From: JBilderback (jiva_at_humboldt1.com)
Date: 2001-05-13 17:30:10
> Hi !
>
> About the soundcard idea: could somebody maybe find out how they are accessed
> in Windows and Linux ? I think it wouldnīt matter at all if we could use
> only high sampling rates like 40kHz. It would be possible to do some filtering
the specification section on the $20 soundcard that I posted some time back said
that the sampling rate was programmable ... but did not say what the minimum
sample rate was. someone else pointed out that the higher sample rate also
offers
an option for recovering multiplexed input channels.
> and convert this signal into a slower datastream accessible by our software.
> The driver-issue shouldnīt be such a problem.
I'll look around for programming data and post links to what I find.
> That leaves us with two problems:
>
> - the bandwith (itīs limited on the lower side so we canīt sample signals below
> maybe 5 Hz)
I gather this is your best guess on lowering the bandpass of the dc blocking
cap.
Is there any other way to isolatate the input section? Even if high frequencies
are lost? (This project doesn't care too much about the chipset abilities to
20khz)
> - linear isolation. Distortion and so on could limit the number of usable bits.
> Maybe it could be calibrated using a signal generator or something like that.
>
The person who's pre-amp design was posted earlier appears very serious into
EEG. I suppose you could write him about any problems with his isolation
technique. The biggest issue I saw (very limited peak) with his design was
the +6 -6 battery power supply. 8 batteries and the holders would add a lot
of bulk/cost to the unit.
> ________________________________________________________________________
> ________________________________________________________________________
>
> Being of a non-technical nature, I don't understand half of what's
> being said on the sound card discussion, but I think we should bear
> these points in mind:
>
> 1. Performance should not be compromised. If using a soundcard means
> we cannot train at certain frequencies then we should not use a
> soundcard. I think that to really take off our product should be a
> perfect eeg system.
>
It might be an alternative (very low cost) system. I haven't done my
homework (rather forgot what I've read) but I think not all eeg work
requires the near 0hz frequencies.
> 2. How much cheaper would using a soundcard make things? Personally,
> I would view $300 as "cheap" for an eeg system. Obviously, I would
> prefer to see a cheaper system, but not by sacrificing quality, so if
> we're talking about a difference between $150 and $180 (for example)
> then I would say the price difference is not worth the compromise.
That is the question.
There are no system costs estimates for any of the project options yet.
(as near as I can tell). My completely unqualified "gut feeling" is that
50 manufactured units of the RS232 system will run close to the cost
of a brainmaster.
>
> It may be that I have missed the point here as I don't fully
> understand the sound card discussion, so if this post doesn't make
> sense then please ignore it :)
>
As much input and discussion as possible is desirable as far as I'm
concerned. However I don't want it to derail existing efforts so
tell me to shut up if it's counter-productive.
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:29 BST