Home built EEG

Chipstein chipstein at aol.com
Mon Aug 8 22:09:03 EST 1994


In article <sjacobs.240.00170058 at maroon.tc.umn.edu>,
sjacobs at maroon.tc.umn.edu (Steve Jacobson) writes:

>I am building an EEG machine and intend to digitize the amplified analog
signals for input into a 
computer.  I am considering using a 16 bit A/D for each channel.  The
question I have is: Is a 16 
bit A/D overkill?  Is 12 bits sufficient.  What kind of voltage range am I
liable to encounter on 
the EEG electrodes?  I intend to use isolation devices between the
front-end 
processor and the main computer.

16 bits is definitely overkill, since an average scalp EEG is about 50 uV
peak-to-peak and the noise is around 2 uV.  Although 12 bit resolution is
nice, I have done a lot of research with an ancient 8-bit A/D.  Isolation
is necessary only if you plan to study patients connected to intravenous
lines or gadgets, or if for some reason the subjects can't be kept away
from earth ground.

Noise is in fact the main problem.  In addition to a good high-impedance
instrumentation amplifier, you should boost the analog signal out of the
millivolt range to avoid noise generated by the computer.

________________________________
Charles M. Epstein, M.D.
Emory University




More information about the Neur-sci mailing list