TITLE: 07/12 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Commissioning
INCOMING OPERATOR: None
SHIFT SUMMARY:

As the shift ends:

Jason is solo in the PSL

Jenne, Hang, Craig, Georgia, and Robert in the CR

Stefan, Phillipe, and Kara out at TCSY
 

LOG:

15:37 Jim & Hugh out to CER

15:43 Jim & Hugh back

15:53 Phillipe and Kara out to the LVEA

15:57 Robert, Phillipe, and Kara will be going into the PSL enc to move an accelerometer

16:25 Stefan, TVo and Daniel out to TCSY

16:42 Jason out to PSL enc

17:05 HFD on site to inspect RAFAR - will be driving up and down the arms

17:53 Jeff B into the cleaning area and Optics Lab

18:18 Fil out to LVEA - PSL closet

18:33 Jeff B back

19:00 Robert & Co. back

19:13 Jason back for lunch

19:15 Stefan and Co out for lunch

21:05 Stefan, TVo, and Daniel out to TCSY

21:08 Kara and Phillipe out to LVEA - accelerometers

It looks like someone changed the IMC-MC2_TRANS_SUM_OFFSET on July 10th around 23:30:00 UTC.  Was there a reason for this?  The input to the filter bank is off, and the offset is on and set, which seems like it's meant to make the sum look like the IMC is locked with 10W of input power. 

The consequence of this change is that the IMC guardian was unable to recognize that the IMC had lost lock (since it looks at this sum channel to determine if the IMC is locked), and so the loops were still engaged, and the integrator at the top stage of MC2 was just integrating up, railing the top stage actuators of MC2.  Not the end of the world, but not so great either.

I'm reverting the settings back to their nominal values.

I've also added a clause to the is_locked() function for the IMC case that if the power injected into the vacuum is less than 0.1W, assume that the IMC is unlocked - hopefully this will catch situations like this in the future.

Checked the alignment of the 70W beam towards the pre-modecleaner.  After some minor adjustments
things locked up.  The pre-modecleaner servo locked okay but starts oscillating when the servo
gain slider is increased beyond 0 dB.

    The alignment to the reference cavity was also tweaked up.

    The last optic that turns the beam into the ISS photodiode box was replaced with a 30%
reflecting plate beamsplitter, a Thorlabs BSS11.  This reduces the amount of light incident
on the photodiodes without having to dump most of it inside the box.
The beam incident on the quadrant photodiode was re-centred.  The DC output of the
photodiodes was set to ~11.5 VDC.  The intention being that when the ISS is locked, the
locked DC output of the photodiodes will be 10 V, or ~3 mA of photocurrent.

TITLE: 07/12 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Commissioning
OUTGOING OPERATOR: None
CURRENT ENVIRONMENT:
    Wind: 3mph Gusts, 1mph 5min avg
    Primary useism: 0.01 μm/s
    Secondary useism: 0.10 μm/s
QUICK SUMMARY:

16:20 Peter out to the PSL - PMC and FSS are locked. The light pipe is open and there is light on the IMC REFL camera

Daniel, Thomas, Stefan

- Looking at the FLIR image of the heating pattern on the TCSY table we found that the CO2 beam half-missed the IRIS (picture 1).
- Also, on the TCS table the steering mirror circled in blue (picture 2) was very close to clipping.
- Since we were able to see the unique signature of the annulus mask on the HWS, and the TCS design is essentially an imaging system mapping the mask onto the ITM with a magnification of 23, we decided to align only optics before the mask in round one, and use the FLIR to get a good mask illumination (see picture 3). We did that by moving the red circled mirrors (1 & 2) in picture 2.
- This made the FLIR image look quite nice (picture 3), and we still see the annulus mask signature on the test mass, which we centered a little bit using the periscope to picomotors (animated gif 4).

- However, the lower left part of the annulus mask still seems clipped (animated gif 4), and the central heating mask still produces a fringe-like pattern (animated gif 5). So we are planing to do a beam walk tomorrow.
- For that purpose we wrote a "1 minute heating" script that allows us to get a quick turn-around beam position view on the ITMY using the HWS. We added a corresponding button the the TCS_HWS.adl screen. The two animated gifs were taken with that script.

J. Oberling, P. King, R. Savage

This morning Peter continued to align the current FE laser, and ultimately got a pretty good looking beam out of it.  The only problem is that it isn't 35W, but ~31W.  Try as we might, we were unable to get more power out of the FE this morning.  This beam was put through the 70W amplifier and after some quick alignment (no mode matching was done), we had 75.2W out (albeit questionable beam quality, Peter has the pictures from the WinCam).  So even with power out of the FE down from nominal, we are still able to get >70W out of the 70W amplifier. 

At this point we needed to make a decision about whether or not to swap the FE.  After some discussion among ourselves and with Daniel, we decided to continue on with the current FE laser; this means we are NOT swapping the FE laser.  The main drivers for this decision were the current beam quality out of the FE (much improved by Peter's several hours of alignment work over the last couple days), and the fact that neither of the 3IFO FE lasers are swap-in ready.

This afternoon was spent checking the beam quality in various places to make sure we weren't fooling ourselves about the beam profile.  Using the leakage beam from the 70W amp that is directed to the bullseye PD, Rick set up a rail so we could check the beam profile over a distance.  Good news is that the beam shape remained constant, the only change was in its size (as expected).  The bad news is that yes, the beam quality is questionable (as mentioned above, and not entirely surprising given that we have done no mode matching work for the 70W amp).  Using this rail, Peter took the opportunity to measure the beam propagation out of the 70W amp.  We then looked at the beam as it enters the PMC, and it didn't look right (had 2 peaks, was definitely not Gaussian.  Unfortunately I don't have a picture of it); out of an abundance of caution we removed the ISS AOM to prevent any damage.  We inspected some of the optics and found a few that showed evidence of contamination; AMP_WP04, AMP_M03, and L01.  AMP_WP04 and AMP_M03 were simply dirty (and were therefore cleaned), but L01 had a damage spot almost dead center of the lens.  This lens was of course replaced.  Tomorrow we will continue working to recover the rest of the PSL, starting by confirming the beam quality at the PMC is improved after the lens swap and optic cleaning.  Still to do (in no particular order, items in bold represent the bare-minimum required to get a beam into the IFO):

• Measure FE beam propagation
• Mode match FE to 70W amplifier
• Measure 70W beam propagation after mode matching
• Align beam to PMC
• Lock PMC
• Reinstall and align ISS AOM
• Mode match 70W amp to PMC
• Recover FSS and ISS

We currently plan on having a beam available to the IFO by Friday morning, and will need to use Tuesday maintenance days to continue tweaking the mode matching to improve PSL performance.

Jeff K., Jim, Hang

We tried to implement the ISI suspoint to top mass feeding forward installed by Jeff yesterday (LHO:42851; for more details, see the note by Edgard T1800301).

Using scalar FF coefficients for ISI-L_2_M0-L and ISI-L_2_M0-P, we were able to achieve some reduction of the M0 L & P motions at ~ 0.2 Hz (the freq which we tuned the FF for). See the first attached image.

In the measurement we excited the EY ISI second stage with a broadband white noise and then measure the SUSPOINT_L to M0_DAMP_L/P_IN1 transfer functions. The pink and cyan curves were WITHOUT FF, and the red and blue ones were with the FF ON.

We can see that the coherence (top-right panel) was reduced at ~ 0.2 Hz for both M0-L and M0-P. The oplev pitch motion (bottom-right) was also reduced at the same frequency (the oplev rms was a bit misleading as we excited the ISI with white noise so the main resonance at ~ 0.5 Hz dominated the rms; in reality the total rms was dominated by the microseismic). We have not yet fine-tuned the coefficients so the performance can be improved in the future.

However, we noticed that the optimal FF coefficients appeared to have a strong frequency dependency, and this was what limited our subtraction to a narrow band. This was a bit surprising as we was expecting the coefficients to be essentially a constant according to the model (T1800301). The inferred coefficients based on the measurements were plotted in the third and forth plots, for ISI-L_2_M0-L and ISI-L_2_M0-P, respectively.

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

Details:

Instead of putting in the FF filters based on first principle, we decided to do a completely measurement based FF design. 

To do so, we first did 6 measurements: 

    two for ISI SUSPOINT_ETMY_EUL_L to M0_DAMP_L(P)_IN1, and 

    four for M0_TEST_L(P) to M0_DAMP_L(P)_IN1.

We denote the ISI drive as d ISI_L and top mass drives as d M0_L(P). The osem sensor responses are denoted by d O_L(P). 

Our goal is to find FF coefficients d M0_L(P) / d ISI_L that satisfies the equation shown in the second attached image (eq_ff_coeff.png). The solutions can be easily obtained from our measurements, and are shown in the third and forth attached images. 

Unexpectedly, we saw a strong frequency dependency in those coefficients which was not expected from the first principle calculations. 

For now we only put the inferred values at 0.2 Hz, which led to the subtraction results shown in the first image. We will do a more careful measurement to verify the frequency dependence, and if the feature persists we will try to do freq-dependent FF to optimize its behavior. 

Today I finished hooking up the LED source, thanks to Fil getting me cables yesterday, and then started aligning to the ALS-M11. This is the mirror that will be swapped out for a polarizing beamspiltter soon. Alignment to M11 looks good, and I tried to match distances between lenses as close to the table in Aidan's alog42171, though it seems that some of our lenses are not the same focal length. In particular, HWS-L1 (labeled layout attached) is different that what was used on EY, but I'm guessing this is on purpose since the path to the periscope is further at this [X] end. I will of course look into this to make sure it is correct.

Attachment 1 - ETMY Optical Layout. EX has not been updated yet so I am using EY to reference.

Attachment 2 - High shot of current table.

Attachment 3 - Label on HWS-L1.

TITLE: 07/11 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Commissioning
INCOMING OPERATOR: None
SHIFT SUMMARY:
LOG:

15:49 Richard and Fil out to HAM5 to look at temp laser

15:55 TJ out to EY - work on table

15:56 Phillipe ad Kara out to CER and floor

15:57 Fil and Richard back

16:18 Jason ou to PSL enc

16:22 Vanessa out to LVEA for cleaning

16:45 TJ back

16:50 Robert out to LVEA

17:11 Richard to EX looking for parts

17:35 Richard heading to EY - didn't find what he was looking for at EX

17:54 Nico going to EY

18:01 TVo and Daniel out to LVEA to put a LASER curtain around TCSY table

18:03 Joe M in with tour group

18:16 TJ out to EX to install some mirrors on the optics table

19:15 Stefan, Dan and TVo out to TCSY

20:20 TJ to EX

21:36 Phllipe and Kara out to LVEA

22:00 SAFETY MEETING - HEAT RELATED DANGERS

22:44 Betsy and Travis out to MX

HF range on all appear to be nominally ok

Jeff suggested adding the st1 & st2 blnd and sc nodes to the ISI overview, I'm attaching my first pass. When I emailed Arnaud about it, he said he didn't like it cause he doesn't have 3/4 nodes (per chamber) installed yet. I'm kind of liking it, but I haven't checked this into the SVN yet, so we may have to arm wrestle for it...

The ETMY BRS enclosure has a T240 mounted on the BRS table, in hopes that the tilt subtraction will be improved by co-locating the seismometer and the tilt sensor. Krishna has a couple of alogs looking at the differences in the subtraction, but both of the measurements he posted were for relatively low winds (his max was ~22mph). We were waiting for more wind before deciding to make the switch, then kind of forgot about it. But looking shortly after Krishna's alog, there was a pretty windy time Nov 16th 2017, after Hugh got good insulation on the seismometer and we were happy with the mounting of the STS. The first attached plot shows the difference in subtraction between the PEM sensor (the T240 on the BRS table) and the ground STS. This plot is pretty similar to Krishna's (should be, it came from his script, just with new data), I think it shows that subtraction is still improved with winds into the 30's. The second plot is new, it shows the product & RMS of each of the two ground supersensor signals with sensor correction filter (i.e. (PEMT240-BRS)*SC_filter & (GNDSTS - BRS)*SC_filter). This should be the dominant control signal for the ISI at low frequency and between .01 -.1 hz the rms is reduced up to 30%. It's possible this could be improved some with gain tweaks. 

   This morning added 100ml water to the PSL crystal chiller. As Patrick noted in aLOG #42772, the crystal chiller filter has picked up a bit of yellow. This could be due to the several changes made to the crystal chiller cooling circuit over the past month. There is one more large change (moving the FE cooling circuit to the end of the manifold) in the works. Will monitor filter condition as these changes are completed.   

   Added 100ml to the Diode chiller as well. The diode chiller filter looks clean and clear. 

J. Kissel, D. Barker
WP 7704
IIET Ticket 10975
ECR E1800201
Tech Note T1800301
Implementation Design SEI aLOG 1343

As per all the above documentation, I've installed a new control infrastructure path designed for the BSC ISI ST2 GS13s, projected into the basis of the test mass QUAD suspension point's motion, to be fed-forward to the top mass (main chain, M0) OSEM actuators. This is motivated by the coherence seen between these ISI sensors (appropriately projected) and arm cavity control signals (e.g. the ASC DOFs; see LHO aLOG 42751), and the hope to reduce the control bandwidth of the arm cavity control be re-allocating the reduction of motion to the local QUADs.

This involved 
- modifying the top levels of all test mass ISI front-end simulink models,
    /opt/rtcds/userapps/release/isi/h1/models/
        h1isietmx.mdl
        h1isietmy.mdl
        h1isiitmx.mdl
        h1isiitmy.mdl
to add IPC senders for pitch (longitudinal was already being broadcast),
- modifying the top levels of the test mass QUAD suspension front-end simulink models,
    /opt/rtcds/userapps/release/sus/h1/models/
        h1susetmx.mdl
        h1susetmy.mdl
        h1susitmx.mdl
        h1susitmy.mdl
to both add receivers and displace the inputs to the QUAD_ITM or QUAD_MASTER library parts down by two, to account for the new SUSPOINT inputs on those parts, and
- modifying the M0 block (and subsequent inputs to it) of the above mentioned
    /opt/rtcds/userapps/release/sus/common/models/
        QUAD_MASTER.mdl
        QUAD_ITM_MASTER.mdl
to include the new frequency-dependent matrix for feed forward filters (ISIFF).
While in there, I took the opportunity to clean up the summation of ISC-related signals into a single "ISC SUM" block, instead of what had previously been an ASCSUM block and a couple of cobbled together LSC signals.
Screenshots of the various changes are attached below.

I also have since added the beginnings of MEDM infrastructure for the filter path to the ETM overview screens (I'll complete the ITMs tomorrow). Screenshots of these changes are below as well. These have been committed to 
    /opt/rtcds/userapps/release/sus/common/medm/quad/
        SUS_CUST_QUAD_M0_ISIFF.adl
        SUS_CUST_QUAD_OVERVIEW.adl

I've not yet had the chance to update the SDF system to absorb the new channels / infrastructure. I'll do this once I complete the MEDM screen modifications for the ITMs tomorrow morning.

Also, while mucking around with the overview screen, I've 
- compactified the new abundance of violin mode filters in to a sub-screen (see LHO aLOG 42752, LHO aLOG 39539)
    /opt/rtcds/userapps/release/sus/common/medm/quad/
        SUS_CUST_QUAD_L2_DAMP.adl
- fixed a long-standing bug in the M0 (highest) Bounce & Roll mode (frequency) damping filters' input matrix.
    /opt/rtcds/userapps/release/sus/common/medm/quad/
        SUS_CUST_QUAD_M0_DAMP_MATRIX.adl

To receive updates at LLO
- svn update your MEDM screen infrastructure in the following folder,
    /opt/rtcds/userapps/release/sus/common/medm/quad/
This should be done first, because of the ol' RCG "feature" that when a new filter bank gets created, it turns on as a pass-though. Which means when you restart all of the models, the SUSPOINT GS13s are being piped straight to the output of the QUAD's DAC -- except for the master switch. So, it's convenient to have a functional MEDM screens to turn off the gains or the outputs or something of this new filter infrastructure before you restore the suspension to nominal damping operation.
- svn update the common corner of the suspension models,
    /opt/rtcds/userapps/release/sus/common/models/
which will import the new QUAD and QUAD ITM MASTER library parts.
- svn update the H1 corners of the ISI and SUS repos, such that you have examples of what needs to be done at the top levels of the L1 models,
    /opt/rtcds/userapps/release/sus/h1/models/
    /opt/rtcds/userapps/release/isi/h1/models/
- Update the top level of the isi and sus ?tm? models to add the senders and receivers, respectively of the SUSPOINT pitch (and longitudinal) signals. You'll also need to shift all of the inputs to the main library parts down by two, or else you'll get terrible mis-mapping of signals.
- Reconcile your SDF system for the QUADs and associated ISIs
- Bring your guardian managers of the QUADs and ISIs to SAFE and (at least) ISI_OFFLINE, respectively.
- Compile, install, and restart your models
- Turn off the new ISI FF path, before restoring the output of the suspension.
- Accept all new channels and changes to the output into the SDF system.
- Commission the new FF system at our leisure!

Daniel, Thomas, Stefan

- We added a button to the TCS-HWS medm screen that plots the contour difference between a reference time and the current time. For now the reference time is taken from /opt/rtcds/userapps/release/tcs/common/medm/referencetimeITMY.txt

- We really want a streaming version of this working, but the current streaming scripts are inoperational (and some of them interfere with the EPICS logging)

- With that we did a quick shot with the CO2 laser - it didn't look good... (plot 1) It almost looks like CO2 heating fringes... we will have to go back to this.

- We next wanted to verify that none of this is the fault of the HWS, so we did a ring heater run. That one looked reasonably well (plot 2), both in circularity and centering (note that the centering is sensitive to SR3 alignment).

- The optics are now cooling down over night - we will look at the CO2 heating again tomorrow morning.

This morning, in response to FRS 10964 (alog 42658), I switched the ETMy OpLev laser back to the original diode laser.  To aid in diagnosing the abnormally high electronics noise seen in this OpLev, I switched the gains back to their original settings in a staggered fashion:

• 17:13 UTC (10:13 PDT): Switched 12dB whitening gain OFF
• 17:29 UTC (10:29 PDT): Switched 6dB whitening gain OFF
• 17:44 UTC (10:44 PDT): Switched back to the original diode laser

After switching back to the original laser, I turned the 3rd whitening filter stage OFF and deactivated the 3rd stage of digital compensation; this change was accepted in SDF.  Not knowing the state of ETMy alignment, I did not re-center the OpLev; this can be done at a later date either when necessary or when a good alignment for ETMy is acheived.