Correct me if I’m wrong or if I’m missing something, but this is what I’ve read so far:
Elevation mask too low: satellites near the horizon (greater ionospheric delay and multipath errors ) will be included in the computations
Elevation mask too high: satellites with ‘good enough’ signals can be excluded, which can degrade overall satellite geometry
I get that the ‘right’ solution is a satellite mask value where the observations/sessions are fixed, but is there a trend as to which option yields better results?
i.e, in cases where the observations do not have fixed solutions, which should I try first, a higher or a lower elevation mask?
There is no simple answer to this, as it depends on your environment.
Another approach (though only available through post processing in RTKpost) is to define different SNR for different elevation angles. That works pretty good, and takes out fewer otherwise good sats.
Your understanding of the elevation mask is correct.
Usually, the default setting in 15 degrees is the best choice in reasonable conditions. If you can’t get Fix solution, the first thing I’d recommend is paying attention to the environment and hardware setup.
If these are ok, then you can try changing masks angles. My experience shows that higher angles work better in urban areas and lower for highlands.