Restarted cdsfs1 file server after RAID controller "locked up" over the weekend.

Summary: The HIFO-Y feature at 70 Hz is produced by both ALS periscopes on ISCT1. The features below 1 Hz are coherent with ground motion and OSEM sensors. We did not identify the dominant source of noise between 0.8 and 2 Hz, though this band contributes little to the HIFO-Y RMS. We may be able to reduce the RMS by using cylindrical periscopes that have resonant frequencies closer to 250 Hz, but we will also likely get a free factor of about two reduction in RMS (in the 60-80 Hz band) after noisy installation activities cease.

Vibration coupling to HIFO-Y was reduced by establishing a science mode and a commissioning mode for PSL air handling (Link). There remain two frequency bands that contribute significantly to the HIFO-Y RMS, 60-80 Hz, and 0.3-0.7 Hz.

Figure 1 shows spectra for the PEM sensors that we recently installed whose signals have the highest coherence with the HIFO-Y signal above 1 Hz. The PSL was operating in science mode. The accelerometer that we mounted temporarily on one of the two similar ALS telescopes on ISCT1 (red trace), the one that carries infrared light, has high coherence with and similar shape to the 50-100 Hz band noise in HIFO-Y (red and black traces). Further excitation tests on the ISCT1 table confirmed that the 70 Hz feature is associated with the lowest resonant frequencies of both ALS periscopes. Neither the green nor the infrared periscope appears to dominate, and I could find no particular source of, e.g., backscattering noise. It may be that the noise is associated with the differential motion of the two periscopes in the locked path.

The other coherence features in the spectrum are, at about 180 Hz, one of the higher-Q resonances of the PSL periscope – perhaps driven by vibrations at the 60 Hz harmonic. These 180 Hz vibrations may be smaller after installation, and if they are not suffieciently smaller, we may be able to damp at this frequency. The 12 and 13 Hz coherence regions are produced by table sway resonances of ISCT1. I am less certain about the feature between 5 and 6 Hz, but there is a building resonance at that frequency excited by wind. The seismometers, some distance from the PSL and ISCT1, also show strong coherence with HIFO-Y at 5.5 Hz, supporting the building resonance interpretation.

Coherences with features in the low frequency region are shown in Figure 2. Seismometer and OSEM signals have high coherence with the HIFO-Y signal, except in the band between 0.8 and 2 Hz. We did not find the source of noise in this region.

It may be useful for the DetChar group to look for coherence in this 0.8-2 Hz band (I only guessed the most important channels) and, if none is found, search for upconversion between the 0.1-0.7 Hz band and, say, the 1.2-1.8 Hz band.

It may be possible to reduce the ALS RMS by replacing the periscopes with periscopes that have higher resonant frequencies. We can probably get from 70 Hz to about 250 Hz by using the cylindrical periscopes from iLIGO. Stefan also suggested that we use a single periscope for both the red and green beams, reducing differential motion. I placed an accelerometer on the double periscope, also on ISCT1, and found the lowest resonance to be about 60 Hz. It is not clear to me whether putting both beams on a 60 Hz periscope or keeping them separate on 250 Hz periscopes, would result in the better RMS.

We may get a free factor of about 2 reduction in RMS contribution from the periscope motion after installation activities cease, because Figure 3 shows that ground motion in this band was about a factor of 2 lower during S6. After installation we would expect the ground motion to drop back down to the S6 level, reducing the need to change periscopes.

Figure 4 is a photo of the cylindrical periscopes from iLIGO that we gathered together near the squeezer bay with Corey’s help. None of them are complete, some parts would probably have to be remade.

Robert Schofield, Stefan Ballmer, Emily Maaske, Terra Hardwick, Vincent Roma, Brian Dawes, Tristan Shoemaker

(Alexa, Kiwamu, Keita, Daniel, Sheila, Robert, Stefan)

Last night, at 17:50 local time, we turned off the BSC6 clean room...

We observed a sharp step shortly afterward (probably due to a quick realignment) on both green reflected light and transmitted light, followed by a 15h slow drift.
We then decided to realign the REFL diodes at EY. But surprisingly, to lock the green light, we had to move the ITM pitch by 0.7urad, ETM pitch by -1.6urad, and TMSY by -8urad to get back to a good alignment.

Meanwhile we also had to flip the EY PDH servo sign. We don't know why that happened.


Image A:     The the green refl diode, ALS-Y_REFL_A_DC power is falling off the PD over the duration of about 15h.
Image B:     We are also losing a little bit on ALS-Y_REFL_B_DC
Image C & D: The clean room transient is also visible in the TMSY OSEMs as a step. The HPI IPS's show a glitch, but no step. (The large step on HEPI is the lock loss.)
Image E:     The corner green power (ALS-C_TRY_A_DC_VOLTS) and beat note RFMON actually stepped up initially, but then dropped over ~15h.


For tonight we switched the clean room on again around 5pm

There is no signal coming out of the I mon at the end station PDH demod.  Since there was a PDH signal on Q mon I swapped the cable so that Q goes to In 1 on the PDH board.  Both I mon and Q mon are on the scope visible from the ceiling cam right now, both 500mV/div.  I also switched the phase shifter from local to remote control, and roughly tuned the phase to 186.92 degrees.

I set up a delayed measurement for MC3 :

It will start this week end on SUNDAY evening from 5:38 pm, and will run for ~ 3 hours

MC3 locking and damping filters will be turned off as well as the alignment offsets

The h1asctt model was installed back on 7/9/13 (alog 7031), but has gone without filters or a safe.snap file till now.

I created new filter and safe.snap files in the userapps area, adapting the most recent LLO version, and linked to them as follows:

• /opt/rtcds/lho/h1/target/h1asctt/h1ascttepics/burt/safe.snap -> /opt/rtcds/userapps/trunk/asc/h1/burtfiles/h1asctt_safe.snap
• /opt/rtcds/lho/h1/chans/H1ASCTT.txt -> /opt/rtcds/userapps/trunk/asc/h1/filterfiles/H1ASC.txt

I then loaded the filters and BURT restored the safe.snap.

To get the safe.snap file working without errors I had to remove a number of lines corresponding to the two IPC signals I had to remove when I adapted l1asctt.mdl because they weren't yet available at LHO (H1:OMC_PZT*_TTSUS_TEMP, *=1,2). I put these lines in h1asctt_safe_TTSUS_TEMP_stuff.snap next to h1asctt_safe.snap so they can be easily restored if necessary.

The new files h1asctt_safe.snap and H1ASC.txt (but not h1asctt_safe_TTSUS_TEMP_stuff.snap) have been committed.

Work at End-Y (Accelerometer Installation) - Bryan
Work at End-X (ETM Reaction Chain Locking) - Travis
Work on Arm Cavity Baffle Assembly (LVEA West bay) - SLC group
Work on Squeezer (LVEA East bay) - Corey
Replace dewar level gauge for CP4 at Y-mid - Kyle/Gerardo

The raw minute trend data which was damaged by writing incorrect GPS times last Wednesday has been repaired for h1nds0.  The data files have been written to the SATAboy in /frames/trend/minute_raw/minute_raw_1058139783 directory, and h1nds0 now accesses these.  So it is possible to get minute trend data for the past week without a big hole in the data.  

Note that the data on h1fw1 (and h1nds1, which is used by default) has not been repaired yet.  Plans are to repair h1fw1 early next week.

As a quick check of the upper sections of the wire loop in the reaction chain, I suspended the PenRe (recall the ERM was already suspended) from the UIM.  Although not a true test of the chain due to cabling interferences with the LSAT, I wanted to verify that no error large enough to justify a replacement loop existed.  In fact, it looked as good as I have seen on any previous Quad.  Satisfied with the results, I locked the PenRe and ERM for the weekend.

The DAC buttons on the SUS_CUST_QUAD_OVERVIEW.adl screen have been semi-broken for some time. They did link to automatically generated DAC monitor screens but typically the wrong ones - always DAC0 or DAC1, even though there are up to five different cards.

Therefore I created a master SUS_CUST_QUAD_DAC_MONITOR.adl screen that would show all channels for all levels of a quad, in both the user and IOP numberings.

To facilitate this, I added new macros to the sus*_overview_macro.txt files for *=itmx/itmy/etmx/etmy/quadtst as follows:

• UDAC0/UDAC1/UDAC2/UDAC3/UDAC4 are the user model numbers of the DAC cards, always 0/1/2/3/4 (or 0/1/2/3/3, if there are only four total).
• DAC0/DAC1/DAC2/DAC3/DAC4 are the IOP model number of the DAC cards, e.g., 0/1/2/6/7 for ITMY.
• UEDC/UEUL/UELL/UEUR/UELR are the user model DAC channels for the ESD.
• EDC/EUL/ELL/EUR/ELR are the IOP model DAC channels for the ESD.

(While I was at it, I corrected an error I discovered in the file for ITMX: UPUL/UPLL/UPUR/UPLR had been 2_* but should have been 3_*.)

I linked the new screen to all the DAC buttons on SUS_CUST_QUAD_OVERVIEW.adl and SUS_CUST_QUAD_R0.adl. Since the ESD channels were now referenceable, I put them on the SUS_CUST_QUAD_OVERVIEW.adl as well.

The new screens are committed - LLO should svn up /opt/rtcds/userapps/trunk/sus/common/medm.

JimW got the main ~600lb iLIGO Leg Element Mass on the Optical Table sitting on 6 (3 corners of 2) pieces of viton.  This required the forklift--that need is now over.  Further payloading is required to float the Table--ongoing.

(Robert, Stefan)

I calibrated the REFLAIR_A_RF9 diode:
 - gain: 82Hz/(2*1000cts)  (see Alexa's elog from last week, but the amount of light we get changed slightly)
 - Cavity pole compensation zero at 82Hz.

With the PSL in science mode, here is the noise performance we get. We get a significant improvement between 2Hz and 60Hz due to the lower PSL noise. Also the noise at 0.44Hz is slightly lower on the PDH signal (not surprising - it's caused by alignment fluctuations).

The traces are:
 - Red:   REFLAIR_A_RF9 in low bandwidth configuration (on center fringe)
 - Blue:  ASC-y_TR_A_NSUM_OUT in low bandwidth configuration (on half fringe)
 - Green: REFLAIR_A_RF9 in high bandwidth configuration (on center fringe)
 - Gold:  Best curve before PSL work (July 8)

- The total RMS of the red trace is 3.5Hz, 3Hz above 0.5Hz.

HAM3-ISI was half way tripped this morning. Tripping time is 8:49am.

Watchdogs were reset, and isolation loops turned back on. 

Only damping was running on HAM3-ISI between tripping time and now.

Mitchell Robinson, Scott Shankle, Mike Vargas, Thomas Vo and Apollo's Tyler

Manifold Cryopump for ETMx -
Tyler modified 20 FHCS 1/2-13 and submitted for Class A clean and bake. Reviewed specifications for 'Made in India' group of screws for acceptability. Determined parts are ANSI qualified SSTL.
ECR E1200612 was submitted and approved. Scott submitted design of spacer. Part fabrication in process; aim is to Class A by installation on Tuesday.


Arm Cavity Baffle for ETMx -
Started upgrading Arm Cavity Baffle from BSC8 to install in BSC9. Completed Spring Blade Assembly.

(Corey, Keita)

Went out to the Lab in the afternoon.  With an autocollimator, we aligned the gold-coated alignment mirror (located at input of Telescope).  Next up, made sure an iris was aligned (via a second autocollimator at the "output" of the small Secondary mirror (this is on "top" of the Telescope).

We then removed the upper autocollimator and mounted 180deg, such that it can be pointed at a "laser table" nearby.  Used the upper autocollimator to position an iris on the laser table.  The IR laser was then used to align optics on the laser table and then point the beam to the Telescope from the laser table.  The Secondary Mirror's position was adjusted to make the output beam size acceptable---this was done only roughly.

Next up, we want to finely align the irises (via multiple Watec cameras), and then continue with alignment.  All of this happens as we expect to receive (perhaps in the next few days) some new mirrors which will hopefully replace our less than desireable F1 Mirror.

The Telescope Assy was covered with a "tent" for a night with a big desicant pouch inside to keep humidity down (in the hopes of preserving Mirror coatings).

Daniel Halbe, Josh Smith, Jess McIver

Summary: strong, semi-periodic transient ground motion is propagating up the SEI platforms and down the suspension stages at ETMY. Cause of the ground motion is not yet determined. Effect on the HIFOY signal is not yet evaluated. 

Glitching in the top stage of the ETMY BOSEMs was first identified by Daniel Halbe (see Spectrogram_SUS_Longitundinal_M0_BOSEM_July2.png). These glitches are seen in all DOFs of the suspensions and seismic isolation platforms, have an irregular periodicity of about every 10-20 minutes, a duration of a few minutes, a central frequency of 3-5 Hz, an amplitude in Stage 1 T240s of the ISI on the order of a thousand nm/s (~ um/s) away from baseline noise, and have been occurring since at least June 12, 2013. They are not seen in ITMY suspensions channels. 

For a table that traces these glitches across each DOF and up the stages of seismic isolation to the top stage of the suspension, see: https://wiki.ligo.org/DetChar/HIFOYETMYglitching > Normalized spectrograms (PSD) of the periodic glitches for 1 hour 10 min

Daniel also found them in the lower stage ETMY OSEMs: https://wiki.ligo.org/DetChar/SpectrogramsOfH1AllMassQuadSUSETMY

And Josh Smith traced them to excess ground motion using a representative top stage BOSEM channel (see EMTY_top_stage_BOSEM_pitch_correlation_to_excess_ground_motion.png). 

These glitches have a strong correlation with local ground motion and significant correlation with ground motion near the vault. There appears to be faint correlation with ground motion near MX and the LVEA that merits further investigation.  
(See the normalized spectrogram Top_stage_BOSEM_ETMY_longitudinal_glitching.png and compare to normalized spectrograms Ground_motion_PEM_{location}_spectrogram.png of the same time period)

For additional plots of ground motion at various locations around the ifo during these glitches, see again: https://wiki.ligo.org/DetChar/HIFOYETMYglitching
(If you are unable to see some of the plots on this page, please see the instructions under 'Normalized spectrograms (PSD) of the periodic glitches for 1 hour and 10 min'). 

Note that the reported units of counts are incorrect for all plots (a bug in ligoDV) - these channels are calibrated to nm/s for inertial sensors or to um for BOSEMs and OSEMs.