I had an issue with the wiper script on h1ldasgw1 (it was not running) which meant that new /frames-1 eventually became 100% full having been running since 2nd November. I have started the wiper and h1fw1 is writing frame files again.

This means that the default NDS (h1nds1) has no full or second trend data during the data gap. These data can be obtained by changing your NDS to h1nds0, which has no data gap.

Minute trend data (acquired by h1tw1) was unaffected by this.

The data gap on h1fw1 is:

See https://alog.ligo-la.caltech.edu/aLOG/index.php?callRep=37028

Thanks to site personnel for setting this up.

I added 250ml to the Xtal chiller.

Everything looks normal here. There was an incursion - a tour and some incidental 45Mhz amp mod troubleshooting - done last Thu.

• 15:15UTC - Bubba, ChrisS, MarkD, TylerG - LVEA to install HAM5 and HAM6 doors
• RO water alarm, some vacuum alarms about pressure and instrument air, non-critical
• a couple EQs in the last 5 hours
• nothing else to report

Making a few manual adjustments to VBOs in VPW fow tomorrow.  

Inspected purge air.  Cheryl V. had noted low supply on Friday -> I observed low flow when processed air being routed through left drying tower and typical flow when on right tower.  I had noted this behavior in previous log entry.  I suspect that this started when last the tower desiccant had been changed.  Don't have a flow meter in system so only can make qualitative observations.  It may be that the left tower had too much desiccant added or perhaps the two towers have different media size?  In any case, it seems that the left tower is more restrictive to flow than is the right tower.  For post O2 in-chamber activity, so many doors have been removed as to make this situation mute.  The measured dew point is still typical for the "maximum demand state" as has been the case post O2 (-30C < dew point < -25C).  

We should evaluate the cause of this when next the Purge Air system can be shut down.  

Measured dew point at OMC vent/purge check valve = -28C.  

1405 hrs. local -> Leaving site now.

Starting from the input beam model described in alog 30126, I expanded the modal model to include the new arm cavities and new SRC.

The optics location data is based on:
- https://galaxy.ligo.caltech.edu/optics/ for optics ROC and glass thicknesses
- E1400205 for BS, ITMX, ITMY, SR3, SR2 and SRM locations (SR2 location moved by 5cm towards the BS compared to E1400205)

While the code is a bit lengthy, it only builds simple structures describing sections of the interferometer:

SRCrt = 

    lambda: 1.0640e-06
         q: -1.8528e+03 + 4.0141e+02i
         N: 8
      dist: [0 0 24.4907 15.4657 15.7441 15.7441 15.4657 24.4907 0]
      ifoc: [-0.0010 -2.3180e-04 0.0555 -0.3113 -0.3522 -0.3113 0.0555 -2.3180e-04]
     label: {'ITMback'  'ITMlensr'  'SR3'  'SR2'  'SRM'  'SR2r'  'SR3r'  'ITMlens'}

XARM = 

    lambda: 1.0640e-06
         q: -1.8377e+03 + 4.3420e+02i
         N: 2
      dist: [3.9945e+03 3.9945e+03 0]
      ifoc: [8.9119e-04 0.0010]
     label: {'ETMX'  'ITMX'}

YARM = 

    lambda: 1.0640e-06
         q: -1.8356e+03 + 4.3599e+02i
         N: 2
      dist: [3.9945e+03 3.9945e+03 0]
      ifoc: [8.9012e-04 0.0010]
     label: {'ETMY'  'ITMY'}

Here the relevant fields are:
q: input beam parameter
N : number of optics
dist : distance between the optics (d/n if in glass, i.e. NOT optical thickness d*n, since this is only for the modal calculation); one more than optics
ifoc : inverse focal length (either 1/f for lenses or 2/R for reflective optics)1 label: label of optics
 

The cavity overlap in the model is quite good:
Mode-matching of PRC to average arm: 99.9976 %
Mode-matching of PRC to X arm: 99.9988 %
Mode-matching of PRC to Y arm: 99.9959 %
Mode-matching of SRC to PRC: 99.8424 %
Mode-matching of SRC to X arm: 99.8136 %
Mode-matching of SRC to Y arm: 99.7878 %
Contrast defect between x and y arm, incl Schnupp asymmetry, but without reflectivity imbalance: 5.0033e-06
 

Note that without the 5cm SR2 move that was just completed the SRC looks significantly worse:
Mode-matching of SRC to PRC: 96.8874 %
Mode-matching of SRC to X arm: 96.7677 %
Mode-matching of SRC to Y arm: 96.666 %

The code is located at /ligo/home/controls/sballmer/20171207/beamCalc.m. It uses custom MATLAB libraries in /ligo/home/controls/sballmer/lib. The library includes a number of mode propagation plotting and overlapping tools.

Stefan, Thomas

In light of the new ITMX and ETMs being installed before O3, Stefan thought it would be a good idea to try to estimate the new contrast defect by calculating the fundamental mode of  each arm cavity before and after the swap using data from the core optics galaxy page and propagating the beams to a common point and calculating the contrast defect via this equation:

Two methods:

- Stefan used handwritten MATLAB code,which is located in ligo/home/controls/sballmer/20171207,  takes into account the arm cavity modes and the Schnupp asymmetry.

- I used Finesse with the aLIGO design and modifying the optics to the most in-situ numbers for ROCs (attached .kat parameter file and custom function to find eigenmodes).  This model includes the arm cavities and the Schnupp asymmetry, as well as the thermal lens which is approximated at f=34.5km and the michelson length.

When looking at the contrast defect on the HR surface of the arm cavities, both models are consistent, which makes sense because its the simplest calculation.  However, when propagating the beam to a different point such as the beamsplitter, things get more complicated because of the lenses caused by the AR surface of the ITMs and beamsplitter are estimates at best.  Although there is a slight difference in the q-parameters between the two models because of these effects, the way we calculated the contrast defect is the same.

The parameters we used were as follows:

Both simulations used the same ROC numbers, arm cavity length of 3994.5 meters and Schnupp asymmetry = 0.08m

Before Swap

After Swap

Results

Finesse

Stefan's Code

In the end, the interesting part which both models  agree on is that the contrast defect should be better with the new test masses.

It is important to note that we've only really considered the modal overlap, this has not incorporated the reflectivities of the ITMs which could make a difference in the end results if they are substantially imbalanced.  

Vacuumed in HAM2, using this combination of vaccum hoses:

• tiger vac
• vacuum hose covered in a new clean C3 cover
• adaptor that attaches to about 30 inches of plastic hose (~1.2inch, wiped outside and IPA run through the inside)
• Kapton tub

The Kapton tube was class-A, so I used it to vacuum horizontal surfaces in MC3, MC1, PR3, IM1, and IM3 cages, and I did use it in contact with the metal.

I kept close track of the space under the +Y MC3 blade spring, and from this test area I can see that larger particulate are picked up by the vacuum, but smaller particulate are not, and when I wipe with a IPA wipe, the small particulate are removed.  Picture attached.

I put IM1 on EQ stops and used the vacuum to remove a large particle on the upper portion of the face of IM1, and this was successful without any contact between the optic and the Kapton tube.

I used the same technique to remove large particulate from the MC REFL periscope 2 lower mirror, which was very contaminated.  Not all of the particulate that I could see would come off of the optic, so some may be scratches that have been present since install.

Given my conclusion that vacuuming leaves small particulate, I wiped over all of the surfaces that I could reach in HAM2 from both the West and East doors.

PRM stage M2 has OSEM flags on the HR side of the tower that do not go into OSEMS, and the -Y flag has shifted out of place, and will need to be restored.

The really big story is that IM1, IM2, and IM3 shifted in alignment between 50 and 1500urad, and I have only one explanation for this, which is that the 4 8/32 bolts that hold the top plate to the side plates are loose, and will need to be tightened.

These bolts are covered by the brackets that hold the blade spring EQ stops at the top of the tower, so the brackets will need to come off to give access to the remaining 4 loose bolts.

This also reveals that the root cause of the shift I saw in IM3 after the B&K, a shift  I recreated myself by tapping with a dog clamp on the top plate, are these 4 hidden loose bolts.

   00-:25 (16:25) Cheryl is still in the LVEA working on the closeout of HAM2. She will post her LVEA exit in the aLOG when she is finished.

I checked the alignments of the IMs since I finished working yesterday, and found that IM1, IM2, and IM4 recoverd their alignments when restored this morning, but IM3 did not, it has a significant change in alignment in pitch and yaw.

IM3 change in pitch is ~-1000urad, and change in yaw is ~-1500urad, however the alignment sliders are at (0,0), and have been since yesterday, so there's no apparent reason for the change.

smoke test at EndX - Brian Lantz, Jim Warner, Siddhesh Pai, Brijesh Pant
Nov 29 2017, about 2:08 pm local
used 1 2B candle from superior signal

burn is ~60 sec.
https://youtu.be/XbouhfAK1Mg

smoke candle is ~10ft in front of the fence,

The smoke candle is about 11 ft. up off the ground, close to the vertical center of the fence material. The picture below shows the pole we used w/ Jim as scale standing next to the fence - The smoke emitter will be placed at the height of the blue tape on the pole.

We took several videos, the one from Brian's phone is linked above.

The smoke trail clear to the fence, but often difficult to see downstream of the fence. Why?
1. partly due to blue sky behind early trail and clouds behind late trail. note this for future film angles
2. trail is darker behind fence - might be some shade?
3. trail is getting bigger.
4. the place where the candle burned though was next to the post.

 - next tests we might try 2 candles together, and try to shoot against blue sky. i'm not sure I have enough candles to use 2 each for all my desired tests.

We see that the smoke moves through the fence. It seems like the fence might be "softening" the edges of the trails my generating lots of short scale mixing, but the structure at the ~1 foot scale seems unaffected by the fence.

Is the air moving more slowly behind the fence? Maybe? I sort of looks this way, but I can't understand how this would be true, if none of the air flow is getting diverted up-and-over or sideways-and-around the fence.
1. It is not going up over the fence -> if you assume the same density, then conservation of mass suggests that the speed on the two sides needs to match.
2. It might be slipping sideways instead of going over - this could allow slower speeds and would not be visible from this camera view -
3. It sort of looks slower - this could be the result of the smoke getting further away from the camera as it moves downwind. how to watch for this at EY?
 - maybe have a camera on the ground at the corner of the building looking up?

4. The structure of the plume moves right though this fence. The fence can't be adding large, immediate changes to the vertical velocity

More videos and pictures will be added, and we are planning to do some additional tests on Friday.

Jim set up an anemometer a bit (something like 50 feet) upwind of the fence to record the wind speed. Data download is pending.