|
Hi Graham,
no I haven't yet tried to compare the
results directly to the original Opera decoder - I haven't even
downloaded Jose's software yet.
Here I'm still working on the carrier
self-focussing part, which seems to greatly improve the response to slightly
drifting or fading signals. There's also still some uncertainty about the
SNR calibration, which is based on the energy of the reconstructed spectral
peak.
During the last two nights I had selected
incoherent (ie. power only) detection because I tried to catch UW8SM who is
not phase continuous across dashes (see zoomed sections in attached
spectrogram). I think this is also the way Jose's decoder works, so I
wouldn't expect a large sensitivity gain from the correlation. To
reap the benefit of coherent correlation, a phase continuous and
stable transmission is required. To avoid having to decide beforehand,
I intend to let the software automatically adapt the matched
filter width to the each spectral peak.
To be able to monitor the full band I
have now set this up as a postprocessing tool for SpecLab's FFT export. Each
262144 point complex FFTs covers 125 Hz across a 35 minute interval (477
µHz resolution). A new file is generated every 10 minutes so there's some
overlap. One FFT file is about 1.6 MB ASCII, summing up to 100
MB of data per night - yes we're literally searching for needles in
the haystack ;-) Last night I've changed the procedure from offline to
real-time processing, so I can now view a spot within a few
minutes of the end of transmission. Actually detections often start to
appear already when the signal is about half way through.
When things have progressed a bit further I intend
to make the correlation software available to all.
Best
73,
Markus (DF6NM)
----- Original Message -----
Sent: Saturday, January 26, 2013 5:11 PM
Subject: Re: LF: Re: Re: Op32 correlation detections: UW8SM et
al
Marcus
Do you run the Opera decoder during , or post
, these tests to indicate the point
at which your process takes over in terms of
s/n - decode level ?
How big is the audio file for such a
observation ? could one be sent via the
web or would a remote observer need to
run the decoder locally ?
73 -G..
Sent: Saturday, January 26, 2013 11:55 AM
Subject: LF: Re: Re: Op32 correlation detections: UW8SM et
al
... consistent detections from
Andy throughout last night:
Jan 25-26:
frequency
peak focused Opera
end time corr. callsign
137542.088Hz 13.4dB 33.4dBaf -34.2dBOp
18:36:55 20.8dB DF2JP
137566.957Hz 4.5dB
31.8dBaf -35.9dBOp 19:04:10 18.3dB UW8SM
137566.953Hz 8.9dB
33.7dBaf -33.9dBOp 19:52:01 20.8dB UW8SM
137566.972Hz 10.1dB 33.8dBaf
-33.9dBOp 20:39:50 21.1dB UW8SM
137557.023Hz 5.1dB 33.4dBaf
-34.2dBOp 21:06:01 19.6dB RN3AGC
137566.950Hz 9.5dB 31.7dBaf
-36.0dBOp 21:27:41 20.5dB UW8SM
137566.967Hz 6.5dB 32.5dBaf
-35.1dBOp 22:15:32 20.2dB UW8SM
137556.947Hz 8.5dB 31.7dBaf
-35.9dBOp 22:28:04 20.3dB RN3AGC
137566.977Hz 4.6dB 28.7dBaf
-38.9dBOp 23:03:23 16.1dB UW8SM
137556.942Hz 11.1dB 32.6dBaf
-35.0dBOp 23:37:35 20.8dB RN3AGC
137566.994Hz 8.9dB 31.9dBaf
-35.8dBOp 23:51:12 19.0dB UW8SM
137566.974Hz 5.7dB 30.3dBaf -37.4dBOp 00:39:03 19.4dB
UW8SM
137566.963Hz 9.1dB 31.0dBaf -36.6dBOp 01:26:54 20.3dB
UW8SM
137566.987Hz 8.2dB 32.8dBaf -34.9dBOp 02:14:43 20.5dB
UW8SM
137566.984Hz 8.7dB 32.7dBaf -34.9dBOp 03:02:34 21.3dB
UW8SM
137566.977Hz 8.6dB 32.6dBaf -35.1dBOp 03:50:25 20.4dB
UW8SM
137566.981Hz 10.0dB 30.3dBaf -37.3dBOp 04:38:14 21.2dB
UW8SM
137566.967Hz 9.2dB 31.4dBaf -36.2dBOp 05:26:05 21.1dB
UW8SM
137566.971Hz 4.2dB 27.1dBaf -40.5dBOp 06:13:54 16.1dB
UW8SM
137566.976Hz 5.6dB 29.1dBaf -38.5dBOp 07:01:45 18.4dB
UW8SM
73, Markus (DF6NM in JN59NJ)
| 
