I'm finishing up my last set on PRM (HAM2).  Jim is going in to start work on rebalancing the HAM3 ISI.

Alexa, Stefan,

To verify that the Green QPDs on TMSX all have comparable segment gains, we steered the beam on all 8 segments. The whitening gain was set to 30dB.

QPDB:
Seg 1  14060 cts
Seg 2  14000 cts
Seg 3  13900 cts
Seg 4  14300 cts

QPDA:
Seg 1  15900 cts
Seg 2  17500 cts
Seg 3  17100 cts
Seg 4  16200 cts

QPD A was somewhat more difficult - depending on the QPD B pointing, the beam can clip before reaching the peak value on QPDA.

The gains are withing 10% of each other.

We need a single DOF TF on each of the optics we are cleaning next week in order to run a comparison against  "on the fly" after pulling the First Contact.  So, I am going to run a few this morning to make new references.  I'll take MC1, MC2, MC3, and PRM data.

The ITMx TFs failed last night so we are running them with DTT right now.

J. Kissel, S. Aston

I've installed damping filters using exactly what L1 is using (see LLO aLOG 9724). It may not be the most awesomely tuned filters and gains, but it's good enough for now.

L = -6.00
T = -3.00 
V = -5.00 
R = -0.40
P = -1.50
Y = -3.00

Each DOF should have filter banks FM1(0:30,30), FM5(normX) and FM10(Ellip50) engaged. (Note there are other experimental filters lying around in some FMs of some DOFs. Try not to get confused. When we advance to Level 2 filters, we'll clean these up.)

The second loop power stabilisation photodetector array was installed in HAM2.  Unfortunately the beam
used for the alignment of the photodetector, the reflected beam from MC3, had the wrong polarisation.  One
consequence of which is that the expected signals levels on the quadrant photodiode were a lot lower.  This
will not be the case when the real beam into the interferometer is restored.

    The beam into the array was aligned through through the input alignment aperture such that the output
of the eight individual photodiodes was peaked.  In the end the output of one of the eight diodes was 
somewhat lower than the others.  This is not because of a problem with the photodiode but because the
input beam alignment was not optimised.  The parameter space of one end of the input beam alignment can
be explored by using the Picomotor driven mirror - the last mirror in the optical path to the photodetector
array.

    More details to follow in a later entry.

J. Gleason, P. King, R. McCarthy, O. Puncken, R. Savage

[Jeff B, Andres R, Koji A]

Prior to the OMCS TF measurements...

The OMC was fully cabled up (i.e. OMC DCPDs, QPDs, PZTs, DCPD Preamps, 5 BOSEMS)

The height and balance of the OMC glass breadboard on the OMCS were adjusted.

Then BOSEM positions were adjusted so that the flags are centered in the aperture.
The BOSEM longitudinal positions were adjusted so that the flags blocks half of the light.

A. Pele, J. Kissel

After I had quickly glossed over the front-end model for the H1SUSOMC and declared it error-free, I sent Arnaud and Fil on a wild electronics goose chase for two hours before Arnaud came back in and found the flaw in the digital system in 5 minutes (cf LHO aLOG 8563). Mea cupla, gents. 

We've fixed the error in Simulink, and now that transfer functions are complete (see LHO aLOG 8565), I've re-compiled, re-installed, re-started, and re-stored (4R'd) the SUS to functionality. Further, I've confirmed the BIO monitor / readbacks work like a charm.

Next time, I'll follow Arnaud's advice: try to drive an offset out to the suspension and confirm no displacement from the OSEM speed-dials before declaring the BIO non-functional. My better judgment was polluted by too many encounters with turned off BIO drivers (which freezes their EPICs values), and the PUM driver low-voltage problem (see LHO aLOG 1278).

J. Kissel, K. Arai

After Arnaud and Fil graciously debugged the binary I/O for the H1 SUS OMC (see LHO aLOG 8563), I began and finished a round of DTT transfer functions to clear the suspended dynamics. While the frequency dependence looks great and as expected, there appears to be an anomalous scale factor of 5.5 in all degrees of freedom. Since the scale factor is identical in all degrees of freedom it implies something is Arai (teehee, couldn't resist) with the electronics or digital chain and not something inherent to the dynamics.

As such, I've vetted the digital portion of the chain:

- Measurement calibration factor (i.e. post-processing measurement analysis)
    - L1 uses 0.26193 [(m/ct) / (N/ct)] (see, e.g., LLO aLOG 9570) -- is the same as used here to calibrate this measurement, and L1's measurements match the model well (certainly not off by a factor of 5.5). Therefore the post-processing scale factor is the same (It's not that we forgot the 18 vs. 16 bit DAC factor of 4 or that we have the wrong coil driver transconductance or coil-magnet gain compensation, etc.).
    - (Not attached, but) I made the same plots but dividing the measurement by a further factor of 5.5 (exactly, which is how I got the number), and the measurements line up with L1's measurements excellently, and which all match the model to within ~1.0%.

- EPICs Values
    - The OSEMINF Calibration Filters ("to_um") are turned ON (and the maltab processing code has meas.sensCalib = true)
    - OSEM2EUL and EUL2OSEM matrices are as produced by make_susomcs_projections.m, which is what L1 uses to install their matrices.
    - No excess gain in the SENSALIGN matrix; an identity matrix as expected
    - No excess gains in the TEST filters
    - The OSEM sensor normalization gains in the OSEMINF bank are roughly 1.3 to 1.4, as expected, no gains close to 5.5
    - The COILOUTF filter magnet polarity gains are +/- 1.0 as expected
    - The DRIVEALIGN matrix is no longer in the TEST excitation path, so that can't screw us up 

- Foton Filters
    - The OSEMINF filters have the expected DC gain (10:04 has a DC gain of 1.0, "to_um" has a DC gain of 0.02333)
    - The COILOUTF filters are all engaged and all have a DC gain of 1.0
    - The DAMP filters are not installed, but they're inconsequential to this undamped measurement


In summary, I confirmed that everything digital in the transfer function path is OK, and I confirmed that all post-processing is identical. So -- I've been wrong that it wasn't a digital problem once already this evening, but -- I don't see anything that could be wrong other then the analog electronics chain. We'll investigate more tomorrow morning.

If we're really pressed for time (i.e. we must install tomorrow or man-power to install evaporates) -- after we install the damping filters (quick), and get a set of damped transfer functions (1.5 hours), I vote we punt debugging this until after install. Unless anyone knows of something obvious...

ITMX undamped TF will run overnight from ~1.50 am on opsws6, and will be running for 2*4hours (Main+Reaction chain)

[Jeff K. Fil Koji Arnaud]

Koji wanted to run some TF tonight for OMC but switching the coil driver filter state from the BIO screen wouldn't change its readback.

After Fil made sure that the analog filters were being physically enabled by the digital switch (thank you Fil), we realized the problem was on the model side, because BIO_M1_MON_In was grounded.

Since it's only a readback issue, Jeff started the TFs, and we will recompile the model tomorrow morning.

Looked at the connections to the H1 ITMX ESD starting at the airside of the vacuum feedthrough.  There is no evidence that there is any connection being made to ANYTHING inside the vacuum system.  In the process of analyzing this, I installed the Delrin insert on the airside of the vacuum feedthrough and very carefully connected the coaxial cables.  Truly no way it's connected on the inside of the vacuum flange.

(Alexa, Stefan)

• This morning we had noticed a swap in QPD A and B between in-vac and the medm screens/front end. This was due to swapped cables from the isc whitening chassis to the AA chassis (in particular cable 11, 12).  The cables are now properly connected.
• When we shined a flashlight onto the DCPDs on ISCTEX, the readouts were all mixed up. First, we noticed that the cables inside the table to the feedthrough were swapped between the Laser-IR and the Fiber Trans. We also noticed that fiber trans and rejected where swapped between channels in the system manager. This has all been corrected for. The DCPDs now follow the specifications of T1100472-v11 from feedthrough through to the software (ie: Laser-IR: 1, Fiber Rejected: 2, Fiber Trans, 3, Spare: 4).
• The BBPDs on the table were checked and are OK
• We measured the 532nm polarization to be s-pol at the bottom of the persiscope.

Summary: Some optic(s) might have moved after the TMS was moved out of the lab.

After the beam centering on QPDs is engaged, the QPDs were nicely centered but the beam spot on the 2ndary was about 3mm off centered to the left, the beam is clipped on the F2 already, and only half or less of the beam makes it to the ETM.

If the TMS ISC table is as was adjusted in the lab, the beam spot should be slightly to the right and a few mm down (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=7285), so there's a huge difference between what we expect and what we are observing now. There are two possibilities:

• We're fooling ourselves, e.g. QPD is looking at a ghost beam, imbalanced analog gain between quadrants etc.
• One or more optics moved on the TMS ISC table, e.g. something was loose, somebody bumped on the mirror holder etc.

In the latter case we somehow need to realign the green and IR beam to the telescope and recenter the green and red QPDs, all in situ. I'm working on a procedure to do this.

Checking that we're not fooling ourselves.

Sensing electronics offset? -> Exonerated (changing whitening gain didn't change the beam position though it increased the QPD signal)

Beam clipping downstream of Mad City Labs PZTs? -> Exonerated. No clipping in chamber, and the beam looks OK in chamber.

Different gains for quadrants? -> Asked Stefan to check by bringing the beam in one quadrant, one by one.

Ghost beam?

Wrong polarization (which might produce strong ghost beams)?

Identifying what moved and why.

We don't have any good idea to pinpoint what moved except that it should be downstream of the first steering mirror, and either upstream of the secondary or in the QPD path. Optics mounted at the edge of the table are more likely to be bumped.

Loose mount? -> Exonerated. Wiggled all relevant optics (except the fixed top periscope mirror on the TMS ISC table, which I overlooked and should be checked later) by touching the barrel of the mirrors using fingers, but after each attempt the beam position didn't change.

Bumping? -> Not fully rejected. I didn't apply strong pressure nor any strong impact to optics, so bumping-strongly theory is still alive.

Drift? -> Not that likely. We have observed during HIFO-Y a huge alignment drift correlated with ambient temperature (triggered by clean room on/off) and nobody knows why this happens. But at that time we were probably talking about the radius of ITM over 4km = 17cm/4km ~ 40urad drift, while this time it is 3mm on the TMS ISC table, so it's on the order of a few mrad. This is a much larger effect than the drift in the past.

• Pitch: 976 µrad down
  • Target: 971 µrad down (619 µrad optic pitch + 352 µrad FC correction)
  • Error: 5 µrad down
  • Tolerance: ±50 µrad

Jeff, Jim, Dave

Jeff found that we had not upgraded the end station sus models (quad, tms and aux) to matlab2012b on monday. He did the upgrade, we compiled, installed and restarted the models:

h1susetm[x,y],h1sustms[x,y],h1susauxe[x,y]

Prior to this change the end stations were running 2010a top level mdl files (and 2012b common models) built with 2.8. As expected, no DAQ restart was required as INI files did not change.

The IMs and MCs have been burtrestored to the automatically generated snap file from monday 11/11 at 20:00

Attached are 10 min trends of the IMs osem signals in the euler basis (Long/Pitch/Yaw) during alignment on monday, and today after restoring the snapfile.

picture 1 : Monday 11/11 from 19:50 to 20:00 PT

picture 2 : Today 11/14 from 10:20 to 10:30 PT

No major shift is seen when comparing both trends