Summary:

There is a broad peak in the DARM spectrum around 5kHz which is consistent with the TEM10 being rung up at that frequency. 

• The transient behaviour is consistent with the simulated change in g-factor (and HOM spacing) of the cavity due to self-heating of the optics causing the ITM and ETM surface curvatures to change
• The observed peak in DARM shifted down in frequency after the ETMY ring heater values were changed (as expected).

Details:

The attached MATLAB file produces an animation that shows the DARM spectrum around 4700 - 5500 Hz. It updates this spectrum for 50s intervals around the time that the IFO is locked. Very quickly we see a broad peak (around 50Hz wide) form in DARM around 4900 Hz. This peak moves up in frequency as the lock is maintained.

I have the online simulation of the HOM spacing (frequency vs time) shown in the top plot. As the anmation steps forward in time, we can start to find the maxima of the broad peak in DARM. I've taken the location of these maxima and plotted them in the FREQ vs TIME plot (I've only picked the location of the maximum value of the broad peak and tried to exclude all times before the lock is acquired and the frequency starts to change). The transient behaviour is quite consistent with what the simulation predicts. 

The two figures correspond to two recent locks but different ETMY ring heater settings:

1. GPS TIME: 1160974817: ETMY RH = 0W
2. GPS TIME: 1161055457: ETMY RH = 1.1W (aLOG 30685)

After the ETMY ring heater is turned on and stabilizes, we expect the HOM frequency to shift down by about 60Hz, according to the simulation. We see a commensurate change in the location of the peak in DARM.

If this is the TEM10 mode in the arm, then what is ringing it up?

Time 2:

More to come ...

(Download and run the MATLAB file to see the animation).

model restarts logged for Thu 20/Oct/2016 No restarts reported

model restarts logged for Wed 19/Oct/2016
2016_10_19 09:33 h1fw0
2016_10_19 10:13 h1fw0
2016_10_19 10:22 h1fw0
2016_10_19 10:59 h1fw0

Unexpected fw0 restarts, logs indicate slow file system access problems. Power cycled h1ldasgw0 (not shown) and h1fw0. No subsequent restarts as of time of this alog (49+ hours).

Made a small modification to a pre-modecleaner servo that reduced the boost gain by
lowering its corner frequency.  Installed the servo and couldn't lock the pre-modecleaner.
After about ~15 minutes of trying, I gave up.  The original pre-modecleaner servo was
re-installed.  With it I noticed on the CCD camera that the alignment had shifted.
Re-aligned the beam into the pre-modecleaner, checked alignment onto the locking
photodiode, re-locked the pre-modecleaner.

Bake of RGA is going good thus far, still "inching" toward 180C < temperatures < 200C everywhere excepting the tapering-off to 90C at turbo inlet.

Lock loss at NLN, same as last night, DHARD pitch went into oscillation. Since Terra only needed high power and not NLN, stopped there for subsequent locks.

07:10 UTC Sheila to LVEA to check PSL enclosure air conditioning
07:15 UTC Sheila back
07:28 UTC Attempting to lock
08:00 UTC Stopping at INCREASE_POWER (50W) for Kiwamu and Sheila
08:26 UTC Moving to NLN
08:32 UTC NLN
08:36 UTC Lock loss
08:53 UTC PRMI to DRMI transition worked
09:06 UTC Lock loss almost immediately upon reaching DC_READOUT
09:12 UTC I'm trying to turn off the ISS autolock, but something keeps turning it back on.
09:41 UTC DC_READOUT. Waiting for roll modes to damp.
09:49 UTC Moving on
10:01 UTC NLN
10:05 UTC Lock loss. DHARD oscillation before hand.
10:10 UTC Turning off ISS autolock turns on the first loop integrator? Turning on the ISS autolock turns off the first loop integrator?
10:30 UTC Stopping at INCREASE_POWER (50W). Terra taking PI measurements.
11:00 UTC Lock loss (PI mode 27)
11:21 UTC Stopping at INCREASE_POWER (50W). Terra taking PI measurements.
12:10 UTC Lock loss (PI mode 27)
12:14 UTC IR not found for diff. Found by hand.
12:34 UTC Lock loss just after reaching DRMI_LOCKED.
13:00 UTC Stopping at INCREASE_POWER (50W). Terra taking PI measurements.
14:16 UTC Changed ITMY ring heater upper and lower requested power from 1.25 W to .25 W per Terra's request.
14:23 UTC ETMY violin mode 9 rang up. Terra changed the damping gain to damp it.
15:03 UTC Lock loss. Trouble locking PRMI. Catches but does not hold.

Summary: ETMY PI 18041 Hz - aliased down from 47495 Hz, Mode27 - is (more readily) dampable if the OMC DCPD is used as error signal.  ESDs provide adequade force if actuating on mode while it is < 4 orders of magnitude above OMC-PI_DCPD noise floor (~10^4 magnitude in OMC-PI_DCPD channel), ~2.5 orders of magnitude above QPD noise floor.  

- - -

Previously we had been unable to damp Mode27 (alog 29685) and turned off ETMY ring heater (alog 29702) to have less optical mode overlap with this 47495 Hz mechanical mode. Wednesday night I turned the ETMY ring heater back on to revert back to the bad zone and more thoroughly test our ESD damping capabilities. This mode has the highest gain during the transient time.

Over the past 20 hours and about 10 locks, I've found we have the actuation force to damp this mode if it is caught early ( < 4.5 orders of magnitude above noise floor in OMC signal ) which is much more likely when using the OMC DCPDs. All PI damping nominally uses error signal from TransMon QPDs. The OMC DCPDs also see mechanical modes and with an SNR ~15x the QPDs (see here for example from tonight of Mode27 seen in both error signal paths). We have avoided using DCPDs as error signal becaues the coupling path is not well understood and we see gain sign flips and much less steady phase changes. However, the high SNR seems necessary for PI with high gain that rings up quickly so that the damping loop can actuate against it more immediately while the mechanical mode amplitudes are still lower. This and other PI in the 47kHz ring up especially quickly (and twice) during the first hour transient sweep. Mode is very responsive and easily damped with a gain sign or phase change under 4 orders of mag above noise floor (1k magnitude in H1:OMC-PI_DCPD_64KHZ_AHF_DQ); above 4 is gets less responsive and is essentially non responsive by 5 (10k mag). So while we do have an error signal from QPDs within the effective damping range, the smaller window combined with high PI gain/rapid ring ups make the QPD signal inadequate. 

Operators: I've left Mode27 in the new OMC DCPD scheme. Patrick and I found a pattern of needing to change the gain sign twice during the beginning of the lock and adjust phase after that during the next hour or so. This mode responds very quickly, so be ready to immediately revert gain signs if it turns out to be the wrong direction. 

Note: Rang up with single quadrants (LL or UR) a few times and saw slightly less effective actuation than with LL+UR, though still need to process actual ring up time constants to compare effect. I switched once to LR, UL drive but many modes went crazy and broke lock so didn't try again tonight. Will repeat at lower power to get relative quadrant coupling for a given mode. 

NPRO pump diode output power and NPRO output power for the past 12 months.  The power decay for the pump diodes
has more or less levelled out.  We no longer really have any margin left for the pump diodes.  At this stage of
their life, increasing the diode current might precipitate their demise.

    The NPRO power may have levelled off too.  We should still plan on replacing the NPRO during the mid-run
break.

We had the 4735 Hz violin mode ring up bigly ~13:30 UTC. Was able to damp by flipping the sign; it's still fairly rung up so incoming operator should keep an eye. This is ETMY MODE 9.

Per Sheila's request, we just reduced ITMY ring heater 1.25 W --> 0.25 top, 1.25 --> 0.25 bottom at 14:16 UTC.

Keita, Sheila, Kiwamu

This afternoon Keita and I did another test of opening and closing DBB shutters, since Keita realized that there are multiple shutters that matter.  The results are in the screen shot.  We only see two shutters on the PSL layout, (SH01 in the 35 W path to the DBB and SH02 in the 200W path) and on the photos documenting the table, but there must be a third shutter, perhaps inside the DBB box. We did not test switching to the 35W beam because that caused a lockloss last night. Apparently the shutters used here are Thorlabs SH05, which has an aluminum blade according the the thorlabs website. 

When changing between shutter states today we saw a broad band change in the DARM noise throughout our 200Hz-1kHz lump, (this is a little different from what we saw last night).  However, we saw 3 different noise states in DARM depending on the shutter requests, shown in the attachment.  

We guess that the shutter which is controlled by the epics channel "PSL-DBB_SHUTTER_DBB" is inside the DBB box itself and not on the layout.  It is hard to explain the table above.  For example, if no beam is selected, and both beams are really blocked before they reach the DBB, why would closing the shutter inside the DBB matter?  

Kiwmau and I went inside the PSL, placed beam dumps in the paths to the DBB.  We placed a "black hole" beam dump (no cone in the middle) in the HPO path (a 250 mW beam between M3 and M12 on the layout).  Looking at that beam with an IR card, we could see a corona around it, pictures will be attached to this alog.  This corona is scattered around, hitting the black baffle near the DBB apperature and other things.  We also placed a black glass beam dump upstream of the front end laser beam path the the DBB, just before the shutter. 

Update:

After waiting out the Japanese earthquake, we relocked.  The lump was smaller in the first moments of the lock.  After a few minutes the peaks reappeared, but the peaks still changed when we opened and closed the shutters, in a way that is repeatable although the plot is confusing since the overall level of noise was changing probably with the thermal state.  (each time we opened the shutter, things got worse than they had been).

We do not understand how changing the shutter state impacts the DARM noise, although we think we have ruled out scattered light.   Kiwamu thought that perhaps the change in the noise could be due to the change in the diffracted power when we move the shutter (see Keita's alog 30679). We tried a test of changing the diffracted power, which unlocked the IFO.  It could also be through the same electrical coupling that means the diffracted power changes when we open the shutter. 

TITLE: 10/20 Eve Shift: 23:00-07:00 UTC (16:00-00:00 PST), all times posted in UTC
STATE of H1: Down. Earthquake.
INCOMING OPERATOR: Patrick
SHIFT SUMMARY: Been stopping at Increase Power so far. Couple of locklosses. One was due to 9.32Hz rung up (DHARD loop related) and one due to PI mode27 (18043Hz) rung up really fast.

LOG:

01:18 Lockloss -- 9.32Hz rung up

04:05 Kiwamu and Sheila to PSL

04:25 Lockloss -- PI rung up (fast). Was able to find the right phase where the mode became stablized but it was too late.

05:34 Unable to lock due to a 6.2M earthquake in Japan. Take IFO to Down.

[Stefan, Jenne]

We looked at the cross-power spectrum at 10W and at 50W, to see if we would be able to see the smooth "lump" at several hundred Hz if we do as PeterK suggests and lock with 10W, no HPO on Monday.

Our conclusion is that at least at the beginning of a 10W lock, the answer is no lump.  While the shot noise makes it difficult to see whether the lump is there or not, by taking a long cross power spectrum (DCPD_A vs. DCPD_B) we can dig down much deeper. 

In the attached screenshot, we show the 50W and 10W calibrated DARM spectra (labeled "asd"), the shot noise level assuming the spectrum is shot noise limited at 1340Hz ("shot") and the cross-power-spectra calibrated to amplitude units ("cross-asd").  Note that the 10W shot noise almost limits our ability to see the lump, but the cross-asd tells us that the lump isn't there.  The 10W spectra are taken with 15,000 averages, while the 50W spectra is only 1,500 averages.  10W data is from 20:56:00utc today, while the 50W is from our best time last night around 06:44:00utc. 

We'll look more at the data later, but we are currently back at 10W, and at the beginning it looked like the lump was there (spectra and quick cross-power-spectra), but 10 min later it is starting to go away.  This is definitely something thermal-related, and not relating to the different amount of polarization we let through the rotation stage + PBS. 

TITLE: 10/18 Day Shift: 15:00-23:00 UTC (16:00-00:00 PST), all times posted in UTC
STATE of H1: Locked at Increase Power

SHIFT SUMMARY: Locking was mostly fine today. Needed some care at DRMI acquisition after high power locklosses, otherwise PI was PI

Few people went to mid stations, Peter did a test of PSL make up air, otherwise no other work done in VEAs.

PI StripTool on nuc6 now has two StripTools. Upper StripTool (PI_monitoring.stp) handles modes with frequency under 17k Hz, lower (PI2_monitoring.stp) with frequency above 17kHz.

We've seen more modes than fit on a single StripTool and with the upcoming TCS changes, it's probably best to keep an eye on all possibilities. Both StripTools are found in /opt/rtcds/userapps/release/isc/h1/scripts and can be opened from the PI Overview medm screen. 

Reports: 

- DBB shutter open results in a huge noise bump issue

- peaks are smaller, but still there wtih DBB closed off

- PMC high voltage noise: (HV drive coherent, better readout--> peaks visible, realignment of PMC improved peaks / but hump is still there)

- coupling mechanism from PMC length offset to DARM unclear

- PMC UGF was 600Hz instead of 5kHz. going to (nominal) 5kHz made DARM noise worse.

- Reducing PMC loop gain further is not helping enough

- projection ~x2 below noise bump.

- PMC wrong polsarization light? (would be almost resonant, and therefore sensitive to PMC length noise) not clear how the coupling downstream of PMC works.

- Jitter spectrum looks smooth - thermally diven? Is so, other modes also produced... 20 mode...!

- picomotor steering into PMC to test...

- SRM: alignment affects jitter bump... !?

======================================================================

To do:

- 9MHz missing in reflection (x5 missing: (2W-50W we are loosing 2.5x)) - > find 1W of 9MHz (what was the conclusion for REFL9 gain?) 

- Low noise operaion at 12W, 24W (available in guardian)

- ER9 TCS configuration?

- PMC pole down

- Donut jitter verification? How?

Things to copy from LLO:

- Black glass for ETM camera to eliminate back-scatter

- ESD bias noise / zero crossing * run with ESD offset /reduced bias

- Compensation plate alignment ?

- PCAL switch (too much noise?)

"classical" tunings to be redone at operating point:

- A2L / ASC tuning

 -Aux loops feed-forward

=======================================================================

Timeline:

- HPO: Earliest: Monday 10/24

        - Option: 12W low power test first

- ER10 start LLO Oct 31 / LHO 10 days later

========================================================================

Lockloss ~19:30 UTC was due to rapid rise (about 10 min into 50W) of new ITMY PI at 18136 Hz. It's now become MODE24 and was damped the following lock with the PLL scheme. 

First saw mode in the live broadband monitor (found on PI medm screen under the StripTool button). I identified the arm by looking at SUS-ETMX/Y_PI_DOWNCONV_DC2_SIG_OUT_DQ spectrum at time right before lockloss. I identified test mass by alternating matrix elements to send damping signal to ETMY (with no result) then ITMY. Screenshot of PLL, broadband monitor, and StripTools during successful damping are attached. 

Here is a table showing the current LHO simulink mdl files which have local modifications not checked into svn and/or have pending updates from the svn repository.

Note that the SUS MASTER common files are currently incompatible with the H1 sus models and should not be updated until we are ready to apply Stuart's changes.

The first attached screenshot is a series of DCPD spectra from the last several months.  When we first turned on the HPO, the noise in DARM did not get any worse.  The noise stayed similar to O1 until July 16th, the lump from 200-1kHz started to appear in late July, when we started a series of changes in alignment and TCS to keep us stable with a decent recycling gain at 50 Watts.  

PMC

Last night, Gabriele pointed out that the noise in DARM is coherent with the PMC HV.  The HV readback had been mostly some white noise (probably ADC noise), until the last few weeks, but has been getting noisier so that some of the things jitter peaks show up in it now (SEcond attached screenshot, colors corespond to the dates in the DCPD spectrum legend.  This may be related to the problem described in LLO alogs 16186 and 15986.  The PMC transmision has been degrading since July, which could be a symptom of misalingment.  Since July, the REFL power has nearly doubled from 22 to 38 Watts, while the transmission has dropped 28%.  The PMC sum has also dropped by 4%, roughly consistent with the 3% drop in the power out of the laser.  Peter and Jason are planning on realigning the PMC in the morning, so it will be interesting to see if we see any difference in the HV readback.  

TCS + PRC alignment:

The other two main changes we have had in this time are changes in the alignment through the PRC, and changes to TCS.  These things were done to improve our recycling gain and stability without watching the noise impact carefully.  In July we were using no offsets in the POP A QPDs.  We started changing the offsets on August 1st, after the lumpy noise first appeared around July 22nd.  We have continued to change them every few weeks since then, but generally moving in the same direction.  

The only TCS change that directly coresponds to the dates when our noise got worse was the reduction of ETMY RH from 0.75 W each on July 22nd, the other main TCS changes happened September 10th.  It would be nice to undo some of these changes before turning off the HPO, even if it means reduing the power to be stable.