I attempted to repeat the out-of-loop ALS_COMM frequency noise measurement from the old days.
The main motivation is checking if the new high finesse green arms suppression of frequency noise is seen in ALS, or if other noise sources dominate.
We've already looked at IR frequency noise using ALS_COMM as the out-of-loop sensor, now we're basically doing the reverse.

Procedure
This measurement feels a little weird, as red light is locked to the arm in a pretty roundabout way:
0) Get decent alignment of green into X-arm, then input align red.
1) Lock ALS_COMM.  This starts feeding forward to the IMC length, which in turn controls the IR frequency.
2) Use ALS_COMM PLL offset to park IR on a resonant X-arm fringe.  This effectively locks the IR to the cavity via the green PDH.
3) Sweep COMM offset over the fringe to get a rough calibration of your IR PDH PD response.  I used REFL_9_I_NORM and calibrated LSC_REFLBIAS into Hz to be consistent with the past measurement.
4) Return to the IR fringe and take a PDH spectrum.

Results

Out-of-loop ALS_COMM Frequency Noise RMS = 1.7 Hz

Calibration: 46.2 Hz/cts
(REFL_A_RF9_I_NORM counts peak to peak = 0.910, IR X-arm pole = 42 Hz.  I'm not sure why the old spectrum used 80 Hz for the IR arm pole?  I followed the usual cavityPole/Vpp method which has the build-in factor of two because you're using the HWHM cavity pole over the FWHM peak-to-peak PDH signal)

In short, literally everything is way better, so we win a factor of 30 reduction in ALS frequency noise.  We see some of the old noise peaks we saw four and a half year ago, like those from 200-500 Hz.  But the entire spectrum is much cleaner, and lower overall by about a factor of 5 even at mid frequencies, much more at low frequencies (< 1 Hz).  The significant cleanup at high frequency wipes out most of the RMS from before, which was about 60 Hz.  

This measurement flies in the face of my previous in-loop ALS COMM frequency noise measurement.  I'm thinking that the in-loop frequency noise measurement is running into increased control noise at high frequency, to fully understand we'll need a noisebudget.

Notes
COMM beatnote = -9 dBm, pretty low.  The Xarm green is badly pitched onto ISCT1 with the new X-arm IR alignment, we should fix when there is somebody else around.