High Pass Filter Error

Hello everyone,

I'm having some problems in solving a problem I have with an High Pass filter. I just want to filter a signal with an High Pass Filter with a cut off frequency of 5 Hz. This is my simple script:

F_cutoff=5;
ii=1; % I'm just taking the first row to have a simple example.
fs=400;
u_G_Interp(ii,:)=highpass(u_G_Interp(ii,:),F_cutoff,fs); % u_G_Interp is a simple time signal matrix where each row is the signal that I want to filter. The dimension is [2923 x 1024]

If I use F_cutoff > of a certain value that depends case by case I don't get any error but If it is lower than a certain value I have an error like this:

Error using designfilt>parseAndDesignFilter (line 516)
Size inputs must be integers.

Error in designfilt (line 224)
  [err,requestedResponse,parseParams,h] = parseAndDesignFilter(inputParamValueNames, varargin{:});

Error in highpass>designFilter (line 194)
opts.FilterObject = designfilt(params{:});

Error in highpass (line 97)
opts = designFilter(opts);

I'm not an expert on filtering so I don't understand where exactly the problem is. Is there any constraint in using these filters that I'm not considering?

Yhank you in advance for your help.

Davide

Answer 1

[Star Strider]

The problem may be with ‘u_G_Interp’, and since we do not have it to work with, we cannot determine that.

Assuming it is a double array, try this:

u_G_Interp = randn(1923,1024);                                          % Create Signal Matrix

F_cutoff=5;
fs = 400;
[u_G_InterpFiltT(ii,:),df] = highpass(u_G_Interp(ii,:), F_cutoff, fs);  % Filters Signal Column-Wise, So Transpose To Filter Rows

.

[Davide Mastrodicasa]

Thank you for your answer.

I tried with your example and it works perfectly but I don't get the point on why it doen't work on my dataset. Is there any constraint on the function that can led to this error?

You can find a row of my dataset in attachment. On this data, with a cutoff frequency of 2 Hz, I get the error I posted in the question. My data set, as you said, is a double array.

[Star Strider]

The only problem I can see with respect to ‘F_cutoff’ is that it must be less than fs/2 which is the Nyquist frequency. I have tried it with ‘F_cutoff’ values ranging from 0.001 to 199 and it has not thrown any errors. The values for it that threw the error were not posted, so I have no idea what the problem could be.

This runs for me without error and produces the expected result:

LD = load('Data.mat');
u_G_Interp = LD.u_G_Interp;

ii=1; 
F_cutoff=5;
fs = 400;
L = size(u_G_Interp,2);
t = linspace(0, L, L)/fs;

figure
plot(t, u_G_Interp(ii,:))
grid
xlabel('Time')
title('Original Signal')


Fn = fs/2;
FTuGI(ii,:) = fft(u_G_Interp(ii,:))/L;
Fv = linspace(0, 1, fix(L/2)+1)*Fn;
Iv = 1:numel(Fv);

figure
plot(Fv, abs(FTuGI(Iv))*2)
grid
xlabel('Frequency')
title('Fourier Transform')

u_G_Interp_Filt(ii,:) = highpass(u_G_Interp(ii,:),F_cutoff,fs);

figure
plot(t, u_G_Interp_Filt(ii,:))
grid
xlabel('Time')
title('Highpass Filtered Signal')

.