Regarding Observation 2400834 …
This is weird - this observation has an unusually large number of data frames (302) - altho only 12 display in the dk3wn tlm decoder. Anyway, when I grab the IQ file from the pi gs 2400834.raw it is actually the iq for VZLUSAT1 2400835 cw !
The post obs scripts is just:
# cat /etc/default/satnogs-client
# cat /var/lib/satnogs/pos.sh
/bin/mv /tmp/iq.raw /tmp/$1.raw
wonder why the one off. Ending timestamp on the web of 2400834 is 2020-06-18 15:07:43, for 2400835 is 2020-06-18 15:13:55 and the timestamp of 2400834.raw is (EDT) Jun 18 11:13 or 15:13Z which matches 2400835 exactly, and 6 minutes after 2400834. Somehow they got conflated in the temporal proximity and the grifex obs raw iq was lost
Anyway - wonder about all the extra data. Looking at https://www.qsl.net/k/k4kdr//how-to/jpg-from-raw-data.pdf and the data there is an ffd8 start of image and ffd9 end of image
$ xxd -p grifex_2400834.kss | less
Trying in a hex editor - there were interestingly 128 c0c000 kiss frames between those - but appearently not an actual image as it is full of invalid jpg ‘markers’ ffxx . Giving up.
Also - the data w/o kiss frames is 76502 bytes long, so by sheer statistics the words ‘ffd8’ and ‘ffd9’ have a good odds of appearing randomly I think any given 16bit word has an even chance of appearing in any randomly distributed series of 65536 16bit words.