I’m pretty sure this is the answer.
I tried a lot of changes yesterday between the rain, and couldn’t find anything definitive.
I ran the Spektrum scan with both mounts and there was not much difference. Neither had any stand-out interference that only showed up on one of the locations but not the other. I plan to re-test next weekend when the weather is better and can try more changes.
I changed to separate 978MHz and 1090MHz LNAs (previously used one that boosted both).
I moved those LNAs to the roof.
I changed the antenna feed to 6’ of LMR400 into a simple splitter to the 2 amplifiers, then 25’ of LMR240 to the SDRs.
After that change, the only difference between the mast and no mast was the dB of the received signal. On the AC unit itself gave me a slightly hotter signal, while on the mast seemed to be down maybe couple of dB (looking at the graphs and too-strong signal %).
Aluminum tape coated PVC mast didn’t change anything, for the record.
So with the LNAs moved, and better cable, the difference seems to be rather diminished.
I’ll keep playing around with things though, and plan to replace the 1090MHz RTL-SDR blog V3 with an air-spy mini. I plan to get separate 1090MHz & 978MHz antennas eventually.
On a possibly related note (but probably not), the RTL-SDR Blog V3 dongles I have (both of them) seem to be completely broken with a gain of 48.0. If I set it to 49.6 it works fine (obvious way too much gain though). If I set it to 44.5 or 43.9 or anything else it also works (showing planes and messages). However it seems that both dongles, if set to a gain of 48.0, don’t show any messages at all. Very odd.