John's explanation has cleared up the physical picture I have of what is happening. Two or more coherent signals traveling over different static paths and interfering at the receiver can't produce new frequencies. All they can do is enhance or reduce signal strength, but there won't be any new harmonic content. The path length has to be changing/moving to induce a Doppler shift.
If I assume a steady 10mph wind, pushing an air mass that can
somehow reflect/scatter RF, I estimate a Doppler shift of about 7
Hz at 432 MHz. Stronger winds and higher carrier frequencies
induce bigger frequency shifts. I am unsure if atmospherics can do
this, but such a shift is consistent with the fuzz I remember
seeing on the FT8 waterfall. If the wind is steady, the signals
should still decode. If the air mass is not moving steady,
then the Doppler shift will fluctuate and and decodes may not
happen. FT8 is not tolerant of frequency instability beyond ± 1
This problem should have been reciprocal, however, and Jay seemed
to be decoding us just fine. This makes me suspect there was a
problem in our setup. I'm quite sure the audio level coming into
WSJT was set below the point of clipping, although it might have
been getting close. The indicator on the GUI turns red to warn the
op of this condition, but I don't recall seeing that.
Didn't save any traces, so just going by memory. I think the JT65 ghost signals (yes, they were weaker) were spaced at periodic intervals in frequency, suggesting a nonlinearity like clipping somewhere in the demodulation or decoding.
On 9/17/19 11:02 PM, John Klem wrote:
Actually, I'm saying the multipath you describe (time-invariant coherent interference) should not produce what you saw. If I understand the physics correctly...