Evan, Stefan,
We re-measured the frequency noise coupling to DARM from the four sensors we have: REFL9 (in-loop), and POP9, REFL45, POP45 (all out-of-loop).
Plot 3 shows that transfer function in m/ct. (For REFL 9, the calibration from Evan 1.5e-7W/ct, see e.g. 30286. We didn't calibrate the others in W yet.)
Plot 1 shows the projection of all 4 sensors to DARM. The fact that POP9 disagrees with the others - but is coherent with them (Plot2) suggests that what limits our sensing is not frequency noise.
Finally, Plot 4 shows the relative calibration of the four sensors for frequency noise.
The xml files are in
/ligo/home/controls/sballmer/20161103/FreqNoiseProjection.xml
/ligo/home/controls/sballmer/20161103/REFLPOPV2.xml
Also, for reference, the frequency noise injection was done between 21:36:50 and 22:02:50 UTC on Nov 03 2016.
Here is a frequency coupling TF calibrated into meters per hertz, using the above data.
The calibration uses the whitening gain (12 dB), the digital gain (0.36 ct/ct), the ADC gain (216 ct / 40 V), the demod gain and transimpedance (2600 V/W), and an assumed CARM plant with a dc gain of 13 mW/Hz and a pole at 0.63 Hz. The numbers for the plant come from OLTF budgeting from O1 (so these numbers are quite old and may have changed).