J. Kissel We've discovered that the H1 SUS MC2 compensation filters are incorrect for the M2 stage coil driver, having never been changed after the analog coil driver was swapped out for a beefier drive (see LHO aLOG 9956). It turns out, by lucky coincidence, we've been operating the new coil driver in a state (State 1) where it's frequency response is flat, and the compensation filters -- though wrong -- cancel each other to also be flat, so the overall crossover loop gains remained unaffected. Just to be sure, however, I too a "pre" measurement of each crossover of the IMC_L loop gains. I attach the results, and summarize below. Crossover UGF [Hz] Phase Margin [deg] MC_L/MC_F 14.5 39 M2/M3 7 38 M1/M2 0.125 36 These numbers are comparable to LLO's numbers (see, e.g. LLO aLOG 4331). Templates can be found here: /ligo/svncommon/SusSVN/sus/trunk/HSTS/H1/MC2/Common/Data/ 2014-02-11_H1IMC_M1-M2_Crossover_OLGTF.xml 2014-02-11_H1IMC_M2-M3_Crossover_OLGTF.xml 2014-02-11_H1IMC_MCL-MCF_Crossover_OLGTF.xml
For posterity, the IMC loop had the following in its path: Input Mode Cleaner, Common Mode Servo Board Settings Common Path - H1:IMC-REFL_SERVO_IN1GAIN = 9 [dB] - p:z 40Hz:4kHz compensation filter ON - p:z 1kHz:20kHz first stage boost ON (second and third stage boosts OFF) - p:z 4kHz:17kHz filter ON Fast Path - H1:IMC-REFL_SERVO_FASTPOL = + - H1:IMC-REFL_SERVO_FASTGAIN = -6 [dB] - p:z 140kHz:70kHz highpass ON Slow Path - p:z [100 100]:[10 10] "generic filter" LSC Model Settings IMC_L Filter - FM1; antiWhite = p:z [10 10]:[100 100] (compensates for "generic filter") Gain = 1.0 LSC-MC Filter - All filters OFF (when locked) Gain = 1.0 SUS MC2 Model Settings All coil driver states set to 1. M3_ISCINF - FM 6; notch_R3 = notch at highest roll mode of MC2 - FM 7; notch_V3 = notch at highest vertical mode of MC2 Gain = 1.0 M3_LOCK - FM 3; z40:p140 - FM 9; CLP300 = Chebychev low pass at 300 [Hz] Gain = -300 M2_LOCK - FM3; z1:p0.1 - FM4; z1:p100 - FM10; ELP70 = elliptic low pass at 70 [Hz] Gain = 0.1 M1_LOCK - FM1; z0.01:p0 - FM2; z:p0.01,0.1 Gain = 1.0
I repeated the MCL-MCF crossover measurement that Jeff made since I could not get it to agree with my model. I made this measurement with the SR785 by injecting a source into excitation B through the slow path of the MC common mode board and looking at the ratio of test 1 and test 2. The results still differ from my model (likely something missing in the model..), but more suprisingly I found that this measurement showed a factor of 2 greater in magntiude in comparison to Jeff's open loop transfer function. See attached picture.