It's a shoddy lock, it only lasts for maybe 0.5 sec or so and unlocks, and after 0.5 sec or so it locks again, and this lock/unlock cycle is perfectly in sync with the alignment fluctuation, but it locks nevertheless!

The alignment is not that great, but when it locks we can see a very weak 00-mode spot on the ETM surface at the center of the mirror. Right before it unlocks, the spot shifts in PIT (so my assessment about YAW being a problem might have been wrong).

The real breakthrough was to lock the PLL. You would think that the PLL doesn't matter that much if you lock the laser to the arm, but in reality the frequency noise of Prometheus is so big that it masks things badly. Higher order mode peaks were there but indistinguishable from each other and the demod signal looked like a hopeless random noise hell masking the peaks. As soon as the Beckhoff model was fixed (I don't know why it broke in the first place) and we locked the PLL, we were able to see multiple resonances clearly.

After Jax and Elli somewhat refined the alignment, we turned the PDH lock loop on, and with some tweaking in the gain and sign I was able to find the beam spot on the ETM.

The DC level of the reflected light on the PDH PD is still fluctuating by a large amount, and we confirmed that the fluctuation is NOT from the clipping on the diode by using a power meter in front of the diode. It's not clear where the clipping is happening, it might be the Faraday, or upstream. I'll measure the power fluctuation in the Hartman sensor path tomorrow.

This is good enough of a victory for today.