EEG filters are extremely difficult to design because of the frequency definitions. They are best filtered with active filters in hardware.
These look reasonable, and are the best I can design:
frq_pb = [2.1 4; 4 8; 8 13; 13 31; 31 51; 51 120]; % Passband Matrix
frq_sb = [frq_pb(:,1)-2 frq_pb(:,2)+2]; % Stopband Matrix
Fs = 1000; % Sampling Frequency (Hz)
Fn = Fs/2; % Nyquist Frequency (Hz)
Rp = 30; % Passband Ripple (dB)
Rs = 10; % Stopband Ripple (dB)
for k1 = 1:size(frq_pb,1)
[n,Ws] = cheb2ord(frq_pb(k1,:)/Fn, frq_sb(k1,:)/Fn, Rp, Rs); % Chebyshev Type II Filter Order
n = max(k1+1, real(n));
[b,a] = cheby2(n, Rs, Ws); % Transfer Function Polynomial Coefficients
[SOS{k1}, G{k1}] = tf2sos(b,a); % Second-Order-Section For Stability
end
for k1 = 1:size(frq_pb,1) % Plot Filter Passbands
figure(k1)
freqz(SOS{k1}, 4096, Fs)
set(subplot(2,1,1), 'XLim',[0 150]) % Set X-Axis Limits To Improve Visibility
set(subplot(2,1,2), 'XLim',[0 150])
end
Experiment with them to get the result you want. You can also experiment with using designfilt to see if it can produce a better design. I use a Chebyshev Type II FIR filter here, but other designs and IIR filters may also be appropriate.
