findpeaks returns wrong width

I have a pulse that repeats 4 times with same period (attached oscilloscope data points). I try to isolate the 4 peaks and respective width imposing that the minimum distance between peaks is very close to the inverse of pulse frequency. I use the code pasted below. As you can see in the resulting table, the width of the third peak is the distance (at half prominence) between the signal and a border vertical line that MATLAB consider to separate two neighbour very close peaks. However to avoid this kind of situations I used the clause 'MinPeakDistance' with a value that is smaller but close to the period of the pulses. Nevertheless the function gives a wrong value for the width of the third peak because still a border with a close peak is considered. Why? How can we correct the problem?

function Results=FindPeaks(file)
    folder='replace this string with the folder path where the data file is';
    D=csvread(strcat(folder,file),11);
    A1 = D(:,2);
    t1 = D(:,1);    
    t = t1;
    A = A1;
    
    %number of expected peaks:
    %frequency of periodic signal
    f=7.0025*10^6;
    %sampling interval
    s=4*10^(-12);
    NExp=round(length(t)*s*f);
    
    %plot raw data with all peaks
    figure(1)
    plot(t,A1);
    findpeaks(A,t,'Annotate','extents');
    
    %create a table with relevant peaks, considering FWHM of the
    %prominence. 
    [pks,locs,fwhm,proms]=findpeaks(A,t,'Annotate','extents','MinPeakDistance',1.4*10^(-7),'MinPeakHeight',0.136,'Threshold',1e-7,'MinPeakProminence',0.1);
    PkTime = locs;
    Results = table(pks, PkTime, fwhm, proms, 'VariableNames',{'Value','PeakTime', 'FWHM', 'Prominence'});
    NAct=length(pks);
    disp(NExp)
    disp(NAct)
end    

Answer 1

[Star Strider]

If you have the Signal Processing Toolbox, use the pulsewidth function for this:

D = readmatrix('test.csv');
A1 = D(:,2);
t1 = D(:,1);
t = t1;
A = A1;

Idx = ~isnan(t);
t = t(Idx);                                                 % Eliminate ‘NaN’ Values
A = A(Idx); 

[Width,InitX,FinalX,MidLvl] = pulsewidth(A,t);

figure
plot(t, A)
hold on
plot(InitX, ones(size(InitX))*MidLvl, '>g')
plot(FinalX, ones(size(FinalX))*MidLvl, '<r')
hold off
grid
text(FinalX, ones(size(FinalX))*MidLvl, compose('\\leftarrow %5.3E', Width))

Producing:

The plot simply illustrates the result.

.