Please ignore alarms from Y2-8 pressure gauge. It is reading zero which may be from lack of solar power.
Hang, Dan Brown
As there is no dedicated guardian for powering up from 20W to 35W we'll list what we do here:
The reason you need to wait between each step is that the adjust power guardian state adjusts the PRCL gain as we thermalise. If you go too fast then it increases the gain too much and you'll see some peaks appearing in the lengths around ~110Hz. This could be done in a more neat way but we ran out of time.
Once we reached 35W the PRG became noisy. This coincided with the length loops becoming noisier between 10-100 Hz, from the low PRCL gain perhaps. This was happening at 28W before too but some TCS tuning managed to fix it. Here, you can see 45 refl becomes very noisy once powered up. This seems to coincide with a noisy PRG and bad PRC2/CHARD coupling. I tried the CO2's and they didn't help, then I tried increasing the common ring heater powers in steps. I didn't have Danny's new RH filters on at this stage. I remembered to switch them on when I got to the ETM. To switch these on you need to set the RH guardians to the filtered state, not nominal.
For TCS tuning I have been aiming to keep the REFL9Q and REFL45 signals near zero and quiet, increase the PRG, RF18, and AS45. Also keep an eye on PRCL gain as this can drop quickly with some differential TCS changes. Each time I try the CO2s first to see how they improve these metrics (right now we've only got 0.6W on CO2X left to reduce). As we're trying to go for more power I then gradually increased the ITM/ETM RH in steps observing how things change.
Some more notes from trying high power tonight:
Was probably too early to declare 35W stable it looked better yesterday, we've had a few fast locklosses between 32-35W and two slow 0.1Hz ring up. The slow ring ups show up in the INP/PRC1/CHARD control signals. 30W seems fine though. I was starting to have a look at the PRC1 sensing matrix and CHARD couples strongly and gets worse with increased power.
DRMI has felt a bit slower to acquire today at times. Today I've been dropping back to PRMI and MICH fringes a few times to double check the alignment was ok as the flashes were bad. BS/PRM has needed tweaking up a bit each time. I haven't had to wait ages for DRMI though, the worst was 5 minutes, the last 3 I had were 1-2 minutes. The new TCS settings have more CO2 power at 2W than before to compensate the increased ring heater power for 35W. PRMI looks messier than before. Perhaps that's why it's more sensitive to DRMI alignment now.
I get the feeling that the time it takes to get from lockloss to acquiring DRMI plays a part to. The self heating decays much faster than the CO2 heats back up again, which means we can end up dipping below an acceptable thermal state making DRMI harder to acquire as the flashes aren't as good. I feel the fastest acquisitions happen when ALS wasn't dropping all the time and FIND_IR was fast.
The unoptimised 2W TCS state is apparent in the RF18 buildup which doesn't get much better than ~68 now (was 80 prior to these changes). It's probably worth spending some time tuning TCS and CO2 alignments for 2W and find out what it likes given some ring heater setting.
I had reduced CO2X at 30W input from 0.6W to 0W, it didn't seem to like being reduced. I tried 0.3W and given I have had more lock losses than yesterday I have just set it back to 0.6W. As ITMY absorbs more it might be that it prefers having additional CO2X to compensate for the differential self heating. You could also differentially tune ITM RH more, increase RHY and decrease RHX, but I didn't have time to try it.
I missed the 7,2 quake from Palau warning which put an end to my tests, the interferometer held on for about 10 minutes before it gave up. I pushed the big red button when we lost lock, not sure if it was too late to push it by this point?
Note that wind and microseism have been exceptionally quiet the last few days which has helped getting to high power no doubt.
Hang, Dan
Last night after the lockloss we initially had some issues with the fast shutter not passing the test. Checking this morning this was because the AS beam was misaligned and there wasn't any light on the QPDs to do the test with.
The cause was SR3 pitch, looking at the oplevs and OSEMS it dropped after the lockloss and never came back... Adjusting the sliders values by 15 puts us back at the previous oplev/osem value. SR3 usually gets a kick during a lockloss however it always comes back. The fact it didn't this time seems odd.
This morning we have been trying to get the interferometer aligned, green arms lock fine but when we try initial alignment DOFs 1, 2, and 3, diverge way more than usual at the start and then won't converge.
We temporarily commented the fast shutter check in PREP_FOR_LOCKING so that it continues with setting up the suspensions. Otherwise the beam is clipped through the OFI and the power is not high enough for the fast shutter to complete it's check. This should be moved to some later stage perhaps, however we don't know what would be appropriate so we'll leave this fix to someone who does.
Even after completeing initial alignment and then adding the fast shutter check back into PREP_FOR_LOCKING it won't pass the test. The levels of light it is looking for on AS B and A are between 2500-6000. At this stage on single bounce we only see about 1000 and 5000 with 10W SRY, so we're not sure how this is using such high levels for checking the shutter works.
During ENGAGE_ASC_FOR_FULL_IFO and the dither loops are converging SRC1/2 struggles to keep up in yaw and we lost lock a few times. We manually helped it along until everything converged nicely.
We also wanted to update the initial alignment references as we found last night and today that the two positions are significantly off in yaw. While the arm was locked and converged we open the green shutters and moved the green QPDs offsets to zero the WFS signals and then updated the initial alignment green camera references,
TCSY chiller got down to ~6.8cm, topped it up with 1200ml back up to 10mm.
Hang, Dan
It's getting late so this is a quick post, more analysis to come tomorrow.
Today we powered up to 30W and did some fine tuning of the TCS to improve the CHARD sensing matrix rather than putting in a new one. I adjusted the CO2s to optimise the REFL9I, RF18, PRCG until they stopped improving anything, then I started increasing 0.4W common ITM RH and 0.2W ETM common RH. This made all the outputs much quieter and helped fix our CHARD/PRC2 sensing matrix coupling. We started to have a 3mHz oscillation in CHARD/PRC1/INP1 but it didn't cause too much of an issue. We went upto 33W and then we lost lock.
This lock loss was a bit messy as all the ring heater values had changed so the the DRMI thermal state was a bit mismatched. Even though the mismatch was very strong DRMI still locked in less than a minute, although RF18 build up was ~50 instead of ~80. After the RHs had settled and I had found a rough CO2 setting for 2W DRMI was locking ok, RF18~60.
We had issues getting through engaging full IFO ASC, though that was from changes we made in error today. Hang can comment on what we moved around.
Once we were back up and locked again we got all the way to 35W with little issue, and seemed stable enough that we could have stayed there. All the ASC loops were quiet. PRCL optical gain didn't drop as much as it used to with these TCS settings so we had to adjust the PRCL gain increments in ADJUST_POWER. However it still gets low at 38W where we had a fast lock lost almost straight away.
Going from 33W to 35W we increased the power by ~6% and the arm power increased by ~4% looking at TR photodiodes.
One odd thing we did see was that the input power was oscillating at 80mHz looking at MC2 trans, with corresponding oscillations in the arm powers, not sure why.
I took the TRY values from the 35W lock yesterday to predict where we might level off for injected power. We've yet to try tuning the soft offsets at 35W or fully optimise TCS to maximise the PRG so these predictions could be very off.
If the trend continues as a quadratic then the break even point is 70W of injected power, which seems optimistic.
There is also the measured and predicted PRG values and the projected arm power, which predicts we'll reach a maximum arm power around 45W
Previously the pump laser has been drifting in and out of multimode. Today I came in and found it constantly running multimode (an obvious symptom being low green production out of the SHG). So I put the current half way between where the multimode starts and clamp current (and set the temperature to keep PSL and pump beat note at 158 MHz beat note at zero MHz offset on the TTFSS temp servo). The multimode was monitored with the SHG GR PD (Thorlabs) while scanning the SHG cavity. Below is a few data point regarding this particular shitty laser.
| Current (A) | Temperature (deg C) | |
| Before (constantly running multimode) | 2.134 | 29.69 |
| Now (no more multimode) | 2.193 | 29.65 |
| Multimode starts (approaching from clamp) | 2.137 | 29.68 |
| Clamp | 2.250 | 29.59 |
| Mode hop (beat note disappeared) | 2.050 | 29.75 |
These numbers were read of the controller box LCD screen. The current readback on the medm screen aren't the same.
Keeping the temperature the same I tweaked the current to figure out MHz/A calibration
| Current (A) | Beat note (MHz) |
| 2.193 | 164.2 |
| 2.136 | 30.2 |
| 2.250 | 308.2 |
| 2.222 | 235.2 |
This set of data yields 2.5MHz/mA (+- 0.1MHz).
Then I keep the current the same, and tweaked the temperature
| Temperature (C) | Beat note (MHz) |
| 29.65 | 164.8 |
| 29.63 | 89.1 |
| 29.62 | 48.2 |
| 29.69 | 290.8 |
| 29.72 | 361.1 |
This yields 34.21MHz/0.01degC (+-6.6MHz)
If I calculated things correctly here this means we are +70 MHz away from multimode at the moment. The temperature would only have to drift .02 deg C to go back to multimode. Could there be a region beyond mode hop that works?
There should be more mode hop free regions at lower currents (so you can get away from the max current).
911 has gone down statewide. The police called and told us to use emergency number 3730911 when on site.
We tuned some dtt templates for radiation pressure compensation adjustments.
For DHARD PIT and CHARD YAW, the templates currently are at:
/ligo/home/hang.yu/Desktop/ASC/common/dtt/DHARD/data/DHARD_P_10W_HBW_ref.xml
/ligo/home/hang.yu/Desktop/ASC/common/dtt/CHARD/data/CHARD_Y_10W_HBW_ref.xml
Those were taken at the high ASC bandwidth configuration. For DHARD PIT the boost is on and CHARD YAW the boost is off. The excitations used are saved yet due to dtt's issue in ramping the gains, it might be better to send in the excitations through awggui. When do the tuning, one only needs to check the high freq part to make sure the secondary suspension resonance stayed at ~ 1.7 Hz for PIT and ~ 1.6 Hz for YAW.
For the CHARD_P and DHARD_Y, we don't have templates as detailed as the above ones (which covered from 0.5 to 3 Hz). However, one could use
/ligo/home/hang.yu/Desktop/ASC/common/dtt/DHARD/DHARD_OLG_Y_broadband.xml
/ligo/home/hang.yu/Desktop/ASC/common/dtt/CHARD/CHARD_OLG_P_broadband.xml
and again try to match the location of the secondary sus resonance. This would effectively reduce the plant back to the 10 W input level where most of our ASC loops were designed for.
==================================================
While taking the data for DHARD PIT in the 0.5-1 Hz band, we noticed the suspension plant seemed to be strange at around the main resonance at ~ 0.9 Hz. Specifically, if one focus on the middle-right panel and the green and pink traces, one would see that when the TF hits the main sus resonance, the phase instead of decreasing by 180 deg, it actually increased by 180 deg...
To see the issue more clearly, one could see DH_P_OLTF.pdf attached to this entry. Here the TF is calculated from <exc, in1>/<in2, exc>, where <> stands for taking the cross-spectrum, and therefore it should be an unbiased estimator. Again we saw a weird phase behavior around the main sus resonance at ~ 0.9 Hz, in both the 10 W data (without any RPC), and in the 20 W data after engaging the radiation pressure compensation path.
Dan, Hang
Using the dithering lines we set up for the past few days for sensing matrix measurement, and the suspension calibration provided in DCC:T1100378, we calibrated the spectra of the ASC error points into physical rad/rtHz. The results were attached.
The best sensor we have is AS45 for DHARD, and the sensing noise is ~ 1e-14 rad/rtHz. While a single QPD has a sensing noise similar to AS45, when we use a QPD combo for CHARD to decouple CSOFT, the sensing noise get slightly worse at ~ 3e-14 rad/rtHz level. The soft loops' sensing noise is about a factor of 10 higher than the hard loops at a few x 1e-13 rad/rtHz level.
On the other hand, the REFL sensors are significantly noisier than AS45/TR QPDs. Using the REFL sensors the CHARD sensitivity is only 1e-12 rad/rtHz at 10 Hz and the slope suggests that it is not shot noise limited. This indicates the necessity of doing CHARD blend from the noise point of view.
For future reference, the calibrations are (from L3 physical angle times the output matrix, to the error point of each dof after the input matrix).
DHARD PIT: 6.8e+10 [ct/rad]
DSOFT PIT: 1.0e+5 [ct/rad]
CHARD YAW (DC, REFL combo): 2.0e+10 [ct/rad]
CHARD YAW (AC, TRQPD combo): 4.3e+10 [ct/rad]
CSOFT YAW: 2.1e+5 [ct/rad]
The /opt/rtcds file system disk space used has dropped from 96% to 46% today due to the rotation of the ZFS snapshots out of the reduced look-back time period. On 8th December I compressed most of the target_archive area and reduced the snapshot look-back from 31 days to 19 days. The non-compressed files have now rotated out of the ZFS snapshots.
Next I need to complete the target_archive compression and restore the ZFS look-back to 31 days.
CDS Systems Report (Thu Dec 27 08:00:01 PST 2018):
Thu Dec 27 08:00:01 PST 2018
-- Disk Usage -----------------------------------------
h1fs0:
Size 899 Used 416 (UsedDS 402 UsedSNAP 13 ) Avail 483 Use% 46 %
Hang, Dan Brown
When powering up and thermalising the coupling of PRC2 into CHARD becomes orders of magnitude greater than CHARD back into itself which seems to be the cause of our 3Hz instabilities. At 25W we injected some lines in and looked at how the coupling changed, after it reached some steady state I took 5 minutes of data and processed it offline to compute a new input matrix for PRC2 and CHARD. I used DC5_Y to try out the new input matrix, as can be seen in the DTT window here, blue (new) is looking a lot better than red (old).
We lost lock trying to switch over the matrix, so just update the guardian with these new values. When engaging ful IFO ASC it is a lot more noisy then before, when you get to CHARD_BLEND though everything quietens down and the dither loops catch up.
| Name | Dec 27 2018 03:58:22 UTC | Dec 27 2018 07:49:13 UTC | ||||
|---|---|---|---|---|---|---|
| Value | Alarm Status | Alarm Severity | Value | Alarm Status | Alarm Severity | |
| H1:ASC-INMATRIX_Y_10_11 | 0 | NO_ALARM | NO_ALARM | 0.08296 | NO_ALARM | NO_ALARM |
| H1:ASC-INMATRIX_Y_10_13 | 0.2 | NO_ALARM | NO_ALARM | 0 | NO_ALARM | NO_ALARM |
| H1:ASC-INMATRIX_Y_10_15 | 1.3 | NO_ALARM | NO_ALARM | -0.4889 | NO_ALARM | NO_ALARM |
| H1:ASC-INMATRIX_Y_10_9 | 1 | NO_ALARM | NO_ALARM | 3.4 | NO_ALARM | NO_ALARM |
We put the old sensing matrix values back into DC5_Y to see how they behaved as we powered up in steps, a comparison is shown here. As can be seen the new matrix is overall more noisy than the old, however at 28W PRC2 is much lower than before. At 30W we didn't reach a complete thermal steady state as it got too bad, so we still need to fine tune this more tomorrow. The hope is we can get steady at 30W and then do some TCS tuning.
Here's the 28W sensing matrix. The darker the red the lower the coherence between the excitation and the measurement, cell values are "magnitude (phase in degrees)". The sensing matrix is gradually becoming degenerate as we go higher in power.
This input matrix may work:
| REFL_A_RF9_I | POP_X_RF_I | |
|---|---|---|
| PRC2 | 0.000 | -12.996 |
| CHARD | 1.000 | 12.996 |
Overnight we stayed locked at 28W for ~11 hours or so. When I came in I started tuning some TCS and lost lock when the PRCL gain dropped too much. First plot is PRCL gain, second PRCL->SRCL, third SRCL gain (46082).
Dropping CO2Y down to 0W at 28W increased the PRC gain and RF18. REFL 9Q got much quieter but trended away from 0. I then tried decreasing CO2X from 0.6 to 0.3W. This had the inverse effect on REFL9Q and RF18, reducing both, but seemed to increased the PRC gain slightly. Not sure why REFL45 becomes so much noisier when we power up to 28W.
I removed the SRM dither code from ISC_LOCK in PREP_ASC_FOR_FULL_IFO and replaced it with the CSOFT/CHARD/PRC1/PRC2 lines so they switch on automatically.
Hang, Dan Brown
Only EY was down from this dolphin crash so this recovery was quicker than some of the others.
Initial alignment went fine except for PRC align. It seems the PRC1 filter changes made the other day (46151) made it unstable for initial alignment, probably too much gain. In PREP_PRC_ALIGN Guardian we switched off all the filters apart from FM1 (-20dB), it converged slowly and with FM1 off it still works.
Only one ALS lockloss during LOCKING_ALS, DRMI locked in about 20s.
Took about 2 hours until we were up again - although most of that was checking CO2Y table for leaks and figuring out what was up with PRX initial alignment rather than the Dolphin crash itself.
I've created FRS12051 to cover this crash. The ticket has been closed with an estimated 1.0 hours lost commissioning time.
Contrast Defect Ratio at 2 W = 8.4 +- 0.2 ppm Contrast Defect is as follows:The electric fields associated with this equation are shown in the fourth attachment. When DARM is locked on RF at 2 watts and the OMC is locked, we get about 11.8 mA, or 14.0 mW of power on the OMC DCDPs. This means the DARM offset is around 29 pm according to this plot. During this measurement, we move the DARM offset from 29 to -29 pm and see what happens. Procedure: 1) Lock OMC while DARM is locked on RF 2) Zero the QPD offsets for OMC ASC by switching to OMC dither alignment, zeroing, and moving back to QPD control 3) Zero the OMC LSC dither control signal by offloading to PZT2, then stop the feedback, parking the OMC on the carrier fringe 4) Change DARM1 TRAMP to 120 seconds 5) Move DARM1 OFFSET through zero (in this case from 9e-5 cts to -9e-5 cts) 6) Re-engage OMC LSC by stopping input, clearing history, unholding output, and restarting input. This will find the fringe again quickly for the next measurement. Parameters during measurement:
Input Power = 1.89 W PRG at 0 pm Offset = 46.8 +- 0.1 DCPD Responsivity = 0.858 A/W DCPD Quantum Efficiency = 0.98 OMC DCPD Dark Offset = 7.2e-5 +- 3e-5 mW (Found at the last time the IMC was OFFLINE) OMC DCPD NULL/SUM = 0.0060 +- 0.0015 Measurement GPS Time = 1229475480Differences between 0 and 29 pm DARM offset:Value 29 pm 0 pm 29 pm/0 pm ------------------------------------------ PRG 46.5 47.0 0.99 TRX [cts] 1502. 1518. 0.99 TRY [cts] 1555. 1572. 0.99Assumed to be one:- IMC Transmission - Input Faraday Transmission - Output Faraday Transmission - OMC Transmission - Mode MatchingNot yet considered:- RF45 Transmission through the OMC - Source of contrast (differential lens)Dark Port alogs: 45753 DARM offset calibration 45734 OMC DCPD Sum/Null Matrix
