Up and observing for a bit while it looks like livingston is about to be up, then I'll drop the intent bit to do some tasks from commissioners.

TITLE: 11/05 Eve Shift: 23:00-07:00 UTC (16:00-00:00 PST), all times posted in UTC
STATE of H1: Aligning
INCOMING OPERATOR: TJ
SHIFT SUMMARY:  IFO has been up and down all night for various reasons.  After 2 of the 4 locklosses, ITMy swung around for a long time, slowing relocking.  Stefan looked for the cause but was unable to find one.  Also, the ASAIR and POPAIR signals have been oscillating during full lock.  Matt and Stefan are looking into the cause of this, but as of the end of my shift have not come to a conclusion.  I ran A2L for 2.5 of the locks, the 0.5 being that I was halfway through the 2nd script when we lost lock.  Stefan said that it was useless to run A2L with the AS and POP oscillations, so I haven't run it for the last lock.  Microseism is still above the 90%ile level.  Wind is calm.  Locked at ~70MPc.
LOG:

23:59 Updated LSC and ASC SDF OBSERVE.snap files per Hugh's request.  If the Owl shift operator has a lockloss, he'd like to update the LSC and ASC DOWN.snaps as well.
 

Evan, Stefan, Matt (writing)

We've been seeing for a few days the bouncing of the harmonics of the 1kHz violins (exact harmonics, not higher violin modes).  Watching these on a fast spectrum shows an ~18s period, in which the harmonics appear out of the noise and grow by more than a factor of 10 as the beating at 1kHz reaches a maximum.  The 1kHz peak (in which the violin modes are unresolved) is only changing by a factor of about 5 (see attached).

We noticed that the sidebands buildups started to oscillate at the microseism once we switched to low-noise (lower ASC bandwidth ?). We suspect something with ITMY or its suspension is mistuned, because it would hardly damp after4 lock-loss. Not sure whether it is related.

Stefan, Kiwamu, Matt

To test the theory that the ILS is gain limited, and that more gain might help with DARM noise, we measured the loop and used the TF input to add some gain at low frequency.

The open-loop TF was as previously reported (see 31121 and photos to follow).  We had computed that an SR560 with a gain of 10 and a pole at 100Hz accompanied by a 10kHz passive low-pass would make for a good test without destabilizing the loop (see attachment 1).  With this test-rig attached the TFB and TFIN ports on the ILS front panel (photos to follow), we relocked the IFO and found that the coherence with DARM had been reduced as expected (see attachment 2).  Noteabley the PMC HV signal was significantly reduced (black to brown in attachment 2), indicating that this gain was preventing noise in the PMC.  Unfortunately, this simple test was not able to apply much gain to the 1kHz, where the highest coherence with DARM is seen.

Kiwamu has offered to investigate the possibility of using a new CDS FE running at 65kHz to implement a tailored gain profile which will suppress the 1kHz peak and other HPO features while keeping the ILS loop stable.

Sheila, Matt

We wanted to see if there was anything interesting in the voltage between the ESD cable grounds and the chamber grounds (similar to tests done at LLO), so we installed a couple of SR560s which we clipped to the ESD grounds and the chambers at both end stations.  The SR560s have a gain of 1000 and a LP at 10kHz.

A little coherence with DARM is seen around 280Hz and 140Hz (see image).  Robert says that these are likely to be due to fans.

These SR560s can be recovered whenever someone is next at the end stations (we don't  want to disturb the BRSs at the moment).

[Sheila, Jenne]

We were interested to see if the scatter peaks could be significantly reduced by moving the IMC WFS offsets, and if they're repeatable lock-to-lock.  Sheila moved DOF1_P to +350 counts and DOF2_P to -100 counts.  The DOF1 move was significant, but the DOF2 and DOF3 moves that we tried weren't.  We advocate leaving in the DOF1 offset, but not bother with the other two.  In the attached screenshot, yellow is Sheila's reference from before changing offsets, and purple is with the offsets.  The spectrum looked pretty much the same with and without the DOF2 offset, but this spectrum had it in.  We note that the IMC trans power went down a bit with the DOF1 offset and the IMC REFL camera looked pretty non-symmetric, so maybe this isn't so ideal, but the recycling gain went up and some peaks went down.

In the attached spectra, the rectangle noise between 410-440 Hz is an injection on the input pointing PZT.  Sheila's offsets improved this by not quite a factor of 2.  The improvement at the 260Hz peak is a little over exaggerated - it was bouncing up and down and we happen to have caught it at a relatively low point.

I started trying the yaw offsets, but broke the lock.  For the next person who tries this, I had gone to -350 in DOF1 yaw.  -300 was stable, but we lost lock shortly after I went to -350.  I think negative was the wrong direction, although I wasn't seeing all that much effect.  It seems like it's worth trying this some more, although if the peaks are gone after Matt's efforts tonight on the PSL ILS loop then maybe it's not so critical. 

As was brought up at this afternoon's commissioning meeting, I have computed the test mass spot positions over the last year. T(0) on the plot is 10-Oct-2015, the date of the first A2L measurement that I have data for.  The data through the end of O1 has been seen before, but I haven't looked at it since then.  The spots have been moving a whole lot, so probably I should be better about updating these plots more often.

The 3 attached plots all show the same data, but with differing amounts of extra info.  The dots are the spot positions, and the magenta horizontal line is the average position of all the measurements taken through O1. 

The second plot adds to this a series of vertical lines of notable events, either for the IFO or for the alignment.  The third plot is the same, but zoomed in to the last 75 days. 

The lines are from the following "events":

1. 2015-09-15 00:00:00 (start of O1)
2. 2016-01-19 00:00:00 (end of O1)
3. 2016-04-02 00:00:00 (no HPO, but after HAM1 grouting work and subsequent realignment)
4. 2016-05-17 00:00:00 (first lock with the HPO)
5. 2016-07-11 00:00:00 (ER9 end)
6. 2016-07-16 00:00:00 (Good July spectrum from Sheila's G1602240)
7. 2016-09-02 00:00:00 (Big POP_A offset move, resetting of initial alignment setpoints, to improve 50W recycling gain)
8. 2016-09-26 00:00:00 (Moved IMs closer to their O1 OSEM values, trying to keep rest of IFO alignment roughly constant)
9. 2016-10-08 00:00:00 (Bad October spectrum from Sheila's G1602240)
10. 2016-10-13 00:00:00 (Moved POP_A and SOFT offsets again to improve 50W recycling gain more)
11. 2016-10-23 05:25:00 (TCS and alignment settings back to July values)
12. 2016-10-27 05:00:00 (Alignment recovery after PSL shift - IM3 now close to actuation edge)

You can see that after #7 many of our spot positions moved pretty significantly, although they continued to move after event 7 and I just didn't find / remember an event to explain this specifically.  But, we'd been doing lots of alignment searching during that time.  Also, after #11 when we went back to our July TCS and alignment values our spot positions largely went back to where they were before the big POP_A move.  We could try more to get closer to the O1 spots, and that may be somewhat helpful.  Food for thought.

The Hartman codes for both ITMs had been off for some days in order to upgrade the computer to Ubuntu 14. The new codes, which are written in python, were started at around 21:40 UTC today. No major issues so far. We did not check the Hartman beam alignment with respect to the cameras yet.

As Sheila pointed out, back in May we had a 25W lock that lasted >5h and never showed any indication of jitter peaks.

The attached plot 1 shows spectra from May 17 (25W), July 16 (40W) and Nov 3 (25W). Note that the May 17 spectrum is taken 4h into the lock.

Plot 2 shows inputy power and range of the May 17 lock.

Plot 3 shows all TCS settings of the May lock.

Plot 4 shows all TCS settings currently used.

Based on this, I would like to turn off the ITM ring heaters.

The HPO length locking servo is gain limited, and the error signal has coherence with down-stream sensors, including DARM.  Decreasing the ILS gain increases the coherence with DARM at several peaks above 200Hz (see attached).  We can't increase the gain much with the current servo shape (see 31121, servo board here), but a small gain increase reduces the coherence with DARM.

In the image: BLUE is low gain (servo gain 3.0 V), GREEN is nominal gain (servo gain 1.0 V), RED is max gain (servo gain 0.0 V)

- After the demod board was fixed, we could not transition to DRMI 3f. Stefan found that there was no 3f signal to transition to. He and Richard went out and disconnected/reconnected cables and it came back.
- Evan changed something in the ALS guardian code that had to be fixed.
- There was a 6.4 earthquake in Chile.
- I ran through an initial alignment near the beginning of the shift.
- Sheila told me that although there is less sensitivity to alignment in locking PRMI, there is currently something not quite right about the PRMI acquisition and we have a better chance of acquiring DRMI than PRMI now.
- I tried to go from PRMI to DRMI, but guardian would not let me. Sheila said she has fixed this now.
- We made it to NLN at 22:56 UTC.

15:07 UTC Robert to PSL enclosure to 'move hot things around'.
15:10 UTC Richard to PSL ISC racks
15:38 UTC Richard back. Demod chassis swapped with spare.
15:40 UTC Robert back
15:55 UTC Kyle to LVEA to take measurement with tape measure
16:08 UTC Kyle back
16:10 UTC Gerardo to LVEA to check ion pumps
16:14 UTC Peter to diode room to take inventory
16:17 UTC Robert and Richard to LVEA
16:23 UTC Richard and Robert back?
16:29 UTC Peter done
16:37 UTC Robert back
16:42 UTC Richard to LVEA to get Torque wrench

Earthquake: 6.4
25km ESE of Curico, Chile
2016-11-04 16:20:43 (UTC)

16:53 UTC Richard to LVEA to swap back in original demod chassis
17:21 UTC Richard back
17:29 UTC Keita done zeroing dark offsets on LSC photodiodes
17:57 UTC Starting initial alignment
18:38 UTC Done initial alignment. SRC had trouble locking.
18:51 UTC Lost lock twice going to LOCK_DRMI_3F. Stopping at DRMI_ENGAGE_ASC.
21:14 UTC Sheila to LVEA
22:01 UTC Robert to end Y to pick up speaker from VEA
22:33 UTC Robert back

Took 6 minutes to overfill CP3, first set the manual fill to 36%, noticed the temperature drop slow, so I opened the valve a bit more, to 50%.  See attached for more.

LLCV remains at 20%.

The demod chassis was fixed. One of the chips was not fully seated. Keita updated the dark offsets. I tried locking but had trouble on PRMI. I ran through an initial alignment. After that I lost lock twice in a row on the transition to LOCK_DRMI_3F. I stopped at DRMI_ENGAGE_ASC and Keita checked the phases related to the fixed chassis. He said they looked fine, but noticed that H1:LSC-REFLAIR_B_RF27_I_OFFSET and H1:LSC-REFLAIR_B_RF135_Q_OFFSET had been changed. He determined that they must be changed in guardian and asked me to find out where. I believe I have found the relevant code in ISC_DRMI.py under ZERO_3F_OFFSETS, but I'm not sure how to address it. The commissioners are in a meeting.

class ZERO_3F_OFFSETS(GuardState):
    index = 110
    request = True
    @assert_mc_locked
    @assert_drmi_locked  
    #@nodes.checker() 
    def main(self):
        ezca.switch('LSC-SRCL1', 'OFFSET', 'OFF')
        # Zero the 3f offsets
        # [FIXME] either shorten the averaging or run the offsets in parallel,
        # or both average the current offsets and put them in.
        offsets = cdu.avg(5, ['LSC-REFLAIR_B_RF27_I_INMON',
                                   'LSC-REFLAIR_B_RF27_Q_INMON', 
                                   'LSC-REFLAIR_B_RF135_I_INMON',
                                   'LSC-REFLAIR_B_RF135_Q_INMON'],
                               )
        # write offsets
        ezca['LSC-REFLAIR_B_RF27_I_OFFSET'] = -round(offsets[0], 3)
        ezca['LSC-REFLAIR_B_RF27_Q_OFFSET'] = -round(offsets[1], 3)
        ezca['LSC-REFLAIR_B_RF135_I_OFFSET'] = -round(offsets[2], 3)
        ezca['LSC-REFLAIR_B_RF135_Q_OFFSET'] = -round(offsets[3], 3)

    @assert_mc_locked
    @assert_drmi_locked  
    #@nodes.checker()  
    def run(self):
        return True

This is a comparison between the ISS, ILS and PMC signals before (REF traces) and after the changes in the electroncis and the modulation depth, see 31095.

A few observations:

• The ILS HVMon signal is unchanged except above 5 kHz, where the new whitening makes a difference.
• The PMC HVMon signal now shows good coherence with ILS HVMon above 200 Hz.
  Even so there is a gain of 3 of additional loop gain at ~1.5 kHz and higher the peaks are higher than before.
  At frequencies above 1kHz the signal looks significantly better, maybe more than one would expect from the modulation depth change.
• The in-loop PDA signal from the first loop ISS looks the same.
• The out-of-loop PDB signal on the other hand now shows the same peaks as the ILS/PMC length signals—even so the overall noise level is lower.
• The feedback signal to the ISS AOM stayed very similar.
• There was coherence between the HPO power and PDB before, which is now gone.
  
   

Matt, Evan

As has been noted before, the DARM residual these days is usually microseism-dominated, and it is getting worse as we move into winter.

We installed a new boost (FM2 in DARM1) to give >40 dB more suppression at the microseism. The performance during yesterday's 25 W lock is shown in an attachment.

Attached is a 60 day trend of PT140 which is a one of the new Inficon BPG402s?  IP7 and IP8 have been a steady 5000 volts for this time period.  Is this a gauge thing?  I haven't been intimate with what Gerardo, John and Chandra have learned regarding the behavior of these new wide-range Bayard-Alpert/Pirani hybrids but this slope looks "not insignificant"

I ran the current version of the calibration pipeline over a stretch of O1 data to reproduce the kappas and compare to those in the C02 frames.
The filters file used was aligocalibration/trunk/Runs/O1/GDSFilters/H1DCS_1131419668.npz, as suggested by the calibration configuration page for O1:
https://wiki.ligo.org/viewauth/Calibration/GDSCalibrationConfigurationsO1#LHO_AN2
The agreement looks quite good. Time series plots of the kappas and the cavity pole are attached. The start time used here was 2016-10-04 12:41:19 UTC (GPS 1127997696).