[Eeglablist] Frequency resolution of newcrossf

[Arnaud Delorme]

Dear Bethel,

> I have:
> srate=1926Hz
> winlen=[0,1063]%i.e 2048 samples. So frequency resolution should be 1926/2048=0.9404
> 
> pop_newcrossf( EEG, 1, 1, 2, [0  1063], [0] ,'type', 'phasecoher', 'title','Channel 1-2 Phase Coherence','padratio', 1,'freqs',[1,70],'detrend','on','rmerp','on’);

‘winsize' is the important parameter. This is the size of the sliding window. This determines the minimum frequency resolution when taken together with the number of cycles.

> " 'padratio'  = FFT-length/winframes (2^k)                    {2}
>                       Multiplies the number of output frequencies by dividing
>                       their spacing (standard FFT padding). When cycles~=0, 
>                       frequency spacing is divided by padratio.”

Padding ratio is FFT padding. It is a way to increase the number of frequencies returned (although it does not truly increase the resolution). I suggest you look at some of the lectures on the EEGLAB online tutorial.

Best wishes,

Arno