A couple of shots here.
The TCXO is quoted as 0.3ppm, not cheap. I got similar results with my usual 26MHz TCXO, not quite so accurate, but fitted that to see if there was any difference with a 27MHz clock.
The WSPR shot shows the U3S with 27MHz reference. The U3S set at 144.90500. Reference frequency 27.000000. Park 3, no park.The faint continuous line is 144.90500 GPS-derived. RX is FCDPro+/HDSDR/IQ out to WSPR2.12
Of concern are the spurious signals, I have not checked for the cause. I see similar on HF but they are 40dB down and of no real concern. These 144MHz results are simply using the U3 with a 10m LPF, just stray coupling so maybe that's the reason.
The spurii spread out from the fundamental at 50Hz spacing. This sort of spacing has often been blamed on the 50Hz powerline frequency but I tried the U3 with a battery and it looked the same. I wonder how it is produced?
73 Alan G4ZFQ