Evan G., Jeff K., Kiwamu I., Darkhan T.
Summary:
We have processed the DARM open loop gain and sensing function measurements and built a model of the loop that we believe sufficiently represents our knowledge of the parameters. In addition, the actuator coefficients have been remeasured using the Pcal. In summary, our model reproduces the DARM open loop transfer function to within +/-10% in magnitude and within +/-5 degrees in phase from 10 Hz to 1 kHz. Caveat: we have not yet assessed how this translates to the overall calibration uncertainty. Actuation coefficients for L1, L2, and L3 are within +/-5% of their O1 values. This model is used in the recently updated CAL-CS model (alog 28178).
Details:
The DARM measurements made recently (alog 28107) were processed using the updated DARM model code. Included in this new model is a functional form for the SRC detuning (alog 28150), and the fit to the measured data gives updated cavity gain, cavity pole frequency, time delay, and detuning spring frequency. The uncompensated OMC whitening filters are updated (alog 28087). Since the L1, L2, and L3 driver compensation has been updated (alogs 27150 and 27180), we removed all the measured and compensation zeros and poles, except for the uncompensated high frequency zeros and poles of the L3 stage (alog 27619). The AA/AI downsampling filters are using the RCG 3.0 64-16k filter coefficients (alog 27173).
Attached are figures showing (in order) the measurement and model comparison DARM open loop gain transfer function, the measurement and model comparison of the sensing function, and the L1/L2/L3 actuation coefficient measurements calibrated using the Pcal.
It was realized that since we no longer read out H1:CAL-DARM_ERR_WHITEN_OUT_DQ for the for the Pcal to DARM transfer functions--instead we are reading out H1:LSC-DARM1_IN1_DQ--the measurement to model comparison for the sensing function should not include the 1 16k clock cycle delay. However, the model for the sensing function should include this 1 16k clock cycle delay.
We tried a few different values for unknown sensing and actuation delays but did not attempt to optimize this any further.