Dear LF Group,
Thanks for the reports on the WOLF beacon overnight. The beacon closed down
at 0600utc Unfortunately, I drifted off to sleep shortly after starting it
up ... all this beaconing is getting a bit exhausting! So I have only just
seen the comments posted overnight.
As people have noticed, WOLF is quite critical in terms of setting-up
required, and neccessary frequency accuracy and stability. The software
expects the signal frequency to be within hertz of the specified value, so
the receiver has to be accurate to this kind of level. The frequency drift
has to be lower still, so much more demanding than WSPR. Wolf transmits a
data frame of 10 baud BPSK lasting 96 seconds, which contains a 15-character
message with highly redundant encoding. The 96s frame is repeated back to
back. If the SNR is good enough, the WOLF decoder can successfully decode
the message from only 1/4 of the complete frame after 24s. On the other
hand, if the signal is very weak, it allows several frames of the signal to
be integrated over a period of time, potentially making it much more
sensitive than WSPR. After every 96s, WOLF makes a new attempt to decode the
accumulated signal, which is what each line of text on the decoder display
represents.
For this to work, the timing of both transmitter and receiver have to be
quite accurate as well as the frequency, so that the pattern of bits in
successive frames does not get "out of step". This is governed mostly by the
sample rate of the sound card, and for WOLF to successfully decode very weak
signals, this needs to be calibrated within a few tens of parts per million,
which is rather better than sound card manufacturers specify their products.
The calibration utility in WOLF can do this by measuring the frequency of of
an accurate audio tone, unfortunately, not every amateur has a suitably
accurate audio source. I am using an audio source made from a TCXO and
digital divider for calibration, made some years ago specially for this
purpose. The problem with using an off-air signal such as 198kHz to
calibrate the sound card is that the actual audio frequency coming out of an
SSB/CW receiver then depends on the error in the receiver's oscillators, and
the measured frequency depends on the error in the sound card sample rate,
so there are two unknown errors that have to be resolved. So while an
off-air signal can be used to calibrate the received frequency with more
than adequate accuracy, the sound card sample rate is a bit more tricky.
The sample rate error will probably be small enough without calibration when
the signal is strong, where successful decodes can be obtained from short
periods of signal, but is likely to prevent the successful accumulation of a
weak signal over many frames. Also, the experience with WSPR shows that
quite large timing errors can appear for unknown reasons with "some
computers". I think W1TAG has a good point about WOLF being possibly less
satisfactory at 500kHz than 136k, due to the more rapidly changing phase in
the propagation path at the higher frequency. The signal strength of the
beacon in the UK and nearby during the last few days has undoubtedly been
too high for a good assesment of the weak-signal performance of WOLF - if
anyone is interested, I could try a low-power beacon this weekend for the
benefit of UK stations.
Cheers, Jim Moritz
73 de M0BMU
|
