evan.hall@LIGO.ORG - posted 04:45, Thursday 23 April 2015 - last comment - 12:56, Thursday 23 April 2015(18020)
More recycling gain work
Sheila, Gabriele, Evan
Tonight we spent some more time refining the locking sequence with good recycling gain. Some lessons learned:
- It seems that saving the green spot positions on the ITM cameras is not enough to consistently bring us to a good recycling gain. Instead, it seems that the spot position on the POP camera and POP QPDs is a better indicator; a high spot seems to give a mediocre recycling gain. To reduce the amount of ITM steering required in full lock, as part of the initial alignment (with IR locked to the X arm and with WFS closed around IM4 and PR2) we moved ITMX in pitch by about 2 µrad to bring the beam lower on POP A. This brought us to an okay recycling gain (about 34 W/W) in full lock. It remains to be seen how repeatable this is.
- In addition to the bounce and roll mode work this afternoon, we spent some time damping violin modes on ITMY, since they were badly rung up. In particular, 501.750 Hz is damped with a gain of +200 and 0°, and 501.811 Hz is damped with a gain of -200 and 0°.
- We found that the steering we did with the ITMs tended to cause an ~0.5 Hz angular instability (seen in the ITM oplev pitch signals) when we powered up from 3 W to 10 W. Since ITM pitch is not servoed, we suspect this is radiation pressure coupling caused by beam miscentering.
- Increasing the power to 10 W tends to spoil whatever ITM alignment work is done at 3 W (we see the transmitted arm powers drop). We again suspect radiation torque.
- We were able to engage the new LSC boosts (DARM, MICH, SRCL) without issue.
- On dc readout, we injected broadband noise into SRCL. 10:55:15 to 10:59:15 UTC is done with feedforward off, and 11:04:00 to 11:09:00 UTC is done with feedforward on. In both cases it seems the coupling is still nonstationary.
This evening we had a look at which signals could be used for ITM ASC. We dithered each TMS to find acombination of signal which is insensitive to TMS motion, the attached screen shot shows the combination we ended up with. We has to move the picomotor on TMSY (M14 only). While doing this we noticed that there is a bug in the software so that according to the epics record of the position it looked like we were moving M3 rather than M14. We drove down to the end station and saw that indeed, the right motor was selected in the driver. We also see that the beam on the QPD reacts as we expected to moving the motor, so this must be correct.
For anyone who tries to lock in the morning
We are currently using the guardian as is through the state resonance. Sometimes there is an oscillation that gets rung up in this state and dies down, we have been waiting for this to ring down before moving onto analog CARM in our sucessfull locking attempts tonight. We have also been skipping the states BOUNCE VIOLIN mode damping and INCREASE POWER. We have also been using no whitening on the OMC DCPDs since the violin modes are still rung up and we haven't finshed re commisioning all the damping.
If we had more enegry...
we might try more violin mode damping, or investiagting why we aren't able to power up. Our last idea on this was to adjust the alignment to get a good recycling gain before attempting to power up, taking small steps and re-adjusting the alingment as we go. Another option would be to commision the ITM loops to do this job for us, using the error signals in this screenshot.
Also, we are leaving the intent bit undisturbed, even though this isn't a great lock. We are missing some ASC cut offs which make the noise worse below 30 Hz, we are at 3 Watts, and we have our bothersome 90 Hz shelf still.
We measured the response of the TRANS QPD error signals to ITM motions. Here is the result, in QPD units over alignment slider units:
0.05
0
-0.10
0
Right after the power increased from 3 to 10W (still no DARM offset and the IFO was locked with RF), all TMs moved by a large angle in PIT, mostly in SOFT:
ETMX +0.2 to 0.3 urad, ETMY +0.9, ITMX +0.5, ITMY +0.6.
It's possible that this is indeed DC radiation pressure pushing optics and the ASC fails to restore things back because we don't have the full DOF controlled, twisting other controlled optics to make the servo happy. If that's the case you need to be able to get back the recycling gain just by pushing ITMs manually even if you get an instability by doing so.
Also, if the angular optical spring is messing with the stability of the ASC, it looks better to me to switch to true HARD-SOFT based business instead of optic-based.Update: Evan and Gabriele think that the instability is coupled to the centering.