I started some analytic calculations of how OMC mirror motion would add to the noise in the BHD. I want to make some prettier plots, and am adding the interferometer so I can also compute the noise due to backscatter into the IFO. However, since I've pushed the notebook I wanted to post an update. Here's the location in the repo.
I used Koji's soft limit of 0.02 degrees additional phase accumulation per reflection for p polarization.
I'm still not satisfied/done with the solution to this, but this has gone too long without an update and anyway probably someone else will have a direction to take it that prevents me spinning my wheels on solved or basic questions.
The story will have to wait to be on the elog, but I've put it in the jupyter notebook. Basically:
It's clear to me that there is a way to optimize the OMC, but the normalization of my DARM referred noise is clearly wrong, because I'm finding that the input-referred noise is at least 4e-11 m/rt(Hz). This seems too large to believe.
Indeed, I was finding the noise in the wrong way, in a pretty basic mistake. I’m glad I found it I guess. I’ll post some plots and update the git tomorrow.
Began setting up fiber assembly for OMC testing:
-Aligned fiber mount to maximize transmission through fiber
-Adjusted polarization at output of fiber to minimize s-polarized output.
fiber input: 56.7 mW
fiber output:43.2 mW
s-polarized output: 700 uW
Friday, Nov 11th, 2022
Setting up OMC #1 for transmission measurements:
The laser beam was aligned to the OMC cavity. The OMC cavity was locked and the transmission measurements were recorded.
Monday, November 14, 2022
Camille and Koji did a "deep cleaning" of OMC#1:
1) Applied First Contact to the mirror surfaces. Removed first contact after ~10 minutes.
2) Acetone scrub of the mirror surfaces with a cotton swab.
3) Applied First Contact again. Removed after ~10 minutes. We left the FC paint on for the work on Thu.
The foggy spot on the input mirror was unchanged after the first round of First Contact. But the foggy spot came off during the acetone scrub.
The four cavity mirrors in OMC #1 were each scrubbed using acetone and a cotton swab.
Then, the four mirrors were painted with First Contact (picture attached). The First Contact was allowed to dry for 20 minutes, then removed while using the top gun.
21 March 2023
We made slight adjustments to the beam expander lenses in the ROC setup. The position of the second lens was moved slightly (a few mm) to improve the collimation of the beam. The beam profiler was used to measure the beam size at various distances (measurements attached). This will be used to characterize the beam divergence.
This beam was reflected off the curved mirror and the beam profiler was used to measure the beam size at various positions near the focal point. This process was repeated for various curved mirrors (measurements attached). These values will be used to determine the ROC of each mirror. ROC=2*FL
22 March 2023
Beam profile measurements were continued for more of the curved mirrors.
Mirror sn07 was repeated to verify that Camille and Thejas get the same focal length measurement (plot attached).
[Camille, Stephen, Thejas]
Curved mirror sn02 was used to test the method for collecting Zygo measurements on the curved mirrors. The curved mirror was mounted with its back surface against a reference flat. The reference flat was pitched/yawed until its fringes were nulled. Then a measurement of the surface profile of the curved mirror + flat mirror together was taken.
The curved mirror was rotated in 90deg increments and the measurements were repeated. (5 measurements in total were taken, with the curved mirror's fiducial in the 12:00, 3:00, 6:00, 9:00 and 12:00 again positions.) The curvature minumum was seen to clock as expected with the rotation of the mirror.
The attached figures show the surface profile of the central 8.5 mm diameter of the mirror (central with respect to the coating edge). Also attached is a plot of the surface profile across the line drawn in the figure.
[Camille, Thejas, Stephen]
We modified the Zygo setup for measuring the sagitta of the curved mirrors. A mirror at 45deg was used to reflect the interferometer beam down towards the surface of the table (see picture). A fused silica flat was placed horizontally with the surface of the table and was used as our reference flat. The back surface of the curved mirror and the top surface of the reference flat were cleaned using top gun and/or swabs. Once it was verified that the surfaces were clean, the curved mirror could be easily placed on the surface of the reference flat.
Once the curved mirror was placed on the reference flat, the fringes of the reference flat were nulled using the 45deg mirror. After nulling the flat's fringes, the data was recorded. The curved mirror was then rotated 90deg clockwise. The measurment was repeated with the curved mirror's fiducial located at 12:00, 3:00, 6:00 and 9:00. The 12:00 position was measured twice to ensure repeatability. (A drop of first contact had been placed at the edge of the optic to indicate where the fiducial arrow is. This helped with clocking alignment.)
The already-characterized aLIGO C7 mirror was measured to verify the setup. After verifying agreement with the results in T1500060, this process was repeated with all the remaining curved mirrors.
The data was analyzed using Thejas's python script (separation distance between mirror center and curvature minimum, angular position of curvature minimum.) Those mirrors with a large spread in the measurements will be remeasured.
We repeated Zygo measurements (using the same setup and method as below) for curved mirrors sn07, sn11, sn12, sn18, sn19, sn25, sn26, and sn30.
sn11 and sn25 still show a large spread in angular measurements (see attached.) This is attributed to the low decentering values for these two mirrors (0.072mm and 0.158mm, respectively).
Summary of Zygo setups
Initial Zygo Setup:
Our initial Zygo setup consisted of a flat transmission sphere with the 0.5" curved mirror mounted against a 1" flat mirror.
The bottom part of the gluing fixture was attached to a mounting plate using two screws. The 1" reference flat was placed on the gluing fixture. The reference flat was inspected with a green flashlight to ensure that there was no dust on the mirror surface. Any dust was removed using top gun. If any dust remained after using top gun, it was removed with a swab.
The back surface of the curved mirror was inspected and cleaned using the same method (flashlight inspection, followed by top gun if necessary, followed by swab if necessary).
After ensuring that both surfaces are clean, the back surface of the curved mirror was placed on the front surface of the reference flat. The fiducial of the curved mirror was positioned at 12:00. (12:00 is defined as the top of assembly.) The two mirrors were held in place using a mounting plate with a 0.4" aperture. The mounting plate was fixed to the bottom part of the gluing fixture using two screws and a spring for each screw (see attached picture).
The mounting plate holding this assembly was then attached to a optical mount with tip/tilt adjustments (see attached picture).
This assembly was placed facing the Zygo transmission flat (see attached picture) and the mount was pitched/yawed until the fringes on the 1" reference flat were nulled. After nulling the fringes, the data was then recorded.
The mounting plate was then removed from the tip/tilt mount and dissassembled so that the curved mirror could be rotated so that the fiducial is in the 3:00 position. The procedure is then repeated and the data recorded.
This was repeated again with the fiducial in the 6:00, 9:00 and 12:00 (again) positions.
Review of this data shows that the positions of the curvature minimums was not reproducible with sufficient precision. A teflon mounting plate was added to clamp the 1" reference flat more securely to the gluing fixture (See attached pictures). Data was collected in the same manner (twice with the fiducial at 12:00 and once with fiducial at each of the positions 3:00, 6:00, and 9:00).
Additional data collected still failed to produce reproducible results and the removing/remounting process of the curved mirror was time-consuming, so we attempted a new setup for the Zygo measurments.
Final Zygo Setup:
The new setup used a fold mirror mounted at 45degrees to direct the Zygo beam downwards into the plane of the table. A 3" flat was used as our reference flat. The reference flat was placed on some lens tissue parallel to the plane of the table. The same inspection and cleaning method was used to ensure that there was no dust on the reference flat (flashlight inspection, followed by top gun if necessary, followed by swabbing if necessary).
The back of the curved optic was inspected and cleaned using the same method. The curved mirror was placed on the 3" reference flat with the fiducial at the 12:00 position. (12:00 here is defined as the direction ponting towards the Zygo instrument.) (See attached picture of this setup.)
The fold mirror was pitched/yawed so that the fringes on the 3" reference flat were nulled. (An additional advantage of this setup is that more surface of the reference flat was viewable.) After nulling the fringes, the curved mirror was picked up and replaced a few times to verify that the fringe pattern on the curved mirror appeared reproducible. The data was collected with the fiducial at the 12:00 position. This process was repeated with the fiducial at 3:00, 6:00, 9:00, and again at 12:00.
Results from this setup were reproducible and this setup was used to measure the surface profile of all the curved mirrors.
The ThorLabs MDT694B piezo driver was returned to the OMC lab.
Borrowed for PZT DC Response Shadow Sensor Setup (see Attachment 1):
Current Location: Downs 227
The op-lev setup was modified slightly (picture attached). The He-Ne laser was replaced with one with a lower divergence angle. The stage that the PZT/mirror stack sits on was replaced with a setup that allows for tip/tilt adjustements. The PZTs were driven from 0 to ~92V at 0.5Hz. The total path length from the PZT stack to the QPD is ~1.6m.
The following PZTs have been measured in the current oplev setup: 37, 38, 39, 40, 41, 42, 43, 44, 46, 47, 48.
The DC response of the PZT was measured using the following setup (pictures attached):
The output from a HeNe laser was transmitted through two lenses forming a beam reducer, giving us a beam size of 0.34mm. The PZT was placed on a labjack and a 62g washer was placed on top of the PZT. The HeNe beam grazes the top of the washer and the beam is monitored by a photodiode. The labjack height is adjusted so that the PZT stack clips half of the beam signal.
In order to avoid scatter as the beam grazes the top surface of the washer, an extra layer of kapton tape was placed on one side of the bottom of the washer, giving it a slight tilt so that the beam is clipped by the washer at the front surface of the washer (surface closest to beam source).
The PZT was driven from 0 to 150V at 0.5Hz using a triangle function. The drive function and the photodiode response was recorded using a spectrum analyzer.
The DC resposne was measured twice for each PZT: before the reliability test and again after the reliability test.
Calibration of the DC response measurement:
The setup was calibrated using a height gauge with a dial indicator to monitor the labjack height. (A PZT was not needed for the calibration; only the labjack and the washer.) The labjack was adjusted so that its height is such that the entire beam just passes over the top and the photodiode response was recorded.
The labjack height was increased in 10um increments, with the photodiode response recorded at each increment. The height values were plotted against the photodiode response values. The slope of this line was used to convert the photodiode response from mV to um.
To test the reliability of the PZTs, they were driven for 10^7 cycles. (At 100Hz, this is ~28 hours.) The drive function was a 100Hz sine wave with peak-to-peak amplitude of ~100V. The drive function was monitored using an oscilloscope and recorded at the start of the test. A thermal imaging camera was used to record the temperature of each PZT at the start of the drive and at ~10 minute intervals during the first hour of the drive. The temperatures of the PZTs are expected to rise slightly (1 or 2 degC above ambient) and then level off during the first hour or so.
At the end of the drive (10^7 cycles), the temperature of each PZT is recorded again and the drive function on the oscilloscope is also recorded to verify that the peak-to-peak amplitude did not change.
Results for PZT reliability tests:
Batch 1: SN37, SN38, SN39, SN44
Batch 2: SN41, SN42, SN43, SN46, SN47, SN48
The reliability tests were performed using the procedure described in https://nodus.ligo.caltech.edu:8081/OMC_Lab/?mode=full&reverse=0&reverse=1&npp=50&Subject=Summary+of+PZT+reliability+test+procedure
Results for PZT reliability test:
Batch 3: SN31, SN32, SN35, SN40, SN45
This reliability test was performed using the procedure described in https://nodus.ligo.caltech.edu:8081/OMC_Lab/?mode=full&reverse=0&reverse=1&npp=50&Subject=Summary+of+PZT+reliability+test+procedure
*Note: The drive function used here was a triangle function (not sine function).
While using mirror SN14 and mirror SN30 for CM1 and CM2 respectively, we monitored the two beam spots on FM2 and continued to see ~5mm of vertical displacement between the two spots. We swapped the subassembly containing SN30 for the one containing SN02 and we see that the pitch misalignment is resolved. We will proceed to lock the cavity using the following mirrors:
FM1 --> SN11
FM2 --> SN17
CM1 --> SN14
CM2 --> SN02
Plans for this week:
1) Verify electronics setup for cavity locking.
Current state: Output from function generator is going directly to laser driver (not currently using the Newport Servo Module). We will need to set up the servo module.
2) Finish setting up the table optics so we can monitor the reflected beam (need to set up photodiode and CCD camera to monitor this beam).
3) Once the cavity is locked, we can use the steering mirrors to optimize the cavity while monitoring the CCD cameras for the transmitted and reflected beams. We need to walk the beam so that more of the TEM00 mode is transmitted. (Currently still seeing a lot of higher order modes transmitted, as shown in attachment).
We rechecked the alignment this morning and re-optimized the resonance a bit. There was some horizontal drift in the alignment from yesterday.
Plan for this afternoon:
Camille, Thejas, and Masayuki will meet with Koji in the OMC lab to go over electronics setup. (Need to set up EOM driver? Laser is not sweeping.)
Yesterday, we measured the particle count in the enclosure to ensure that the lower setting on one of the HEPAs is still acceptable. The particle count is still zero for all measured particle sizes (0.3um, 0.5um, 1.0um, 2.0um, and 5.0um).
We also relocked the cavity and repeated the optimization efforts in https://nodus.ligo.caltech.edu:8081/OMC_Lab/584. The reflected signal was around 110mV (compared to 80mV on Tuesday).
We used a CCD camera to view the beam spots on the curved mirrors (pictures attached). We will compare the spot positions to the scatter plots for these mirrors and try to steer the beam spots accordingly.
(*Note: We did see some stray light scattering on the edge of FM2. We will further examine this and see where the stray light is coming from.)
We lowered the setting to MEDIUM on the HEPA furthest from the entrance. (The two HEPA settings are now LOW and MEDIUM.) We measured the particle count for all particle sizes at different locations in the enclosure (next to the OMC fixture and at the edge of the table closest to the entrance). All counts read zero.
[Camille, Thejas] 4 August 2023
-REFL PD signal (unlocked): 3.5V
We optimized the alignment of the steering mirror on the REFL PD.
-REFL PD signal (locked): 1.5V (initially) --> 0.15V after we optimized alignment into the cavity.
-We finished setting up the table optics so we monitor the transmitted beam with a photodiode. (Previously, we were only monitoring the transmitted beam with a CCD camera.) A mirror is used to steer the beam to the PD.
-We checked the beam spot positions on the mode-matching lenses and adjusted the fiber mount. (It is centered on the first lens. It is somewhat off-center on the second lens [attach picture].)
-We optimized the alignment to reduce the REFL PD signal when locked. (We slightly pitched the periscope mirror, adjusted the fiber mount, and slightly walked the beam with the periscope. Lowest achievable REFL PD signal was 0.15V
Our mode-matching ratio is ~95.8%
-We unlocked the laser and used the frequency offset to sweep and observe shape of the transmitted modes on the CCD camera. TEM00 modes are the most prominent. The other observed modes appear to be 1st/2nd/3rd order rectangular modes. [attach pictures of modes]
Plans for this week:
-If mode-matching ratio is acceptable, we will proceed to move the cavity axis between CM1 and CM2. (The axis needs to move towards FM1 and FM2.)
-If mode-matching ratio needs to be improved, we will re-characterize the beam waist position and re-verify the location of our beam waist. (We will temporarily place a mirror or beamsplitter after the mode-matching breadboard, so we can steer the beam away from the OMC optics and use the beam profiler without disrupting the OMC optics.)
We will plan to look at the reflection CCD image this afternoon.
With the current level of the mode matching, the spot positions can be optimized. I'd work on it first as the mirror RoC and loss depends on the spot positions on the curved mirrors.
I'd also look at the reflection CCD image:
With the previous OMC, we saw the reflection to be 60~70mV. We saw ~85mV with the previous alignment when I went down to the lab.
Today it is ~150mV.
- If this increase is coming from the worse reflection of the cavity (lossy cavity somehow), the reflection CCD image should definitely show TEM00 mode.
-> The cavity needs to be cleaned
- If this is coming from the mode matching, the image seen with the room lights off should show (somewhat symmetric) higher-order modes.
-> I'd try moving the lenses.
7 August 2023
REFL PD signal (unlocked): 3.42V
REFL PD signal (locked): 0.16V
While monitoring the REFL PD and the REFL CCD, we adjusted the the fiber coupler and the steering mirrors to optimize alignment into the cavity. Attached is the image we observed on the REFL CCD camera. This image shows some higher-order modes in the reflected beam. To further improve this, we might consider to move the lenses.
7 August 2023
After optimizing the alignment, we used the CCD video cameras to observe the beam spots on CM1 and CM2 (pictures attached). For both mirrors the beam spot is off center in the direction away from FM1 and FM2.
In order to steer the beam axis towards FM1 and FM2, we rotated CM1 clockwise (as viewed from above) and we rotated CM2 counter-clockwise (as viewed from above).
We aimed to steer the beam axis by 1mm, so we displaced the micrometers on both curved mirrors by 10um.
After displacing the micrometers, we attempted to recover TEM00 using only the fiber coupler and the steering mirrors, but we were unable to achieve this. We made an additional adjustment to the CM2 micrometer so that the beam reflected from CM2 would better overlap the incoming beam on FM1. After doing this, we were able to observe TEM00 and proceeded to lock and optimize using the fiber coupler and steering mirrors. (REFL PD was 0.12V, REFL CCD looked similar to the one in https://nodus.ligo.caltech.edu:8081/OMC_Lab/597.
We observed the beam spots on CM1 and CM2 again and didn't observe any noticeable change (pictures attached).
We repeated this process. (We rotated CM1 clockwise and CM2 counter-clockwise each by the same amount.) We tried to recover TEM00 using only the steering mirrors and fiber coupler, but were unable to find TEM00. We made an additional small adjustment to the CM2 micrometer and found TEM00. After locking and optimizing, the beam spots on the curved mirrors still appear to be in the same locations.
We inspected surfaces of the cavity optics with a halogen lamp and noticed a particulates on optic HR surfaces. We cleaned all 4 optics with top gun, this did not seem to reduce the cavity loss upon locking. The cavity loss seemed to be higher with refl power now at 600 mV.
So we proceeded to apply First Contact on all the four optics.
The bottom surfaces of the prisms and tombstones were swabbed with IPA. We also cleaned the breadboard positions for these four optics with an IPA swab. (The CM1 mirror looked better after cleaning (no scatter observed).)
After cleaning the optics were returned to their positions on the breadboard. The spot positions on the curved mirrors still look roughly centered, but the steering mirrors need to be optimized.
When viewed through the IR viewer, the spots on the curved optics are not as bright as they were before FC. (The brightness is more or less equal to the brightness on the FMs.)
We observed the transmission PD which was mainly dominated with TEM00 modes, and also some higher-order modes:
HOMs: 0.174V and 0.054V
--> 3.2% mode-mismatch
REFL PD: 3.6V (unlocked)
REFL PD: 0.340V (locked) (We estimate we should be able to reach 0.115V for the same level of mode-matching.) (The refl mode shape looks similar to last time on the CCD (some TEM00 mode).)
This implies there is still ~200 mV of loss from the cavity.
This morning, we continued the efforts from http://nodus.ligo.caltech.edu:8080/OMC_Lab/602.
While monitoring the REFL PD, we walked/optimized the alignment after cleaning the optics with FC. After optimization, our PD signals were:
REFL PD: 0.11V (locked)
REFL PD: 3.3V (unlocked)
We also have a picture of the REFL CCD (attached). The REFL CCD does not appear to show so much TEM00 as it did previously.
We also checked the mode scan on the transmission PD and recorded the signals of the TEM00 mode and another higher-order mode that was still observed.
--> ~1% mode mismatch.
From this we estimate that we should be able to reduce the REFL PD signal to ~30mV.
We used the CCD video cameras to record pictures of the beam spots on the cavity optics (attached). The scatter looks much better post-cleaning and the beam spots are closer to center. However, we can still steer the beam closer to center (need to move away from FM1 and FM2).
10 August 2023
We want to evaluate the loss in the cavity, so we recorded the power measurements needed to enter into the power budget analysis (see attached picture with recorded values). We collected three sets of measurements to be averaged.
The screenshot below shows the output from the python analysis. (OMC throughput values are surprisingly high: 100.0%, 103.3%, and 99.8%) We plan to try to improve the mode-matching and re-evaluate the power throughput.
11 August 2023
We cleaned the cavity optics again using the top gun and First Contact. (FC was used on the S1 and S2 sides of the FMs and on the S1 side of the CMs.) The bottom surfaces of the optics were swabbed with IPA. The breadboard was also swabbed with IPA where the optics are positioned.
The optics were returned to their same positions on the breadboard.
On Monday, we will finish slight realignments to recover TEM00 lock. We will take pictures of the beam spots and see if we can improve mode-matching before taking more power efficiency measurements.
For Set 1 of the data in https://nodus.ligo.caltech.edu:8081/OMC_Lab/604,
VREFL(unlocked) = 3.226V
VREFL(locked) = 0.13V
Pin = 20.56mW
Fraction of light that is reflected (mode-mismatched) = 0.135/3.226 = 4.03%
Pjunk = 0.0403*20.56mW = 0.83mW
From T1500060 Section 3.3, "The incident beam power to the cavity (Pin) can be split into the mode-matched (coupled) and mode-mismatched (junk) light power (Pcoupled and Pjunk, respectively)."
Pin = Pcoupled + Pjunk
20.56mW = Pcoupled + 0.83mW
Pcoupled = 19.73mW
This suggests that our cavity has nearly no loss, and the mode-matching efficiency is ~96%
However, this mode-matching efficiency is very different from the mode-matching efficiency determined from our transmitted PD signal on https://nodus.ligo.caltech.edu:8081/OMC_Lab/603.
From the PDtrans signal, the TEM00 signal is ~7.0V
There was only one higher-order mode observed with a signal of 0.070V.
0.070/7.0 = 1% mode-mismatch
[Camille, Thejas, Masayuki]
This afternoon we finished the realignment that we started after the FC cleaning in https://nodus.ligo.caltech.edu:8081/OMC_Lab/605.
We wanted to try to improve mode-matching before taking new power measurements. We used the signal from the transmission PD to characterize the mode-matching. We observed the TEM00 peak and one additional HOM peak:
TEM00 signal: 6.9V
HOM signal: 0.095V
--> mode-matching efficiency is ~1.4%
We observed the REFL CCD and include an attached picture. We recorded pictures of the beam spots using the CCD video camera (pictures attached).
We took one set of power budget measurements (measured values and outputs are shown in the attached screenshot).
The fraction of light that is reflected is
0.012V/3.35V = 3.6%
This is very similar to our previous data.
Similarly, our reflected power, incident power, and transmittd power are very similar to our previous values (Prefl=0.79mW, Pin=21.82mW, and POMCT=20.73mW)
This would seemingly indicate that we have very little loss in the cavity, however we still plan to further investigate the 3.5% loss observed by the REFL signal.
[Camille, Thejas, Koji]
16 August 2023
We met in the lab to try to understand the mode-match discrepancy we see in our measurments. Adjusted the fiber coupler and the periscope to minimize the REFL PD signal. (REFL PD signal was 0.116 when locked.) The shape of the beam on the REFL CCD looked the same as in https://nodus.ligo.caltech.edu:8081/OMC_Lab/607.
We observed the transmission spectrum on the scope to identify higher order modes and side bands (need to attach plot). We closely examined the signal intensity of the weaker peaks in addition to the stong TEM00 peaks and exported the data from the scope. We also locked the cavity on the other modes to observe the shaped of these other modes (we see some pitch and yaw misalignment in the other modes).
The intensity signals of the other modes estimates ~1.8% mode-mismatch. (Still does not explain the 2% discrepancy between we mode-mismatch we calculate in our power budget analysis.)
We also varied amplitude of the phase modulation (from ~8-17dB) but this showed no improvement to the REFL PD signal.
Our plans moving forward:
-Center the beam path through the lenses to try to improve the mode-matching
-Further reduce REFL PD signal (~70mv?)
-Quick check: Attenuate the TRANS PD signal and compare ration between TEM00 signal and other modes.
16 August 2023
We reconvened in the afternoon to begin realignment of the beam path through the mode-matching lenses. Before doing so, we placed two iris to mark our current beam path to the OMC. (one iris after the steering mirror, one iris right in front of the OMC (picture attached))
We made a few slights adjustments to the fiber coupler and the lenses: We used a level to adjust the height of the second lens so that it is at the same height as the first lens. We slightly adjusted the height of the fiber coupler mount so that the fiber height matches the height of the center of the lenses. We translated the fiber coupler slighly to adjust the centering while maintaining the same distance from the first lens.
After centering the path through the lenses, we repositioned the periscope mount and the steering mirror accordingly so that the beam path hits the centers of these mirrors.
Tomorrow, we will lock the cavity and repeat power measurements to determine if there is any improvement to the mode-matching.
18 August 2023
We continued the work started on https://nodus.ligo.caltech.edu:8081/OMC_Lab/609. The beam is well-centered through the mode-matching lenses. We used the periscope to optimize cavity alignment while locked on the TEM00 mode.
We checked the steering on the REFL PD and the TRANS PD to make sure both are aligned.
The REFL PD signal was 3.0V (unlocked) and 46mV (locked). (This is the lowest REFL power we have had with this cavity.) A picture of the REFL CCD is attached.
We also checked the intensity of the HOMs on the mode spectrum (pictures attached). The TEM00 signal was ~7.2V while the observed HOMs had a signal of 23mV and 2 mV.
We proceeded to take 2 sets of power budget measurements (measurements and screenshot attached). After running the measurements in the power analysis script, we have and OMC throughput of ~99% and mode-matching efficiency of 98%. This seems to agree better with our mode-spectrum. (The excel spreadsheet with the analysis is attached as Attachment 5.)
Thanks for clarifying that. We will repeat some power measurements and check the output offset voltage to the laser.
This is WOW!
Excellent mode matching work.
The measurement is still consistent with the low loss even with the different mode-matching level.
99.5%? The IFO commissioners will cry.
Edit: Wait a sec. The incident * mode matching = 20.14mW. This is the cavity-coupled power.
And you have the transmission of 9.78*2=19.56mW.
The ratio of these is ~97% and not 99.5%. Did I miss something?
=> Ah, understood. You have the incident power measurement with a significantly different reference voltage from the one at the transmission measurement. (4.21V vs 4.11V)
This is because the laser output power depends on the laser PZT feedback.
The quick hack to reduce this is that check the laser PZT feedback voltage (on the Thorlabs driver) right before the unlock, and bring the "output offset" close to that value after unlocking.
This brings the laser frequency back to the one during lock. At the same time, the laser freq is now close to the cavity resonance. So reading the unlocked REFL voltage, you need a bit of care.
We checked the length-to-angle coupling of each PZT by monitoring the position of the transmitted beam on the CCD camera. The CCD camera was placed behind the steering mirror that guides the transmitted beam to the PD. We used a ThorLabs piezo controller to actuate the PZT.
We first tested PZT2. We increased the voltage to PZT2 in 50V increments from 0V to 150V. We did not observe any change in the position of the transmitted beam. We monitored the signal of the TRANS PD on the scope and did not see any change. (The signal was between 191-195V.) We monitored the REFL CCD and did see changes in the beamshape, which was expected (see pictures). The REFL PD signal also increased slightly with PZT actuation (see attachment).
We repeated this process for PZT1, which showed similar results (see attachment). We did not observe any movement in the position of the transmitted beam. Increasing PZT voltage shows increasing pitch misalignment in the REFL CCD and increasing REFL PD signal.
PZTs 31, 32, 35, 40 and 45 have had their DC responses measured (pre-reliability test) and have been through the reliability test (see 564). We still need to measure their DC responses post-reliability test.
Camille and Thejas will plan to do these measurements tomorrow afternoon (15 Sept. 2023).
Notes for setup:
-The setup will be the same as 542 and 551.
-0.5 Hz triangle wave 0-150V
-8s aquisition time (128Hz sample rate) on SRS785 spectrum analyzer
-62g washer on top of the PZT
-Same laser, collimation lenses (beam size 0.34mm at edge of washer), and photodiode
-Record photodiode response and voltage to the PZT
22 August 2023
We used the network analyzer to measure the FSR of the cavity using the method described in section 3.2.1 of T1500060. We locked the OMC cavity and maximized transmission the TEM00 mode. (REFL PD signal was ~45-50mV and REFL CCD looked the same as in 610). We adjusted to input offset on the servo module (REFL PD signal ~95mV) and recorded the transfer function between the modulation signal (channel R) and the transmission PD signal (channel A). (See attached picture of transfer function and phase.) We fit the FSR data to the code to get a value of 264.658982 MHz.
We also recorded the TMS of the cavity (with 0V to the PZTs). We measured the horizontal and vertical mode spacing separately. After maximizing transmission of TEM00, we then used the fiber coupler to misalign in the vertical direction first (REFL PD signal ~100). Using the network analyzer, we observed the peak at ~58 MHz. We then misaligned the mirror that steers the transmited beam to the PD. We clipped the transmitted beam so as to maximize the peak at ~58 MHz. (See attached spectrum.)
We recoved vertical alignment and then repeated this process for the horizontal direction. (See attached spectrum.)
Analysis in the next elog entry.
test_22-08-2023_160812 --> FSR
test_22-08-2022_165728 --> TMS vertical
test_22-08-2022_170543 --> TMS horizontal
[Camille, Thejas, Masayuki, Koji]
Previously, we were using the function generator to drive the laser without using the servo module. On Tuesday, we incorporated the servo module (output from function generator to sweep input). Slow laser freq scan: 8Hz, 3 Vpp ramp. We were able to see the TEM00 mode after increasing the span or by adjusting the offset from the servo module (usually between 50-90 V on PZT driver). If the span knob is set to zero, drive from the signal generator to the laser PZT driver is suppressed (This was the reason why we couldn't scan the laser frequency through the servo module last week). the EOM drive freq is ~ 31.23 MHz, 13 V
Once the cavity was locked, we set up a photodiode to monitor the reflected beam. (Reflected beam signal was ~3.4V).
It was observed that the control signal from the servo module to the laser was noisy. HEPA air filter was the source and we reduced the speed of on of the HEPAs (close to the entrance).
Process for optimizing alignment once cavity was locked:
-First maximise the power on the reflection PD using the steering mirror infront of the PD.
-Use cavity steering mirrors to minimise reflected PD signal.
-When PD signal is low like ~0.3V or less, switch to fiber output alignment.
-Continue to optimize using fiber adjustment. (Best was ~80mV).
-Make sure that the reflected light is still coupling 100% into the PD using the steering miiror.
-Check reflected beamshape. Since the OMC cavity is a critically coupled cavity the Transmitted light = Incident light, & Reflected lgiht = 0 at resonance and when the mode-matching is perfect. Since we have a some amount of light reflecting, current mode-matching efficiency = 3.4-0.080/3.4 i.e. 97.6 %. May be we can translate the mode-matching optics bench to improve this at some point.
We then set up two more cameras so we can monitor the beam spot positions on the curved mirrors.
We also moved the CCD w/ ND filter that was monitoring the output at FM2. We moved it so that it is monitoring the leaked transmission through CM2 instead (no need to use ND filter).
Next steps for this week:
-Save data from camera showing the beam spot positions on the curved mirrors.
- Look at the scatter plots, and steer the cavity beam spots on the mirrors as needed (refer to the cavity matrix)
- Beam transmitted through the cavity should availble for incidence on the PDs at the tranmission side. If needs be steer the cavity axis to attain this.
-Measure power transmission through the cavity. The target efficiency is ...
- If qualified measure TMS, FSR else swap subassemblies.
Attached are the analysis results from the measurements in 616.
FSR: 264.657354 +/- 0.003444 MHz
Pitch TMS: 58.45691858660249 MHz
Yaw TMS: 58.55821902092523 MHz
The attached plot shows the HOM spectrum with their sidebands. We see that there is overlap between TEM00 and one of the 9th-order modes which means this higher order mode will resonate with the TEM00 carrier.
We estimate that by increasing the FSR to ~266.7 MHz, we will avoid this as shown in the next attached plot (HOM_scan.pdf).
This will require us to decrease the cavity length by 16mm (4 mm each mirror). We plan begin adjusting the micrometers this afternoon.
[Camille, Thejas, Masayuki]
8 August 2023
We continued our efforts to steer the beam spots on CM1 and CM2 towards FM1 and FM2. We rotated CM1 clockwise by displacing the micrometer 20um (2 small divisions). We rotated CM2 counter-clockwise by the same amount by displacing the micrometer 20um.
We then walked the beam with the periscope until we recovered the TEM00 mode. Once recovered, we locked the cavity and continued to optimize using the periscope and the fiber coupler output. We checked the beam spot positions with an IR card and viewer and recorded images with the CCD video camera (images attached).
We also monitored the transmission PD on the scope (green trace in attached picture, pink trace is the laser freq sweep) to compare the voltage signal from TEM00 (4.3V) and the voltage signal from the higher-order mode (76mV). (1.7% from mode-mismatch = 76 mV *100 / 4.3 V + 0.76 mV) This means we should be able to resonate the cavity at 0.017*3.4 V ~ 70 mV of refl power with the current level of mode-mathcing
Our refl power (refl.jpeg) shows significant relfection of about 400 mV, suggests the cavity is lossy. (See scatter on CM1). A separate elog will look at the scatter plot.
For this afternoon, we will plan to clean the optic surfaces with First Contact and see if this reduces the cavity loss.
That is an expanding fire pillow, also known as firebrick. It is used to create a fire block where holes in fire-rated walls are made and prevents lab fires from spreading rapidly to adjacent labs. I had to pull cable from B254 to our labs on either side during a rather narrow window of time. Some of the cable holes are partially blocked, making it difficult to reach the cable to them. The cable is then just guided to the hole from a distance. With no help, it's not possible to see this material getting shoved out of the hole. I can assure you that I took great pains not to allow the CYMAC coax to fall into any equipment, or drag against any other cables.
Suspension cage and transportation box: 250.8lb
Suspension cage and transportation box: 150.2lb ==> 100.6lb ==> 45,630 g
Metal Breadboard: 7261 g
Glass Breadboard and transportation fixture: 16382 g
Transportation fixture only: 9432 g ==> 6950 g
Added mass (up to now): 300 g ==> 7250 g
[Joe, Phillip, Koji, Stephen]
*draft post, please add anymore info if I missed something*
General notes about working with this set up. The lens on the CCD can come off quite easily, as you just change how much its screwed on to change the focus. Care should be taken that you don't know the template with this as well, as the camera is quite close to the template (and near the edge of the bench!). Also be mindful of the PZT wires, as they can pull the mirrors out of position.
Attachment 1 shows the position of the spots on the mirrors A14 and PZT11. The spots are about 3mm ish from the centre of the curved mirror in the vertical and horizontal direction.
Attachment 2 sketch of mirror positions.
Attachment 3 shows the postion of the spot on PZT13. The spot is less near the edge than on PZT11, but its still 2mm ish from the centre of the curved mirror in both directions.
To move the beam horizontally we can use the alignment matrix in appendix C of T1500060. However since we don't have control over the pitch of the mirrors, moving the spots down could require us to inspect the glass breadboard/prisms for dust. We suspect that PZT could be the culprit, as we could not see newtonian rings between its base and the glass breadboard. One way to test this idea is just to clean the bottom of the PZT with acetone, and see if that improves the spot position. If we don't have to do any work to realign it, then this was not the issue.
Koji pointed out that the spot in attachment 1 is very near the edge of the optic, so shifting the beam horizontally could also fix the vertical issue.
[koji,philip, joe, liyuan, steven]
*still need to add photos to post*
PZT 11 was removed and inspected for so dust/dirt on the bottom of the prism. We saw a spot. We tried to remove this with acetone, but it stayed there. (Attachment 2, see the little white spec near the edge of the bottom surface of the prism)
current micrometer positions:
Swapped PZT for PZT 22, cleaned the bottom and put it into position of CM1. We saw a low number of newton rings, so this is good.
We got a rough initial alignment by walking the beam with the periscope and PZT 22 mirrors. Once we saw a faint amount of transmission, we set up the wincam at the output. The reflected light from the cavity could also be seen to be flashing as the laser frequency was being modulated.
Once it was roughly aligned, using the persicope we walked the beam until we got good 00 flashes. We checked the positions of the spots on the mirror with the beam card. This looked a lot better in the verticle direction (very near the centre) on both curved mirrors. We locked the cavity and contiued to align it better. This was done with the periscope until the DC error signal was about 0.6V. We switched to the fibre coupler after this.
Once we were satisfied that he cavity was near where it would be really well aligned, we took some images of the spot positions. Using these we can work out which way to move the curved mirrors. Koji worked this out and drew some diagrams, we should attach them to this post. [Diagram: See Attachment 1 of ELOG OMC 350]
We made the corrections to the cavity mirrors
The scatter from CM1 looked very small, it was hard to see with a viewer or CCD. We had to turn up the laser power by a factor of 3 to begin to see it, indicating that this is a good mirror.
Once this was done, the spot positions looked uch nearer the centre of each mirror. They look pitched 1mm too high, which might be because of the bottom surfaces of the prisms having a piece of dust on them? For now though it was good enough to try take the detuned locking FSR measurement and RFAM measurement.
To see the higher order mode spacing, we misaligned them incoming beam in pitch and yaw so that the TM10 and TM01 modes were excited. The cavity transmission beam was aligned onto the photodiode such that we could make a transfer function measurement (i.e. shift the beam along the photodiode so that only half of the beam was on it, this maximises the amount of photocurrent).
attachment 1 shows the fitting of the detuned locking method for measuring FSR and cavity length/
I saved this data on my laptop. When I next edit this post (hopefully I will before monday, although I might be too tired from being a tourist in california...) I want to upload plots of the higher order mode spacing.
in units 20um per div on the micrometer [n.b. we reailised that its 10um per div on the micrometer]
CM1 inner screw pos: 11.5
cm1 outer screw pos: 33.5
cm2 inner screw pos: 11
cm2 outer screw pos: 13
the cavity is currently 3mm too long, move each mirror closer by 0.75mm
CM1 inner screw pos: 11.5+37.5 = 49
cm1 outer screw pos: 33.5+37.5= 71
cm2 inner screw pos: 11+37.5 = 48.5
cm2 outer screw pos: 13+37.5 = 50.5
The screws on the micrometers were adjusted to these values.
cleaned cm1 (PZT 11). There was a mark near the edge which we were not able to remove with acetone. On the breadboard there were 3 spots which we could not remove with acetone. Once we wiped the mirror and breadboard we put the mirror back.
FM2 (A5). The prism looked quite bad when inspected under the green torch, with lots of lines going breadthways. We thought about replacing this with A1, however this has had the most exposure to the environment according to koji. This has a bit of negative pitch, so would bring down the beam slightly. We decided to continue to use A5 as it had worked fairly well before. The breadboard was cleaned, we could see a few spots on it, they were cleaned using acetone.
FM1 (A14). Near the edge of the bottom surface of the prism we could see some shiny marks, which may have been first contact. We attempted to scrape them off we tweezers. The breadboard looked like it had a few marks on it. These were hard to remove with the acetone, it kept leaving residue marks. We used isopropanol to clean this now, which worked much better. The sharp edges of the breadboard can cause the lens tissue to tear a bit, so it took a few rounds of cleaning before it looked good to put a prism on. The mirror was put back onto the breadboard.
The cavity was aligned, then we realised that 1 turn is 500um, so its still too long (1.75mm long). The FSR was 264.433Mhz, which is
CM2 still showed quite a bit more scattering than CM1, so we want to move this beam.
want to increase by 1.7/4 = 0.425, so
we tried to align the cavity, however the periscope screws ran out of range, so we changed the mircometers on CM2. We tried this for quite some time, but had problems with the beam reflected from the cavity clipping the steering mirror on the breadboard (to close to the outer edge of the mirror). This was fixed by changing the angle of the two curved mirrors. (We should include a diagram to explain this).
The cavity was locke, the FSR was measured using the detuned locking method, and we found that the FSR = 264.805 MHz, which corresponds to a cavity length of 1.1321m
we took some photos, the spot is quite far to the edge of the mirrors (3 to 4mm), but its near the centre vertically. photos are
123-7699 = CM2
123-7697 = CM1
[Koji,Philip, Liyuan, Joe]
We moved the curved mirrors to these positions:
inner = 0.807mm
outer = 0.983 mm
inner = 0.92 mm
outer = 0.85 mm
To do this so that realignment was easier, we moved the screws in steps of 5um. We alternated which mirror we adjusted so that we could monitor with a wincam how well aligned the beam into the cavity was. We only moved the cavity mirrors a small amount so we could still see higher order mode flashes transmitted through the cavity (e.g.TM03 modes). We would then improve the input alignment, and then move the cavity mirrors some more. Once the mirrors were adjusted according to http://nodus.ligo.caltech.edu:8080/OMC_Lab/190422_195450/misalignment4.pdf the spot positions looked near the middle of the curved mirrors (using a beam card). We began beam walking but we ran out of range of the bottom periscope screws in the yaw dof. We tried using the third screw to move the mirrror in both yaw and pitch, hopefully this will let move the mirror such that we can use the just the yaw screw. This screw also ran out of range, so we decided that the cavity needed a small adjustment.
The curved mirrors were moved slightly (>5um) and then we tried to get alignment. By using the fibre coupler translation stage, we move the beam side ways slightly, and then tried to get the periscope mirrors back to a position where the screws could move the mirrors. Once we had an ok alignment, we checked the beam. It looked like it was pretty close to the centre of the curved mirrors, which is where we wanted it to be.
We then tried locking the cavity, although the error signal was quite small. The adjusted the input offset and gain of the servo (there is apparently some problem to do with the input and output offsets). Once the cavity was locked we could make the final adjustments to aligning. We still ran out of range on the periscope. We decided to move the breadboard with the fibre coupler and mode matching lenses on it. Because we knew that the cavity was aligned such that the beam hits the centres of the curved mirrors, we could regain flashes quite quickly. We saw the error signal go down, but eventually this decrease was just to do with the beam clipping on the periscope mirrors. We moved the spot back to where we ok aligned, and slid the periscope so we were not clipping the mirror. This worked very well, and then optimised the alignment.
We then tried to improve the mode matching.
We took photos of the spot positions (quite near the center) and made the detuned locking measurement. The fitting of the data (attachment 1) wsa 1.1318m (what error should we put here?).
I think the order we did things in was:
need to add spot positions.
Mirrors: PZT11,PZT22, A14, A5
Mode matching = 97.72%
15.66-> 15.30mW coupled.
~100uW for QPD
->15.2mW in cavity
Trans = 14.55mW -> 95.7% transmission
The flat mirrors were the ones with the most scattering, so we thought about how to improve it. We tried to move the first flat mirror by pushing it with our finger so that he beam would move along the optic. We tried this a couple of times, however the second time we moved it we lost our alignment and could not retrieve it. We looked at the mirror and we could see quite a lot of newtonian rings. We could see a small fibre on the glass bread board. We cleaned the optics base and the gbb, and we could get the alignment back. The beam was aligned to the cavity, the spots no longer hit the centre of the CM2.
We measured the power budget again.
mode matching = 1-47/2680 = 0.9824, 98.2% mode matching
same p_normalise so
15.66-> 15.34mW coupled.
~15.24mW in cavity
transmission = 14.45, so 94.8% transmission.
Koji noticed that FM1 wasn't touching the template correctly, so he re-aligned the cavity.
Afternoon session - UV Bonding (E1300201-v1 procedure 6.4.4 "Gluing" using procedure in section 7.2 "UV Gluing")
Wiped down UV PPE, UV Illuminator, and UV Power Meter
Applied Optocast 3553-LV Epoxy to sample fused silica optics, to test quantity of glue needed and to become familiar with the process and tools. Philip and Joe each created a successful bond. Joe's had 3 visible spots in the bulk of the bond. Acetone was used to scrub some residue of epoxy from the surface near the OD, which was likely cured. Short duration exposure (seconds) to acetone at the perimeter of the bond did not yield any weakening of bond.
While test pieces were bonded, Koji was making some adjustments to the cavity alignment in preparation for gluing of the steering mirror BS1.
Koji noticed that the spring clamp was causing pitch in the BS1 mirror, so he recommended that we utilize the "restrain by allen key" technique to load the mirror during curing.
Once aligned, we tried taking the BS1 mirror out of the template and then putting it back. We did this twice and both times the cavity needed realigning (with the curved mirrors as well as the input steering periscope). Why is this? Since the mirror was touching the template it should not have become misaligned right? Maybe the template moves slightly? I think before glueing in the cavity mirrors we should find out why probably? Koji took a look and claimed that a few optics may have been unconstrained.
Planning between Koji and Joe led to placement of 5 drops of epoxy on the BS1 surface, to match the bonding area. At this point we noticed that the template was not secured very well, by poking down on it we could see it move. This might explain why we are becoming misaligned very easily. Once the prism was back on the board, Koji used allen keys to move around the prism. This was done until we could align it again (i.t looked too pitched). The beam was aligned back into the cavity, and the UV light was used to cure the bond. The reflected DC when locked was
so it looks ok still.