Hi Stefan, Graham,
I think Graham's statement is not that far
off.
From memory, I believe that the best
Australian receptions of DK7FC peaked around 12 dB SNR in 11 mHz FFT (eg
VK1SV on March 18). The noise bandwidth is 16 mHz ie -18 dBHz or -52
dB vs 2.5 kHz. Thus the SNR would have been -44 dB in 2.5
kHz, corresponding to -48 dB on the Opera average-power scale. There was
slow QSB which limited the time window of best SNR to some 20 minutes, but
otherwise no visible indications of multipath Doppler
or chirp.
According to Rik's evaluation, Opera-2 decodes
reliably at -23 dB carrier power in 2.5 kHz. Slowing to 32
minutes gains 12 dB, thus the Op-32 threshold is -35 dB in 2.5 kHz, or
-39 dB on the Op scale. Often Opera has been
claimed to go 2 dB further down (to -41 dB Op scale), but I believe this
may be based on sporadic "lucky" decodes due to favourable noise
statistics. BTW these do show up with other modes as
well, for example WSPR occasionally gets -32 dB spots, way below
its nominal threshold of -29 dB.
The Opera "deep search" correlation method is still
under development, but so far it looks like it can indeed reliably detect
and identify coherent Op-32 signals 10 dB below standard Opera decoding (ie -49
dB Op scale). This means that in the
best-case, an Op-32 signal from Stefan could have been
detected in VK with a theoretical 1 dB margin.
For comparison, JT9-30 should decode down to -42 dB
/ 2.5 kHz, ie. 7 dB better than the standard Opera decoder, and 2 dB
"short of VK".
Of course we have to keep in mind that these dBs
are more or less ballpark numbers, and the whims of propagation and
background noise may add (or more likely subtract) several of them at any
time...
Best 73,
Markus (DF6NM)
Sent: Thursday, December 27, 2012 9:49 PM
Subject: Re: LF: Fw: WE6XGR 0P8 498KHz DIAL 2200 Z TO
NIGHT
Am 27.12.2012 21:18,
schrieb Graham:
[...]
I think using the correlation 10 dB gain ,
Markus has found using OP32 , you would
have a confirmed decode in VK by now and
possibly a OP32 decode as well
Did he?
:-)
Markus, is that realistic?
73, Stefan
, where M-C is looking for a
pattern match , the decode engine is extracting
data , but from the other angle a
confirmation is the only criteria : )
73 -G..
