From: Jim Peters (jim_at_uazu.net)
Date: 2001-12-03 10:11:01
JBilderback wrote:
> restating the question: if there's say 5uv of high frequency noise
> riding on
> a 100uv EEG signal then what is gained with 12-16 bit data sampling ??
There are several possible reasons (from theory here, not necessarily
practice). For one, if the noise is within a well-defined frequency
range, you can filter it away, so it doesn't necessarily mean that
those lower bits are worthless. Also, if it's the other way around,
and you have a huge 60Hz wave and a 5% signal (I've not the experience
to know if this really happens) those extra bits are even more
important to get a good resolution on the 5% signal once you've
filtered away the unwanted signals. (Someone else please add to this
if what I've said isn't always true in practice).
The other advantage of having more resolution than you really need is
that you don't have to be so fussy about getting the signal gained to
full-scale all the time -- you can give yourself head-room that can
absorb sudden changes in signal amplitude without risking the signal
clipping.
For instance, in recording music, you would never adjust the gain to
give full-scale signals at normal levels. If you were recording a
singer, you would blow the whole recording as soon as they went for a
big note. So you effectively leave a few bits spare at the top to
absorb these changes in signal level. (You don't notice this with CD
recordings because all the signal level changes have already been
ironed out using audio compression.) For this reason, with audio it's
better to record at 24-bit resolution, even if you're aiming for a
16-bit final result.
Hope this helps -
Jim
-- Jim Peters (_)/=\~/_(_) Uazú (_) /=\ ~/_ (_) jim@ (_) /=\ ~/_ (_) www. uazu.net (_) ____ /=\ ____ ~/_ ____ (_) uazu.net
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:32 BST