I have created a placeholder h1sqz model and started it on h1lsc0. It has dcu-id=76, a processing rate of 16384Hz and runs on the 7th core of this new 10 core machine.

I have not yet added h1sqz to the DAQ. I'll do that when we next restart the DAQ.

I have added the model to the CDS Overview MEDM (see attached).

I believe this completes the new parts needed in CDS computing for Squeezer install.

Krishna, Jim

This is a slight variation on an earlier duty cycle analysis by Jim. I'm trying to establish how the new ISI-Stage1 control scheme implemented in O2 at LHO benefited the interferometer. As a reminder, in O1, we only used feedback from the Stage 1 seismometer and switched between the 45/90 mHz blends to combat microseism/wind respectively. In O2 we used 'tilt-subtracted' feedforward at low-frequencies and 250 mHz blends as the nominal configuration on all platforms including the HAMs. The data lives in:  SeiSVN/seismic/Common/Data/LHO_O1_O2_duty_cycle_data

The first attachment shows plots for duty cycle versus wind for O1/O2. It uses the minute trends of ISC Lock State and the ETMY windspeed (max) signal. The first page simply shows the distribution of wind - fraction of time windspeeds were in a given bin (bins were ~2 mph) during O1 and O2. The second page shows the fraction of the time the interferometer was locked at a given windspeed. Not only is there a clear improvement in O2, but the curve looks flat up to a windspeed of ~30 mph unlike in O1. The overall duty cycle in O2 seems to have suffered a bit, possibly due to other reasons. Pages 3 and 4 show similar plots, but only comparing the 45 mHz blends used in O1, which are still the default configuration at LLO. Again, it is interesting to note the downward trend on page 4 for the 45 mHz blend, which suggests that even 10-20 mph winds would begin to impact the interferometer.

The second attachment has very similar plots for duty cycle versus microseism velocity, using the band-limited-rms ITMY_Z (max) signal in the microseism band. The O2 configuration looks better once again and there is a similar trend of nearly flat duty cycle up to ~1300 nm/s velocities in O2. The distribution of the velocities looks odd/different, partly because of the inclusion of Hanford summers in O2, which are very quiet in the microseism.

WP7162 Arnab, Dave:

This morning we upgraded h1lsc0 to a faster model (same model as h1suse[x,y]). The reason for the upgrade is not that we need the faster cores per-se, rather we need the additional cores provided to run the new h1sqz LSC squeezer model. The original was a 6 core machine, allowing 4 user models. The new is a 10 core machine, allowing 8 user models.

To expedite the install, we took the DTS x1susex computer, added a second GeFanuc 5565 card and used that for h1lsc0.

Unfortunately despite our best efforts, the upgrade glitched the corner station Dolphin fabric, and the models had to be restarted on all the Dolphin'ed corner station front ends, which includes the PSL.

S. Dwyer, J. Kissel

While inspecting HAM6 regarding the broken OMC REFL Beam Diverter Dump, we took the opportunity to check out the fast shutter.

We saw several things of concern:
    (1) The cabling that emanates from the shutter itself looks (and has been previously confirmed to be) very close to the main IFO beam path. Koji indicated in his LHO aLOG 28969 that "the clearance does not look great in the [above linked] photo but in reality the beam is clearly away from the wire."
    (2) One of these wires (the wire closer to the IFO beam) has a kink in it, but no visible burn marks, so we suspect this is from mechanical manipulation during install.
    (3) OM3's readout cables are kinked in a stressed position to make room for the fast shutter, which is a little forward (-Y, away from the beam splitter) of the position indicated in the drawing likely because of this interference.
    (4) Some small fleck of particulate on the inside of the "toaster" face, on the HR (back towards OM1) side of the shutter.

I attach picture collections of these things (admittedly the particulate pictures are not great -- we tried lots of times to get a good picture but failed).

Hopes:
   Re (1): Can we find a way to route these cables below the beam line, instead of surrounding it?
   Re (2): Same as (1)
   Re (3): Can we replace OM3's OSEM cables with a 90 deg backshell, so as to clear room for the fast shutter and relieve stress on the cable?
   Re (4): hopefully this is not from the HR surface of the fast shutter optic

S. Dwyer, J. Kissel

This completes the investigation of the broken beam dump found in HAM6 (see LHO aLOG 38918) -- the beam dump has been identified as the OMC REFL Beam Diverter Dump. 
This beam dump captures light *down stream* (toward the OMC) of the fast shutter.
The dump was broken in a relatively clean vertical fracture which lines up with the dump's set screw, and when reconstructed appears to show a small pock-mark from an apparent small laser blast.

While there's no way to prove why it broke, we have two main suspicions:
   - The black glass is secured to the dump's mount with metal set screws. It has been suggested that all such black glass should be secured with PEEK set screws. If not, these metal screws create un-due stress on the glass, especially if over-tightened.
   - During observation, it has become standard to leave the OMC REFL path's Beam Diverter CLOSED, i.e. blocking the path from hitting the OMC REFL QPDs and/or exiting HAM6 onto ISCT6. Thus, during some high-power lock loss in which both
      - the fast shutter protection failed, leaving the OM2 > OM3 > OMC > OMC REFL path exposed, and
      - the OMC was unlocked sending lots of OMC REFL light down the REFL path (instead of through the OMC, if it were locked).


Picture highlights and labeled drawings are attached as .jpgs, and a more complete collection of pictures are compressed as a .pdf.
Regarding the metal set screws on this dump: a survey of other similar beam dumps in the chamber indicate that *all* such dumps in HAM6 are secured with metal set screws (see 2017-10-11_HAM6_BumpDump_SetScrews.pdf).

Open question:
   - Was the beam diverter closed when the shutter failed and killed the OMC DCPDs in Aug 2016 (LHO aLOGs 28820 and 28842)?
   - If closed, did we inspect this path / dump when we went in to fix the DCPDs?
       This picture from the Corey's Resource Space collection show that the dump is at least intact then.

TITLE: 10/12 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Planned Engineering
OUTGOING OPERATOR: None
CURRENT ENVIRONMENT:
    Wind: 13mph Gusts, 7mph 5min avg
    Primary useism: 0.02 μm/s
    Secondary useism: 0.19 μm/s
QUICK SUMMARY:

As Dave mentioned (38991), we have started the model for running the squeezer ASC, although we don't have cables to the actual hardware yet.  

This is a simple model, which takes in the 16 AS_A and AS_B RF42 signals and processes them for sending to ZM1 and ZM2.  The attached medm screenshot isn't finished yet but shows the basic functionality. 

Most of the difficulty was with getting the channel names to be what we wanted.  We wanted to make this a new model rather than adding more parts to the ASC model because the ASC model is already so big that we run into problems with the RCG.  Because the squeezer WFS are really just an additional demod of the AS WFS, I wanted their naming to be consistent with the rest of the AS WFS channels (H1:ASC-AS_A_RF42  ect). 

The model is broken into two blocks, one of which has the top name ASC and just has the standard WFS parts in it, the other is actually two nested blocks, SQZ with a common block called ASC inside of it which has the matrices and servo filters.  This was done so that the channel names would be SQZ-ASC.  As Dave mentioned we had to change the model name from h1sqzasc to h1sqzwfs to be able to use the name ASC in the model without causing a conflict with the actual ASC model's DCU_ID channel.  

The models and the medm screen (still in progress) are in the squeezer userapps repo.  

G. Moreno, J. Kissel, N. Robertson, B. Weaver

Gerardo and Norna finished installing and cabling up the temporary OSEMs on the current OFI for testing. I helped them buzz out the wiring into the ADC and center the OSEMs.
More details later, but I want to get these numbers in the aLOG: 
  ADC Channel            OSEM     Open Light Current         OSEMINF Offset
ADC0_28                         LF      32767                                16383.5
ADC0_29                         RT     30455.2	                             15227.6
ADC0_30                         SD     30319.6	                            15159.8

Note, when we found the system, the temporary cables were hooked up incorrectly to the last DB9 input (analog 29-32, or digital 28-31) of h1sush56's ADC 1's AA chassis (i.e. U33) in the SUS-C7 rack (incorrectly -- according to D1002740). I've moved the cable to the last DB9 input of ADC 0's AA chassis (i.e. U34) as suggested in D1002740, and was able to use the new OFI MEDM infrastructure as designed (see LHO aLOG 38827 for front-end code, aLOG is still pending on new MEDM screens).

On the 5th I opened the soft covers on HAM6 to lock the ISI and noticed that one of the black glass pieces was broken. I'm posting a couple of pictures here. Betsy, JeffK and others have looked to see if we noticed this during the last HAM6 vent, but none of the pictures in the alog show this glass in any detail.