WSPR-15 Vs. WOLF Tests 17 February 2013
Background: Joe Taylor, K1JT, has produced a slow-speed WSPR mode optimized for low and medium frequency use. The original WSPR format works in 2-minute
periods with a signalling rate appropriate for HF propagation where Doppler shift is a significant issue. That mode works well at lower
frequencies, but the more stable propagation offers the possibility of using lower data rates. Joe's new program, "WSPR-X", is delivering copy with
signal to noise conditions about 10 dB worse than with the original mode. Conventional WSPR-2 minimums are around -29 dB, where WSPR-X can produce
copy at -39 dB or a bit lower. The penalty is, of course, time. Instead of the original 2 minute decoding periods, the time is now stretched to
15 minutes. A discussion of the WSPR-X program may be found at K1JT's
Tests on 136 and 474.2 kHz have proven out K1JT's expectations of WSPR-X SNR's in the upper 30 dB range. The question naturally arose as to how
this mode would compare with WOLF, created in 2001 by Stewart Nelson, KK7KA, specifically for low-frequency communication. It uses binary phase-shift
keying at 10 bits/second to send a repeated 15-character message. The mode's strength is that with careful control of frequency stability and
sampling rates, copy can be built up over a series of 96 second periods. It has proven to be not only a good weak-signal performer, but to
work well in the presence of static crashes. The current version of WOLF, as modified by DL4YHF, may be found at the
DL4YHF WOLF GUI web site.
The Tests: Jay, W1VD, ran a series of tests on Sunday, February 17, 2013 on 136 kHz. The idea was to transmit WSPR-15 and WOLF signals simultaneously, summing
the audio in a resistive combiner, and running
the combined audio through an SSB generator on 136.0 kHz. The WOLF signal was created at 800 Hz, putting it at 136.800 kHz, the the WSPR-15 signal
at 1614 Hz, or 137.614 kHz. 50 Ohm step attenuators and a linear amplifier followed the SSB generator, allowing accurate settings of different
power levels. All references to power are to the PEP of the individual signals into the antenna system, and do not refer to ERP. Tests were
made in the daytime, and all but the last were done under fade-free conditions. This was a cold winter day, with an air temperature around 20F (-6C).
Atmospheric conditions were quiet. The receiving site was at W1TAG, at a distance of 72 miles (116 km). The receiver was an Icom R-75, with a
5-foot diameter loop antenna.
W1VD began transmitting at 1530 UTC with 10 watts per signal. This was widely copied, with reports received from over 600 miles. It was agreed that
with the WSPR-15 signal working in nicely-defined 15-minute "chunks," WOLF copy should be restarted at 15-minute intervals, and not allowed to
build up over longer periods. All power changes were made at the 15-minute points. Copy at W1TAG was excellent, with first-line correct copy on WOLF,
and an SNR of +2 dB on WSPR-15. The first power change was to 1 watt at 1715 UTC. WOLF copy was had on the second line, and the WSPR-15 SNR was -4 dB.
W4DEX reported WOLF copy at 4 minutes, 48 seconds, and a WSPR-15 SNR of -32 dB. The next reduction was to 250 milliwatts at 1815 UTC. W1TAG reported
WOLF text on line 3, and a WSPR-15 SNR of -9 dB. In two runs at this power level, W4DEX had WOLF copy at 9.6 and 11.2 minutes, with no WSPR-15 at
W1VD and W1TAG then conducted a series of tests at lower power levels, looking for the threshold of copy over the 72 mile path. Here is a table
of the results:
We feel that we hit a pre-sunset fade on the last run beginning at 2000 UTC. These fades are a regular feature on this 72 mile path. At that point,
it appeared that we had lost the daytime stablility needed for direct comparisons, and decided to stop.
Conclusions: It appears that WOLF has a little edge over WSPR-15 on stable, daytime LF paths under fairly quiet conditions. This
is not bad news for WSPR-15, as it does not require the linear amplification and tight sampling rate control that WOLF does. This new mode places
text-based weak signal at low frequencies in the hands of a lot more operators. Earlier comparisons between WOLF and QRSS show that it is roughly
equivalent to QRSS60 (60 seconds per dot), so this is indeed weak-signal territory. Further experiments will be conducted under noisier conditions
to determine whether WOLF and WSPR-15 hold up outside of the favored winter periods.
Back to Home