Reports until 17:39, Wednesday 25 October 2023
H1 SQZ
camilla.compton@LIGO.ORG - posted 17:39, Wednesday 25 October 2023 - last comment - 15:00, Monday 20 November 2023(73747)
NLG Sweep with IFO

Sheila, Naoki, Camilla

We took FDS, Anti-SQZ and Mean SQZ data at different NLGs: 14, 17, 43, 82 and 123. We think we see evidence of frequency noise. We see less squeezing at high NLGs: around 4.5dB of SQZ at NLG < 43.1 but at NLG 81.9 and 122.8, squeezing at 1kHz reduced to 3.6 and 3.1dB. Attached Plot.

Each time we optimized SQZ angle around 1kHz. Because of this, the low frequency increased noise at high NLGs could be due to incorrect rotation, rather than phase noise. 

 
Time (UTC)
Demod phase
DTT ref#
 NLG
SQZ dB at 900Hz
FDS
21:23:40-21:26:00
152.25
3
13.9
-4.3
FDAS
21:27:30- 21:32:30
228.10
6
13.9
15.2
Mean
21:33:47- 21:38:00
-
7
13.9
12.7
No SQZ
21:40:00 -21:44:50
-
0
-
 
FDS (not optimized) 
21:51:20 -21:54:30
164.57
Deleted
43.1
-4
FDAS (Left ASC off) 
21:56:00 -21:59:45
226.2
5
43.1
19.1
FDS (retake)
22:03:00 -22:06:13
167.42
4
43.1
-4.3
Mean
22:06:45 -22:09:52
-
8
43.1
16.9
FDS
22:37:50- 22:41:00
172.16
9
122.8
-3.1
FDAS
22:42:43-22:45:45
217.67
10
122.8
24.7
Mean
22:46:35 - 22:49:55
-
11
122.8
 
FDS
22:57:30- 22:59:40
169.31
12
81.9
-3.6
FDAS
23:01:20 - 23:02:55
219.56
13
81.9
22.7
Mean sqz
23:03:26 - 23:05:45
-
14
81.9
20
FDS (left IFO here)
23:19:38
154.14
15
17.1
-4.5

Data saved in /camilla.compton/Documents/sqz/templates/dtt/20231025_GDS_CALIB_STRAIN.xml

Time (UTC)
Unamplified OPO OPO_IR_PD_LF
(Scan Seed)
 
OPO trans (uW)
OPO Temp (degC)
Amplified (maxium) H1:OPO_IR_PD_LF
(Scan OPO PZT)
NLG 
(Amplified / (Unamplified Peak Max - Unamplified Dark Offset))
Peak Maximum Dark Offset
21:15
786e-6
-16e-6
80.54 uW
31.446
0.0112
13.9
21:44
 
 
100.52
31.414
0.0346
43.1
22:17
 
 
120.56
31.402
0.160
199.5 (SQZ loops went unstable) 
22:30
 
 
110.5
31.407
0.06617
82 (didn't use) 
22:36
 
 
115.46
31.405
0.09853
122.8
22:52
 
 
110.5
31.407
0.0657
81.9
22:09
769e-6
-15e-6
80.5
31.424
0.0137
17.1

Vicky and Naoki did an NLG sweep on the Homodyne in 73562.

Images attached to this report
Comments related to this report
camilla.compton@LIGO.ORG - 10:49, Sunday 29 October 2023 (73801)
Link

Edits to NLG table in above alog: Last measurement was at 22:09 23:09 UTC. For measuring unamplified we scanned OPO PZT and for amplified we scanned SEED PZT. 

Setup for each SQZ Measurement:

For FDS: Took SQZ_MANAGER guardian to FREQ_DEP_SQZ and rotate SQZ angle to give best squeezing at 900Hz. 
For ASQZ: Turn off IFO ASC and FC ASC, rotate SQZ angle to give worst squeezing at 900Hz. 
For Mean SQZ: Took SQZ_MANAGER guardian to FREQ_INDEP_SQZ and SQZ_LO_LR to DOWN. 
 
To measure unampilifed signal for NLG:
  • SQZ_MANAGER to NO_SQUEEZING and SQZ_OPO to DOWN
  • Scan the OPO PZT with PZT_1 Enable, Use trigger: No (via SQZ Overview > HAM7 > OPO IR > OPO PZT1)
  • Should see mode scan on H1: SQZ-OPO_IR_PD_LF_OUT. Measure the highest peak as "Unamplified" and the zero level as "Unamplified Dark Offset".
  • Only need to this once (we did at start and end)
To change to another NLG:
  • SQZ_MANAGER to NO_SQUEEZING
  • Change opo_grTrans_setpoint_uW in sqzparams.py (higher power = higher NLG)
  • Reload SQZ_OPO
  • SQZ_OPO to LOCKED_CLF_DUAL_NO_ISS  then LOCKED_CLF_DUAL
    • Check it locks with correct OPO TRANS power, may need to adjust SHG power with waveplate
  • SQZ_OPO to LOCKED_SEED_NLG
  • SQZ_CLF to DOWN
  • Adjust temperature to maximize SQZ-OPO_IR_PD_LF_OUT (with SEED_PZT scanning enabled)
  • Record SQZ-OPO_IR_PD_LF_OUT maximum it is "Amplified" signal.
    • Can now calculate NLG = Amplified / (Unamplified - Unamplified Dark Offset).
  • SQZ_MANAGER to SQZ_READY_IFO (with swap back from SEED to CLF)
victoriaa.xu@LIGO.ORG - 15:00, Monday 20 November 2023 (74318)
Link

Attached are fits of squeezing loss, phase noise, and technical noise to this NLG sweep on DARM: we can infer 25-30% total SQZ losses, ~25 mrad rms phase noise, with technical noise almost -12 dB below shot noise.

Losses are tracked in the SQZ wiki and gsheet. Of the 30% total inferred loss, we expect 20%: that is 7.5% injection loss, and 13.7% readout loss.

Remaining ~10% mystery loss is compatible with mode-mismatch: in sqz-omc single bounce mode scans, LHO:73696, we estimate 8-15% mismatch, and we observe the frequency-dependence of the mismatch as we vary squeezer beam profile using PSAMS: 73400, 73621.

In the fits, [loss, phase noise, technical noise] are fit to the measured SQZ and Anti-SQZ, given the measured NLG, using equations from the aoki paper. "Loss from mean sqz" is the calculated loss from the measured NLG and measured mean-sqz dB; it is not a fit, but depends on the calibration from NLG to generated SQZ dB.

Some summary slides here show the progression of our loss hunting, which are so far lining up with Sheila's projections from 73395.

Images attached to this comment
Non-image files attached to this comment