From: sleeper75se (sleeper75se_at_yahoo.se)
Date: 2001-12-05 10:34:39
--- In buildcheapeeg_at_yahoogroups.com, "Jim Meissner" <jpmeissner_at_mindspring.com> wrote:
> Dear Andreas:
>
> I am so glad that program worked for you.
> I was not sure about how to upload and download stuff.
Hello Jim,
first I want to clarify one thing - I did not download egavga.bgi
from your site, I already had it on my system so there was no need.
Upon seeing another post you made, I downloaded your version and it
proved to be corrupted somehow. My bad, I should have checked all
your files properly, and for that I apologize. The other files
(slink12r + .dat files) work perfectly though.
A working version can be found here (second hit in a Google search):
http://www.csupomona.edu/~pbsiegel/www/stuprog/egavga.exe
> The 60 Hz pickup you see is a minor annoyance and I have spent
> considerable time to eliminate it.
> I did not choose the conventional solution of a notch filter at
> 60 Hz or a steep roll-off of the high frequency end because I did
> not want to throw away possibly valuable data.
With today's faster computers, digital filtering makes it possible to
be much more selective, as they don't require any trimming. How about
75dB attenuation (5th order IIR-filter) in a 1 Hz wide stopband
centered around 60Hz (or 50Hz for Europe), would that cause
significant loss of information? Looking at your data, it appears
there's not much going on at those frequencies.
> What you call muscle artifacts may be much more than you think.
> I had to DC coupled the input stage so I could get clean and
> accurate .5 to 2 Hz signals. Look at the Downey file. These
> are the brain waves of a very gifted and quite accurate medium.
> I am a real doubting Thomas and it takes a lot to convince me.
> He had access to information that I had to call my mother to
> confirm. Very impressive and quite humble. He allowed me to test
> him several times. When he was in his mediumistic state he
> produced massive high amplitude Delta waves.
That's pretty cool! :-) Do you have any idea of what amplitude these
delta waves had? On the order of several millivolts or...? Just
trying to figure out how many bits you need to capture them.
> There is a problem with common mode rejection. The circuit that
> I use allows you to get infinite common mode rejection by adjusting
> the balance pot. BUT that assumes a balanced source resistance.
> Logic would tell you that if you had a very high input impedance
> input amp that would not be a problem.
> After a lot of experimenting, I had to install a 10 Meg ohm input
> resistor to ground.
It would be really interesting to see the schematics of a tried and
tested circuit. Did you use a right-leg-drive solution or passive
ground, for instance...
> In a meditative state the skin resistance will rise to 100 K ohms
> and sometimes several meg ohms.
A most valuable piece of information! This is something you won't
find in a medical textbook - doctors don't do research on altered
states.... :-p
> do not see the same source. Therefore the common mode specs may
> be meaningless.
It is true that the CMRR degrades with unbalanced electrodes, but
surely, a circuit with higher CMRR is better? You could at least use
it for comparing different solutions. Assuming of course that the
designer didn't make the mistake of stating the intrinsic CMRR (of
the opamps) that changes as soon as you put them in a network with
resistors and capacitors... :-)
>
> Please tell me more about your application, maybe I can help.
Thanks, that's very kind of you. Ok, here goes... the technical bit
is more evolved (naturally). My blind spots really lie in the
biofeedback-specific issues.
My meditation practice is going nowhere fast, so I figured that if I
can see (or rather, hear, with audio feedback) what's going on, I
could learn to meditate like a zen-monk. Right now, I'm planning a
system with one or two channels. I would put the reference electrode
on Cz and the live electrodes on O1 and O2 or thereabouts to get good
alpha and theta readings. If you think some other placement is
better, please let me know.
I will then write a program that monitors the activity (averaged over
a few seconds perhaps) in different frequency bands and plays sounds
as it changes. An increase in activity produces lower/higher sounds.
Different frequency bands can have different instruments (I will use
the MIDI-synthesizer on my sound card).
I've designed an active electrode that hopefully will do the job, it
is still only a paper product:
http://groups.yahoo.com/group/buildcheapeeg/files/hardware%
20design/PDF_and_PNG_schematics/Active-electrodes.pdf
I only need about 15x15 mm (0.6x0.6 inches) of PCB-surface per
electrode if I use surface-mount parts. I can then place the
amplifiers on my head, near or on the Ag-ACl electrodes I'll be
making.
The specs: Two-opamp instrumentation amplifier, with an integrator in
the feedback loop to get a 1st order highpass response. The gain is
about 4 @ 0.01 Hz, 40 @ 0.1 Hz, 100 @ 0.4 Hz and levels off at 125 @
1Hz. In other words, it is AC-coupled, but low enough to get good
delta-readings down to at least 0.4 Hz. As delta-waves are generally
of higher amplitude, I'm hoping that the lower gain won't be that
much of a problem, even though I get lower signal-to-noise ratio.
With 'perfect' (as in simulated) resistor values I get about 92 dB
CMRR, unless I tweak the right leg driver. Removing resistor R104
gives me a CMRR of about 140 dB, but I'm not sure my 'brain-model' is
accurate enough, so these numbers may be wrong.
The model looks like (if it looks like crud, copy and paste into
notepad)
---------> live electrode (+)
|
V1 = 100uV sine voltage source
|
-----+
| |
V3 Z
| Z 100k 'unbalancing resistor' -
| Z useless as I simulate with TL071.
| |
DRL V2 = as V1
|
---------> reference electrode (-)
V3 = 1.5V 50Hz sine - power line interference.
DRL = right-leg-drive connection
I would appreciate any advice or criticism you have to offer,
particularily regarding electrode placement, and biofeedback
practicalities.
/Andreas
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:32 BST