QOSEM Noise Hunting - Coil Driver & DAC Noise

Summary: It seems like noise from the coil driver and DACs cannot explain the excess noise seen on the BBSS QOSEMs. The old 18Bit DACs would be limiting in R,P,Y DOFs below 10Hz, The new 32Bit DACs and coil driver are subdominant to ISI noise.

This is one in a series of posts investigating the excess noise seen on the BBSS M1 QOSEMs. In short, at both LHO and LLO, the noisefloor of the QOSEMs appears to be around 8pm/rtHz at 100Hz, quickly increasing to 50pm/rtHz by 1Hz; where we expect the noisefloor to be flat at ~5pm/rtHz across this bandwidth. See more details in LHO alog 91466. 

Attached in Figure 1, is a noise budget for the BBSS M1, showing contributions from ISI motion as measured by the stage 2 GS13's (more details in LHO alog 91744), coil driver output noise, and DAC output noise (both 18 and 32 bit variants). Here we see that the DAC and Coil driver noise should not be limiting the QOSEM noisefloor, particularly above 10Hz. The 18bit DAC may be dominant below 10Hz in the R, P, Y DOFs. It's my understanding that LHO is using the 32Bit DACs, while LLO is using 18Bit DACs for the BBSS M1.

The conclusions here is more evidence that it is something in the QOSEM itself limiting the noisefloor of the BBSS M1. The leading theory is that its due to excess intensity noise from the QOSEM LED. We are currently investigating solutions, like upping the gain on the sat amps LED intensity feedback loop. More details can be found in the follow ticket: 39047

ETMX is rubbing (again)

Keita, Elenna, Louis, Sheila, Oli

It turns out that ETMX IS actually still showing signs of rubbing. We thought we had been in the clear but it looks like we had just found temporary relief from it.

Over the course of trying to lock in the afternoon, ALSX started losing lock more and more frequently in similar looking ways. We couldn't find any quick obvious causes and landed on taking ETMX M0 transfer functions again (reminder that ETMX was looking healthy 2 days ago: 91833). I took an M0 to M0 transfer function measurement with the OPTICALIGN OFFSETs ON, and we saw that we were rubbing >:(. We tried different checks like turning the alignment sliders OFF and ON and eventually we stopped seeing rubbing with the OFFSETs ON (appearently what dislodged it is that we put excitations in the M0 OPTICALIGN filter bank?? I don't remember doing this), but would still see the rubbing when we turned the alignment offsets OFF. This became consistant even while we kept messing with L1 COILOUTF OFFSETs, M0 TEST OFFSETs, and R0 TEST OFFSETs. While we were doing all these tests we were occasionally seeing jolts of movement while ramping stuff on and off, something that I had seen last time and what I imagine is whatever is touching catching an edge on something else before pushing past it.

Keita and I eventually were able to find an offset in R0 Yaw that pulled the LL OSEM away from the LL flag and allowed L1 to swing freely with OFFSETs both ON and OFF. However, when we tried turning the alignment sliders off ~30 minutes later, we found that we were rubbing again even though the R0 Y offset was still on. We also then found that we were now rubbing with the alignment sliders ON, which was strange since we had thought that we had gotten it unstuck when the alignment sliders were ON. Turning R0 Y OFFSET back on did unstick us though, so we are currently free-swinging.

I haven't been able to find definitive proof that ALSX was losing lock because of L1 LL rubbing, though. Looking back through all those short locks, we see that it looks like ALSX DOF2 P (TMSX) seems to grow before we lose lock, but it's not larger than anything that it's able to handle other times during the same locks. This observation also doesn't point to ETMX, but the ETMX channels I looked at didn't show anything that would cause ALSX to drop out.

TF Settings:
- In HEALTH_CHECK, except OPTICALIGN OFFSETs ON in some situations
- DAMP OFF

Table of TFs:
Measurements can be found in $(sustrunk)/QUAD/H1/ETMX/SAG{M0,R0}/Data/
Measurements have been svn'd as: r13143 for M0 and r13143 for R0

	M0	R0
Measurement	OPTICALIGN OFFSETs	TEST Y OFFSET	Rubbing?	Comments
2026-09-10_2145   M0   L	ON	0	YES	Rubbing seen on M0 when alignment sliders were ON
2026-09-10_2145   M0   P	ON	0	YES
2026-09-10_2145   M0   Y	ON	0	YES
Got L1 LL unstuck for when the OPTICALIGN OFFSETs were ON
2026-09-10_2245   M0   L	ON	0	NO	After messing around with different offsets we suddenly were no longer rubbing on M0 when the alignment slider offsets were ON
2026-09-10_2245   M0   P	ON	0	NO
2026-09-10_2300   R0   L	ON	0	NO	Not rubbing on R0 either!
2026-09-10_2300   R0   P	ON	0	NO
2026-09-10_2310   R0   L	OFF	0	YES	Alignment sliders OFF showed rubbing on R0
2026-09-10_2310   R0   P	OFF	0	YES
2026-09-10_2342   M0   L	OFF	-25000	NO	Found the right R0 offset that made it so we were unstuck with alignment sliders both OFF and ON
2026-09-10_2348   M0   L	ON	-25000	NO

Vent Recovery, HAM5-HAM7 Relay Tube Done, HAM7 Ongoing

(Jordan V., Travis S., Gerardo M.)

Installation of the relay tube that connects HAM5 to HAM7 is done.  For the installation we started with the flange on the HAM7 side, only hand tighten the bolts, then moved to the HAM5 side flange.  For that side the compression tool made it easy to handle the bellows, however when we released the bellows no spring back was noted from the bellows because we noticed a gap between both of the flanges involved, the gap was not noted during the removal of the relay tube, we had to compress the bellows some to be able to separate the flanges.  Once the flanges were closed and torqued the volume of the relay tube was pumpdown with a small can turbo backed with a leak detector.  The pressure internal to the relay tube dropped fast, and yesterday afternoon the pressure was low enough to allow us to leak test the joints with the new copper gaskets, and no helium signal was detected by the leak detector above the background that was <1.0X10^-10 Torr*l/sec.

Taking the relay tube volume to a lower pressure
After more relay tube pumping the pressure was low enough to incorporate the its volume to HAM7, this configuration took the pressure down at the the relay tube without affecting much the pressure inside HAM7, see attached plot.  At this time we introduced its ion pump to HAM7, while pumping with the turbo pump, this configuration remained over night.  Today we checked the pressure at the relay tube and it was low enough to change the configuration again, we isolated the relay tube from HAM7 and introduced its volume to HAM5 + main vacuum volume, this is noted on the attached plot.  This was done remove the scattering that RV1 causes to the interferometer.

HAM7 Pumpdown
Today to help the pumpdown at HAM7, and proceed to a nominal configuration, we isolated the turbo pump and allowed its main ion pump to try and maintain the vacuum pressure on the chamber, at the same time we included the volume of section FC-A by opening FC-V1 to aide with its additional 150 ion pump, FC-A section still has FC-V2 closed at BSC3.  As of this afternoon the vacuum pressure at HAM7 leveled and it is turning around.

Thursday OPS Report

TITLE: 09/11 Eve Shift: 2330-0500 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Commissioning
OUTGOING OPERATOR: Corey
CURRENT ENVIRONMENT:
    SEI_ENV state: CALM
    Wind: 26mph Gusts, 17mph 3min avg
    Primary useism: 0.14 μm/s
    Secondary useism: 0.11 μm/s 
QUICK SUMMARY:
More ETMX transfer Functions are currently being ran to determine if ETMX is rubbing.
VAC team is setting up for this potenial vent.

 

 

 

Summary of today's DRMI attempts

Sheila, Elenna, Oli

Starting with a quick summary that I will update later.

In short, the problem in DRMI appears to be some cross coupling between MICH and PRCL.

Jenne noted yesterday that the BS M2/M3 crossover seems odd, but I locked MICH dark and PRMI today and checked the BS M2/M3 crossover and it is fine, as is the MICH OLG in both states. However, once we lock DRMI, the MICH OLG and the crossover look strange. We cannot proceed through the DRMI locking process and 3f switch because we can't engage our normal DRMI filters because MICH is not stable.

Our conclusion is that this comes from PRCL because we change the PRCL gain and see a change in the MICH OLG. We also have tried adjusting the phase of the REFLAIR 45 diode and this changes the MICH olg.

I was in the process of moving the phase of REFLAIR 45 and removing the PRCL from REFLAIR 45 Q, but then we realized we had a much bigger problem with ETMX, alog incoming on that later.

Unfortunately it seems the REFLAIR phasing that Jenne and I did yesterday was incorrect because we could not lock PRMI with that phasing. I reverted that phasing, and we have tried going to more positive phase on the diode instead, which is also helping.

It would be good if we could decouple PRCL and MICH. After that, I think it would be helpful if we could increase the MICH gain and bring the UGF closer to 15 Hz. Sheila and I think we can bump SRCL up to 35 Hz.

Thurs DAY Ops Summary

TITLE: 09/10 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Commissioning
INCOMING OPERATOR: Tony
SHIFT SUMMARY:

DRMI locking work continued, but the afternoon focused on ETMx investigation/troubleshooting (look for an alog on this later).

The Relay Tube was opened up to the main volume (HAM7 & "downstream" remain isolated and pumping down).
LOG:

• All Day:  Chip-sealing on X-arm continues; just passed the +3km mark (randy)
• 1500-1540 Morning Alignment:  Started aligning of cavities noted in Elenna's alog notes);  Took ISC_LOCK to MANUAL_INTIAL_ALIGNMENT: 
  • Green Arms:  Used INIT_ALIGN & selected GREEN_ARMS_OFFLOADED
    • Y-arm had decent flashes, but X-arm was not as good--tried tweaking a little by hand but SCAN ALIGNMENT took over & finished the alignment.
• NOTE:  After Green Arms, used ALIGN_IFO to select different states for aligning other cavities
  • Xarm IR: (completed WFS CENTERING XARM)
    • When Elenna arrived, we checked this again at XARM_IR_LOCKED (I think this was at 1, so no changes made)
  • PRX: (completed WFS CENTERING PRX),
    • Elenna had me redo PRX alignment:  Since no WFS, because rfpds need to get rephased after DRMI work
    • Hand Tweaks:  Went to PRX_LOCKED & then tweaked PRM to maximize AS_A_DC
  • Mich Dark: (lockloss at MICH DARK ALIGN, so I skipped this after 2nd attempt). 
    • Hand Tweaks:  Then just tweaked BS by hand by taking to MICH DARK LOCKED and tweaking BS to optimize Dark spot for AS AIR (or minimize AS_A_DC)
  • SR Align: (This is when Elenna arrived and helped me)   edit:  (completed to PREP SRC ALIGN)
  • SRY:  Go to SRY_LOCKED
    • Hand Tweaks:  While locking, SRM tweaked-up fringes (highs high/low low) 
    • SRY locks and then continue SRM tweaks to maximize AS_A_DC
  • SR2 (completed OFFLOAD SR2 ALIGNMENT)
  • PRX Again:  Make sure PRM is tweaked up (with Elenna)
    • PRM used to maximize AS_A_DC
• 1500-1548 optics lab (kim)
• 1613-1825 spackle 2hrs (chris)
• 1600-1710 TCS Property sorting (tj,christina)
• 1749-1758 H2 room clean (kim)
• 1815-1845 checking on key for a door (tj,richard)
• 1846-1909 Connect Relay Tube to Main Volume (HAM7 will remain valved off to continue pumping down) [gerardo.jordan]
• 2101 Moving parts from Corner to H2 Bldg & MY (fil + marc joining)
• 2125 Plugging in cable at psl racks for JAC measurement; but it was already connected (sheila)

Thurs DAY Ops Summary

TITLE: 09/10 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Commissioning
OUTGOING OPERATOR: None
CURRENT ENVIRONMENT:
    SEI_ENV state: CALM
    Wind: 6mph Gusts, 4mph 3min avg
    Primary useism: 0.01 μm/s
    Secondary useism: 0.11 μm/s 
QUICK SUMMARY:

Locking work continues around DRMI & with other spots.  Guessing the Relay Tube (with no leaks & MegaCleanroom OFF) is still pumping down.

For the commute, crack sealing dealys on SR240 from Twin Bridges to the ORV park road (not sure where it starts in town)...actually WSDOT notes this chip seal work is from Milepost 8 to 28.

Wednesday OPS EVE Shift End

TITLE: 09/10 Eve Shift: 2330-0500 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Commissioning
INCOMING OPERATOR: None
SHIFT SUMMARY:

The Mega Clean room was Turnt Off. For What!? You may ask? Because we are trying to get closer to our final LVEA temperature for locking the IFO. 
GRB-Short E646174 notification @ 1:24:41 UTC
H1 is now sitting in down er rather Prep for locking..


LOG:
                                                      

Start Time	System	Name	Location	Lazer_Haz	Task	Time End
22:35	vac	VACteam	EY	-	vendor visit to EY	23:50
23:42	VAC	Jordan Gerardo	LVEA HAM7	N	Valving swap on HAM7	23:53
00:40	VAC	Gerardo	LVEA	N	Turning Off the Mega Clean room	00:49

Morning locking to PRMI, oplev damping solution, Afternoon L2P decoupling, DRMI

Sheila, Elenna, Jenne

Morning locking notes

• Jenne completed initial alignment by running green initial alignment and through manual initial alignment for X arm IR, PRX, MICH dark, SR align, and SRY. Except for SR2 align, all alignment was touched up by hand.
• We tried locking PRMI once without oplev damping, however we lost ALS with a large beamsplitter kick.
• The second time we tried locking with oplev damping (using oplev as sensor), and we were successful. PRMI locked for 10 minutes. I touched up the PRM and BS alignment. Sheila still measured a large amount of motion at 0.45 Hz. We think this motion is due to length to pitch coupling on M2.
• I tried engaging the L2P decoupling filter in M2 drivealign. We lost lock due to a 19 Hz ringup in MICH. We are not sure if this was caused by the L2P filter or if MICH is unstable. We will remeasure.
• After that we had two lockloss on the first PRMI flash. This flash is coincident with oplev damping ramping on, but we realized it is because the damping is not on before the LSC starts triggering. I moved the oplev damping to engage as "prep DRMI" in the ISC DRMI guardian, so it will hopefully help.

At this time we held in down to allow relay tube work, and used the time to measure the oplev damping and try switching to using M3 wit for lower stage damping instead.

I measured the oplev damping OLGs, pitch and yaw. The reference traces use the oplev as the sensor for the damping, and the live traces use the M3 wit as the sensor but a different matrix value that Sheila calculated based on this alog.

We then switched to M2 damping using M3 wits as the sensor, and compared the BS spectra with no lower stage damping, oplev damping, and wit damping, comparison here. We see good suppression of the modes and less sensor noise injection on the M3 wits.

Here is a screenshot of the "OLDAMP" settings we want to use on M2. I want to clarify that even though we are now using the M3 wits as the sensor, the damping is still running through the oplev damping bank on M2.

At this point, we decided to break further and allow Filiberto to update the resistors on the M1 QOSEMs, so our lockloss transient from saturation is reduced.

***

After Fil and Oli finished the sat amp changes, we began locking again. We locked PRMI with M3 wit damping engaged before the triggering occurred so there was no lockloss. Once locked, the guardian disengages the lower stage damping.

Sheila measured the BS motion using oplev and M3 wit without the L2P decoupling engaged but the wit damping off. She checked the MICH and PRCL OLGs and we saw that they were fine. MICH UGF is 8.5 Hz, PRCL is 60 Hz.

I engaged the L2P decoupling, and we saw that low frequency pitch motion was reduced. However, there was still a large 0.45 Hz peak. With the  M3 wit damping reengaged, the peak is reduced. See plot

Recipe for successful PRMI: M3 wit damping engaged, L2P decoupling engaged. Once we can reengage MICH ASC, we can turn off the M3 wit damping. I have SDFed the filters for BBSS M2 L2P decoupling 1 2

There is a lot of alignment drift in BS yaw while we sit locked. I had to touch up the aligment by hand.

We stayed in PRMI for 25 minutes.

***

We struggled to lock DRMI, and eventually chose to relock PRMI and fix the alignment. After this, DRMI locked quickly. We lost lock at the 3 f transition, so Sheila went to check we were aligned to REFLAIR 3f diodes. Then, I rephased the diodes while locked in PRMI by injecting a PRCL line and maximizing the height in 27 I, and a MICH line to maximize the height in 135 Q. PRCL, MICH, SDF

With DRMI locked, we saw that the PRCL OLG looked strange, so I also checked the REFLAIR 1f phasing. I made a minimal change to reflair 9 to improve PRCL in I and a 50 deg change in 45 to reduce SRCL in 45 Q. PRCL, SRCL, SDF

Sheila's guardian changes, to improve some problems we faced when locking.

• removed srcl mode hop offset
• if PRMI ASC flags are false, the guardian returns true (doesn't check for ASC convergence), we should also do this for DRMI ASC
• changed PRMI to DRMI transition to just zero gains for MICH and PRCL

***

Jenne and I attempted to check the PRCL/SRCL decoupling in the LSC matrix, and measured a value by injecting a PRCL line into reflair 9 and 45. However, the resulting OLGs looked even worse. Both PRCL and MICH look very strange, possibly indicating cross coupling. Unfortunately I don't have images of the PRCL transfer function but here is the MICH tf.

Seems like the phasing is wrong, but the loops became more unstable when we reverted the phasing changes I made earlier.

Wednesday OPS EVE Shift Start

TITLE: 09/09 Eve Shift: 2330-0500 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Commissioning
OUTGOING OPERATOR: Corey
CURRENT ENVIRONMENT:
    SEI_ENV state: CALM
    Wind: 16mph Gusts, 9mph 3min avg
    Primary useism: 0.02 μm/s
    Secondary useism: 0.11 μm/s 
QUICK SUMMARY:
The Comissioners are Working on DRMI alignment and Locking seems like H1 is going through Green Arms and IR just fine.
VAC Team is currently working on HAM7 valving.

Wed DAY Ops Summary

TITLE: 09/09 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Commissioning
INCOMING OPERATOR: Tony
SHIFT SUMMARY:

H1 Commissioning continued with relatively stable PRMI locks + 2-DRMI locks, and continued locking work (i.e. alignment tweaks, Length to Pit decoupling measurements, BS satamp whitening change, rfpd phasing, checking pointing on REFL A/B)

Relay Tube install started late morning with disassembly on HAM5 and then pumping down the Relay Tube (Gate Valves on both sides of the Relay Tube will remain  CLOSED).  Relay Tube leak checked (no leaks observed, VAC will turn OFF Mega-cleanroom at end of day)

BBSS had a satamp change to address 
LOG:

• 1522 Jenne started  on H1 to get to PRMI locking...
  • Green Arms Manual:  Green arms weren't locking, so Jenne touched up both arms by hand and they eventually locked to progress to Locking ALS; then they went to CHECK IR 
  • Had a 10min PRMI while they were running L to G coupling measurements
  • During Relay Tube work, Elenna ran OLG measurements
• 1712-1856 Relay Tube install, pump-down, & leak check (jordan, gerardo, travis)
  • --> Pumping down and maybe leak check tonight or tomorrow <--
• 1728-1847 Alog down for mainteance! (nyath)
• 1807-2012 BBSS satamp chassis change (fil)
  • 2000-2012 investigating change with oli & going out to rack
• 1813-1847 Taking photos of Dustmon pumps at EX
• 1855-1935 EX & EY glycol fill for HVAC chillers (eric)
• 1900-2000 JAC measurements (jennie)
• 1958 Vendor visit #1 (for VAC team)
• 2119-2142 Leak checking Relay Tube (jordan, gerardo)
  • No leaks for Relay Tube; Mega-Cleanroom lights turned OFF, but will turn OFF Mega-Cleanroom fans at the end of the day (to minimize temperature changes for commissioners)
• 2119-2143 EX switching off Turbo Station (travis)
• 21:42:34 Lights flickered in control room
• 2151 Vendor visit #2 (for VAC team)
• 2159 Checking alignment on REFL A/B PD on ISCT1 (sheila)
  • clicking LVEA Local on Reservation System
• 2224 sorting old tcs property for removal along the X-Beam Manifold (tj)
• 2235-2333 VAC trip to EY with vendor.

Repeating actuator calibration measurements for IMC mirrors

I tried to repeat Masayuki's measurement in this alog to calibrate the JM3, MC1, MC2 and MC3 angular motion into uradians using the IMC cavity transmission, see this alog #91487.

This injects a line in turn on each steering mirror in pitch and yaw and looks for the height of the second harmonic of this line in the IMC transmitted power, as this should depend only on the size of the dither we apply to the alignment and not any existing static mis-alignment.

 

First I paused the IMC lock guardian and then turned off the IMC WFS (take gain to 0).

I then injected a line on JM3 pitch, then JM3 yaw. There were both at 30 counts into H1:SUS-JM3_M1_TEST_P_EXC and H1:SUS-JM3_M1_TEST_Y_EXC respectively.

 

The 2f line is viisble at 22 Hz as seen in the transmitted power ASD for the pitch excitation and the yaw excitation.

I then tried to do the same for MC1. I used 17Hz as my dither frequency and checked this didn't overlap with any modes of the MC1 mirror by checking the DAMP filter bank.

I could not get good measurement coherence at 34Hz with a 30 count excitation, when I went up to 60 counts the cavity unlocked, so I turned off the excitation, unpaused the guardian and brought the mode cleaner to down.

This plot shows the ASD during the 30 count injection with the red lines marking the 1f and 2f lines.

Looking back at the ndscope for the time when I thought the IMC unlocked (the mc trans camera spot flashed), it looks like the cavity did not unlock completely but started oscillating at 1Hz.

The first cursor is when the cavity lock started oscillating and the transmitted signal on IMC-TRANS PD went from 11000 counts to 6700 counts. When I turned off the excitation it went back up to the locked level again.

Second cursor is where I purposely brought the mode cleaner to down and the IMC TRANS PD went to 4000 counts.

 

Summary: I can use a dither on JM3 to measure the calibration of this mirror but to measure the MC mirrors I need to tune my excitation more precisely.

BBSS M1 QOSEM satamp whitening modified (again)

Fil, Oli

Fil went ahead and did another modification to the BBSS M1 QOSEM satamp. This is the same modification that LLO had done (LLO:82137) - on each of the six boards, resistors R902, R1002, R1102, R903, R1003, and R1103 were swapped from 499kOhm to 22kOhm resistors.

Before the modification the satamp whitening was zpk([0.0144],[2.89],1), with compensation filter zpk([2.89],[0.0144],1). Changing these resistors has shifted the whitening zero up in frequency so the satamp whitening is now zpk([0.263],[2.89],1), with the new compensation filter zpk([2.89],[0.263],1). I have updated the whitening compensation filters ( BS_M1_OSEMINF_{F1,F2,F3,LF,RT,SD}_{SUM,X_RAW,Y_RAW} ) in OSEMINF to these new compensation filters.

Before we brought the BBSS back up we noticed that the SUM counts for F1 were railed, and that the SUMs for all other osems were between 18-20k counts. This is with the sum feedback on. When we turned the sum feedback off and just sent a constant current to the LEDs, F1 was still railed, but the other osems all jumped to normal varying SUM values that we had seen months ago prior to turning the sum feedback on - F3 and RT were staturating (like they had been doing back then), and the others were around 30k counts. Switching back to the sum feedback mode, we have been seeing the SUM counts all (except F1) climbing up slowly, and will hopefully stop in their nominal feedback on locations. We aren't sure why F1 is showing the counts as saturated though while in sum feedback mode - it hadn't been saturated before, and either way, turning on sum feedback mode should lower the current to the LED to lower the counts to a reasonable level (between 29k and 30k counts).

SPI P and Y Signals, Now Calibrated with QPDA Dark Noise Fixed

J. Kissel, J. Wright

Jennie and Marc did some dark noise investigations yesterday afternoon, found grounding issues, but then reverted the temporary configuration that improved the situation. 
However, this morning, mis-reading their aLOG thinking that the solution was still in place, I re-took new ASDs excited to see QPDA fixed.
Even though their temporary solution has been reverted, QPDA still seems to be fixed. 

So -- here's some excellent, final answer plots for pitch and yaw, comparing HAM2 and HAM3 rotation against their local sensors.

PITCH
    - Now, as expected, the whitened ADC noise QPDA aka ISIK's HAM3 sensor is the factor of ~5x larger where the QPD is limited by ADC noise above 0.5 Hz, and it has the same inverse-whitening-filter shape as QPDB.
    - Encouragingly, the QPDA ISIK optical lever PIT signal matches the HAM3 CRS RY signal below 0.5 Hz -- so this QPD will be at least a little bit useful!
    - The noise floor of the QPDA ISIK optical lever PIT signal is better than the CRS above ~2 Hz, though the actual platform motion (as reported by the GS13s) is still much lower than that.

    Very interesting science -- 
    - The improvement in platform motion of HAM3 from blending in the CRS is corroborated with the ISIK optical lever signal below 0.5 Hz.
    - The QPDB measure of HAM2 shows that there's a lot more physical motion in PIT in the 1 to 10 Hz region than is reported by the in-loop GS13s. In that region, the QPD's signal is a factor of 10 above the noise floor, so I'm pretty sure this is real signal.

YAW 
    - Again, the two SPI QPDs noise floor now makes sense, with the QPDB (on HAM3) measurement of HAM2 (ISIJ) noise floor being a factor of ~5x better than the QPDA (on HAM2) measurement of HAM3 (ISIK).
    - Thus, the SPI OL YAW measure of HAM2 agrees with the local HAM2 sensors up to ~0.8 Hz, but the OL YAW measure of HAM3 only agrees up to 0.5 Hz.


Measurement time: 2026-09-04 16:47 UTC. I took 25 averages with a 0.01 Hz frequency resolution. Hanning window with 50% overlap.

ISIs were isolated, IMC was locked, still no IFO tho. sensor correction was ON in the nominal WINDY configuration. No earthquakes. CRS is blended in with ISI HAM3.
Spots centered, nominal sum voltage.

DTT Calibration of traces beyond what's done in the front-end is what's quoted in LHO:91809, namely -- just a conversion of nanoradians to radians for sensors already in displacement, and converting the GS13s from inertial sensor units asymptoting to 1 [nm/s] into displacement by inverting the ideal 1 Hz pendulum response. 

Calibration of JAC error signal into m

I wanted to get a rough number for the JAC error signal in Hz, this it to allow Jeff and I to measure the frequency noise of the JAC as an independent monitor of the IMC frequency noise which will allow us to compare the IMC length noise to the SPI length noise.

I think Masayuki calculated this back in March LHO alog #89399  but I wanted to recheck it as the Guardian settings have been changed since then.

I measured the unlocked error signal Vpp at H1:JAC-L_SERVO_IN1_DQ to be 1.01365V.

The value in metres is the half-width at half-maximum of the cavity.

First calculate the HWHM in Hz:

HWHM = FSR/2*Finesse  = 0.595 MHz.

The finesse was measured by Masdayuki and I at Caltech for our JAC which is unit 008 (see E2500324).

The FSR of the JAC I got from T0900616 to be 148.5MHz (couldn't find the exact FSR for this specific unit mentioned anywhere so used the one quoted in the PMC design document).

This means to measure the frequency noise we need to multiply the signal at 

H1:JAC-L_SERVO_IN1_DQ by 0.587 e6Hz/V to measure frequency noise.

 

ISI HAM2 and HAM3 Driven in X for SPI H23 Calibration Run

J. Kissel

After tuning up an excitation with color that roughly matches the shape of the quiescent HAM2 and HAM3 displacement, I drove ISI HAM2 and ISI HAM3 in X as a calibration run for H1SPIH23. I'll process and post the results and conclusions tomorrow. 

MEASUREMENT CONFIG NOTES:
Input Mode Cleaner was LOCKED, but no IFO.

The SPI 
   - QPDs were well centered, with their latest calibration installed. Their normalized spot positions remained well below 0.1 [V/V] throughout the excitation. QPDA's dark noise is still terrible; that hasn't yet been resolved. QPDB should be functioning as normal.
   - MEAS and REF IFOs had their nominal contrast of ~75% and 100% respectively, and they remained as such all throughout the excitation.
   - phase unwrapper algo was reset prior to each measurement run to be sure.

HAM ISIs were isolated the entire time, with sensor correction ON. HAM3 was using the CRS in RY.
Blend configurations are screenshotted and attached.

All Suspensions in HAM2 and HAM3 were DAMPED and aligned. PR3 had its Estimators L, P, and Y ON.

(big big 6.something EQ at 11:23 UTC in Alaska / Alutians)
Reference Time  2026-09-03 17:09:49 UTC 
HAM2 excitation 2026-09-03 18:34:14 UTC - 18:58:09 UTC
HAM3 excitation 2026-09-03 18:59:31 UTC - 19:23:26 UTC

Excitation parameters and frequency dependence attached in the screenshots.
Used the same exact excitation for both platforms.
excitation bandwidth was from 0.005 to 200 Hz, but I rolled off the excitation band at either end with a 5th order elliptic with corners at 0.01 and 50 Hz.
Frequency resolution was therefore 0.005 Hz.
Took 10 avgs.

Corner station Sensor correction mode was WINDY during the HAM2 excitation.
Interestingly? Sadly? there was an 5.3 mag Alaskan earthquake in the middle of the HAM3 excitation (started at 18:59:31 UTC). That automatically changed the corner station sensor correction configuration to CONFIG_EQ at 19:05:17 UTC, then back to CONFIG_WINDY at 19:15:18 UTC.

The typical total RMS of each ISI X motion during the the excitation was ~1e-6 [m]_RMS. 

DTT Templates and results are saved to the following DTT templates
   /ligo/svncommon/SeiSVN/seismic/Common/SPI/Data
       2026-09-03_170949UTC_H1SPIH23_DIFF_DISP_ASD.xml
       2026-09-03_183414UTC_H1SPIH23_DIFF_DISP_ASD_HAM2Drive_X.xml
       2026-09-03_185931UTC_H1SPIH23_DIFF_DISP_ASD_HAM3Drive_X.xml

SPI Readout vs. ISI Sensors: A First Comparison

After the angle calibrations were applied to the SPI readout channels, I grabbed some data to check whether the SPI is roughly agreeing with the ISI sensors. I used the supersensor OUT channels for the ISI sensors (these are the signals behind the blending filters, before the summation for the blended supersensor). Screenshots from diaggui are attached.

For the differential X degree of freedom (SPI length), the visual agreement is remarkably good.

The optical lever (angle) measurements are harder to interpret. First of all, QPD A has an as-yet-unresolved noise issue, which Jennie is currently investigating. Aside from this, the results look very encouraging, since the CPS and GS13 at least somewhat agree with the SPI.

These spectra should be regarded as a first result, but not yet final. We will post a better analysis once the error has been resolved.

Note: If anyone has suggestions for which sensor channels would provide a better/more appropriate comparison than the supersensor OUT channels used here, please comment, your input is welcome.