Takano

Abstract

I tuned the gain of ETMX actuators not to oscillate at the crosover frequency. Now FPMI is back.

Detail

As reported here, ETMX actuuators were unstable at some crossover frequencies and due to this we could not lock ALS DARM. To reduce the instability, I tried changing the relative gain of the actuators. Figure1 shows the transfer function from ETMX_ISCINF_L to DARM_OUT with various gain of ETMX TM actuator. From these results I chose the gain of 1.3. To match the gain of the total actuator, I compared the value around 5 Hz and estimated the factor as 0.677 (Figure2), and this factor was implemented in "gain" filter of each stage (Figure3, FIgure4, Figure5).

After that, I engaged ALS DARM and successfully locked without no oscillation. I measured the open loop transfer function of ALS DARM and the result was almost the same as the previous measurement (Figure6). I also measured OLTF with a lead filter impleneted by Ushiba-san, and thanks to this filter the phase margin was improved. However, RMS of the feedback filter got larger (~ x1.2) as shown in Figure7, so I didn't turned on the filter after all.

Next I tried to hand over the error signal to IR. I remember that last time we could lock FPMI, LSC-REFL_TRNORM_MTRX was weird; (1,2) component was 0 instead of 0.5. I heard that this setting may be for PR Xarm. I also found that in FM2 of LSC-DGCARM there was a gain factor of 2 (Figure8). Anyway I removed the gain factor, and with the nominal matrix values hand over was successful. Finally I proceeded to FPMI 1f lock, and lock was completed without any problem. I'm not sure whether this setting is suitable for PR Xarm.

After locking FPMI I measured the OLTF of DARM, and the result was almost the same as the previous measurement (Figure9). Apparently the phase margin was small, and further improvement is necessary.

I also proceeded to the next state, which requests suspension guradians to LOW_NOISE, but LOW_NOISE state of type-A suspensions were not stable and lock was lost. I skipped the guardian request and moved on to OMC_LOCK. To inject 15mW  power to OMC, I set DARM offset to 0.1 (by default: 0.04), and OMC was successfully locked. Finally I tried to hand over RF signal to OMC trans, but I was not able to do it. For DC readout further tuning may be necessary.

Next

• Improvement of DARM OLTF
• Transition to DC readout (less prioriry)
• Working on PRFPMI recovery