At 05:21 PM 11/14/2005, John wrote:
WD2XES is running WOLF at the 5 bit/sec rate on 137.422 kHz with 50 watts.
2005-11-14 23:56:23 >WOLF -r 8100 -f 799.95 -t 1.0 -w 0.0000
t: 48 f: 0.000 a:-0.3 dp:147.0 ci:15 cj:221 XES 50W WOLF 5 -
t: 96 f: 0.000 a:-0.2 dp:148.7 ci:15 cj:221 XES 50W WOLF 5 -
t: 192 f: 0.000 a:-0.4 dp:150.1 ci:15 cj:221 XES 50W WOLF 5 -
t: 384 f: 0.000 pm:334567 jm:519 q: 10.1 10.5 XES 50W WOLF 5 -
Looking at the waterfall display, I see a fair amount of QSB on your signal
tonight John. Probably means quite a bit of flutter in the
freq/phase measurements.
Years ago when we were testing COHERENT BPSK on HF we found the optimum rate
over many different sets of conditions was MS25 (40 bits per second). Most
operators don't type fast enough to overflow the buffer at that rate, and the
HF phase flutter is acceptable. At lower rates copy would degrade
(more on some
nights than others) due to flutter. We are trying to measure the phase during
each bit time, so of course we get the best answer when phase stays
constant over
the bit time. When the frequency changes due to doppler it results
in phase changes
at the receiver (that were not transmitted), so we get more
errors. Same result
with mistuning. At MS100 (10 symbols per second), a tuning error of +/- 2.5 Hz
produces an error in the measured phase of +/- 90 degrees, which gets
us exactly
to the halfway point so the receiver cannot decide which phase was transmitted
regardless of the SNR. The slower the rate (more msec per bit), the
more critical
the tuning gets. I don't know what the optimum signaling rate
happens to be for
137 Khz at night, but I'd be inclined to guess somewhere around MS50
or 20 symbols
per second. We're all looking forward to testing WOLF at the faster speeds.
Bill VE2IQ
| 