[Ushiba, Lucia, Anamaria]
We started with a locked simple Michelson to work on ASC MICH loop. But we've been having trouble centering the ASC WFS QPDs. Last night's long lock puts the beam on the QPDs at P,Y (0.5, 0.5) for both QPD1 and QPD2 despite us centering them in bright sinple Michelson a few hours before. We decided then to move some output beam optic to bring the beam within PZT centering range, so we could make progress on the MICH loop. Then we had planned to go align the QPDs in situ, for full ifo positions.
However, when we moved SR2 and OMMT1 we saw the power on all four AS port PDs go up (2 ASC QPDs and 2 LSC PDs). This worried us for clipping the ifo output beam so we went to single bounce ITMX configuration and surveyed the SR2 and OMMT1 ranges. The attached two plots show the moves of SR2 and OMMT1 and their effect on the four PDs. There is a caveat that the ASC QPDs are almost off the photodiodes so sometimes when they drop, it's not a real loss of power at the AS port, which is why we trust LSC ones primarily (and they are DARM sensor).
The following table shows the increase in all PD power.
| Values | starting point | after SR2 move | after OMMT1 move |
| ASC QPD1 sum | 2600 | 5800 | 5960 |
| ASC QPD2 sum | 1000 | 2300 | 2300 |
| LSC PDA1 DC | 800 | 1930 | 1960 |
| LSC PDA2 DC | 600 | 1380 | 1410 |
| SR2 position P, Y | -20, 259 | 120, 270 | 120, 270 |
| OMMT1 position P, Y | 0, 0 | 0, 0 |
-1500, -450 |
We also checked what the range is for these two actuators before we start clipping again:
SR2 P: 90->200
SR2 Y: 255->285
OMMT1 P: -500 -> -2000
OMMT1 Y: -700 -> 100
OMMT2 is not usable (no OSEM signals so probably no coil driver connection).
We calculated that with these moves, in order to properly center the ASC QPDs for full ifo we need to go (in beamsize units):
QPD1 P,Y: -1, 0.6
QPD2 P,Y: -1, +1
We will do this move tomorrow morning as well as check the whole AS table in case such a move made us close to the edge on some optic.
The DARM gain was reduced by a factor of 2.5 to relock FPMI. There may be some DARM improvement but it's too early to tell.
This measurement was done with TWR_DAMPED state.
I modified the real-time model of the payload part in order to allow us monitoring the sum of the actuation signals in the IM degrees of freedom. The motivation of the change is the measurement of the transfer function from IM-Y and IM-P into DARM.
- In the medm screen the monitror is between the sum block and the EUL2OSEM matrix.
- The monitor is actually a filter bank, used to achieve a suitable naming convention.
- The real-time model compiles successfully but I still need to install it whenever the suspension is available.
I am calculating the noise coupling due to the damping loop using the theoretical model.
Here, I will submit the preriminary result.
The attached is the DARM and expected ETMX mirror displacement (L) due to the ETMX_MN_MNOLDAMP loop and ETMX_IM_TMOLDAMP loop.
Though something seems to be worng since the estimated displacement is larger than DARM, it might gives you rough idea where the control noise would be.
* Next:
- Find what makes the noise estimation larger than the DARM.
I modified channel name and sampling rate os this list.
Finally, I can keep monitoring status of SammaryPage to be green.
[Yamada]
I measured TFs of ETMX cryopayload.
Measured DOFs are following:
TML, TMP, TMY, IML, IMP, IMY, MNL, MNP, MNY, MNR, MNT.
Data can be found in users/VISsvn/TypeApayload/TF/Measurement/20191119.
Today's results are references.
We'll track drifts cased by temperature increasing.
Temperature distribution is also attached.
We'd like to measure these TFs every 10K.
This measurement was done with TWR_DAMPED state.
With Yamamoto-kun.
I write this entry because it was reported in the morning daily commissioning meeting a cable from the input/output binary card may have been disconnected from one or more coil drivers. The symptom was that it was not possible to control them remotely.
We went t the tunnel and saw no cable was disconnected then we realized we just had to change the STATE REQUEST setting in the I/O control medm screen from -1 to 0.
I investigated the DARM displacement noise from 40Hz to 50Hz.
Fig.1 showed the spectrogram of the DARM displacement. The amplitude of the 46Hz (and 45, 47, 54Hz) have time variance. Also, it has some coherence with each line amplitude.
Fig.2 showed the IXC microphone, there are some common lines, but amplitude variation seems independent.
Anyway, we should turn off hte air compressor (and if possible air cooler)
Successfully measured (pic.1):
A_{ETMY}/A_{ITMY} = (DARM1_IN1 / ETMY_TM_LOCK_L_EXC) / (DARM1_IN1 / ITMY_TM_LOCK_L_EXC)
=6.021 (this is consistent with the past value measured by Yamamoto-san in 11252)
However, when measuring for
A_{ITMY}/A_{BS} = (MICH1_IN1 / ITMY_TM_LOCK_L_EXC) / (MICH1_IN1 / BS_TM_LOCK_L_EXC),
I found that only A_{BS}'s slope seems strange (pic.2).
I suspected existence of some filters in the path of the excitation signals, but there seems no filters.
So today I couldn't complete propagation of actuator efficiency from BS to ETMY.
I want to measure them again on another day, including calibration of absolute value of A_{BS}.
During this night, the vacuum gauge at EXA was increasing exponentialy.
(currently, it's > 1e-4 Pa)
The temperature change in EXC booth is less than 0.1 degC.
PR2 tilt oplev P and Y started crazy when (or, after) the X arm engaged resonance...
By the way, have this trend-oscillation also been obeserved when you did not do the dithering of PR2? (for example, the dithering freq is too high for PR2???)
In case this happens every time the cavity is aligned, this might be due to some ghost beams by the light source of the oplev itself, for example + unfortunately (as you may know, no cares for those ghosts there, as far as I know), although such big changes were not observred in the trend of tilt-SUM and len-SUM; even a fraction of light might be enough to rise "pitch and yaw signals" even though it won't change the SUM so much.
So I inserted the whitening filter zpk([10;10;10], [100;100;100], 1.0, "n") on the INVC and TM filter banks.
The inverse whitening filter is also inserted on the diaggui template.
-
I implemented alignment control with using dither and the TMS QPDs. The control scheme is as follows:
PR3: Dithered at 19.2 Hz(P) and 11.1 Hz(Y), and demodulate the GRX TRANS.
PR2: Dithered at 13.3 Hz(P) and 16.8 Hz(Y), and demodulate the IRX TRANS.
IMMT2: Dithered at 9.1 Hz(P) and 7.2 Hz(Y), and demodulate the IRX TRANS.
ITMX, ETMX: The DC QPD signal at TMSX is fed back.
Yarm, BS: No control -
As shown in the attached figure, PR2 and IMMT2 oplev moving a lot. This indicates three hypotheses:
1. PR2 oplev itself is moving
2. IMC output beam is moving and the IMMT2 and the PR2 follow it.
3. IMMT1 is moving and the IMMT2 and the PR2 follow it.
The dv plot shows IMMT1 oplev signal, IMMT TRANS QPD signal, and PR2, IMMT2 oplev signals. IMMT TRANS QPDs show that 3 is not likely. We are not sure about IMMT1, but it does not seem to move so much although the noise is huge. I feel 1 is most likely and it can explain our unstable FPMI. - The other oplevs are surprisingly stable in DC. We can trust them at least within a couple of hours.
- Yarm is not controlled in angle. I implemented dither controls into BS and ETMY, and it does work with the single arm. However, this loop seems unstable with the FPMI. We can try more about it if we need it, but I don't think it's worthful to spend time on that.
- Still, the guardian is not good enough and we need to tweak the resonance scan by hand. I think we can solve it by using additional digital CARM with aux. signal such as TRANS, or normalized REFL.
- (Yarm ASC implementation)
- ASC with FPMI
- MICH control noise reduction.
PR2 tilt oplev P and Y started crazy when (or, after) the X arm engaged resonance...
By the way, have this trend-oscillation also been obeserved when you did not do the dithering of PR2? (for example, the dithering freq is too high for PR2???)
In case this happens every time the cavity is aligned, this might be due to some ghost beams by the light source of the oplev itself, for example + unfortunately (as you may know, no cares for those ghosts there, as far as I know), although such big changes were not observred in the trend of tilt-SUM and len-SUM; even a fraction of light might be enough to rise "pitch and yaw signals" even though it won't change the SUM so much.
We tried to turned ON the dewhitening filter for the TM stage of ETMY.
But there is no efficient improvement (See red and green curve on attachment 1).
After then we planned to turn ON the dewhitening filter also for the TM stage of BS because MICH noise becomes closing to the current DARM sensitivity (See red and magenta curve on attachment 2).
But anti-dewhitening filter was not turned ON on the COILOUTF filter bank.
In the case that the BIO cable connected with the coil driver, filter modules 1-3, 6-8 should be turned ON such as SRM. But on the BS model, they are not turned ON (See attachment 3).
According to lscpi command, 3 BIO cards are already installed on k1bs.
So we need to check the cable connection in the mine.
I also checked for BS, SR2, SR3.
Just a memo for PEM team
There is some coherence between AS ACC and DARM displacement at 124-126Hz.
I compared the DARM displacement and coil drive volt monitor for the ETMY MN.
60Hz may come from circuit noise, but there is no other peak in the coil driver volt monitor
[Yamada]
As reported in #11522, damping filters contribute DARM. So we made roll-off frequency low in cryopayload damping filters (ETMY, ITMX, ITMY, except for BP filtes, ), #11571.
I compared 3 periods: Nov. 13th 22:04:00 (no roll off), Nov.13th 23:15:00 (ETMX_MN_MNOLDAMP_P roll off) and Nov.17th 22:32:00(all filtes for ETMY, ITMX, ITMY roll off).
Comparing Nov.13th 22:04:00 and 23:15:00, we can cleary see reduction of contribution of damping filter in DARM signal.
Comparing Nov. 13th 23:15:00 and Nov. 17th 22:32:00, we can see reduction of coherences, however, the effect was unvisible in DARM signal.
I also checked for BS, SR2, SR3.
I found that 24Hz noise in DARM disappeared, I checked the IY0 channel and found 6th Nov. 16:40 (7:40 UTC) suddenly disappeared.
I couldn't find what task hunted this noise... (I will ask technitian tomorrow)
