I posted LHO:81917 regarding the calibrated ASC coupling functions. At the time, my results seemed wildly large and I was certain I had made some calibration error somewhere, and indeed I had. Lee reached out last week since he is working on something similar for his optimal controls work, see LLO:77901. These are the calibration errors I made:

• my old calibration for PUM: 20 V /2**20 ct [DAC] * 0.268 mA / V [drive strength] * 0.0309 N / A [force coeff] * 70.7 mm [lever arm]
  • the bolded term is technically correct because we do have a 20 bit DAC, however, there is a sneaky factor of 4 gain filter in the coil output filters bank, which was added to avoid everything changing by a factor of 4 when the DACs were upgraded from 18 bit to 20 bit
• There is an additional euler to osem gain value of 3.5355, and factor of 4 for four coils

The full counts of drive to Nm of torque conversion factor is therefore: (20 / 2**20) * 0.268e-3 * 0.0309 * 70.7e-3 * 4 * 3.5355 * 4 = 6.317e-10 Nm/ct

Lee also pointed out instead of using the modeled free suspension plant, I should be using the radiation pressure modified plant. This is correct, however for the purposes of calibrating the coupling function the effect is mostly the same, since we know that the rad/Nm transfer function is the same at 10 Hz within a few percent for zero power and high power.

However, for completeness, and because it matters for other calibrations, I did this instead:

• Gabriele fitted HARD loop OLGs to determine the hard plants at high power in 77857 and 71848. These fits are in counts, so they need to be scaled to real units
• Using the same modeled free plants, in T2300299, I scaled the high power plant to the modeled plant at 20 Hz
• I used this scaled high power plant to calibrate the m (DARM)/Nm (ASC) coupling function to m (DARM)/ rad (ASC)

The end result is much more sensible, resulting in a coupling function around 30 Hz that is about 1 mm/rad for both pitch and yaw. This is still "high" in the sense that Matt and Lisa assumed a coupling on the order of 0.1 mm/rad in T0900511.

I went a step further to check the linearity of the coupling. I measured the transfer function of ASC to DARM during the noise budget injection times. However, the noise budget is usually calculated with an excess power projection, so we have both quiet and injection times taken. Using the same calibration method, I compare the excess power coupling function with the linear transfer function coupling function. They appear to be nearly the same, showing that the ASC coupling is dominated by linear behavior.