Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.


Ben
Posts:
9
Registered:
9/11/12


digital filter overflow oscillation
Posted:
Jun 20, 2013 7:16 AM


Hello,
I am experiencing what looks like overflow oscillations when generating the spectra of some EEG (brain signal) data. The code is something like this:
%% Sample spectra code tmp = psd(spectrum.welch('Hamming',1024,50), eeg.data(channels(k),:), 'NFFT', nfft, 'Fs', eeg.samplerate); plot(tmp.Frequencies, 20*log10(tmp.Data), 'k', 'linewidth', 1); %% End code
The output generally looks like very rapid oscillations about the mean power per frequency bin. The question is, how is it possible in Matlab to reduce the probability of generating overflow oscillations? Googling and scanning (for instance) the book "Filter Design for Signal Processing Using Matlab and Mathematica" turns up a whole lot of analytic approaches, but I can't translate them into Matlab in the given time.
Any insights would be much appreciated.



