Most likely from PEM injections, HEPI BS seemed to trip first and then we lost lock 1.2 seconds later.

M. Todd, S. Dwyer, T. Shaffer

After the 1Hz ring up causing a lockloss we decided to move LOWNOISE_ASC to just before NLN and gave it a 5min timer before it goes and sets gains. This should hopefully give the IFO enough time to thermalize past MAX_POWER and keep the higher gains in the ASC loops until the 1Hz oscillation comes back down.

Thu Nov 06 10:16:33 2025 INFO: Fill completed in 16min 29secs

Last checked in alog87617, Closes FAMIS27430

The OUT building fans look the same as during the last check.

The CS fans, MR_FAN1_170_{1.2} look to be ever so slightly increasing in motion over the past 3 days, the rest look just as they did during the last check.

TITLE: 11/06 Day Shift: 1530-0030 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Lock Acquisition
OUTGOING OPERATOR: Ryan S
CURRENT ENVIRONMENT:
    SEI_ENV state: USEISM
    Wind: 26mph Gusts, 19mph 3min avg
    Primary useism: 0.06 μm/s
    Secondary useism: 0.46 μm/s 
QUICK SUMMARY:

IFO is LOCKING at LOCKING_ALS

Wind and microseism were high last night and microseism is still quite high and over the 90% threshold. H1 was finishing an automatic initial alignment when I arrived.

Relocking now.

TITLE: 11/06 Eve Shift: 0030-0600 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Observing at 19Mpc
INCOMING OPERATOR: Ryan S
SHIFT SUMMARY:

H1 was locked the entire shift.  There was one quick drop due to an "automatic" PMC Voltage adjustment.  Microseism is still high and mostly stable at same level for the last 20hrs.
LOG:

• 0327-0330 SQZ lockloss due to PMC having too high of a voltage (see attached)---in contrast to last night when it was too low.  It relocked automatically and PMC Voltage went from 93 down to 30V.
• 0348-0533 Cleaning SPI optics & organizing in Optics Lab (corey, with remote H1 alerts on my phone enabled)

Attached are the plots for all four Test Masses.  Closing FAMIS 28430.

Attached are monthly TCS trends for CO2 and HWS lasers.  (FAMIS link)

NOTE:  The change for ITMx HWS is due to its SLED being swapped on Oct 7 (alog 87353).

TITLE: 11/06 Day Shift: 1530-0030 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Observing at 148Mpc
INCOMING OPERATOR: Corey
SHIFT SUMMARY: We had 3 lock losses from Close_Beam_Diverters today. The first reminded us that the issue is still present, even though we didn't have a problem getting locked last night. For the second, we tried to delay when the OMC whitening came on, thinking it was getting engaged too soon while the beam diverters were closing. The third we spaced out when the beam diverters closed, so they wouldn't close all three at once. It lost lock just after the first one closed. We then relocked and went to observing with the beam diverters open, but this ended up being a rather short, 1hr, lock. On the final attempt of this shift, we tested out moving the beam diverters and ended up moving the ISC_LOCK state that closes them to just after engaging ASC for full IFO. We have been observing since. See alog87970 for more beam diverter investigations.
LOG:

• 2027 Observing with beam diverters open
• 2121 Lock loss
• 2340 Observing with beam diverters closed           

TITLE: 11/06 Eve Shift: 0030-0600 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Observing at 148Mpc
OUTGOING OPERATOR: TJ
CURRENT ENVIRONMENT:
    SEI_ENV state: USEISM
    Wind: 18mph Gusts, 15mph 3min avg
    Primary useism: 0.04 μm/s
    Secondary useism: 0.57 μm/s 
QUICK SUMMARY:

H1 continues to have a rough time with locking....where the pesky HAM1 Beam Diverters have been a focus.  Additionally, in the background, this all happens with very high µseism!  (peaked 17hrs ago and slowly dropped a tiny amount since then.  Looks like we might have just finished riding through a small wind storm.  At the moment, H1 has been locked for almost 1hr.

Observing at 2340UTC with the beam diverters closed this time. We have moved the ISC_LOCK state the closes the beam diverters to right after Engage_ASC_For_Full_IFO. 

On our way up we did some testing of the beam divereters, see alog87978. Both HAM1 beam diverters show a similar response in the GS13s to the one in HAM6, but the one in HAM6 is 10x smaller in magnitude. 

We are back to Observing at 2027UTC but we have the REFL, AS, and POP beam diverters still open. See alog87970 for more info.

[Alicia Calafat, Joan-Rene Merou, Sheila Dwyer, Robert Schofield, Jenne Driggers]



Scope
We analyze violin harmonics and their intermodulation products in H1:OMC-DCPD_SUM_OUT_DQ, and estimate the quadratic mixing coefficient β for those products. Peak detection and confirmation use an Ansel-inspired adaptive-threshold approach on a stabilized series, combined with simple height floors. We then confirm analytically expected intermodulation frequencies and compute beta from amplitude products of paired parent lines and their intermodulation.



Calibration lines at 15.6, 16.4, 17.1, and 17.6 Hz:







The full-band ASD (about 10 to 1610 Hz) shows violin harmonics near 500 Hz (v), about 1000 Hz (2v), and about 1500 Hz (3v), together with the labeled calibration tones. This context motivates the intermodulation search at violin plus/minus calibration, and at violin-violin sums and differences across the band.







Problem and physical context
Small-signal readout model (mA domain, OUT):

y(t) = α s(t) + β s(t)^2

where s(t) is the photocurrent at DCPD SUM (mA), α is the linear gain, and β is the weak quadratic nonlinearity coefficient (mA-1).

Single violin tone at fv and calibration tone at fcal:

s(t) = X sin(2 π fv t) + Z sin(2 π fcal t)

Square s(t):

s(t)^2 = X^2 sin^2(2 π fv t) + Z^2 sin^2(2 π fcal t) + 2 X Z sin(2 π fv t) sin(2 π fcal t)

Use identities:

sin^2(u) = (1 - cos(2u)) / 2    2 sin(a) sin(b) = cos(a - b) - cos(a + b)

Decomposition

  - Self terms:

    X^2 sin^2(2 π fv t) → DC and a line at 2 fv.

    Z^2 sin^2(2 π fcal t) → DC and a line at 2 fcal.

    - We ignore DC (0 Hz).

    - The 2 fv component is used inside the violin–violin families as the i = j "sum" case (often labeled vv_h2). Estimator: βself(v) = A2v / (Av Av).

    - The 2 fcal component is not used for β (no "cal–cal" family here), though it may be visible.

  - Cross term (interaction):

    2 X Z sin(2 π fv t) sin(2 π fcal t) = X Z [ cos(2 π (fv - fcal) t) - cos(2 π (fv + fcal) t) ] → mixing lines at fv ± fcal with amplitudes ∝ β X Z.

Spectral amplitudes from PSD (same convention for all lines)

PSD has units mA2/Hz. Let Δf be the bin width. For a narrow line:

Abin(f) = sqrt(PSD(f) · Δf) = ASD(f) · sqrt(Δf) [mA].

For robustness we integrate ± 1 bin around the peak index:

Abin ≈ sqrt((PSD[i-1] + PSD[i] + PSD[i+1]) · Δf).

Quadratic mixing scaling used for the estimator:

Amix ∝ β · AL · AR.

Per-line β estimators used here:

  - Violin ± calibration:

    β = A(v±cal) / (Av Acal) [mA-1].

  - Violin self-mix (i = j) at 2 fv (vv_h2):

    βself(v) = A2v / (Av Av) [mA-1].

  - General violin–violin (i ≠ j) sums/diffs:

    β = A(f_i±f_j) / (Af_i Af_j) [mA-1].



How the scripts work
  - Intermodulation explorer: input long spectrum (f, PSD); derive ASD; estimate band medians; detect violin peaks with an adaptive threshold on a stabilized series plus simple height floors; propose and confirm intermods at violin ± calibration and violin–violin sums/diffs; export plots and a JSON summary.

  - Beta estimator: read the JSON and the same spectrum; reconstruct A_left, A_right, A_mix via ± 1-bin PSD integration; compute beta_hat = A_mix_bin / (A_left_bin · A_right_bin); summarize by families and produce plots.



Current intermodulation detections
These are the retained detections after applying our thresholds and pairing rules. They are the inputs used downstream to compute and summarize beta by family.

  - Detected violin peaks kept: v = 20, approx 2v = 25, approx 3v = 30.

  - Intermod candidates before pairing: v plus/minus cal total = 154; violin–violin (sums and differences) total = 761.

  - Paired intermods kept for beta: v plus/minus cal = 22; 2v plus/minus cal = 18; 3v plus/minus cal = 18; vv(fund) = 386; vv(2v) = 247; vv(3v) = 128.



Beta results (paired families)

  

    

      
Family
      
Paired count
      
beta median [1/mA]
    
  
  

    
v ± cal
22
0.638
    
2v ± cal
18
0.450
    
3v ± cal
18
0.286
    
vv (fund)
386
10.824
    
vv (2v)
247
1.586
    
vv (3v)
128
13.329
  




Plots

1) Calibration-violin explorer


Calibration tones align as symmetric sidebands around violin harmonics at the predicted offsets (v ± fcal). This validates the pairing rules used for beta estimation and shows that v ± cal families are present at v, 2v, and 3v.

2) Violin-violin explorer


Sums and differences of violin peaks form dense loci at the predicted frequencies. Violin-violin intermods are abundant in all three bands.


Meaning of "violin–violin": all sum/difference pairings between detected harmonics (v, 2v, 3v), both same-harmonic and cross-harmonic, are tested as (fi + fj) and abs(fi - fj) (i.e., v+v and abs(v-v), 2v+2v and abs(2v-2v), 3v+3v and abs(3v-3v), v+2v and abs(v-2v), v+3v and abs(v-3v), 2v+3v and abs(2v-3v)); only those landing in the displayed band are plotted.

3) Violin and intermods (zoom panels with thresholds)


Zooms around about 500, about 1000, and about 1490 Hz show violin peaks, calibration sidebands, and violin-violin intermods. Thresholds used (representative):

  - Band medians (ASD): about 9.29e-08 (v), 8.12e-08 (2v), 8.38e-08 (3v).

  - Violin peak floor (ASD): about 5.3e-06.

  - Intermod minimum height (ASD): about 1.3e-07.

Adaptive selection tracks local noise; loosening/tightening these settings trades recall vs precision. The chosen values recover the strong violins and yield a physically plausible intermod count for stable beta.

4) Beta histograms by family


Violin-violin families peak at larger beta than calibration mixes, as expected for a quadratic mechanism scaling with the product of parent amplitudes. Broader tails appear near thresholds and in crowded regions.

5) Beta by intermodulation family


The hierarchy is clear: vv(fund) and vv(3v) give the largest medians; vv(2v) is smaller but still above all calibration mixes. This is consistent with larger and cleaner A_left x A_right for violin–violin pairs than for violin ± calibration.

6) β(+cal) versus β(-cal)


v and 2v show near-diagonal symmetry; 3v is imbalanced, consistent with leakage or pairing sensitivity at the third harmonic.

7) Amix versus (AL · AR) - quadratic check


Largest deviations sit near thresholds and in crowded sub-bands.

8) β versus Amix with thresholds


Scatter grows as Amix approaches thresholds.



Next steps
  - Repeat the analysis in the ADC domain (counts) for H1:OMC-DCPD_SUM_IN. Use the OMC_DCPD filter magnitude file (cols: freq, DCPD A mag, DCPD B mag) to convert counts<->mA and recompute beta. Comparing OUT (mA) and IN (counts) will test for any domain-dependent bias in beta.

TJ had another lockloss around the time of the beam diverters closing this morning, it happened during OMC whitening.  Following up on what Jenne and Elenna saw last night 87962  , I looked at HAM1 GS13s during the lockloss. 

In the attached screenshot the first time cursor shows when the CLOSE_BEAM_DIVERTERS guardian state ends and the guardian moves on to switching the OMC whitening state, the HAM1 ISI is still shaking from the beam diverters when the OMC DCPD gains are changed.  I've saved this template as sheila.dwyer/ndscope/LOCKLOSS/beam_diverter_lockloss_check.yml

I've added a 10 second timer to the close beam diverters state, so that these things will be separated in time.