Finding different length pulses

Hi

I have attached a screen shot of some pulses with diiferent lenghts my question is what is the best way clean these up and show them.

Thank You

David

Answer 1

[Star Strider]

Note that ‘clean them up’ is ever so slightly ambiguous.

Assuming that the objective is to remove some of what might be considered ‘noise’, first calculate the Fourier transform of the signal, then using that information to determine where the ‘noise’ frequencies begin, design a lowpass filter to remove the undesirable higher frequencies. (For all of these, the sam;ing intervals must be constant. If it is not, use the resample function to interpolate them to a constant sampling frequency first.)

The Fourier transform is straightforward (assuming ‘x’ is the signal vector):

Fs = ... ;                                      % Sampling Frequency (Hz)
Fn = Fs/2;                                      % Nyquist Frequency (Hz)
L = numel(x);                                   % ‘x’ Is A Vector
xm = mean(x);
xfft = fft(x - xm)/L;                           % Subtract Mean & Normalise By ‘L’
Fv = linspace(0, 1, fix(L/2)+1)*Fn;             % Frequency Vector
Iv = 1:numel(Fv);                               % Index Vector

figure
plot(Fv, abs(xfft(Iv))*2)
grid
xlabel('Frequency (Hz)')
ylabel('Magnitude')

With that information, use the lowpass function to design the filter, or use command-line arguments in a script to design your own filter. This is straightforward with the Signal Propcessing Toolbox functions.

[David Jones]

Hi

Thanks once again for getting back, your quite right it was slightly ambiguous. I am after the data in square wave if possible I have been told that 1 is twice the lenght of 0 I do need to clean it up, but keep the timing information.

Thank You

David

[Star Strider]

My pleasure!

I am not certain that I understand what you want to do.

If you want to convert all the pulses to square waves, and if ‘x’ is your vector, simply do:

sqwvx = sign(x);

All the positive values will be +1, all the negative values will be -1 and any 0 values remain 0.

Example —

t = linspace(0, 61, 100);
x = randn(1, 100);
sqwvx = sign(x);

figure
plot(t, x)
hold on
plot(t, sqwvx, 'LineWidth',1.25)
hold off
grid
ylim(1.1*ylim)

.

[David Jones]

Hi

As always great Job, I am new to Matlab and in the previous message you mentioned resample I had a look at this but dont understand I have 16000 samples sample rate 5mS can you please explain in a little more detail

Thank you for all your help

David

[Star Strider]

My pleasure!

Thank you.

Essentially all digital signal processing requires that the signals be regularly sampled (constant sampling intervals). If your signal already has a uniform sampling rate (all sampling intervals are the same), then you do not need to use the resample funciton. If they are not, there is no guarantee that the results of any signal processing action will be in any way correct.

The resample function interpolates signals with varying sample times to signals with constant sampling times, allowing normal signal processing procedures to be applied to the resampled signals. (It actually does more than interpolate, since it also uses an anti-aliasing filter to remove spurious signals that might result from the resampling process, the resason that it is superior to interp1 for signal processing purposes.)

Answer 2

[David Jones]

Excellent

Thank you

[Star Strider]

As always, my pleasure!