Stefan, Jay, John,
the "96%" number shows the temporal overlap
between the receive period (ie the duration of the currently
used FFT, about 35 minutes) and the identified Op sequence (33
minutes). With 10 minutes spacing between subsequent FFTs, the sequence
does not always fit completely into one of the slots. so it says that 96%
of the sequence was included in the evaluation. However the missing
part has only neglegible impact on the detection
sensitivity.
Ideally Opera can be treated as an AM
signal, with a central carrier and modulation sidebands around it. The "2mHz"
bandwidth figure refers to the bandwidth of the carrier, which should be as
small as possible. Opds applies some smoothing to the power spectrum and then
tries to measure the -10 dB bandwidth of the central peak. Stable and
phase coherent signals consistently show less than 3 mHz bandwidth. An
intermediate width up to about 30 mHz typically indicates a coherent
signal but with a slight thermal drift. Even higher bandwidth (~ 100 mHz)
are mostly due to incoherent keying, ie random phase dashes caused
by stopping the TX oscillator or divider during gaps.
Opds internally uses an "autofocus" concept
similar to synchroneous demodulation, where the central spectral peak is used
as a phase reference. Narrower carriers produce better
demodulated SNR. For fading or incoherent signals, the phase has to be tracked
faster or even on a dash-by-dash basis, which is much less
efficient.
The "dBOp" column is showing SNR according to
José's Opera scale, which is approximately based on average power. It shows 4
dB more negative values than the standard WSPR scale, ie. carrier power in 2.5
kHz. A marginally decode with WSPR-15 would need -38 dB, and an Opera
signal with same PEP would then show as -42 dBOp. For a coherent signal, the
Opds-32 threshold should be around -50 dBOp, which in theory is 8 dB
better than WSPR-15 and 11 dB better than standard Opera-32.
Please be aware that the SNR figures shown in
opds results can sometimes be inaccurate. With an incoherent
signal, often not all of the carrier power is captured during the
bandwidth measurement, producing a low or invalid SNR
reading. The SNR measurement also doesn't work well for strong
signals (> -20 dBOp), eg for DK7FC who should really be plus several dB
here.
So why was Stefan's TA signal not stronger last
night? My guess is that TA propagation just didn't extend into
central Europe: While UK and duch stations received Bob well on 74 kHz, little
or nothing at all apperared on Hartmut's and my 74 kHz
grabbers.
Best 73,
Markus (DF6NM)
PS The weather has improved here,
so I have put out the TX antenna for a possible joint TA session
tonight.
Sent: Friday, October 18, 2013 2:36 PM
Subject: Re: LF: T/A OPDS DK7FC
Hi John, Jay, Markus,
OK. Well, it seems to work. But i can't
value the results. Why is it only 96% and what is the meaning of 2 mHz
here?
Seems the S/N is rather low. Condx must have been poor.
Were there
other reports from US stns?
73, Stefan/DK7FC
Am 18.10.2013
13:05, schrieb John Andrews:
Stefan, Markus, Jay,
Results from last
night:
2013-10-18 05:09:25 DK7FC 5981km
137560.016Hz 2mHz -42.8dBOp 96% 20.6dB
2013-10-18
04:29:25 DK7FC 5981km 137560.016Hz 2mHz
-44.8dBOp 96% 17.8dB
2013-10-18 03:49:23 DK7FC
5981km 137560.016Hz 2mHz -44.0dBOp
96% 20.3dB
2013-10-18 03:09:23 DK7FC 5981km
137560.017Hz 2mHz -43.7dBOp 96% 19.0dB
2013-10-18
02:29:23 DK7FC 5981km 137560.016Hz 3mHz
-47.5dBOp 96% 17.9dB
2013-10-18 01:49:23 DK7FC
5981km 137560.016Hz 3mHz -46.4dBOp 80% 16.8dB
John, W1TAG
