Correcting a y-shift when filtering signal in the frequency domain

I am trying to apply a low-pass filter to a signal ro remove noise.

I have applied the filter and corrected for the x-shift as shown below:

filtertype = 'FIR';
Ts = (time(end)/1000)/N; %sampling interval. time is in ms 
Fs = 1/Ts; %sampling frequency
Fpass = 10; 
Fstop = 50;
Rp = 5;
Astop = 100;

LPF = dsp.LowpassFilter('SampleRate',Fs,...
                        'FilterType',filtertype,...
                        'PassbandFrequency',Fpass,...
                        'StopbandFrequency',Fstop,...
                        'PassbandRipple',Rp,...
                        'StopbandAttenuation',Astop);
                    
output = step(LPF, voltage);

%plot
fig1 = figure ();
plot(time, voltage, time,output);
ylabel('voltage');
legend('Raw Data', 'Filtered Data'); 

%correcting for x-shift
grpdelay(LPF,numdatapoints,Fs);
delay = mean(grpdelay(LPF));
tshift = time(1:end-delay);
vshift = voltage(1:end-delay);
output(1:delay+1) = [];

%plot
plot(tshift,vshift,tshift,output);
legend('Original Signal','Filtered Shifted Signal');

However, there is a y-shift in the filtered data which I cannot seem to prevent (see attached PNG, with similar data with 6 traces before filtering and 6 after filtering), no matter how I change the filter parameters.

Is my method of filtering incorrect, and thus producing this shift? Or how can I correct for the shift?

Answer 1

[Mathieu NOE]

hello Josh

when it comes to smooth a noisy signal, I usually go for smoothdata. Code is simple and it works well , no delay between output and input

% %%%%%%%%%%%%%%%%
output = smoothdata(voltage, 'gaussian' , 3500);
plot(time,voltage,time,output);legend('Raw','Smoothed');
title(['Data samples at Fs = ' num2str(round(Fs)) ' Hz / Smoothed with smoothdata' ]);

[Josh Selfe]

Thank you Mathieu, this looks great. Appreciate your help.

[Mathieu NOE]

My pleasure

would you accept my answer ?

thanks