bandwidth speed mode 4 decimation 1/4: 104, 134 or ?

Post Reply
tom_campbell75
Posts: 18
Joined: Fri Dec 04, 2009 12:34 am
Location: Center for Visual and Cognitive Neuroscience, North Dakota State University

bandwidth speed mode 4 decimation 1/4: 104, 134 or ?

Post by tom_campbell75 »

Dear all,

Please I ask for you help in answering a question about bandwidth on the ActiveTwo Speed Mode 4 with decimation 1/4 that leads to a sampling rate of 512 Hz. Various sources of information seem contradictory.

We would like to report the bandwidth used for recording in a manuscript. We believe that the actiview interface displayed this:
http://mindbrain.ucdavis.edu/people/cam ... 9.png/view

This points to the idea that the bandwidth is DC to 104 Hz, which I think Coen points out in a previous thread is approximate.

However, each EEG channel includes a Prefiltering header value that reads HP: DC; LP: 134 Hz.

The spec (1.7) in the manual reads DC - 400 Hz but I take it that this drops following prefiltering and decimation.

The spectrum of the raw EEG data (prior to any other filtering but necessarily after the prefiltering) is a follows.
http://mindbrain.ucdavis.edu/people/cam ... m.png/view

There seems to be hegemony of what people report in the literature when recording 512 Hz with biosemi, e.g. 100, 134. So please what do you think is the bandwidth? I look forward to your responses.

Kind regards,

Tom Campbell.

Coen
Site Admin
Posts: 1124
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Post by Coen »

The -3 dB frequency for the low-pass filter given in the header is always correct and accurate, and equal to the value indicated in ActiView during recording.

Older versions of ActiView (4.xx and 5.xx) used a 3rd order sinc filter in the decimating routine. This filter has a -3 dB cutoff at approx 1/4th of the sample rate.

Current version of ActiView (6.xx and7.xx) use a 5th order sinc filter in the decimation routine. This filter has a -3 dB cutoff at approx 1/5th of the sample rate. The same 5th order filter is used in the ADC. So, the -3 dB cutoff without decimating (decimation = 1) is always approx 1/5th of the sample rate. The response of the 5th order sinc filter is found at http://www.biosemi.com/faq/adjust_filter.htm .

Best regards, Coen (BioSemi)

tom_campbell75
Posts: 18
Joined: Fri Dec 04, 2009 12:34 am
Location: Center for Visual and Cognitive Neuroscience, North Dakota State University

Post by tom_campbell75 »

Dear Coen,
Thank you very much.
Best regards,
Tom.

Post Reply