From: Jim Peters (jim_at_uazu.net)
Date: 2002-01-26 19:41:45
Jim Meissner wrote:
> Dear Jim Peters:
>
> > The next is from "downey.dat", at a point where there seems to be
> > alpha and delta activity (if I've got the rate right).
>
> In all the files you will see a small (and sometimes not so small)
> amount of 60 Hz pickup. That will give you the "calibration" you
> need. The sampling rate was about 130 Hz and not well controlled.
>
> I am fascinated with the work you are doing. I clicked on the
> files, but could not figure out what I was looking at? Please tell
> me what to look at the files with? My computer showed it as a BMP
> file. Help?
They are PNG files, and your web-browser (IE?) should be able to
display them without problem. Did you see the display with the
numbers along the left hand side and the waveform along the top ? You
might have to use the scrollbars to see all of it. If you don't see
it, perhaps you need to cut and paste the links into Internet
Explorer. I don't know Windows, so I really have no idea what you
have to do, but your browser should be able to handle it.
What you are looking at is a snapshot of the program running. It is a
bit like an oscilloscope in that the data is drawn from left to right,
and on reaching the right, the drawing point flips back to the left.
The waveform display at the top is exactly like that.
The rest of the screen is the display of bands. The top band is the
highest frequency, with the lowest at the bottom. They are equally
spaced in log-frequency space with 6 bands per octave. This means
that there are six bands between 2 and 4Hz, six between 4 and 8Hz, and
so on. The numbers on the left are the band centre frequencies in Hz
(which are correct only if I got the sampling frequency right).
Because there is a delay caused by the filters on each band, with
longer delays for lower frequency bands (which are narrower), the data
generated at the current moment doesn't appear in line with the
waveform at the top -- instead it appears lagged a little bit behind.
The level generated 'now' actually belongs to an earlier time. That
explains why the drawing points curve away downwards.
Really, since I'm working from a file, I could calculate all of this
ahead of time and make the drawing points all line up, but I'm trying
to emulate the situation of working in real time where there is always
a delay according to the width of the band.
If you use all bands with the same width (as in a windowed FFT), the
delay for all of them would be the same. I'm getting different delays
because all of my bands are different widths.
What you see to the right of the drawing points is data from just
before the drawing points wrapped around right->left last time. This
is clearest in the first screen where you can see the old data to the
right was at a lower frequency.
I hope this isn't too confusing. It's hard to explain.
This is just a prototype to help me understand how these filters
behave. You could see it a bit like a waterfall display turned on its
side. The display for the user would obviously have to be very much
simpler than this.
Jim
-- Jim Peters (_)/=\~/_(_) Uazú (_) /=\ ~/_ (_) jim@ (_) /=\ ~/_ (_) www. uazu.net (_) ____ /=\ ____ ~/_ ____ (_) uazu.net
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:37 BST