Sheila, Adam,
We were curious about the DARM residual and how much of the linear range of the error signal it is using in a typical lock. I found a DQ'd channel that looks like the calibrated residual, K1:CAL-CS_PROC_DARM_RESIDUAL_DBL_DQ. The attached plots show the spectrum and rms of this residual from Tuesday night (see http://klog.icrr.u-tokyo.ac.jp/osl/?r=11927). The DARM sensitivity (GW channel) is also included.
As one can see the RMS of the residual is 4x10-11 m, which is approximately 1/4th of the peak of the error signal (which is at the cavity pole 16Hz). I realize I'm comparing metres to Hz, so here is the calculation.
vres = vlaser * Lres / Larm
vres = 3x1014Hz * 4x10-11m / 3x103m
vres = 4 Hz
This seems quite large, so we think we should boost the DARM loop to suppress this more.