sheila.dwyer@LIGO.ORG - posted 02:06, Wednesday 22 July 2015 - last comment - 06:40, Wednesday 22 July 2015(19821)
work tonight
Jenne, Sheila, Evan
- repeatedly locking with wind gusts around 30mph tonight.
- closed beam diverters (now included in gaurdian in LSC_FF step). This did not visibly improve the spectrum. As far as those of us in the control room right now are aware, there isn't krytox on the HAM1 beam divereters yet. If there really is not krytox, we risk getting these stuck if we keep flipping them.
- reran A2L script since gains were lost (accpeted new gains in SDF)
- switched bias sign on ETMY and ETMX. For ETMY we added logic to the gaurdian that checks the sign of the bias and adjusts the sign of the gain in drivealign to match. For ETMX, we did not flip the sign of the gain and were able to lock anyway, we think this could be because the sign flip is taken care of by the linearization.
- ISS 2nd loop is now engaged by the ISC_LOCK guardian. To do this we have let the ISC_LOCK guardian manage IMC_LOCK. If anyone wants to use the IMC_LOCK gaurdian indepentently now, they may have to set it to auto.
- We rang up both the bounce and roll modes very badly on ETMY, enough to cause locklosses in the final stages of the CARM offset reduction. This was with an amplitude above 10^-11 in the RF DARM step. We edited the gaurdain to check the monitors before moving on, and to lower the gain an wait if they are high in some states (BOUNCE_VIOLIN DAMPING and DARM WFS). We can do a better job of gain adjustment using a decorator, but may want to wait until the TrueRMS parts are fixed.
- Tonight we have had several incidents where the state PREP_TR_CARM ends with no IR in the arms. There is logic in the TR_CARM state that resets the filter states and clears history so that you can just go to manual, and reselect PRE_TR_CARM. This has been working fine to recover, where before we would have lost lock. We need to figure out why this is happening so much.
A2L notes:
The new A2L values hadn't been saved to the safe snapshot, so when there were boots and burts earlier today, the values were lost. This is why we re-ran A2L for the test masses at high power tonight. Those values were accepted in the SDF system. This should be done every time the script is run. Since I don't know if it's possible to do this in a scripted way, we'll have to do this accepting by hand for now.
In addition, the script a2l.py was modified to be able to handle the BS in addition to the test masses. The only change for this was an "if" statement at the beginning to define the drivealign channel names (they're in L2 for the quads, but M2 for the BS). We found that excitation amplitudes of 300 cts for pitch and 150 cts for yaw give us an SNR of about 100 in OMC_DC, so those have been added to the GetOptList.py dictionary-creator. Sheila tells me that the BS M2 is roughly 3 times stronger than the quads' L2 stages, and these excitation amplitudes are about 1/3 the quads' values, so that all makes sense.
The A2L is finishing up right now for BS, so in the next few minutes we'll have all 4 test masses and the BS all decoupled via Hang's new script. EDIT: Pitch worked, but Yaw isn't finding a happy place, so I'll have Hang look at this with me tomorrow.
I also added some lines in the "disable dither" function of a2l.py (which runs if you ctrl-c out, or when the measurements are finished) to reset the CLKGAIN excitations to zero for both the ASC oscillator and the LSC demodulators. The script already zeroed the output matrices, so the signals weren't going anywhere, but it made me unhappy to have mystery oscillators running unnecessarily.
I tried to get an OLTF of dHard pitch at 24 W, but somehow I rang up a 1.0 Hz instability and had to abort and bring the power down to 3 W. Strangely, even at 3 W this instability took a minute or so to ring down. I tried lowering the master ASC gain to 0.8 temporarily, but this did not make the ringing go away (so it is probably not a loop instability).
At 3 W I was able to get a preliminary measurement of the dHard pitch loop down to 1 Hz. More refinement required.