Duffey and I discussed this on the long ride home today.
Thinking about it some more, the idea of multipath is that two or
more (coherent) signals from the same source travel by slightly
different paths to the receiver. This path difference only has to
be a fraction of a wavelength. This causes interference and mixing
in the detection circuit, producing ghosts of the digital signal
in the audio waterfall. This is exactly what I saw when we were
running FT8 and especially JT65 (the synch signal is very obvious
in the traces). Seems this would be become a greater issue as the
frequency goes up into UHF. All that said, FT8 should have been
able to successfully decode the baseline signal and any
harmonics. Or maybe the frequency separation of the ghost signals
was not sufficient and confused the decoder.
I am convinced the equipment was not at fault, as we had been
working very successfully with various digital modes throughout
The question of reciprocity is a good one. W9RM's signal was
very up and down, but his digital signals refused to decode
independent of the widely varying receive level. My hunch is that
the reflection(s) occurred closer -- much closer -- to our QTH
than his. Beam divergence and diffraction would substantially
weaken the reflected/scattered signal by the point they reached
him, allowing decodes there. Also the lower gain, wider aperture
of the rover antenna may make us more susceptible. But we were at
an ideal location with almost nothing on the near horizon that
suggested a reflection source. We experimented over the course of
about 30 minutes with different digital modes and especially beam
headings to no avail. I would have wagered that moving the beam
around would have fixed it. And lost.
I have never had this problem (that I can remember anyway) on 6
or 2. My experience with FT8 on UHF has been entirely this past
weekend, or about 3-4 QSOs total.
On 9/16/19 9:40 PM, John Klem wrote:
I am very curious - what characteristics of these signals led you
to conclude multipath was the problem? What was painfully obvious
Should multipath be non-reciprocal?
Getting an audio recording of this kind of signal for analysis
would be really interesting.