8345 CW active now 2019z 12/12/19. no interpreter running.
BTW, thanks for all the work you put into this thread Josh.
Glad you find it interesting, it's my tribute to those who taught me, and there are many.
On cw interpreters, here's an video on the subject comparing several;
https://www.youtube.com/watch?v=MRuFn1nnuoYI find that decent code with no multipath can be correctly copied by the usual sorcerer and etc software decoders, but once a bad fist, qrm/qrn, or multipath appears, it takes a human to make sense of the mush. It's odd but you can hear multipath on a cw note after you've been doing it a while.
Worse yet is Russian Navy doesn't seem to issue keyers to its forces, it issues straight keys, meaning every operator is going to have their own element timings.
Found out the Bubnov is sitting off the coast of Morocco at the moment, so plenty of opportunity for multipath from there to here. Hearing RIW Moscow or RCV Sevastopol is sure to have multipath too.
Speaking of Moscow and Sevastopol, 12464 and 8345 are the primary ship to shore side of the duplex channel, the shore to ship side are;
RIW Moscow 9145, 11000, 14556
RCV Sevastopol 15586, 7566
I've compared several decoders; sorcerer in two versions, K500, WCODE, CW Get, and CW Decoder and they all pretty much perform about the same, perhaps CW Get is a bit better, while WCODE will translate Cyrillic morse - wich I am not good at at all. I let them run overnite or when I'm away or listening to something else with another rig.
One thing I noted is most all of them want about 15 percent cpu time, a bit high to my thinking, setting CW Get to 50 cw signals reduced it to 4 percent, much more to my liking.
CW Decoder dl;
https://download.cnet.com/CW-Decoder/3000-18511_4-75325058.htmlSorcerer dl;
http://www.kd0cq.com/2013/07/sorcerer-decoder-download/One thing I find useful is to set the input freq of the decoder app to twice the cw note freq.
For example, I always employ 600Hz bfo offset, all cw copied ota will be at that 600Hz tone. Setting the decoder to 1200Hz will allow the second harmonic to trigger the decoder rather than the actual tone at 600Hz plus all the band noise, of course one must make sure a second harmonic is present in the app for this to work - you will verify it in the fft when signals are present. The 1200Hz second harmonic will be pretty much completely free of band noise.
I've been thinking about a means to a better decoder app, one that perhaps employs a pll to;
1 detect a sinewave
and
2 to ensure phase coherence wich will also enhance noise rejection (noise is incoherent) and define elements.
(Not that I know how every decoder works)
Then to have a buffer to copy elements into and look for the averages of an individual senders keying, dits, dahs, spacing, then apply these globally to the message for its entirety -- the above is why humans are better cw decoders than machines as we do all this without thinking about it.
I suspect that a decoder written with these features could decode cw that's beneath the noise floor, and humans can't do that. Smolinksi, I'm looking at you.
You could even determine/identify individual senders this way but that has been done for ages.
Lol using massive compute power of the most complex and modern design to decode the simplest of digital codes, dits and dahs, ones and zeros.
Show Posts
