[Eeglablist] Filter roll-off (not just for Butterworth filters)

[Steve Luck]

>
Hi.  In response to the question about filter order for a Butterworth  
filter, it is important for EVERYONE doing EEG/ERP research to realize  
that a steeper roll-off will mean that the filter will cause more  
temporal distortion of the waveform.  This is true for all standard  
(FIR) filters, not just Butterworth filters.  The most fundamental  
principle in filtering is that precision in the frequency domain is  
inversely related to precision in the time domain.  That is, the  
narrower your passband and the sharper the roll-offs of a filter, the  
more the filter will spread the signal in the time domain.  Unless you  
are primarily looking at frequency-domain phenomena, it is usually  
better to have a broader passband and a more gradual roll-off.  So,  
unless you are looking at true EEG oscillations, you will probably  
want a low-order filter.

This is the opposite of the advice people usually give.  But people  
are usually thinking about the frequency domain and forgetting about  
the time domain.  If you don't think about what filters are doing in  
the time domain, you may end up creating artifactual peaks and  
oscillations in your waveforms.  If in doubt, pass a known signal  
(like a single cycle of a square wave) through your filter and see how  
the filter distorts the waveform.

For a detailed overview, see Chapter 5 in An Introduction to the Event- 
Related Potential Technique.

Steve Luck


> From: "Wambua Kazi" <wambua.kazi at gmail.com>
> Date: July 29, 2008 10:05:13 PM PDT
> To: eeglablist at sccn.ucsd.edu
> Subject: [Eeglablist] Pros & cons of Butterworth filter order
>
>
> Hello,
>   I am trying to figure out how to bandpass filter an EEG data set.   
> From skimming the literature it appears that Butterworth filters  
> are  commonly used for filtering EEG data and strike a good  
> compromise between the steepness of the transition band and ripples  
> in the pass and stop bands.  It's also clear to me that the higher  
> the order of the filter, the steeper the transition band.  Are there  
> any cons to using a higher order Butterworth filter?
>        thank you for your help,
>             -Wambua
>
> wambua.kazi at gmail.com
>

--------------------------------------------------------------------
Steven J. Luck, Ph.D.
Professor
Center for Mind & Brain and Department of Psychology
University of California, Davis
267 Cousteau Place
Davis, CA 95618
(530) 297-4424
sjluck at ucdavis.edu
--------------------------------------------------------------------


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20080730/312ff560/attachment.html>