ASAIR beam

Finished the alignment work with the ASAIR beam on ISCT6. The periscope had to be moved and the top mirror lowered. This allowed the beam to travel down periscope and hit the 2 steering mirrors on the table. Couldn't find a camera, so the beam is currently dumped.

Also installed the tube that connects the viewport adapter to ISCT6 and removed the Lexan protectors.

The feet of ISCT6 have been set down. The new table position still needs to be marked on the floor.

Alignment of the IO path on the PSL: clipping and misalignment through the IMC mode matching lenses

Images I took yesterday show that the 2nd reflection from IO_MB_L1 is clipping on the output aperture of the EOM.  The beam is clearly seen on the housing on the +Y side of the output aperture.  The beam should be going cleanly back through the EOM and be easily identified and dumped on or near the EOM steering mirror mount, such that there's no possible path for the beam to re-enter the optical system.   The images also show that the beam is not centered on IO_MB_L2, and should be centered, and the beam is currently mis-centered to the -Y side of the lens.  Both the clipping and mis-alignment show that the beam line from the EOM through the IMC mode matching lenses has not been restored to an alignment that satisfies the original design of the optical system.

Attached is document, T1800243-v1, that shows my IR images taken yesterday.  

• page 1: the alignment fiducials that I installed, and are used to identify and evaluate the current input pointing into the IMC
• page 2: my images of the EOM output aperture, from 3 different angles, which all show the 2nd reflection from IO_MB_L1 clipping on the +Y side of the aperture
• page 3: my image of the beam path from the EOM through IO_MB_L1, IO_MB_WG1, and IO_MB_L2, and my close up of IO_MB_L2, which shows the mis-centered beam

Being mi-centered through the IMC mode matching lenses has the potential to modify the shape of the IMC input beam, as well as increasing the effects of any small beam mis-alignment through the lenses, compared to having the beam better centered.

ETM HWS LED installation day 3 - Improved mode-matching and started initial Hartmann sensor measurements

[Marie, Aidan, Alexei, Dan]

Summary

Yesterday we improved the LED return beam from the ETM to the point at which we could attempt to characterize the noise of the Hartmann sensor and run a measurement of the ring-heater. This was predominantly achieved by (a) improving the mode-matching of the LED beam to the in-vacuum optics and (b) imaging of the ETM.

Details

Our first approach yesterday was to try optimize the imaging of the ETM on the HWS camera. This was attempted using the usual method of injecting a YAW oscillation (10urad amplitude, 0.03Hz period) into the ETM and monitoring the centroid of the return beam on the Hartmann sensor. The amplitude of the oscillation on the centroid should go to zero when the Hartmann sensor is at the conjugate plane of the ETM. However, we moved the Hartmann sensor longitudinally over a range of approximately 30cm and saw no noticeable change in the amplitude (which remained at 10 pixels, or 20 pixel pk-pk) on the Hartmann sensor. 

Additionally, we measured the distances (as best we could) between the various optics in the Hartmann sensor path. The results are:

Parameter	Measurement	Design [T1000717]
Source to Lens 1	0.762 m (+/- 5mm)*	0.9036m
Lens 1 focal length	-0.5589m (specification)	-0.5589m
Lens 2 to Lens 2	1.0858m (+/- 5mm)	1.079m
Lens 2 focal length	+2.2361m (specification)	+2.2361m
Lens 2 to Lens 3	0.6112m (+/- 5mm)	0.610m
Lens 3 focal length	+2.2361m (specification)	+2.2361m

* The distance from the Hartmann sensor to Lens 1 was estimated to be approximately 2"-4" shorter than this. Nominally, the source and the Hartmann sensor should be at the same distance from Lens 1.

The first take-away from this was the Hartmann sensor appeared to be approximately 20cm - 25cm too close to Lens 1.

The input beam from the LED source was approximated by a beam 4mm radius and converging to a focus 1.5m from the source, see TCS eLOG 210. Knowing this, we could inject this beam into a paraxial beam propagation of the full ETM/TMS system to estimate the beam size through the system. The remaining relevant parameters are reproduced here from T1000717-v5.

Parameter	Value
Lens 1 to TMS secondary mirror T2	4314mm
Secondary mirror, T2 ROC	-200mm
Secondary mirror, T2 to Primary mirror, T1, distance	1902.6mm
Primary mirror, T1 ROC	4000 mm
T1 primary to ETM HR distance	1429 mm
ETM ROC	-2245m
ETM ROC apparent (as viewed from outside)*	-1537m

* Assuming n = 1.4607 for 532nm.

Beam propagation (with 1.5m convergence)

We propagated the nominal input beam (4mm radius, +1.5m focal point) through a paraxial beam propagation (it's not a remotely Gaussian beam) from the source through to the ETM and then back to the Hartmann sensor. The results are shown below. Inside the vacuum system, the optics are all (mostly?) 1" diameter on the TMS table. A 1" diameter optic at 45 degrees has a clear aperture horizontally of the order of 16mm across. So, the beam propagation is also shown with the scale limited to 8mm. This makes it very obvious that the return beam is likely significantly clipping horizontally on the TMS & in-air optics.

The return beam looked something like this:

 

Beam propagation (with 0.87m convergence) - improved mode-matching

I changed the input beam to 0.87m convergence. And the model suggested significant improvement in the nominal mode-matching. It suggests that the input and return beam are well balanced, that we have a good retro-reflection from the ETM HR surface and that we are not likely to be clipping on any of the 1" optics. The beam size is approximately a factor of 1.7x smaller than the limiting aperture size.

 

This change was facilitated by adding a 9mm spacer to move the 1" diameter, 125 mm focal length lens further from the 20 mm lens inside the fiber launcher.

Initial system: S_input - S_lens = S_output

S_input = S_output + S_lens = -1/1.5m + 1/0.125m = 7.33 diopters

Therefore input beam looks like it's coming from a point 1/S_input = 0.1363m behind lens

S_out_target = -1/0.87m

S_input_target = S_out_target + S_lens = -1/0.87m + 1/0.125m = 6.85 diopters

Therefore input beam needs to like it's coming from a point 1/S_input_target = 0.146m behind lens.

Therefore, we need to add another 9mm or 10mm spacer to move the lens forward. 

We added a 9mm spacer.

Improved return beam

Following this we optimized the alignment of the return beam by eye, and also moved the Hartmann sensor back approximately 20cm to move it closer to the conjugate plane of the ETM HR surface.

The return beam started to look much cleaner. There is still some high spatial frequency noise on it and it looks like it might be clipping somewhat, but it now appears to show the 220mm diameter annular reaction mass hole. The beam was less elliptical and more circular (but not completely). We could also close the iris down in front of the fiber launcher and see a crisp circle of light on the Hartmann sensor (which we couldn't before).

The improved return beam looked like this. The magnification with the measured values is estimated to be 23.5x. I demagnified the annular reaction mass hole [D1500163] (diameter = 222.5mm) by 23.5x and drew it on the return beam (yellow line). I strongly suspect that the circular aperture in this image is the AERM hole.

 

Hartmann sensor measurements

Given that we were getting toward the end of the day, we decided to install the Hartmann plate and run the Hartmann sensor overnight. We initialized the code and saw approximately 300 spots - indicating good coverage. The spherical power seemed to fluctuate around +/- 4E-6 diopters. We also set up the ring-heater to run. We'll summarize these results in another aLOG.

In conclusion, there is definitely room to improve (a) the input beam size, (b) the mode-matching, (c) the alignment to the ETM and (d) the imaging onto the Hartmann sensor. However, these results are very encouraging.

The Hartmann sensor beam looks like this:

 

 

PRMI locking

Sheila, Thomas, Jenne, Craig

We were able to lock PRMI this afternoon/evening, but we needed larger gains than expected. We will still need to spend some time with PRMI checking that phases, build ups and gains make sense.

• We realigned PRX, PRY and MICH with alignments somewhat similar to what Jenne had here: 41986  Our final alignment is in the screenshot.
• We roughly set the demod phase for RF45 by letting the michelson swing and maximizing the Q signal, then locked MICH on the dark fringe.  Thomas fine tuned the phasing using a 21Hz excitation on the beamsplitter, rotating all of the quadrants of AS_A_RF45 by a total of -60 degrees so that the excitation is 2 orders of magnitudes larger in the Q sum than the I sum. The MICH dark OLG compared to a reference before the EOM swap is attached. 
• I rotated the phases of 45MHz signals by -60 degrees: ASC: REFL  A and B RF45, AS B RF45, POP X RF (which is also a 45MHz signal), LSC: POP 45, POPAIR 45, REFL AIR 45, RELF 45.  I also rotated POP90 by -120 degrees, and REFLAIR 135 by 180 degrees. (See comment, some of this was wrong when looking at trends of all the changes and reboots of the last week)
• Thomas also used the beam splitter excitation with MICH locked on a dark fringe to phase LSC POP9, which he rotated by +75 degrees.  Thomas also looked at the REFL 9 signals with MICH locked and an excitation on the beamsplitter, but based on REFL9 no rotation was required.  (See comment) We were able to lock PRX without much trouble (it used REFL9, and the ugf looks fine compared to a reference taken in DRMI.)
• After locking and doing some by hand alignment, Craig and I found that the MICH gain was low.  We edited the guardian to increase the gain for REFLAIR45 to MICH in PRMI by a factor of 2.8. (Open loop plotted with a reference is attached). We also increased the gain for PRCL by a factor of 2.37, screenshot and template also attached. These gain changes are larger than what Jenne predicts here.  We should probably check the phasing of both sensors. (see comment) 

Other details from today:

• We calibrated the rotation stage today, we found that the minimum power angle was unchanged but the max power input we set to 6.9W to get input powers that match the requested power reasonably well.  This should mean that we can use the LASER_POWER guardian instead of typing requests in by hand.  That will make sure that the normalization stays correct.  
• Thomas' alog about increasing the IMC gain by 15dB: 42165
• Fil and company got the high voltage working in HAM6 so that we could open the fast shutter 42162
• We added to the DRMI guardian a gate_valve_flag so that we can use the gain settings for PRMI without the arms even when ETMX is aligned. 
• We reset dark offsets for AS_A_RF45, but didn't accept them in SDF because I couldn't find which SDF screen they are on. 

Finiished GV11 annulus piping (w/mods), gate O-rings no longer leak following bake-out

Gerardo M., Kyle R.

The modified annulus piping was finished being installed on GV11 today.  The new arrangement eliminated some unneeded joints and lengths (runs) of piping and  now everything is much "tidier".  As part of this exercise we vented the gate annulus volume with UHP N2 and observed that neither the inner nor the outer gate O-rings leak now.  Prior to baking CP4, both of GV11's gate O-rings leaked significantly, as well as, GV12's inner gate O-ring.  The assumption being that the Viton had become less compliant after 20 years exposure to vacuum but must have softened after 8 weeks of being compressed and heated.  We are pumping the annulus volume with a turbo overnight and will helium leak check tomorrow.

Next, we will repeat this exercise on GV12.

Measuring beam astigmatism from OMC scans

[ Alexei, Dan Brown ]

The jupyter notebook for all the relevant calculations is on the LIGO git.

I noticed an odd shape of the second order resonance (can be seen in single_fit.png) in the OMC scans related to the SR3 heater testing. This shape was very consistent between scans and only seemed to show up on the second order mode. Recent beam scan measurements [ alog 41683 ] have shown that the beam could be significantly astigmatic, thus influencing the shape of the second order resonance.

Fitting 2 lorentzians to the peak

To test out the hypothesis I tried to fit two lorentzians to the second order peak (corresponding to the TEM02 and TEM20 modes).

The following model was considered

model = a1 * L(t,fwhm,t0-df) + a2 * L(t,fwhm,t0+df)

Where
L(x,fwhm,x0) = 1 / (2 * pi) * fwhm / ( (x-x0)^2 + (0.5 * fwhm)^2)
is the lorentzian function

The fit was performed by using Nelder-Mead to vary [ fwhm, a1, a2, t0, df ] to minimise [ sum(abs(data-model) ]. The resulting fit seems good as can be seen on double_fit.png and double_fit_residual.png.
The model returns the power of the second order modes as

P_20 = a1 * 2 / (pi * fwhm)
P_02 = a2 * 2 / (pi * fwhm)

Numerically the values are P_20 = 0.21 and P_02 = 0.09, which correspond to a mismatch of 13.7% [ using (P_20 + P_02) / P_00 ], where P_00 was measured separately as 2.17 from the same scan.
Note that the naive way of measuring mismatch [ P_2/P_00 ] gives 10.7% or 13.2% if the correction factor of 1.23 from [ alog  41679 ] is applied, where P_2 = 0.233 is the maximum value of the second order peak.

Using the expression for the power loss of due to mismatch from an astigmatic beam [Eq 21 in T1800165-v2 ] we get 13.9% of power loss.

Fitting 3 lorentzians to the peak

The full description of the shape of the second order peak also has to include the contribution from the TEM_11 mode, which resonantes between the TEM_20 and TEM_02 modes. This can be modelled by adding
another lorentzian. The model is now

model = a1 * L(t,fwhm,t0-df) + a2 * L(t,fwhm,t0+df) + a3 * L(t,fwhm,t0)

Where Nelder-Mead tries to minimise the sum of the residual between the model and the data by varying [ fwhm, a1, a2, a3, t0, df ].
The resulting fit is only marginally better, which can be explained by the fact that the allignment loops for the OMC were closed (TEM_11 amplitude mostly couples through misalignments).

The resulting mode intensities are given by

P_20 = 0.199
P_02 = 0.0824
P_11 = 0.0252

The mismatches are

(P_20+P_02) / P_00 = 12.9%
1-sqrt( (1 -  2 * P_20 / P00) * (1 -  2 * P_02 / P00) ) = 13.0%

So accounting for the TEM_11 peak drops the mismatch by about 1%.

Concluding remarks

Special care has to be taken with the fitting as I've found the results can vary wildly depending on the minimizing routine used and the initial conditions of the fit. I don't have enough faith in these fitting algorithms for them to be able to
work robustly without oversight.

It looks like one can extract information about the astigmatism of the beam going into the OMC from OMC scans (contrary to what I may have claimed before) by carefully fitting lorentzians to the measured second order peak.
This may be useful in hunting down and remedying the unkown source of the astigmatism that has been plaguing us.

IMC UGF Adjusted

Sheila, TVo

We noticed that the boost filter would cause the IMC to oscillate so we checked the OLTF again and found the UGF at 4 kHz, so we added +15 db to the servo gain and found a new UGF of about 30 kHz.  This is probably due to the new EOM changing the modulation depth. 

ETMY HWS running tonight - RH coming on at 2W around midnight.

ETMY HWS running tonight for noise measurements. RH coming on at 2W around 11:30PM tonight for 8 hours.

Full report tomorrow.

ETMX In-Air Violin Measurements underway

Betsy, Travis, Gabriele

This afternoon Gabriele was kind enough to help us take the in-air violin measurements on the ETMXQUAD sus.  We set up all of the in-chamber equipment (HeNe laser/QPD setup) and adjusted them to sight across one fiber at a time, while Gabriele ran the spectrum analyzer.  

We were able to obtain data from the first 3 harmonics of the v-modes on 2 of the fibers (took about 2 hours).  Tomorrow we will proceed with the other 2 fibers.  More to come...

PT110 Inficon Guage BCG450 - HAM6

WP 7592

Newly replaced PT110 Inficon Gauge BCG450 was found to have one set of faulty contacts. This is preventing the HV power supplies interlock from enabling and turning on the HAM6 HV power supplies. Faulty contacts were bypassed.

Also to note: I enabled and set the pressure/HV interlock CoE parameters for PT170 and PT110 to 1e-5. This had not been done when they were installed. Both are new BCG450 gauges.

ISCT6 Cabling

ISCT6 enclosure was moved to final location, alog 42101. Cables had to be be redressed to clear one of the clean room legs. Two of the picomotor cables ISC_182 and ISC_183 were too short,cable extensions were added.

Shift Summary - Day

TITLE: 05/23 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Planned Engineering
INCOMING OPERATOR: None
SHIFT SUMMARY:
LOG:

 

15:14 APS on site

15:29 HFD on site for hydrant testing

15:30 Marc and Hugh out to vault/LVEA

15:35 Jeff B out to LVEA to investigate TCS water leak

15:54 Richard out to LVEA

16:01 Jason out to LVEA to investigate TCSY leak

16:20 Jason back

16:23 Karen servicing Y arm

16:24 Dan and Marie out to Squeezer Bay

16:25 Peter out to PSL enclosure

16:30 Richard back

16:35 Chandra into LVEA to valve in RGA

16:37 Fil out to LVEA for HAM6 cabling

16:38 Mark and Tyler to EX to remove arm from BSC

16:50 Jason out to PSL enclosure

16:51 Dan, Marie and Alexie out to EY

17:00 Chandra back

17:15 Aiden out to EY

17:53 Marc and Hugh back and done

17:56 Jeff B back

18:00 Jason and Peter back

18:24 Jeff B out to transition the LVEA to LASER hazard

18:30 Cheryl and Jeff J out to PSL enclosure

18:43 Travis to Optics lab

18:50 Travis out

20:59 Aiden, Aiden, Marie, and Alexie out to EY

21:25 Cheryl back

22:02 Jeff B out to HAM6 to reset dust monitor and mechanical room to check on TCSY chiller

22:09 Kyle and Gerardo to MY

22:12 Jeff B back

22:24 Marc to MX

22:25 Amber in CR with tour

22:27 Chandra out to MY

22:57 Marc back

ISS measurements / work

Transfer functions of the ISS were made for various gain slider settings from 0 dB to 21 dB.
It was quite noticeable that once the gain was increased beyond 16 dB, the loop started
oscillating.  However in the power noise spectra there are a number of features (spikes and
broad peaks) that were not present before.  These are not caused by the noise eater as they
are present when the noise eater is on or off.  The broad peaks are not present in the
spectrum when the light is blocked.  So they may be caused by scattered light.

    More measurements and characterisation is necessary.



  Jason / Peter

fire pump alarms silenced to cell phones for the next 4 hours

Richard, Dave: Due to ongoing fire pump work, these alarms have been silenced to cell phones for the remainder of the afternoon.

Bypass will expire:
Wed May 23 17:41:13 PDT 2018
For channel(s):
    H0:FMC-CS_FIRE_PUMP_1
    H0:FMC-CS_FIRE_PUMP_2

 

PSL ISS Recovery

J. Oberling, P. King

Today we worked on recovering the PSL ISS, and began by re-installing the ISS AOM.  The AOM mount was moved to a place where the beam was ~1mm in diameter (unfortunately I forgot to take pictures, will take some tomorrow and upload as a comment) and installed the AOM.  The mirror that sits downstream of the AOM and reflects the 1st order diffracted beam into a beam dump was also re-installed and aligned.  A power meter was placed in this beam path and the alignment of the AOM tweaked to maximize the power in the 1st order beam.  We then adjusted the ISS offset to find the point where the AOM was diffracting ~2W; this was found at an offset of 8.0.  The offset was then set to 0, and increased to 20 in steps of 1; a power reading for the 1st order beam was taken at each step.  This allows for calibration of the diffracted power graph on the ISS MEDM screen.  Peter has the data and is analyzing it.

The beam from the PMC to the ISS box had to be re-aligned, as we had not re-aligned this path since moving the ISS box to accommodate the damping bars for the new PMC.  This complete, we then roughly aligned ISS PDB.  The PD assembly was then removed from the ISS box so we could re-install ISS PDA (recall that PDA had been removed from the ISS box and installed in place of IO_AB_PD3 prior to O2).  With PDA back on the PD assembly, this was installed back in the ISS box, and both PDs aligned.  They both now read ~10.0 V.  Next step on the ISS (for tomorrow) is to get the loop to close and measure TFs to optimize its operation.

Attached is the calculated fit to the percent diffracted power versus offset slider setting.
Ihe fit is close to a parabola p(x) = 113.18 x^2 - 82.3309 x + 15.0625.

    The power measured in front of the AOM was 67.8 W, and after the neoVAN amplifier was
69.8 W.  The power in front of the AOM was used in calculating the percentage diffraction.

I went ahead and changed the coefficients in the ISS MEDM screen that calculates the percentage diffraction.
Set the reference level to be -1.89 and closed the loop without any problems.

    The offset slider is currently set to 8.00.  Please do not mess with this value.  Only the reference level
(labelled REFSIGNAL) and the gain slider should be adjusted.  At the time of making this log entry the gain
setting has not been determined as that will require a few transfer function and noise measurements.

Modifications to the OMC DCPD ISC Split Whitening Chassis

Filiberto C., Daniel S., Marc P.

Today we successfully applied E1600252v2 to the OMC DCPD ISC Split Whitening Chassis S1101627.  This change better accommodates the violin modes of the main LIGO optics in the vicinity of 500Hz that have a tendency to saturate the whitening chain's filters.  The new filter replaces a 1Hz-10Hz zero-pole, with a 50Hz-500Hz pole-zero on the second filter channel.  LLO had already completed these changes August 2016,  ALOG.

FRS Ticket 6031

WP7583