Hello Domenico,
Very interesting and a good work, my compliments!
I often had the wish to have more options for the WSPR uploads. For
example i'm running 2 instances on MF from the two loops on the tree
(nw/se and ne/sw). Mostly there are two uploads from DK7FC/P for a
single station from both loops. I would like to upload only the decode
with the highest SNR. This should be possible with a script in Linux.
Also i saw that a noise blanker can improve the decode performance
significantly on LF. Of course we expect this but many people seem to
belive that it is already all inside the software. It isn't! And WSPR
is probably not optimised for LF/MF.
However using a NB can cause multiple decodes from strong signals (!!!)
Thus i usually don't use a NB since i'm in the center of EU activity
here and have to deal with many strong signals.
For you it is probably not so dramatic since you are quite far away
from most strong stations.
Did you try to optimise the vtblank parameters?
Maybe it is possible to measure the strength of the noise (in a 1 kHz
BW, 1 min moving average for example) and dynamically change the
vtblank parameters?
A great tool!
There are so many playing fields to operate on. You are working there.
It is nice to see so many different places where things can be worked
out here below 500 kHz...
Since i have a Linux PC and my MF vorbis stream from the tree, i could
use your tools to realise the above functions. But i have no time
unfortunately. To many open projects. But maybe in some time :-)
Tnx es 73, Stefan
Am 28.10.2017 02:21, schrieb Domenico IZ7SLZ:
Hi, LF
On 10/26/2017, Paul N1BUG wrote:
It seems WSPR-X is "conditionally stable". We had a fair bit of
discussion about it on this side of the pond last winter. It works for
some, but for others it doesn't. For me it stops doing anything after a
few minutes and just idly sits there. I am not alone. There are others
with the same issue and some for whom the program silently self
terminates at seemingly random intervals. In my case it's on Windows 10
64 bit. I suppose there are too few of us interested in WSPR-15 to
carry any weight with the developers. :-(
I want to share the results of an experimental whisper decoder
running now at IZ7SLZ, maybe someone is interested:
since some weeks i'm testing the linux version of WSJT-X. It does not
manage the wspr-15 mode but its internal decoding program 'wsprd'
does. So i wrote a shell routine which, by using some linux commands
and programs from the vlfrx-tools package (thanks to Paul Nicholson) ,
is able to use 'wsprd' without running WSJT-X program but only with
the shell 'autowsprlf' (attached).
The result is a nice and stable decoder for wspr-15 and wspr-2 (at the
same time) processing the audio coming from my LF receiver tuned on USB
dial 136.000 kHz. In the attached screen-shot you can see of the
program running. Furthermore, thanks to the excellent noise blanker
P.N.'s program 'vtblank', it seems to be more sensitive compared with
the wspr-x program running on the same audio. I have always reported
2-3 dB more in S/N ratio, and get more decodes.
In the shell file i'm using the program 'curl' to upload the
received spot on the wsprnet.org site.
These are the spots so far received with this setup : http://qsl.net/iz7slz/OPDS/ALL_WSPR.TXT
(sorry if the list is upside-down ordered). Note that some LF spots are
wspr-15 signals (i.e. 171027 2300 2 -35 -0.32 0.1376102 G8HUH IO81
20 ... thanks to G8HUH for transmitting this rare signal !)
I will be happy if some operator wants to test the program. It
is a little bit messed, specially in the timeslot string generation, so
any helps and comments for improving it are welcome.
All you need is a pc with linux (Ubuntu 14.04LTS works fine here) with
'autowsprlf' and following vlfrx-tools stuffs running in background:
- vtcard (to manage the audio card)
- vtresample (for 12000 samples/s audio required by wsprd)
- vtwrite (to record on the hard drive the raw audio)
other programs are running here, but i think they are not necessary for
wspr mode:
- gpsd for getting the precise time from my Jupiter GPS connected at
COM1 (1pps and binary protocol);
- ntpd (configured only for the gpsd daemon not for other ntp servers)
- vttime - for the audio calibration and time-stamping
Other routines are running on same Ubuntu PC for my auto-EbNaut
decoder and for sending audio stream to another Windows's pc where is
running SL and DF6NM's opds,..but this is another story.
I'm at disposal for any helps on the configuration of this setup.
73 all, Domenico / IZ7SLZ
|
|