J. Kissel, J. Betzwieser, M. Wade, C. Cahillane

After finally getting the LHO GST-LAL, GDS, Low-Latency pipeline up and running, we noticed that the reconstruction of DARM_ERR and DARM_CTRL that is shipped over to CAL-CS and stored in the frames was corrupted by numerical precision errors. We had been using the same whitening filter that's used for the DELTA L RESIDUAL, CTRL, and EXTERNAL channels, i.e.
    zpk([1;1;1;1;1],[100;100;100;100;100],1,"n")
but that *over* whitens DARM CTRL and DARM ERR. As such, I've changed the filters to what LLO has installed, a less aggressive,
    DARM_ERR = zpk([10;10],[100;1000],1,"n")
    DARM_CTRL = zpk([5;5],[500;500],1,"n")
I attach two sets of plots:
(1) Bode plots of the new filters, and
(2&3) Courtesy of Joe, Comparison between the raw ASDs of DARM control for a recent lock stretch, and that which is whitened with the above mentioned filters.

We wanted to install the proper parking beam dump on HAM2 before we go high power, and the arm gate valves are closed right now so it seemed like a perfect opportunity. We tried, and failed.

1. Assy in question

https://dcc.ligo.org/LIGO-d1201430

It doesn't show the viewport, so it's not clear how it is supposed to be mounted, except that three hooks grab the outer edge of the flange.

2. The problem

The beam comes out of HAM2 at a steep angle (this picture was taken when we installed the non-final version back in Jan. 2014).

The new beam dump assy seems to be mounted on top of the viewport protector assy. The problem is that the protector is too high, which is kind of obvious from the above picture. For the non-final version, we had to cut the protector ring so the beam comes out of the ring, and a beam dump outside received the beam. We cannot do that with the final beam dump, and as far as the dump sits on top of the protector assy, the beam will hit the protector assy.

As of now, there is a large space between the bottom of the lexan plate and the viewport. There seems to be two versions of the protector assy, one higher than the other, and we're using the higher version. However, even if we use the lower version, it seems likely to me that the beam still does not come to the steering mirror inside the beam dump assy. It should be possible to modify the viewport protector assy such that the lexan plate sits almost flush to the viewport while the beam dump optic table sits almost fluch to the top of the lexan plate. It seems as if this works, but I'm not 100% sure without CAD.

Or maybe it is designed to be used without viewport protector assy. I couldn't find drawing that shows how it's mounted, so I'm asking around.

I told Kyle to hold off to open the gate valve in a hope that somebody offers me an easy solution by tomorrow morning.

Dan, Fil

Today we swapped R23 on the AS_C transimpedance board (D1001974) to increase the gain on the single-ended sum output that is used for the triggering of the HAM6 shutter functions (the fast shutter, and OMC PZT shutter).  The old resistors were 424 ohms, the new ones are 26.7k.  The increase of 63x compensates for the reduced transmission of the OM1 mirror (was 5%, now is 800ppm - half of which goes to AS_C).  The new resistor value allows us to keep the same threshold setting for the shutter logic: 1W into the chamber results in 400uW on AS_C, with a 1k transimpedance resistor and 80% quantum efficiency the threshold should be (400e-6 * 0.8 * 1000 * 26.7e3/4.99e3) = 1.7 volts.  The maximum threshold we can set via Beckhoff is 2V, so this is a good fit.

As Rich pointed out the last time we made this swap, it would be better to account for changes in the gain at the input to the shutter controller, so that the max output of the AS_C sum channel (10V) matches the full range of the PD (10mA).  Unfortunately the Beckhoff readbacks of the PD input to the shutter controller are picked off before the input gain stage, so if we adjusted the gain on that board, the shutter threshold setting in EPICS would not longer agree with the PD input channel.  This feels like a bigger source of confusion than the signal loss we suffer from changing the gain on the output from the transimpedance box, so we've stuck with the old kludge for now.

Also since the timing of the Beckhoff readbacks violates causality in strange ways, it would be good to have a 16k RCG readback of this single-ended sum output from AS_C.  There's a spare channel in the PD interface box on ISCT6.

I have updated the E-traveler for this board (S1301506).

With this change the HAM6 shutter should be functional with the new OM1, we will test it tomorrow to be sure.

Kyle, Gerardo 

Finished bolting HAM6 East door -> Helium leak tested door outer O-ring -> Began pump down of HAM6 -> Pumping HAM5/6 annulus with pump cart at each pump port

On the DTS I was able to reproduce LSC receive errors for RFM IPC channels sent by ISC-EX when changing the run time of the ASC model.

The set-up is: The DTS has a full set of nodes on the X-ARM RFM loop (LSC - SUSEX - ISCEX - OAF - ASC - LSC). I’m running H1’s h1iscex model on x1iscex, it has two RFM sender channels. I’m running H1’s h1asc model on x1asc0. I’ve modified your h1fe3tim16 model on x1lsc0 to receive the two h1iscex channels. So a 16kHz sender, a 16kHz receiver with a 2kHz system in between.

The h1asc model has 8 sending channels. I can vary the usrTime on h1asc by either running with no filters defined, or with the full set of H1 filters defined or with something in between. When transitioning between the two extremes, it takes the ASC about 50 seconds to do so due to the large number of filters and the usrTime varies linearly between 87uS (no filters) and 133uS (all filters). When the usrTime is in the range 105-120uS, the LSC receive errors shoot up into the thousands per second.

I then customized the H1ASC.txt to the number of filters needed to keep h1asc usrTime at 113uS (in the middle of the bad range) and the LSC gets 2046 errors per second, close to the ASC processing rate.

So it looks like the ASC IPC writes are colliding with the ISC writes. This is not quite the same as the H1 fix I put in yesterday, were I reduced the ASC RFM payload size from 16 channels to 8 channels and zeroed the LSC errors, but the mechanism could be the same.

08:18 Jeff B. to LVEA, check desiccant cabinet, recover contamination control supplies
08:46 Joe to LVEA, check forklift battery, eyewash stations
08:46 Filiberto to LVEA, work on LTS dewpoint sensors by H2 enclosure
08:55 Jeff B. done
09:11 Richard to LVEA to check on Filiberto
Richard back
09:40 Corey to LVEA to pick up supplies from HAM6 work
09:41 Peter K., Jeff B. to H1 PSL enclosure to work on FSS servo
09:44 Delivery from Platt electric
09:45 Corey to H1 PSL enclosure with Peter K. and Jeff B.
09:50 Doug to LVEA to take picture of optical lever installation
John and Bubba craning platform in LVEA
09:58 Elli and Nutsinee to end X, HWS work
10:06 Betsy and Andres to LVEA to pick up equipment in east and west bay
10:07 Doug back
10:10 Richard looking at H1 PSL air shower
10:16 Richard done
10:27 Joe done
10:34 Corey done
10:51 Bubba, John drilling holes for platform posts in LVEA
11:12 Gerardo and Kyle starting pumpdown of HAM6
11:13 Travis and Sudarshan to end X, PCAL
11:56 Betsy running undamped transfer functions on PR3 M3, undamped spectra on SRM, SR2, SR3
12:11 Hugh deisolating HAM1 HEPI
12:28 Elli and Nutsinee back
12:40 Vending machine truck through gate
12:54 Tours in CR
13:38 Peter K. out of H1 PSL enclosure
13:58 Corey to squeezer bay, checking for equipment to install beam dump on HAM2
14:43 Jason and Ed to take voltage reading from field box near H1 PSL
14:53 Jason and Ed done
14:56 Bubba back to LVEA to continue drilling holes for platform posts
15:12 Corey and Keita climbing on HAM2 to test fit of beam dump
15:35 Peter K. to H1 PSL enclosure
15:43 Suresh to end X to set whitening filters on optical lever laser
16:30 Corey and Keita done

I took away the state selection and some other stuff that wasn't needed and added a total of 4 message boxes. If there are ever more than 4 messages, a small, black, blinking box will appear around the ALL button to aleart you to click, ALL.

I found damped TFs which Kissel took last Dec when he imported the LLO HLTS filters and applied them to the LHO HLTSes 15730.  Today, I ran the undamped set.  Attached are results and comparisons with other HLTSes.  I think this closes the PR3 acceptance measurement dedt.

Ed, Jason, PK

Eearlier this afternoon the ISS AOM Diffracted power was up to ~39%. Reason, still unknown except for folks were in the enclosure working on FSS. Peter King set the refsignal to -2.5V and brought the diff power to ~8.5%. It's seems stable for now. We'll see.

Nutsinee, Elli

We have aligned EX HWS to within ~+/-5cm from the ETMX image plane.  This is as good as we can get with the current level of motion in the HWS images.  To locate the image plane more acurately, the next step is to understand why there is this variation in the HWS images.

The distances on the HWS path are now:

L1-L2: 610mm (unchanged)

L2-L3: 877mm

L3-HWS:  1593mm

To do this we applied a 0.05mHz, 2microrad yaw excitation to the M0 optic align filter bank, and looked at the motion of the x centroid in the HWS images.  We moved L3 51mm towards L2, the HWS 57mm away from L3, and mirrors M7 and M8 each 80mm to the right to the lengthen L3 to HWS path.  We also adjusted the pitch of ALS-BS 1 to lower the beam on HWS-M1, which was clipping slightly.

See the attached pdf with L4C and IPS TFs from LHO and LLO--look at title and legend.  Main interest here is above 10hz.  We are looking into why.

The filter switch widget in the SDF doesn't jive with diff list.  See attached.  There is no setting difference but the widget suggests there is.

Firefox became too slow to start from controls account. That's because Jamie's firefox script copied all cache files (over 900Mb) and thumbnail files (over 600 Mb) to a temporary directory each time firefox was launched.

That's not all Jamie's fault, because firefox people changed the cache file naming convention in the past ('cache2' instead of 'Cache').

Anyway, Jamie's script was changed such that '[Cc]ache*' and 'thumbnail*' are excluded from rsync, and now firefox starts in a second or two. Also I deleted all caches and thumbnails. The size of firefox controls profile folder was reduced from about 1.7GB to 73MB.

Jamie's script is necessary for controls account because of the lock file management of firefox. Without the script, only one workstation at any given time can launch firefox as controls without manually deleting lock file.

Summary:

Elli and Nutsinee measured the current state of the ETMX HWS layout and the behaviour of a beam transmitted through it. From this I was able to solve for unknown distances in the layout and produce a new optical solution that will image the ETM onto the HWS with the existing lenses.

Diagnosis:

Elli and Nutsinee made some measurements of the ETMX HWS yesterday to determine the conjugate plane. We started with the following measurements:

• lens L1 focal length = 2.2361m, [CVI PLCX-50.8-1030.2-UV-532]
• lens L2 focal length = 2.2361m, [CVI PLCX-25.4-1030.2-UV-532]
• lens L3 focal length = -0.5589, [CVI PLCC-25.4-257.2-UV-532]
• L1-L2 distance = 610mm
• L2-L3 distance = 928mm
• L3-HWS distance = 1401mm

Also, we know the following measurements of the transmon telescope (Data from T0900385 - v06 and G1100873):

• ETM to Transmon Primary Mirror = 1429 mm
• Primary Mirror ROC = 4000 mm
• Primary to Secondary Distance = 1902.6 mm
• Secondary ROC = -200 mm
• (Nominal) Secondary mirror to HWS Lens L1 = 5273 mm (this one is the problem).

This list of values allows us to calculate the ABCD matrix propagating the beam from the ETM to the HWS.

Elli and Nutsinee injected a 2 urad oscillation into ETMX YAW at 0.05Hz. This results in a peak-to-peak oscillation of 8 urad on the reflected beam (pk-pk multiples by a factor of 2, reflection multiplies by another factor of 2). They then measured the pk-pk oscillation of the centroid of the return beam on the HWS in pixels, where 1 pixel = 12 um, for different positions on the HWS. At the conjugate plane, this should be zero.

• For L3-HWS = 1401mm, centroid oscillation = 7.629 pixel pk-pk
• For L3-HWS = 1351mm, centroid oscillation = 6.648 pixel pk-pk

These numbers didn't jive with the nominal optical layout. The only real uncertainty was the distance "Secondary mirror to HWS Lens L1". So, using the values that for the oscillations, I readjusted this value in the ABCD matrix, modeled the expected centroid oscillations and compared the measured oscillations.  "Secondary mirror to HWS Lens L1" = 5728mm gave a result that agreed to better than 1%. 

New Optical Solution:

So now that we "know" the values for all distances, I re-solved for an optical solution that would image the ETM onto the HWS with a demagnification of -20.5x with the existing lenses. The new values are:

• L1 to L2 = 610 mm
• L2 to L3 = 877 mm
• L3 to HWS = 1499 mm

The calculations are attached.

Diagnosis calculation

New solution calculation

XorYDist = (nominal) + (unknownX) = 5.273m + 0.455m

An Fscan of H1:CAL-DELTAL_EXTERNAL_DQ during the April 2 lock is here:

https://ldas-jobs.ligo-wa.caltech.edu/~pulsar/fscan/H1_DUAL_ARM/H1_DUAL_ARM_HANN/H1_DUAL_ARM_HANN/fscans_2015_04_02_06_00_02_PDT_Thu/H1_CAL-DELTAL_EXTERNAL_DQ/fscanH1_CAL-DELTAL_EXTERNAL_DQ_H1_1111928417_1112014818.html

(Clink on the "old" button to see the plots.) The Fscans are used to look for lines in the data.

A 4 Hz comb is still seen in the power versus frequency spectra, and a number of wandering lines are seen in the time-frequency spectrograms.

Dan, Jeff, Keita, Koji

As part of the closeout activities for the HAM6 work, we have taken TFs for all the suspensions in the chamber.  The OMC-SUS, OM2 and OM3 all checked out fine -- current TFs are the same as the Phase3B measurements that Stuart took last week.

OM1 has changed, mostly in yaw.  The frequencies of the fundamental modes in longitude and pitch have shifted slightly -- this is not surprising since we have slightly changed the mass (the mirror) in the pendulum, and the change is small enough that it doesn't prevent our acceptance of the chamber for closeout.

In yaw, the fundamental mode has split into two high-Q peaks.  The mechanism for how this could happen, while not changing the peak structure in pitch or longitude, is not clear to us.  Rubbing would seem to be ruled out.  The eddy current dampers are a possibility, and the symmetry of the system (two dampers oriented horizontally, aligned to the middle of the optic in the vertical direction) seems to imply that a misalignment of the dampers would primarily couple to yaw.  One leading, somewhat unsatisfying hypothesis is that one of the dampers is too close to the mirror holder, and this is coupling the transverse (side-to-side) mode into yaw.  (But, why not transverse to longitudinal?  And why is this coupling coherent with an excitation in yaw?)  Another idea is that the flags have become misaligned relative to the BOSEMs in just the right way to couple this new mode into yaw.  But, this suffers from many of the same complaints as the first idea.

Regardless of the mechanism, the motion is yaw is well-damped using the local damping loops.  We think this means the motion will not be a problem during low-noise operations, and we are comfortable closing out HAM6.  We can fix it during a subsequent vent.

The attached figures are:

Fig 1 - comparison of OM1 Y-->Y TF with Stuart's measurement from last week.  Note that the Apr 3 measurement was in vacuum, today's measurement is in air with the HAM door on.

Fig 2 - comparison of Y-->Y TF for all OMs (top), comparison of the three DOF TFs for OM1 (bottom).  The bottom panel is meant to illustrate that we are not coupling a pitch or longitudinal mode into yaw.

Fig 3 - comparison of undamped/damped spectra for all OMs.  Dashed references are undamped, solid lines are damped.  Differences between the traces for each OM are due to uncompensated gain differences in the BOSEMs - this has been fixed since the plot was made.  The motion in OM1 yaw, quite large when undamped, is about the same as the other OMs when the damping loops are engaged.

Figs. 4, 5, 6 - A yaw excitation impresses the low-frequency peak onto the other DOFs; the structure is evident in all four BOSEMs; the low-frequency peak is present in a L-->Y TF.

Let's bolt on the door and pump down the chamber.

After the ISC crew were finished, Jim unlocked the ISI and made a small adjustment to it's balance.  I wiped a few of the surfaces of the ISI and chamber that I could reach near the door.  I placed a witness plate on the ISI table in the middle of the sea of ISC mirrors nearest the east side of the table.  Kyle, Gerardo, Bubba and I then tacked the door on with 4 bolts.  Kyle turned off the purge air.  Dan/Keita and Jim are running OM/OMC and ISI TFs to see how we fair for pumping down tomorrow.  We're not yet out of the woods.  TBC...

PS Particle counts at the door interface while the door was going back on was

0.3um  10

0.5um  10

1.0um   0

Attached are the Phase 3b damped and undamped TFs of SR3 taken over the last few months.

The damped TFs of M1 are pretty squashed, but this is because there is some pretty heavy damping filtering engaged for commissioning.