[time-nuts] First results of eclipse data recording
John Ackermann N8UR
jra at febo.com
Sun Aug 27 11:20:54 EDT 2017
(This is more amplitude than frequency related, but thought it might be
interesting to the group. If not, sorry for the intrusion.)
I used three software defined radios to record just under 1 TB of data
during the eclipse period from my cottage at Beaver Island, Michigan,
over 8 hours from 1400 to 2200 UTC. Frequencies recorded included the
entire AM broadcast band and four ham radio bands, one of which also
covers WWV at 10 MHz. (Obligatory time-nuts/FMT-nuts: the receivers
were either locked to an FTS4100 cesium standard, or had reference
markers derived from it.)
It's taken a few days just to aggregate and back up the data. Last
night I was able to do the first actual processing.
This initial result is a "spectrum movie" showing the AM broadcast band
from 0.500 to 2.00 MHz over the 8 hour test period, recorded by a Red
Pitaya SDR at 2.5 megasamples/second. It will take more work to
separate noise/spurs from real signals but it does look like carriers
came up out of the noise during/after the eclipse period.
The movie was created with a Gnuradio script reading the recorded data
and displaying an FFT plot, which was captured with a screenshot tool
every five seconds. I then turned the images into an .mp4 movie at 30
frames per second, so the 8 hour period runs by in a bit over 3 minutes.
I used a deep FFT (8192 points) with long averaging to get rid of as
much of the crud as possible.
Blog post with some details here: http://blog.febo.com/wp/?p=202
Get directly to the movie on YouTube here: https://youtu.be/bqRhfCiD8Pw
There's lot more analysis to be done, but I thought this was a necessary
(and interesting) first step.
More information about the time-nuts