I got the measurements locking the 2023B (from the PSL lab) to one of the 2023As in 40m. I have one graph showing the calibrated data compared to the calibrated data of one of the other 2023As locked to the one I used. Another has the individual phase noise of the 2023A and the 2023B (by removing the noise of the 2023A). The individual noise of the 2023B can be found by using the equation
S2023B = sqrt(S2023A2 - SCombined2)
The individual noise looks very close to that of the other Marconis (previous graphs show all 3 2023As are almost identical), so I'm not going to test the 2023B at all frequency combinations.
I also measured the electronic noise of the VCO, only accounting for the gain of 1000 in the feedback loop.