter coefficients on a very narrow time window, which truncates the filter and changes its characteristics.
Second, you do not properly keep track of which indices in the time domain and frequency domain vectors correspond to time point 0 and frequency 0, respectively. This can be done differently, I here chose to always have t(1) = 0 and f(1) = 0, and apply fftshift for plotting only.
Here's my corrected version of your code:
fo = 10; % frequency of the sine wave
Fs = 100; % sampling rate
Ts = 1 / Fs; % sampling time interval
n = 1000; % number of samples
% prepare sampling time vector such that t(1) = 0
t = (0 : n - 1)';
t = t - n * (t >= 0.5 * n);
t = t / Fs;
% prepare frequency vector such that f(1) = 0;
f = (0 : n - 1)' / n;
f = f - (f >= 0.5);
f = f * Fs;
%% sinc filter with bandwidth fo
% define filter in time domain
s = 2*fo/Fs * sinc(2*fo*t);
% transform into frequency domain
Hs = fft(s);
% since the filter is symmetric in time, it is purely real in the frequency
% domain. remove numeric deviations from that:
Hs = real(Hs);
subplot(2, 4, 1)
plot(fftshift(t), fftshift(s))
ylim([-0.1 0.25])
title('sinc, time domain')
subplot(2, 4, 2)
plot(fftshift(f), real(fftshift(Hs)), ...
fftshift(f), imag(fftshift(Hs)))
ylim([-1.1 1.1])
title({'sinc, frequency domain', 'real / imaginary'})
subplot(2, 4, 3)
plot(fftshift(f), abs(fftshift(Hs)))
ylim([-0.1 1.1])
title({'sinc, frequency domain', 'modulus'})
%% Chebyshev filter with bandwidth fo
% define filter in frequency domain
[b,a] = cheby1(6, 0.1, 2*fo/Fs); % 6th order, 0.1 dB ripple
Hc = freqz(b, a, n, 'whole', Fs);
% transform into time domain
c = ifft(Hc);
% determine phase such that phase(1) is as close to 0 as possible
phase = fftshift(unwrap(angle(fftshift(Hc))));
phase = phase - 2*pi * round(phase(1) /2/pi);
subplot(2, 4, 5)
plot(fftshift(t), fftshift(c))
title('Chebyshev, time domain')
ylim([-0.1 0.25])
subplot(2, 4, 6)
plot(fftshift(f), real(fftshift(Hc)), ...
fftshift(f), imag(fftshift(Hc)))
ylim([-1.1 1.1])
title({'Chebyshev, frequency domain', 'real / imaginary'})
subplot(2, 4, 7)
plot(fftshift(f), abs(fftshift(Hc)))
ylim([-0.1 1.1])
title({'Chebyshev, frequency domain', 'modulus'})
subplot(2, 4, 8)
plot(fftshift(f), fftshift(phase))
title({'Chebyshev, frequency domain', 'phase'})
And here's the result:
As you can see, the sinc and Chebyshev filters are similar with respect to the modulus of the frequency response, but very different regarding the phase. The reason is that the Chebyshev filter is a causal filter, meaning that the coefficients in the time domain are constrained to be 0 for t < 0, a natural property for a filter that is implemented in a real physical system.
I'm using the pwelch method in matlab to compute the power spectra for some wind speed measurements. So, far I have written the following code as an example:
t = 10800; % number of seconds in 3 hours
t = 1:t; % generate time vector
fs = 1; % sampling frequency (seconds)
A = 2; % amplitude
P = 1000; % period (seconds), the time it takes for the signal to repeat itself
f1 = 1/P; % number of cycles per second (i.e. how often the signal repeats itself every second).
y = A*sin(2*pi*f1*t); % signal
fh = figure(1);
set(fh,'color','white','Units', 'Inches', 'Position', [0,0,6,6],...
'PaperUnits', 'Inches', 'PaperSize', [6,6]);
[pxx, f] = pwelch(y,[],[],[],fs);
loglog(f,10*(pxx),'k','linewidth',1.2);
xlabel('log10(cycles per s)');
ylabel('Spectral Density (dB Hz^{-1})');
I cannot include the plot as I do not have enough reputation points
Does this make sense? I'm struggling with the idea of having noise at the right side of the plot. The signal which was decomposed was a sine wave with no noise, where does this noise come from? Does the fact that the values on the yaxis are negative suggest that those frequencies are negligible? Also, what would be the best way to write the units on the y axis if the wind speed is measured in m/s, can this be converted to something more meaningful for environmental scientists?
Your results are fine. dB can be confusing.
A linear plot will get a good view,
Fs = 1000; % Sampling frequency
T = 1/Fs; % Sample time
L = 1000; % Length of signal
t = (0:L-1)*T; % Time vector
y = sin(2 * pi * 50 * t); % 50Hz signal
An fft approach,
NFFT = 2^nextpow2(L); % Next power of 2 from length of y
Y = fft(y,NFFT)/L;
f = Fs/2*linspace(0,1,NFFT/2+1);
subplot(1,2,1);
plot(f,2*abs(Y(1:NFFT/2+1)))
xlabel('Frequency (Hz)')
ylabel('|Y(f)|')
pwelch approach,
subplot(1,2,2);
[pxx, freq] = pwelch(y,[],[],[],Fs);
plot(freq,10*(pxx),'k','linewidth',1.2);
xlabel('Frequency (Hz)');
ylabel('Spectral Density (Hz^{-1})');
As you can see they both have peak at 50Hz.
Using loglog for both,
So "noise" is of 1e-6 and exists in fft as well, and can be ignored.
For your second question, I don't think the axis will change it will be frequency again. For Fs you should use the sampling frequency of wind speed, like if you have 10 samples of speed in one second your Fs is 10. Higher frequencies in your graph means more changes in wind speed and lower frequencies represent less changes for the speed.
I had created a 3 three different frequency signal and filter out the signal i don't want. But when i using ifft in matlab, it shows a wrong graph.How to transform my frequency domain spectrum back into my 3 second time domain graph? Below my code is as below:
clc
clear all
Fs = 8192;
T = 1/Fs;
%create tones with different frequency
t=0:T:1;
t2=1:T:2;
t3=2:T:3;
y1 = sin(2*pi*220*t);
y2 = sin(2*pi*300*t2);
y3 = sin(2*pi*440*t3);
at=y1+y2+y3;
figure;
plot(t,y1,t2,y2,t3,y3),title('Tones with noise');
[b,a]=butter(2,[2*290/Fs,2*350/Fs],'stop');
e=filter(b,a,at);
et=(ifft(abs(e)));
figure,
plot(et)
As it is now, et is in the frequency domain, because of the fft. You don't need to fft. just plot(e) and you'll get the time domain filtered waveform. Yo can check the filter performance in the freq. domain by fft though, just
plot(abs(fftshift(fft(fftshift(e)))));
xlim([4000 5000])
Edit:
Your code as it is written on the question has the following bug: at has exactly 1 second of info in it (or 8192 elements). If you plot(at) you'll see the sum of frequencies alright, but they all happen in the same time. This is how to fix it:
clear all
Fs = 8192; % or multiply by 3 if needed
T = 1/Fs;
%create tones with different frequency
t=0:T:3;
y1 = sin(2*pi*220*t).*(t<1);
y2 = sin(2*pi*300*t).*(t<2 & t>=1);
y3 = sin(2*pi*440*t).*(t>=2);
at=y1+y2+y3;
[b,a]=butter(2,[2*290/Fs,2*350/Fs],'stop');
e=filter(b,a,at);
figure,
plot(t,e)
dt=t(2)-t(1);
N=length(at);
df=1/(N*dt); % the frequency resolution (df=1/max_T)
if mod(N,2)==0
f_vector= df*((1:N)-1-N/2); % frequency vector for EVEN length vectors: f =[-f_max,-f_max+df,...,0,...,f_max-df]
else
f_vector= df*((1:N)-0.5-N/2); % frequency vector for ODD length vectors f =[-f_max,-f_max+fw,...,0,...,f_max]
end
freq_vec=f_vector;
fft_vec=fftshift(fft(e));
plot(freq_vec,abs(fft_vec))
xlim([0 1000])