Jamie, Kiwamu

We tried engaging isolation loops in the stage 1 of ITMX ISI, but kept failing for some reason. In both robust- and high- isolation cases, we saw L4Cs ringing up at 10 Hz at the end of the auto-isolation processes and tripping the watchdog.

Checking each step of the guardian isolation process, we finally found that the instability was due to the X isolation loop. In particular, every time when it engages the 3rd boost, this makes the system unstable. Even though, we don't really know what the loop shape is for the X direction, we tried different gains in order to find a OK-stable gain. We ended up with a gain factor of 2. This seems to give us a OK-ish ISI performance in ITMX ISI. We did not try engaging the stage 2 isolation loops.

J. Kissel

Yet more measurements of charge -- I've taken two more data points on H1 SUS ETMY and one on H1 SUS ETMX. 

I managed to start and complete the measurement of ETMX exactly as the EX Ion pump was turned on and the arm was opening. Note, that the green laser had been injecting ~30 [mW] into the vacuum system since last night. The results show little-to-no charge, clustered around zero. We'll get a few more measurements over the next day to see how an open arm, with green light is affected. Remember, we didn't see much charge change between ion pumps valved in vs. valved out, or with green light for ETMX.
      P [V]       Y [V]
UL   42.6    41.6
LL   46.4    40.9
UR   40.3    20.6
LR   28.9    30.6

For the two ETMY measurements, the configuration has remained identical since Tuesday Aug 26, namely with the ion pump open. Indeed, as suspected, the three measurements over the past 48 hours show significant variation in charge, unlike with the pump valved out. I'm still waiting for Borja to get me the previous data in an easily digestible form so I can make a nice comparison plot over time, but for now, I just report the charge over time via the table below.
Pitch
       2014-09-02    2014-09-03    2014-09-03
        1937UTC       1550UTC      2339UTC
UL [V]    104          101           119
LL [V]    129         -5.51         -15.9
UR [V]    110         -22.4         -76.6
LR [V]    104         97.8          101

Yaw
       2014-09-02    2014-09-03    2014-09-03
        1937UTC       1550UTC      2339UTC
UL [V]    149          160          196
LL [V]    72          13.3         -13.1
UR [V]    13.6        -32          7.17
LR [V]    117         112          91.4

(Alexa, Sheila, Keita)

• X-arm locked on green (see Keita's alog 13740 for more detail on alighment).
• We had to realign Green REFL B on ISCTEX.
• We plugged the 24.407079 MHz RF oscillator source back to the RF Amp, with the demod phase at 120.6deg. The ifr was set to some wrong value, and we didn't bother resetting it for now. For reference, the HIFOX configurartion was Mod Freq: 24.407363 MHz, Demod phase: 120.7 deg (alog 10227).
• COMM beat note was well aligned on ISCT1, we measured 1.5Vppk.
• DIFF and COMM demod controls cable was plugged into the wrong slot in the back of the concentrator. (We moved from slot 3 to slot 1).  This is now consistent with the pull list.
• There is a LO mon error for COMM demod. Will investigate tomorrow.

Now they output negative signal when there's light. That's opposite of what I remember, and that's opposite of ITMX baffle PDs.

This is to get a good reference for the corner station alignment.

Prcedure:

Align TMSX, then ITMX, then ETMX, using arm baffle PDs.

Then steer PR3 to get the green transmission in ISCT1.

Then align IR light using PR2 and IM4/IM3 to the arm.

Results:

Green beam locked to the arm, and the green beam was steered to the right location in ISCT1 using PR3.

Good alignment slider values are:

PR3 slider offset used to be (P,Y)=(-279, -80) until this morning, now it's (-245, -178.9), the difference is (34, -98.9) urad.

Due to ITMX motion, the ITMX number might be off by a few urad, and the errors in PR3, PR2 and IM4 should follow ITMX via some ABCD matrix.

The message is still the same as before (without TCS, the mode matching to the OMC in single bounce is expected to be worse for ITMX than ITMY), but for posterity I add here the alamode files I made with the mode propagation from the IMC to the AS port for ITMX and ITMY single bounce configurations. The IO parameters come from previous measurements and models that Paul did. 

~1100 - 1200 hrs. local
Valved-in IP12, valve-out X-end turbo, dumped GV20's unpumped gate annulus volume into pump cart and opened GV20  

~1200 - 1315 hrs. local 
Valved-in IP1, IP2, IP6, valved-out YBM turbo, valved-out XBM turbo, dumped GV7's unpumped gate annulus volume into pump cart and opened GV7

08:41 Hugh to HAM5 to unlock HEPI
09:07 Fire department on site
09:24 Jason starting optical lever work
09:43 Hugh out of LVEA
10:08 Betsy and Travis to quad test stand
11:30 Jeff K. to end Y to check status of illuminator
~13:00 Karen to mid Y
13:30 Fire department on site
13:40 Karen done at mid Y
13:50 Richard and fire department to LVEA
14:05 Richard and fire department out of LVEA
15:00 Safety meeting

- X arm opened (WP 4826)

At the request of local commissioners, I have modified the SUS guardian code to go back to handling the setting of alignment offset values for the ALIGNED state, and have re-introduced the MISALIGNED state which sets the alignment offset to saved misaligned values.  This all was originally in the SUS guardian code, but was removed at LLO due to precieved problems with the SUS guardians handling the offsets.  The LHO commissioners, on the other hand, prefer the SUS guardians to have separate ALIGNED and MISALIGNED states.

SUS guardians have not been restarted yet.  Only SR2 for testing.  We're waiting until after the arm alignment tests are done.

NOTE: ALIGNED and MISALIGNED offset values MUST BE SAVED in order for the SUS guardians to restore to them! This is the big change by going back to this configuration.  When the SUS guardians were not touching the offsets, there was no worry that they would accidentally be reset.  Now that they are being set by guardian, we have to remember to always save the offsets appropriately after tweaking them.  Forgetting to save them could result in the current alignment being lost if the suspension trips or the alignment state is cycled.

As a bonus, the SUS guardians will use the notification USERMSG to notify you if the alignment has been changed and not saved.

Here's the current SUS state graph:

This is all just a stop-gap until we get a better comprehensive solution to the alignment offset problem.  This will likely be discussed in the next SYS meeting, on 9/9/2014.

I noticed that the anti-whitening filter for the LF signal of the resonant RFPDs were all consistently inaccurate. So I fixed it.

before: zpk([2.4], [0.24], 1, "n")

after: zpk([2.4], [0.2], 1, "n")

I did this for all the resonant RFPDs, including REFL_A, REFLAIR_A, POP_A, POPAIR_A,  ASAIR_A and IMC_REFL. This seems to have fixed the issue of ASAIR_LF going below zero watts.

I have updated the GUARD_OVERVIEW screen to reflect the addition of the new node OK bits.  The change can be summarized with the new legend.  Note the third indicator on the right:

The last indicator will be green if the node is OK, or orange if it is not.  This change has been propogated throughout the GUARD_OVERVIEW screen:

Note that most all SUS and SEI nodes are OK, while we still need to set NOMINAL stages for the ALS and IAS ("initial alignment system") nodes.

The white panels are nodes that are currently being commissioned, and should be available imminently (the H1 "IMC" node will be renamed to "IFO_IMC" to match LLO).

Unlocked the WHAM5 HEPI and then ran Range of Motion measurement.  1mm for most DoFs, V1 0.9 & V4 0.8mm.  Elected to not run linearity test, looked good in April.  Attempted to run through commissioning scripts to close the position loops but there is a bad L4C and I'll need to replace that and rerun the transfer functions.

Meanwhile I updated the Cartesian position targets to the more recent stopped time, although these saw only small changes since I updated the targets in July.  And, I've driven the outputs (Cartesian basis in the Isolation filter section) to the target positions.  The rotational DoFs are all within a few 100nrads and the translationals are within a few umeters.  Feel free to tweek the OFFSETs on the Isolation filters if further tuning is needed.

model restarts logged for Tue 02/Sep/2014
2014_09_02 11:27 h1fw1

unexpected restart of h1fw1

K. Venkateswara, J. Kissel

I'm starting to look at how ground rotation affects the ISI stages and what improvements could be achieved with rotation sensor. I considered the data taken on Aug 21, 2014 night. I took three 10k second sets starting from GPS time: 1092707816. Initially I took the following channels:
H1_ISI-GND_STS_ETMX_X_DQ
H1_ISI-GND_BRS_ETMX_RY_OUT_DQ
H1_ISI-ETMX_ST1_BLND_X_T240_CUR_IN1_DQ

I then made ASD and coherence plots as shown in the attached file for each of the three sets. The seismometer velocity data has been multiplied with (2*pi*f)/g to convert to units of radian. Stage 1 had the "TBetter" isolation filters (see alog 11400). HPI was in the 'floating' state.

In the first data set, the ground rotation was quite small and BRS/GND_STS are very quiet in the 10-100 mHz range. Stage 1 however has very large motion in this range. There is no coherence between Stage 1 and BRS/GND_STS up to ~0.2 Hz, after which it shows strong coherence with the GND_STS.
In the second and the third data set, ground rotations were larger as seen in the BRS/GND_STS channels, but Stage 1 performance and the coherence looks identical to the first. It is clear that in this state, Stage 1 noise (in the 10-100 mHz band) is not due to ground rotation.

The final plot shows the coherence between Stage 1_T240_X and Stage1_T240_RY, HPI_IPS_X and HPI_IPS_RY showing very good coherence below 0.2 Hz.

After discussing further with Jeff, and reading P1200040-v40 (by F. Matichard et al, very useful document!), it seems that the above result is expected when the HEPI is in a floating/open loop state. Please look at pages 28-30 which states that HEPI tilt-horizontal coupling corner frequency is ~0.2 Hz when it is in an open loop state.

Here's my rough interpretation of what is happening: Stage 0 and subsequent stages can be thought of as a boat floating on pool of water (or hockey puck on an air-bearing table). If you stand inside the boat and move a table on the boat, the entire boat will move w.r.t. the walls of the pool, from momentum conservation. In addition there is known tilt-horizontal coupling below 0.2 Hz, so linear motion of the boat produces rotation below 0.2 Hz. As the TBetter isolation loop is trying to isolate down to 50 mHz (and amplifying motion in the 10-100 mHz band), the net result is large rotations of Stage 1, which have nothing to do with ground rotation. 

Where is this excess noise coming from? I'm not sure, but there are a number of candidates: it could be displacement sensor noise of Stage 0/1, actuator/force noise of HEPI, thermal gradients across HEPI springs and so on. Regardless of the noise source, I feel that HEPI floating (with fc~0.2 Hz for rotation) and TBetter filter (isolating to 50 mHz) is probably not the optimum configuration for good low-frequency isolation, given that we may now be measuring ground rotation precisely.

I'm very new to this so I hope others will correct me if I'm mistaken. It would be very useful to try out the HEPI closed-loop configuration and HEPI locked rigid states with BRS measuring ground rotation to understand this issue more clearly.

Jeff K. working on charging measurements at end Y
X arm opening temporarily
in vacuum SRM noise investigations
cable pulling for cameras and illuminators
3IFO assembly work
Jason to align PR3 optical lever, then work on HAM4
Jim W. working on ITMX SEI control loops
Fire department taking RAFAR temporarily out of service to allow cable pulling
Safety meeting at 3

PeterK, RickS

Yesterday (Tues) morning, we adjusted the alignment into the PMC to minimize the reflected light level with the cavity locked.
We adjusted the two steering mirrors immediately upstream of the PMC.  Neither was locked before, but we locked them this time when we finished.
Maybe this will help with the alignment drifts we have been experiencing.  Min. REFL DC level was 0.375 V locked, 2.3 V unlocked.

We then did a cursory alignment into the reference cavity and saw a significant improvement using the two mirrors downstream
of the PMC (again, neither of which was locked).  Trans. level 2.08 on MEDM screen.
We looked at the FSS OLTF and adjusted the gains to 15/30 dB Fast/Common.  Even with the common gain at the max level of 30 dB,
the UGF is only about 320 kHz.  Phase margin about 50 deg.

We installed the insulation end caps on the reference cavity.  This may provide some passive improvement in the temperature stability.

With JeffB (operator on shift) we adjusted the ISS reference level to 2.00 to account for the increased light level resulting from the PMC
input alignment and locked the ISS loop.  The diffracted light level was about 7.4 W and the loop appeared to be functioning properly.
This reference level will have to be adjusted to keep the diffracted light level in the 6-10 W range if the PMC alignment drifts (reflected
light level increases).  0.01 on the reference level is approximately 1% (maybe 1.3%) on the diffracted light level.

We need:
AA batteries for the computer keyboards and trackpads.
15 ft. SMA/SMA cable for the RFPD to TTFSS servo module cable so we can move the module to the floor (and get this heat source off
of the table).
10 ft. SMA/BNC cable for the ISS AOM to AOM driver cable so we can move the module to the floor and get this heat source off of the
table surface too.

I did some math to figure out how much ITMX, ITMY and BS may have been moving (in a frequency band of 0.1-1 ish Hz) in their angles according to fluctuation of the DC light observed at the dark port when the Michelson was locked.

(Summary)

Wednesday night (August 27th)

• The dark port DC (ASAIR_B_LF) flucutated by 16% of the bright fringe when the Michelson was locked on a dark fringe.
• ITMX(Y) may have been moving by 11 urad or 22 urad in peak-to-peak.
• BS may have been moving by 5.7 urad or 11 urad in peak-to-peak.

Friday night (August 29th)

• The dark port DC (ASAIR_B_LF) flucutated by 1% of the bright fringe when the Michelson was locked on a dark fringe.
• ITMX(Y) may have been moving by 2.8 urad or 5.6 urad in peak-to-peak.
• BS may have been moving by 1.4 urad or 2.8 urad in peak-to-peak.

(Some math behind it)

Suppose the Michelson is locked on a dark fringe. If an ITM is misaligned by Ψ, this introduces a displacement and tilt in the reflected beam with respect to the one from the other ITM at the BS. The displacement is x = 2 L Ψ and the title is ϑ = 2 Ψ where L is the distance from the BS to the ITM. So we get a small amount of 01 or 10 mode at the dark port on top of the 00 modes. Since the effect on the resultant 00 mode in its power is proportial to 4th power of the displacement and tilt, we assume the 00 mode to vanish because of the locking loop. The only residual we obtain at the dark port is the 01 or 10 mode whose field can be written as

E₁₀ = 1/sqrt(2) ( x/w₀ + i * ϑ / ϑ₀),

where w₀ is the waist size and ϑ₀ is the divergence angle respectively. A factor of 1/sqrt(2) upfront comes from the BS reflection. If we plug the definition of x and ϑ into the equation, we get

E₁₀ = sqrt(2) ( L/w₀ + i / ϑ₀)  Ψ.

Squaring the above, one can get the dark port power as

P = 2 ( (L/w₀)²+ (1/ ϑ₀)²)  Ψ²

Note that P is already normalized by the input beam power or equivallently the bright fringe. The Rayleigh range of the beam around the BS is roughly 210 m (if my math is correct). This gives a waist size of 8.4 mm and divergence angle of 40 urad.  The ITM-BS distance L is about 5.34 m where I averaged out the Schnupp asymmetry. So the dark port power can be now explicitly written as

P = 1.24 x 10⁹  Ψ²

This is the equation I used for deriving the numbers listed at the very top.

For example, if one wants to explain a 16% DC light fluctuation observed at the dark port by an angle deviation in ITMX(Y), the misalignment should be Ψ = sqrt(0.16 / 1.24 x 10⁹ ) = 11.4 urad. In the case of the BS, the effect gets twice bigger due to the fact it affects both X and Y beams at the same time in a constructive manner.