I haven’t had a chance to try the Reach in RTK mode for navigation recently - in this case the Reach was plugged into serial 4/5 on the Pixhawk and was running in single mode at 10Hz, with auto switching enabled.
I have flown with just the Reach as primary gps, in RTK mode, and seen it switch between fix, float and single solutions without any apparent issues - the new nav dynamics option seems to help, but as it only works at 5Hz for the purposes of mapping if you just need the logs it seems preferable to use the Reach in single mode at a higher rate.
I still haven’t looked into the apm logs enough to work out how to tell which gps is actually being used for navigation at any given moment (hence flying with just the Reach, to ensure the primary was not being used). I’m sure the info is there. I feel that some of the problems I was seeing earlier were perhaps more caused by the ground plane (above and in front of the M8N antenna) - perhaps someone more knowledgeable on GPS could comment, but a short mulitipath error from reflections from the plane would seem to be harder for the system to filter out.
What might be good I think, would be a system with two GPS units sharing the same antenna. Don’t know how APM handles dual GPS, but the system could surely fuse the GPS results, which should be very similar if the GPS systems are the same. That seems like it might help with GPS switching if you are relying on RTK precision. Though separate antennas mean GPS compass is an option so perhaps not.
Anyway, will continue to experiment as time allows 