[Lowfer] WSPR on 630 meters tonight

[JD]

Started off this evening with XIQ in the 2328 UTC timeslot, then XXM showed 
up later in the 0030 time slot.  Good clean late afternoon conditions had 
XIQ around -4 or -5 dn SNRr and XXM at +5.  As usual, XXM was around S4, 
while XIQ was likely around S1.  Noise was only in the S1 range as well, 
with excursions to about S2 at times, in a 250 Hz bandwidth.  XIQ was clean 
enough to have worked solidly at QRSS1, if any of us had that available, and 
was generally detectable by ear as well.

Although I could see no difference at the narrow BW on the S-meter when my 
Honda generator was recharging the batteries, its faint broadband hash 
around 630 to 600 meters did degrade XIQ's reported SNR to -8.  When I cut 
it off, decodes again reported -4 to -5.

Two interesting phenomena seemingly related to the software were observed 
this evening while the noise levels were so low:

* When the generator was not on, and noise was minimal and fundamentally 
Gaussian, and the signal was constant in level, the reported SNR for XIQ did 
not vary regardless of whether the receiver IF bandwidth was 1.8 kHz or 250 
Hz.  Under other conditions involving greater QRN and/or QRM, however, I 
have previously noted 7 dB or more improvement at the narrower bandwidth.

* XXM being so far above the wideband noise level seems to make decoding 
appreciably slower!  That puzzling fact might account for the failures of 
XIQ to decode sometimes that I reported Saturday morning.  If the decoding 
step takes more than the 1 minute 50 seconds during which the new WAV file 
is being recorded, WSPR apparently does not attempt to record the next time 
slot for processing.  I was close to that a couple of times with XXM's 
signal this evening, but had no good explanation for it.  Then I noticed 
that if I backed down the receiver output enough to degrade XIQ's reported 
SNR another 3 dB or so, the decodes of XXM sped up considerably and took 
just under a minute then.  It appears that if a signal is either unusually 
weak OR unusually strong, decoding takes longer than average.  One might 
suppose the latter has something to do with artifacts of the hard limiting I 
use, but the input to the sound card was well within safe limits and not 
appreciably different regardless of the receiver output level.  I can't 
determine what, exactly, might be going on.  One can see capture effect 
cleaning up the background noise level when XXM is on by itself.  There is 
no other signal in the detection passband, and WSPR itself is a single tone 
at a time, so there's nothing to intermodulate with.  The harmonic 
distortion products of the limiting are well outside the detection 
bandwidth.  Spectral analysis shows that, insofar as what the software is 
seeing anywhere near 1500 +/- 200 Hz, it ought to be just a big clean 
sinewave.  Wonder what's going on?

By the time you see this, I'll have been back in the field for a while, 
checking out post-sunset noise levels.  If they look promising for any 
potential DX, I'll keep watching...but no guarantees of another all-nighter 
this time.

JD