The default workstation conda environment was updated.  ndscope is updated to 0.20.7.  A list of changes can be found here: https://git.ligo.org/cds/software/ndscope/-/blob/master/CHANGELOG.md.

tmux and some other system commands were dropped from the conda environment.  tmux should now work in the terminal.

J. Freed, J. Warner

We set up a small rack to be put into the optics lab for the SPI build housing (most) SPI electronics and hooked up all the power cables for these electronics (For the build we are using a dual source RF generator to instead of the 80MHz OCXO and Single Sideband Mixer like in the final set up). Before this set up can be used, a RF power meter should to be used to check to make sure the correct RF power is going to the AOMs in the SPI laser prep chassis. 

After applying power to the TIA chassis I also checked that the DB25 on the front panel of varient 2 of the TIAs was correctly outputing -15V on pins 1-10. 

Results:

Pins 1-4,7-10 were outputing -14.8V

Pins 5-6 were outputing -15.1V

This discrepancy in pins power is expected and is caused by the fact that the Variant 3 TIA is the one that is actually powering pins 5-6 through the front panel DB9 cable from varient 3 to varient 2. 

J. Kissel, R. Short, J. Oberling

This is the first installment of aLOGs documenting the setup of a new stand-alone 1064 nm NPRO laser system whose current "end game" intent is to provide ~100 [mW] of fiber-coupled p-pol light to the SPI laser prep chassis.

Step 1: Gathering materials, find what optics / mounts we had vs. what would be need, and physically layout the plan.

Jason lets Ryan and I know that there are three NPROs in the optics lab, two of which are PSL spares that cannot be used. 
The remaining laser is S/N 1661, the laser used in the PSL during O3 which is *functional* but was briefly installed during O4 circa Fall/Winter 2024 then removed from use in the PSL because of reported glitching / incompatibility with the frequency stabilization servo (FSS) issues -- see the bottom of LHO:81391 for a nice summary, and LHO:81409 for record of its removal.

Inspired by the setup at Stanford Sina shared with us, we're looking to build up the following system to accomplish the goal:
    - NPRO (presumed to be elliptically polarized with Is / Ip = 5:1)
    - QWP (to linearize the polarization)
    - HWP1 (to rotate the polarization into horizontal)
    - FI (accepts horizontal linear polarization, to ensure back-reflections from down-stream components don't seed the NPRO causing glitches/mode hopes/frequency noise)
    - HWP2 (to rotate the FI output polarization into the desired amount of vertical polarization -- aka the desired amount p-polarization)
    - PBS (to filter out and dump the unneeded horizontal / s-pol light, and transmit the desired power of p-pol)
    - SM1 (one of two steering mirrors to align the beam into the fiber collimator)
    - L1 (the single-lens mode-matching solution to convert the NPRO beam into what the fiber collimator needs)
    - SM2 (two of two steering mirrors for alignment into the fiber collimator)
    - 50:50 PWR BS (45 [deg] AOI, optimized for p-pol; to provide a pick-off port for live power measurement)
    - Fiber Collimator

Ryan started with a 24" by 12" breadboard that was lying around in the optics lab. He build up a makeshift stand from three posts and dogs in the lower left corner such that the S/N 1661 NPRO projects the beam at 4" height. The 0.5" thick breadboard has a 1 inch hole pattern offset by 0.5" from the edges. I'll refer to this grid as having axes "m" and "n" where the m-axis are the "row" holes running from 0 to 23, and the "column" n-axis holes run from 0 to 11. I chose (m,n) breadboard coordinates so as to not confuse them with traditional beam profile coordinates of (z [propagation distance], x (transverse horizontal), y (transverse vertical)). 

Thus, the NPRO being in the "lower left" means it projects the beam along the m = 3 row, and the front face of the NPRO is sitting at n = 8. We'll call this beam position z = 0.

We then proceeded to gather as much as we could of optical components from the optics lab drawers, and ended up with this pictured preliminary version of the setup.

Step 2. Power up the NPRO.
Here's were we ran into our first snag. Normally, NPROs are paired/tuned with specific controller boxes. However, when Ryan turned on the S/N 1661 laser with the S/N 1661 control box, the crystal temperature readback reported the temperature was quickly, linearly rising well beyond the desired temperature of 24.7 [deg C]. At ~42 [deg C], (but still below the internal automatic watchdog threshold of 50 [deg C]), Ryan knew something was wrong and turned it all off. He repeated the turn on just in case, and it did the same.

After conversing with Jason, we figure it's good enough to run with one of the other controllers for now, and in the mean time figure out how what's wrong and repair the S/N 1661 controller.

So, we're running with the S/N 7974, and things seem fine. 
    Laser Diode Temp Setup:
        Diode 1 33.7 [deg C]
        Diode 2: 33.1 [deg C]
        Laser diode injection current readback 2.08 [A] (for both diodes)
    Crystal Temp setting is 24.7 [deg C]

We measured the output power*** with no optical elements at all as 1.820 +/- 0.005 W (not noisey, but a slow drift around).
Good enough! Onward and upward!

*** Power measured by ThorLabs power meter
	Model S302C
	SN 111149
	Sensing factor of 316.25 [mV/W]
	(last calibrated Feb 3 2012).

J. Freed (with help from Marc and Fil)

I tested the real wiring chain described in D2400111 from PD through the 3 variants of the TIA and it does follow exactly what is described in the link. There were 4 things of note that were found during this test. 

1. All the whitening filters were switched "on" inside the TIA, however our final design document states that only the QPDs need whitening. As such the SPD boards S2500647 and S2500645 whitening switches were set to bypass. I logged this change in the E-Traveler.

2. The D1700116 silicone circuit board for the FFD-200 causes shorts. The main issue is the GAP 5000 pinholes on the PD side of the board. If the PD is flush with the board those pins touch the case of the FFD. Since the GAP pins are also connected to the cathode and anode, this causes the circuit to short. This is simple enough to test without taking out any PDs from the case. If there is "no" resistance between the case (ground) pin and the anode/cathode pins on the DB9 connector coming from the PD, then it is shorting and the system should not be turned on. There are 2 options suggested to fix this, either put insolation between the GAP pins and the FFD case or cut the GAP pins connection to the FFD anode/cathode. For this specific test, the issue was bypassed by putting in a tissue between the FFD and the silicone board.

3. I tried to take transfer functions from all the PDs through the TIA, however the laser I used (the MIT ISC AM Laser) AM modulation was designed for 5 - 200 MHz, not the 4096Hz we are trying to use. As such, I suspect the harsh attenuation (~70dB at 4096Hz AOM_Attinuation.png) I measured was caused by the laser. So I decided it would be better to put off the transfer functions through the TIA until after SPI is built, then use our own laser AOM system to run the test. The laser was just left at DC and was used as the light source to see if signals on the PDs follow the correct wiring chain.

4. I used the laser to check that the signals on the PDs follow the expected wiring chain through the output of the TIA. Ex. Laser on Quadrent 1 of the QPD outputs a voltage on the expected pin on the output of the TIA. All pins were found to be correct.

Jason, Jennie, Betsy

After working for a bit to try to beam dump the second Septum refl beam dump which also is coming off the IMC septum window, we finally came to a solution.  It involved stealing one of the SPI D1800140-07 smallest beam dump panels and also putting it into the same mount as the first panel.  These 2 beams are pretty close to the main beam at this location, but with the beams otherwise hitting the internal metal of the output periscope we could not come up with anything better.  It took us quite a few small adjustments and many pictures to convince ourselves we are not clipping, but the attached picture is the best visual of the situation.  Commissioners agree that this is what we will live with.

S. Muusse, C. Compton, S. Goswami

We started populating and aligning optics on the second table. Most optics are now mounted and the beam is aligned upto M3 as per the fintrace model. We have aligned with the PD pick off in to save realignment later.

Construction begun on the camera can (D1300703) but did not get far as the screws which are fixed into the side panels (D2500165) are not long enough to reach the nuts attached to the other face of the box. Pictures of these components are attached.

We attached the viewport housing on the table which fits as expected. (photos attached)

J. Kissel

Executive Summary
Using 
   - the AxcelPhotonics butterfly diode laser,
   - the "beam splitter / near-field / bypass" measurement setup described in LHO:89123 to rapidly check the beam diameter at z = 0.991 [m] (near field) and z = 5.41 [m] (far field), 
   - elevating the beam height elevated to 5 [inches] to avoid any sort of clipping on other optomechanical setups on the table from other teams, and
   - loosening the lens position set screws only just barely, 
   - paired fiber collimators, in-vac feedthrus (with integrated patch cords), and ex-vac 7.5 m patch cables in-vac feedthrus 

      Beam Path   Char Date    Collimator   Feedthru (Pwr Transmission*)
      MEAS        2026-02-13   S0272503     S3228003 (100%)    
      REF         2026-02-17   S0272502     S3228002 (99%)    

      * Power transmission as reported from SWG:12296

I was "rapidly" tune the lens position, z_lens, for *both* the MEAS S0272503 and REF S0272502 fiber collimators to within 0.010998 (i.e. 11 [mm] - 2 [um]) with uncertainty of +/- 1 [um] , such that 

each fiber collimator sends out a free-space beam whose parameters meets SPI's requirements:
   - the desired waist radius (in both x/y dimensions) , w0 = 1.05 [mm], of within +/- 0.10 [mm], and
   - the desired waist position (in z), z0 = 0.0 [m] to within +/- 0.18 [m].

(where "rapidly" is in quotes: about 4 hours each from "install into measurement setup" and "I'm happy with a final full-position-vector scan). with no signs of strong astigmatism that I saw in the initial setup. (The later point confirms my suspicion that I was clipping on the EOM characterization setup, and there is *no* issue with the laser seed mode, polarization, or the beam splitters).

Support Media

The raw data from the beam profiles of each free-space beam are posted as
    2026-02-17_spi_fc_S0272502_ft_S3228002_7p5mPatchCord_AxcelPhotonicsLaser.txt
    2026-02-13_spi_fc_S0272503_ft_S3228003_7p5mPatchCord_AxcelPhotonicsLaser.txt

where the columns are jammt-friendly format of z Position [cm], Y waist radius [um], X waist radius [um]. 

Notes: 
    (1) as opposed to all previous data sets (e.g. LHO:89047, LHO:86350, LHO:84825) -- I added an additional measurement at z = 0.25 [m] to try to improve the accuracy of the fit beam.
    (2) X is "Axis 1" of the NanoScan parallel to the optical table, Y is "Axis 2," perpendicular to the table. As discussed in LHO:89099, jammt seems to import any three-column dataset such that the first column ends up fit as the "tangential w0," and the second column end up fit as the "w0." With a la mode, the data is typed in a matlab script manually, so X and Y data are modeled separately the entire time without rename, and thus kept consistent. Thus the intentional flip the X and Y axes in the text file such that a la mode and jammt are now both treating w0 = X = Axis 1 = parallel to table, and tangential w0 = Y = Axis 2 = perpendicular to the table.

I imported these beam profiles into both matlab (to run a la mode) and jammt to obtain waist radius, w0, and waist position, z0, fits to the AxcelPhotonics beam profiles, see attached plots:
    S0272502  a la mode  jammt
    S0272503  a la mode  jammt

Some pictures of the measurement setup with the NanoScan head at the new z = 0.25 [m] position (upstream of the beam splitter bypass):
    2026-02-13_FC_S0272503_S3228003_MeasSetup.jpg
    2026-02-17_FC_S0272502_S3228002_MeasSetup.jpg

Some pictures of the clean fiber collimator + vacuum feedthru systems packed up after measurement:
    2026-02-13_FC_S0272503_S3228003_PackedUp.jpg
    2026-02-17_FC_S0272502_S3228002_PackedUp.jpg

Detailed Analysis Results

Here's a table of the fit results from both a la mode and jammt. 


        

            

                
FC + FT S/N
Model
w0x [m]
z0x [m]
w0y [m]
z0y [m]
            
        
        

            

                
S0272502 + S3228002
a la mode
1.0392
-0.065196
1.0427
-0.020038
            
            

                
jammt
1.0392
-0.06521
1.0427
-0.02007
            
            

                
S0272503 + S3228003
a la mode
1.0396
+0.17704
1.0282
-0.030537
            
            

                
jammt
1.0396
+0.17702
1.0282
-0.03051
            
        
    
It's not a typo, the two fitting softwares agree to within the precision of the each display; +/- 10 [um] on both the waist radius and waist position.

From here on in this section, since I intentionally paired up fiber collimators and fiber feedthrus to match the last two digits of serial number, I'll refer to them as either "MEAS = S[...]03", or "REF = S[...]02".

From the table of fits, you see that the statement in the executive summary about the waist position, z0 = 0.0 +/- 0.18 [m], is defined expanded to cover the X axis waist position of MEAS = S[...]03 z0x = +0.177 [m]. But really, the other axis of the MEAS = S[...]03 FC+FT pair is at z0y = -0.031 [m], and the REF = S[...]02 FC+FT pair is within z0x = -0.065 [m], and the best z0y = -0.020 [m].
So the waist positions are really *quite* close to 0.0 [m]. 

Good. So let's use these numbers to do recast the fit (using jammt numbers only) into context:
    (1) Percent Difference between desired waist radius and final measured waist radius:

        REF, S[...]02 :: (w0x, w0y) = ([1.0392 1.0427] - 1.05) / 1.05
                                    = [-0.0102860  -0.0069524] 
                                    = [-1.0% -0.7%]
                    Within 1.050 +/- 0.1 [mm] = 1.050 [mm] +/- 9.5 [%] = [1.15 0.95] [mm]?
                    Both axes waist radius are a factor of 10x better than reqs.
        MEAS, S[...]03 :: (w0x, w0y) = ([1.0396 1.0282] - 1.05) / 1.05
                                     = [-0.0099048  -0.020762] 
                                     = [-1.0% -2.1%]
                    Within 1.050 +/- 0.1 [mm] = 1.050 [mm] +/- 9.5 [%] = [1.15 0.95] [mm]?
                    Both axes waist radius are a factor of 5x better than reqs.
        

     (2) Rayleigh Range:  

         (zR := pi * w0^2 / lambda)

         REF, S[...]02 :: (zRx,zRy) = (3.1886, 3.2102) [m]
         MEAS, S[...]03 :: (zRx,zRy) = (3.1911, 3.1215) [m]

         There's no requirement on where the Rayleigh range sits, but for a waist radius of w0 = 1.05 [mm], we would expect a Rayleigh Range of zR = 3.255 [m], and these values are at most z = -13 [cm] from that or 4% "short" of the target value. This doesn't really matter to SPI, as long as the Rayleigh Range is somewhere in between the ISIK breadboard and the ISIJ reflector 15.427 [m] away. The real requirement is the spot size at the ISIJ reflector, which we need to keep at the design value of 5 [mm] (whose value was determined with the original design waist radius of 1.05 [mm], and the acceptance that we didn't have enough room on the ISIK transceiver breadboard to install telescopic lens solutions to keep the spot size around 1 [mm] both within the transceiver *and* at the ISI reflector). 

      (3) Astigmatism: 

          A := (zRx - zRy) / (zRx + zRy)

          REF, S[...]02 = -0.003376 = -0.3%
          MEAS, S[...]03 = +0.011026 = +1.1% 

          Here, again, there's been no requirement on the astigmatism, but ~1% astigmatism doesn't smell too terrible. 

      (4) Spot size at 15.427 [m]: 

          w(z) = w0 * sqrt(1 + (z/zR)^2) 

          REF, S[...]02 :: (x,y) = (5124.569, 5155.038) [um]
          MEAS, S[...]03 :: (x,y) = (5194.496, 5075.909) [um]

          Recall the design value of waist radius at the ISIJ reflector, z = 15.427 [m], is w(z = 15.427) = 5 [mm]. As discussed in SWG:12273, that ADC noise with the large spot size is limiting the sensitivity of the pitch and yaw readout, as 

          dx/dTheta ~ sqrt(8/pi) L / w(z).
    
          But still, a 1.1% level of astigmatism -- which would result in a different sensitivity / noise performance between pitch and yaw, means that pitch is only ~2.1% worse than yaw.      

Great! We're good to go!

Jennie W, Ryan S, Keita K

Summary: MC suspensions back to good time from 10th of Feb, we should only lock the IMC length loop manually during this installation period to avoid the WFS engaging.

Today Ryan and I had the task of figuring out why the IMC suspensions seemed to be badly aligned on Friday.

We found a good reference time 16:48:29 UTC on Tuesday 10th Feb at the end of the period where Olli, Jenne and I set the IMC axis back to its nominal state before the vent using ASC loops for DOF 1 and 2 of the IMC plus manual moves of the JM3 mirror in chamber as this cannot be moved by an ASC loop as it is not a tip-tilt currently. See photo of the WFS and MC mirror top mass OSEMS here. Note it is important to choose a time when IMC is in the locked state as the MC2 mirror gets mis-aligned when the IMC guardian is offline.

We figured out that the alignment was changed about 19:33:17 UTC on 13th Feb probably by the IMC guardian as Elenna used it to lock the IMC for phase check measurements. See photo, where the vertical cursors are set at the good reference time and today.

After some investigation of the guardian we reminded ourselves that DOFs 3-5 are turned off in the IMC WFS MASTER filter banks. DOF4 and 5 are normally off and DOF3 is intentionally left off because of the replacement of PSL PZT actuator by JM3.

When the IMC guardian is used to lock it automatically triggers the WFS through some logic from the simulink model, and so when Elenna used the guardian to lock it turned on the feedback to DOFs 1 and 2.

Until we install the JM3 tip-tilt and are finsihed most commissioning of the HAM1 hardware we should only lock the IMC manually by engaging the MC-L servo using the common mode servo board controls.

Keita and I zeroed the M1 LOCK fiters for all 3 MC mirrors and I changed the alignment sliders so the M1 osems were back to their values from the reference time on the 10th. See photo of ndscope.

Morning report.

Pictured the beam position on the input and the output side plate of the EOM.

We'll see if it's off or not, doesn't look bad but we'll see once I have time to look at the pictures.

We realigned the beam into IMC, mode mismatch < 0.4%.

I just assumed that the MC alignment itself was good. I found MC2 in misaligned state and changed it to aligned.

We had to work both on PIT and YAW, but mostly in PIT (i.e. the difference in horizontal beam deflection between the old batch crystal that we're using and the new batch crystal is smaller than the difference in vertical deflection). 

Good news is that the mode matching is still pretty good. Attached is the MC2 trans VS time while the PSL frequency was ramped. Assuming that the tiny thing pointed by the green arrow is actually 2nd order HOM peak, the mismatch is 0.036/9.4~0.4%. Consider this as an upper limit, because it's hard to tell if the "peak" is actually a peak, could be noise fluctuation.

Today Randy T and I ran power for the Cross Flow HEPA fans on the platform around BSC2.  The fans are on and turned down to low at the moment.  We also moved the dust monitor from the floor to the platform to monitor dust counts to have a better understanding before pulling the dome.

M. Todd, S. Muusse, C. Compton, S. Dwyer

Today we ran some more OMC scans with the ITM ring heaters on. At first we ran the OMC scan with the 9/45 sidebands on, single bounce off ITMX. Then we turned sidebands off and did both ITMX and ITMY.

Squeezed in a quick photoshoot of HAM7 (right after ISI-unlocking/measurements by Jim + purge air being turned up and before Door Team sealed it up).  HAM7 was open roughly from 11:16am-11:48amPDT for photos.  Since there was only one door off, lighting wasn't great and limited to shots from the -X side of HAM7.  Used Canon DSLR camera followed by iPhone shots; snapped a handful of macro lens shots for fun.  

Photo Album (109photos) uploaded to Google folder HERE.

TITLE: 02/03 Day Shift: 1530-0030 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Planned Engineering
OUTGOING OPERATOR: None
CURRENT ENVIRONMENT:
    SEI_ENV state: MAINTENANCE
    Wind: 10mph Gusts, 7mph 3min avg
    Primary useism: 0.03 μm/s
    Secondary useism: 0.47 μm/s 
QUICK SUMMARY:

Reminder that the LVEA is LASER SAFE EXCEPT around HAM1/2 and at height.

Summary

We improved the IMC mode matching by repositioning JM2 and JM3 based on the calculation from yesterday. The mode mismatch was reduced from about 10% to ~2% after iterative alignment and mirror position optimization. Further improvement is expected with additional calculations and tuning, which will be continued tomorrow.

Details

We worked on improving the IMC mode matching following the calculation from yesterday. As a first step, JM2 was moved by approximately 3 inches; the new position is shown in the attached photos. The alignment to the IMC was performed using the newly placed iris in front of JAC_L2 and the iris after the output periscope, as described in the previous alog. By centering these two irises, the alignment could be brought to the level where IMC flashes were visible. From that point, adjusting JM3 allowed us to easily reach an alignment where the TEM10 content was at the ~10% level.

After achieving a reasonable alignment using a scratched mirror, we replaced it with a newly cleaned narrow-angle mirror for JM2. The scratched mirror was moved to JACR_M1. As a note, the scratch was oriented on the +y side; by keeping the beam closer to the −y side, the impact of the scratch was minimized.

With this configuration, the mode mismatch improved from about 10% to approximately 4%. Since the calculation suggested that further improvement should be possible, we continued tuning by adjusting the JM3 position. Based on the previous calculation indicating an offset in JM3, we first moved JM3 by about 1/2 inch in the −x direction (increasing the L1–L2 distance). This resulted in a degradation of the mode mismatch to about 6%. We then moved JM3 in the +x direction by a total of 1.5 inches (i.e., 1 inch further from the original position), effectively shortening the L1–L2 distance. With this adjustment, the mode mismatch recovered to approximately 2%.

We stopped the work at this point and plan to perform updated calculations tomorrow to guide the next iteration of tuning.

We've assembled the whole EOM assembly including the pivot base/plate assy and the strain relief post and confirmed that the EOM face/side/bottom plates are isolated from the pivot base/plate.

We've tuned the EOM such that S11 parameter at the modulation frequencies are all within 1dB of the bottom of the dip.

We still saw that the frequency shifted for 118 and 45MHz when we tapped the EOM body around, even though there was no appreciable change in 24 and 9MHz. Each jump was small, like 5kHz or 10kHz or nothing sometimes, but they jumped. Surprisingly, at least for 118MHz peak, after the frequency jumped in one direction due to my tapping on the side plate (e.g. the input side), I tapped the other side plate (e.g. the output side plate) and the frequency jumped back into the opposite direction. This was very consistent.

This means that my expectation was wrong. An expectation that somehow something (like coil wires) is caught by something else (like the core by the friction), and that tapping things will release these "something"s into lower potential state and that eventually they all settle.

Realizing this, we stopped tapping, we just tuned as good as we could and stopped.

Tools and parts were wrapped and bagged. EOM assy was wrapped in a foil. These things will be transported to HAM1 tomorrow. 

In the first two attachment, Elenna is tuning the trickiest frequency (118MHz). 

Following pictures show the fully assembled unit from the back (+Y), input (-X), output (+X) and front (-Y). In front.jpg, note that I'm only using one washer for the low profile 10-32 socket head cap screw. I used to use two, but the screw still stuck out just enough to make things tedious and inconvenient when lowering the pivot plate over the pivot base. Though the screw is 0.5" long and not the initially specified 10-32 x 0.375", with one washer it seems to work. 

Next, look at one 1/4-20 screw below the above mentioned 10-32 in caution.jpg. When tightening/loosening that screw, don't use a T-handle wrench unless it's a ball end tool, because you WILL damage the SMA connector thread. Even if you use an L-shaped tool, if you don't pay attention you can damage it. Be careful.

Last picture showcases a happy mood in the optics lab after the successful day.