I am leaving it to the morning PSL crew.

I had another look at the slow drop in sideband powers that consistently breaks lock within 20 minutes of powering up to 50 Watts.  

This time I changed the soft loops to X,Y degrees of freedom (first I made the pitch filters for dsoft match csoft, this worked fine but isn't in the guardian).  I ran dither lines on the test masses and tried to move offsets in the soft loops to keep the spot positions stable, but this didn't help the tanking sideband powers at all.  I tried moving PR2/SR3/POP offsets.  SR3 can help improve the AS90 build up after a PR2 move that helps POP18 but walking these together didn't work out well.  I tried locking at 40 Watts where we have more time to move offsetsm but didn't find any alignments that could help the sideband build ups.  We might have to look at some lensing.  I tried using offsets in the POPX to PR3 loop, but as we expect these only made things worse.  

The fastshutter lockloss checker seemed to think that the shutter had not fired in one lockloss today, but as shown in the attached screenshot, it did fire. Leaving the IFO with Kiwamu for the night. 

After Peter King brought back the laser, we relocked the interferometer.
We had to increase the OMC scan range ( omcparams.scan_vramp ) from 40 to 60 for the guardian to find both sidebands.

Then we again lost it twice during the COIL_DRIVER swtiching because of a 2Hz CSOFT signal (pitch and yaw) ringing up. Lockloss plots of both locks are attached.

Next we switched back to ther old COILD_DRIVER_SLOW switching. Interesting, we still had problems once we stated switching test masses (ETMY  - the first one tried). All the corner optics switched without problems (Actually, SRM and SR2 had 0 seconds matrix ramp time, and still were fine. I set it to 1 sec for now - we should speed that switching up. )

In a second attempt I tried only switching ETMY by itself - with all top mass relief loops off (because the high BW offloading was one of the things we changed). still no luck - same 2Hz instability.
Still something changed with the test masses - it is not just the switching script.

A large earthquake spopped further work. I left the slow COIL_DRIVER_SLOW state as a parallel path in the Guardian for debugging. Also It currently only switches PRM, PR2, SRM, SR2, BS.

Kiwamu will try to do some ISS work with the IMC during the quake.

Attached is a 5 minute plot of the monitor signal for the current of diode box 3.  The initial reported value of 50.14 A is
corrected.  ~44 seconds after the start of the plot, the monitor signal hits 100 A.  Whilst the switching from the true value
to the false value was going on, the error bit on the corresponding Beckhoff terminal did not change state from 0 to 1.  The
output value changed from ~12460 to ~24880 counts.

    Since an ADC most likely does not change its reported value unless the signal it is monitoring changes, the problem would
seem to lay with the output monitoring signal of the power supply.  Without knowing what is in the output monitoring circuit of
the power supply, the problem I have is explaining why it resets itself after a period of inactivity unless it is a temperature
related fault.  The evidence for it being temperature related is purely circumstantial at this stage.

    I would suggest that the power supply be switched out.

Came to site just now to find laser off. 

5pm local

12 sec to overfill CP3 with LLCV bypass valve 1/2 turn open. 

Next fill due Tuesday, Sept. 6.

TITLE: 09/03 Eve Shift: 23:00-07:00 UTC (16:00-00:00 PST), all times posted in UTC
STATE of H1: Commissioning
INCOMING OPERATOR: None
SHIFT SUMMARY: Commissioning and an EQ. Toward the last bit of the night we made it to DC_READOUT_TRANSITION wher the OMC_LOCK Guardian went to the state, WAIT_FOR_AS_LIGHT. Seems that the light on the OMC QPDs was below its threshold of 100. Searching through alogs, Terra and I only found alog22289 that referenced this type of situation. We tried following Jenne's suggestion of making sure the Master gain was 0 (though it was actually 0.05, whoops) and then clearing the history of the ANG filter banks. But it didnt work, and when we tried to clear the history for the POS filter banks, because why not at this point, it just dropped the lock. We are done for the night, leaving it in down.

 

Overall message:

• We had trouble powering up today because of the CSOFT radiation pressure instabilty (dPdtheta).  I reshaped the ASC control filter so we should have 20dB of supression, and this seems to have taken care of the problem.
• We have tried to do the coil driver switching twice today, and broken lock both times, seemingly because of a 2 Hz instability that only rings up durring the coild driver switching.  We might want to separate the test mass coil drivers from the triples, so we can narrow down what the problem is.
• We decreased the gain of PRC2Y from -2000 to -14000 to avoid a not very stable ugf at 2 Hz.
• When we power up to 50 Watts with a recycling gain of around 31, the lock only lasts about 20-30 minutes.  There is a slow drift of sideband powers that cannot be fixed by moving PR2 yaw or BS pit or yaw.  It could be a X arm soft yaw drift based on oplevs, we can try using a dither line to monitor the spot position on ETMX and try adjusting the soft offsets this way.

More about CSOFT PIT:

We were using a control filter that had a pair of complex zero compensating the 0.4 Hz resonance, which meant we had very little gain to supress the instability using ASC.  I changed this to a single real zero, which gives us 20 dB gain at the instability frequency. The filter comparison is attached, and a measurement taken at 2 Watts.  In the green measurement I had made a msitake in the filter at 1.5 Hz, the red measurement is with this fixed and the phase should be unchanged from the old design at these frequencies.

Here are some plots to illustrate the alignment progress described in Jenne's alog 29457

The first attachment shows various powers plotted against input power for a power up with the old PRC alignment (blue) and the new PRC alingment (orange). 

You can see that we are still loosing recycling gain as we power up, but less so than with the old alignment.  We know that we can move the PR2/POP offsets a bit more; we stopped where we did last night because we started to rail the POP PZT and had to realign on the in air table.  You can also see that the sideband powers are not dropping quite as much as with the old alignment, although we start with POP 18 lower. 

The bottom row shows REFL normalized by the input power; the plot below shows normalized refl power vs recycling gain.  (Time during the power up goes from right to left on this plot).  You can see that as the alingment gets worse the interferometer is approaching critically coupled, with the old PRC alignment the reflected power starts to increase again for recycling gains below about 27. 

After one of the locklosses in the morning the IMC would not relock. Clearing some of the IMC WFS DOF histories brought it back.
I often had to adjust the BS in pitch to lock on DRMI.
At some point the X arm alignment ended up such that engaging the WFS would drive the green alignment off. Jenne and I trended the optical levers back for ITMX and ETMX and tried bringing these optics back to those values. This did not seem to help so Jenne suggested that I go through an initial alignment. I did and was able to lock both arms on green without difficulty after that.
After the initial alignment we kept losing lock engaging the DRMI ASC loops. Jenne and I stopped at LOCK_DRMI_1F and minimized the ASC loop error signals by hand. Eventually this allowed us to move on.

16:26 UTC Betsy to LVEA west bay
16:31 UTC Kyle to LVEA to take RGA scans
16:51 UTC Betsy done
17:48 UTC Kyle back, not able to take RGA scans
18:51 UTC Chris sending excitation to ITMX
18:53 UTC Kyle back
20:14 UTC Chris sending excitation to ITMX
21:40 UTC Kyle retrieved computer from LVEA
21:49 UTC Starting initial alignment
22:14 UTC Done initial alignment
22:42 UTC Kyle to mid Y to fill CP3

The laser tripped again.  Suspected cause is the same as the trip from earlier this morning.

    Maintenance Tuesday will most likely be quite invasive.

~1550 - 1615 hrs. local -> Kyle to and from Y-mid 

Opened exhaust check-valve bypass-valve, opened LLCV bypass-valve 1/2 turn -> LN2 @ exhaust in 18 minutes 28 seconds -> restored valves to as found configuration.  

NOTE: I increased CP3's LLCV %open value from 20% to 21% in response to the time needed to fill today

Please when the laser trips, disable the locking for the PMC, FSS, and ISS.  It doesn't make
much sense to keep ramping the NPRO PZT when there's no chance of any of the cavities acquiring lock.

[Sheila, Jenne, Kiwamu, Terra]

We have reset the initial alignment offsets, including realigning several paths on ISCT1, to match the alignment that we liked from yesterday (alog 29430).  So far, we think that it's good.  We have powered up part way twice since, and the power recycling gain no longer drops very much!  It still drops a bit, so we could consider going a bit farther, but this is probably plenty good to move on and focus entirely on low noise efforts.  Now at 40W it is something like 31, rather than something in the mid/low twenties.  Plots coming tomorrow.

In order to do this, we did the following (hardware alignment changes noted in bold):

• Lock using old alignment, at DC Readout move most of the way to the new alignment using PR2 sliders and POP_A offsets.
• Relieve the POP_X centering loop PZT on ISCT1 by *very slowly* moving the knobs on the steering mirror that goes to POP_X.  We were able to do this while locked, and PRC2 was using POP_X in-loop. 
• We then tried to power up with the new alignment to ensure that it was good.  It seemed pretty good (PRC gain stayed roughly constant), but we lost lock due to dPdTheta. 
• For our next lock, we had to lower the threshold at which the POP_X centering loop triggers on, since the beam was partly off the WFS.  (spoiler alert: once we finished all alignment work for the night, this is no longer necessary.  POPX is already quite centered before the centering loop comes on.)
• Once we were relocked with our PR2 slider motion and POPA offsets in place, we set the green input pointing QPD offsets such that the green beams were maximally co-aligned with the IR in the arms, and the green WFS signals were all zero.  This was done at DC Readout, 2W.  We also reset the green ITM camera setpoints.
  • We lost lock while doing this, but the Yarm was already complete.  Sheila did a test InitialAlignment of just the green Yarm and saw that the test masses didn't move according to their oplevs from where we wanted them by more than a few tenths of a microradian.
  • For the Xarm, we put the test masses back to the locations that the oplevs said they wanted to be from the 2W DC Readout lock, and then finished setting the QPD and camera offsets.  The test masses had to be moved by several microradians each in both pitch and yaw to get the unlocked mirrors pointed the same way that they were while being controlled by the ASC loops.
• We centered the green beams on ISCT1 on the green transmission PDs and on the transmission cameras.
• We put PR3 and BS back to the places that their oplevs said they were during our 2W DC Readout lock, then aligned both the ALS Comm and ALS Diff beatnote PDs.
• TJ did an initial alignment for us using the new set points.
• We aligned POPAIR_A and POPAIR_B, although neither was very far off.
• We checked that the DRMI ASC works using the POP_A offsets that we like.  This is now in the guardian, but at some point we should picomotor to recenter POPA in this new place.
  • PRC1 and SRC1 and SRC2 are once again turned on in the DRMI ASC guardian.
  • This makes CHARD come on much more nicely (although we have still had a few locklosses trying to engage these loops).  The gain has been redistributed in the guardian so it also comes on more slowly, but it's apparently still not perfect yet.  At least engaging CHARD makes the recycling gain go up rather than down, which it had been doing until this afternoon (the recycling gain would come back once the PRC1 loop was closed around POPA).
• On our first power up with the new alignment, we lost lock due to the dPdTheta instability.  Kiwamu measured the suppression at the resonance frequency, and increased the gain of the ISS 3rd loop by a factor of 3.
  • Upon our next power up, it seems as though we can't increase the power quite as fast, without causing trouble.  Tomorrow we will try with the ISS 3rd loop gain only up by a factor of 2, and we will also try increasing the power more slowly.
  • Sheila tweaked the CSOFT pitch loop shape so that we no longer have a marginally stable point around 0.45 Hz - this will allow us to increase the gain such that we can help hold against the instability.
• On our next power up, we lost lock with an ISS saturation.  We think this is due to the 3rd loop gain being too high / increasing the power too fast.
• A few minutes later, the PSL shut off.  Commissioning to be continued tomorrow when more photons are available.

Not that this is a surprise or anything, but I was noticing that the light on the OMC trans camera changes noticeably as the IFO thermalizes. 

I've temporarily increased the exposure of the camera to 7000 (usually 569).  You can't see much at 2W, but you start to see it at 20W and 35W - again not so surprising.  What is perhaps more interesting is the way the light there changes after we've been at 50W for a while.  The titles of the attachments include how long we were at each power.  All images were taken during the same lock, so the 2 min at 50W image is after 5 min at 20W, and then 5 min at 35W, so it's not straight 2W->50W on this acquisition.

Since the original images are .tiffs, I've included a screenshot of all 6 images.  Top row, left to right:  2W, 5min@20W, 5min@35W.  Bottom row, left to right:  2min@50W, 6min@50W, 30+min@50W.