Sheila, Camilla
After TJ found out OPO was struggling to get enough SHG pump power to lock 82196, Sheila repeated Daniel's 82057 SHG work as he had increased H1:SQZ-SHG_GR_DC_POWERMON from 95 to 125mW but now it's down to 92mW: Tried steering the beam with picos moves, and by adjusting the temperature it improved to 93mW only.
We went only SQZT0 and Sheila adjusted the alignment into the OPO pump fiber with the OPO unlocked, bringing the the OPO REFL DC power from 1.7 to 1.9mW. We also moved the SHG wave plate further to maximize the amount of light Now CLF could lock with controlmon around 5.8V. Much better and closer to the center of the range so should avoid this mornings 82196 issues.
Homdyne Work:
- Increased SEED power from 0.6 to 70. Subtraction was fine and we adjusted the LO waveplate to balance the LO and SEED powers.
- Measured visibility with the oscilloscope to be (1930-25.6)/(1930+25.6) = 97.4% visibility. No alignment changes needed.
- Found that the issue we saw with the LO unlocking yesterday 82153 was that the LO loop was unstable, we reduced H1:SQZ-LO_SERVO_IN2GAIN gain, see notes below
To take the below data, SEED power reduced back to 0.6mW, and then:
- NLG measured at (Toggle SQZ/HD path and follow 76542)
- Took SQZ_MANAGER to SQZ_READY_HD (FC2 misaligned) and take data on HD DTT
- Change opo_grTrans_setpoint_uW in sqzparams.py, reload OPO guardian, take it to NO_ISS and then back to LOCKED_CLF_DUAL to check it can reach opo_grTrans_setpoint_uW
- Repeat
| Type | NLG | SQZ dB @ 1kHz | Angle | DTT Ref | Notes |
| Dark Noise | N/A | N/A | ref 0 | This and shot noise was noisy <300Hz at times when the OPO was locking or scanning, unsure why. | |
| ASQZ | 11.2 | 14.2 | 241 | ref 2 | opo_grTrans_setpoint_uW = 80uW |
| SQZ | 11.2 | -6.8 | 154 | ref 3 | reduced LO loop gain from 15 to -1 |
| SQZ | 14.3 | -6.1 | 182 | ref 4 |
opo_grTrans_setpoint_uW = 120uW
reduced LO loop gain from -1 to -12
|
| ASQZ | 14.3 | 17.8 | 216 | ref 5 | |
| ASQZ | 16.0 | 19.1 | 210 | ref 6 | opo_grTrans_setpoint_uW = 140uW |
| SQZ | 16.0 | 191 | ref 7 | ||
| SQZ | 16.0 | -6.8 | 190 | ref 8 | increased LO loop gain from -12 to -1 as was looking peaky |
| Shot Noise | N/A | N/A | ref 1 |
Blocked SEED, LO only.
|
We measured ~6.5dB of SQZ on the HD and up to 19dB of ASQZ. Plot attached.
Next time we work on the HD, we should measure the LO loop to know the correct gain to use.
SQZ data set
- Brought opo_grTrans_setpoint_uW back to 80uW, it locked with controlmon at 6. This is now much closer to the center of the range so should avoid this mornings 82196 issues.
- Measured NLG to be: 10.2
- Before starting to take data, paused SQZ_MANAGER, turned off SQZ ASC and SQZ_ANG_ADJUST to DOWN
- FDS data and FIS data all saved to camilla.compton/Documents/sqz/templates/dtt/SQZdata_20250109.xml
- I took 4 minutes of data for FDS, then 3 minutes for FIS but only 2 minutes each for FIS ASQZ +/-10deg as we were running out of time.
- Back to observing reverted the OPO_SERVO_COMEXCEN sdf diff from dither locking, this should be turned off in SQZ_MANAGER, accepted new OPO temp.
| UTC | Type | CLF Phase | DTT ref |
| 19:02:00 - 19:08:00 | No SQZ | N/A | ref 0 |
| 19:14:00 - 19:20:00 (6mins) | FDS SQZ | 177deg | ref 1 |
| 19:21:00 - 19:25:00 (4mins) | FDS Mid + SQZ | 215deg | ref 2 |
| 19:26:00 - 19:30:00 (4mins) | FDS Mid - SQZ | 125deg | ref 3 |
| 19:32:00 - 19:36:00 (4mins) | FDS ASQZ | (-sign) 108deg | ref 4 |
| 19:40:00- 19:43:00 (3mins) | FIS ASQZ | (-sign) 108deg | ref 5 |
| 19:43:15- 19:45:15 (2mins) | FIS ASQZ +10deg | (-sign) 118deg | ref 6 |
|
19:45:30- 19:47:30 (2mins)
|
FIS ASQZ -10deg
|
(-sign) 98deg
|
ref 7
|
|
19:48:00 - 19:51:00 (3mins)
|
FIS SQZ
|
177deg
|
ref 8
|
|
19:51:30 - 19:54:30 (3mins)
|
FIS Mid + SQZ
(check, small low freq glitch at 19:53:49)
|
215deg
|
ref 9
|
|
19:55:00 - 19:58:00 (3mins)
|
FIS Mid - SQZ
|
125deg
|
ref 10
|
In 83022 we found that we had pump depletion during these measurements so that the OPO trs powers were actually lower than reported
I've imported Camilla's data from dtt, subtracted the dark noise. As Camilla notes above, the NLG measurements here don't nicely fit a model for a consistent threshold power. Instead of using the NLG data, I've allowed the threshold power to be one of the parameters that varies in the fit, along with total efficiency and phase noise. In principle one could also use this data to estimate the technical nosie that limits squeezing, but I haven't tried that with just a few points here.
The attached shows the NLG measurements vs what we would expect given the threshold power that the squeezing and anti-squeezing fits suggest. This model suggests that the total efficiency to of the homodyne squeezing is only only 81%, which seems too low.
This script is available here: quantumnoisebudgeting, but needs a lot of cleaning up.
