Therein lies the fundamental problem: except for a few random moments, different GPS units will tend to show different positions all the time, as each will apply compensations to the antenna signals in its own way. The differences between them may be only a few cm or mm, but there will always be relative movement between their outputs. Connect them both to an oscilloscope and view the Lissajou figure they create: no matter how long you leave it, the circle or ellipse you see will always be changing over time, sometimes barely perceptibly, other times relatively quickly. There will never be absolute stability from a single GPS unit, let alone two different ones, but the differences (you hope) will be small enough for your required degree of accuracy.
I replaced a Star 4+ with a second BG7TBL using the same model antennas for both side by side - but I have an antenna cable issue that will take me a few days to iron out before I can make the cables same length.
In the meantime I'm watching the sine waves do their dance relative to one another on an oscilloscope and thinking about what you said.
If two GPSDOs don't stay aligned in phase, but they both keep track of 10 MHz with a fair amount of accuracy, to a pretty small level of resolution, is it possible that the accuracy of the 10 MHz signal for both GPSDOs is more or less independent of the particular state of the phase at their given level of frequency accuracy? In other words, for either (and both GPSDOs) once the phase slips somewhere between 0 and 360 degrees the GPSDO is letting it slip such that it goes back to it's given frequency? Or said differently, the phase slippage for a single GPSDO is just jitter within one 1 Hz (or perhaps jitter within 1 milliHz or 1 microHz or 1 nanoHz or whatever the level of accuracy is that that GPSDO produces)?
And while this jitter bracketing is occurring with one GPSDO, the second GPSDO is doing it's own jitter bracketing within it's own level of accuracy - which could (if the two GPSDOs were identical with identical antenna feeds) be the same level of accuracy but nonetheless be out of phase with one another.
This last part doesn't seem so likely if the two GPSDO's were really identical including the antenna feed. Then they should have the same accuracy and same phase jitter, and the only question is did they startup/lock at the same time (perhaps unlikely) - but at least two identical GPSDOs with the same antenna signals should at least have their 10 MHz sine waves stay relative to one another. I think... maybe?
Trying to get my head around this. Thx