I've just watched Jim and Grahams transmissions both in the same slot for
four periods. And all decoded OK.
At a rough guess Grahams transmission was 1500ms shorter in duration than
Jims.
Thus confirming Jims tedious findings.
This constitues a reasonable bit rate error.
I suppose over a longer period of receiving Grahams signal will tend to
produce more erroneous data and be rejected.
Pulling out the antenna early has limited this accululating error.
The converse is probably true on receive, just how long are stations really
receiving for.
And the net total of both transmission and reception timing errors between
two specific stations yeild this odd outcome where, some stations
combinations receive consistently fine, while others struggle.
Stations that clearly see multiple transmissions, yet fail to decode any are
probably suffering a receive frame timing error.
The questions maybe how much of a timing discrepancy is WSPR prepared to
tolerate.
And/or does it make any attempt to phase lock onto the data stream presented
(if so what's the pull in range).
Just some thoughts for the cooking pot.
73,
Lee M0LMH.
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of James Moritz
Sent: 28 January 2009 23:54
To: [email protected]
Subject: Re: LF: WSPR - G0NBD
Dear Graham, LF Group,
Remembering what I said jokingly to RA3AGC a few days ago about visually
verifying a spectrogram trace of a WSPR signal, and the discussions of
merits of WSPR vs QRSS, I thought this ought to work for diagnosis
purposes - after all, WSPR is measuring the received frequency changes and
the times at which they change, and in principle the same info is displayed
by the spectrogram.
So I spent a "hugely exciting" couple of hours studying your signal with
Spectrum Lab, and comparing it with a locally generated signal from my PC
(which everyone seems to decode OK), configured with the same beacon message
you have been transmitting. Eventually, I realised that the difference was
that your transmissions are a few seconds shorter than the locally generated
signals - in the attachment, the locally generated signals have the dark,
noise free background. This means the transmitted sequence will be a couple
of symbols out of step by the end of each transmission - I expect that could
mess up the demodulation/decoding processes quite a lot. If you have a noisy
signal, the demodulator/decoder might be using effectively just the least
noisy fraction of the signal, with less timing error over a shorter part of
the sequence.
To test this theory, I waited until about 1 minute of your signal had
appeared on the spectrogram, then unplugged the antenna, so that the
remaining minute was white noise. The result was successful decodes! I tried
this 3 times, and it worked 3 times. So your signal does not have to be
noisy in order to decode, there has to be part of it missing in order to
decode!
So that might be why the signals don't decode - whether this is a problem
with the soundcard, motherboard, or software is probably a question for
K1JT! Still, a step in the right direction...
Cheers, Jim Moritz
73 de M0BMU
No virus found in this incoming message.
Checked by AVG - http://www.avg.com
Version: 8.0.176 / Virus Database: 270.10.13/1915 - Release Date: 1/28/2009
6:37 AM
|
