Powering up and other weekend fun

Kevin, Elenna, Evan, Dan

TLDR

• When we lose lock we should do an intial alignment
• We can sit at 78W (74W input into the PRC with 430kW in the arms) but ASC isn't happy with more, lots of 0.4Hz buzzing and buildups were noisy. Not unsurmountable ASC problems we think, will just take time.
• We are losing a lot of 9MHz at POP when we power up, nearly a factor of 8 PRCL gain loss in total - somewhat compensated by increasing the PRCL digital gain. PRCL gain is a measure of how much 9 and carrier are at POP and how well overlapped in shape they are
• Suspect bad 9 MHz sidebands might be causing ASC issues at REFL - optical gain and sensing matrix changes which get worse with more power as 9 MHz sidebands degrade.
• RF18 is a bad measure of the 9 performance, it might also clip in certain alignments, PRCL gain should be used for optimising [1, 2, 3]
• PRCL gain and PRG are anti-correllated with spot positions - a good PRG reduces the 9 and might make sensing problems worse
• Changing the ITM ring heaters this weekend has not been that problematic for lock acquisition - ETM ring heater change may have helped this?
• Common ITM ring heater make PRCL gain worse at 25W, it's not clear how much it helps at 70-80W either without more attempts at changing them

This weekend we wanted to try and see how much more we could inject in power before things became unstable. On Friday we tried going up to 74W but that resulted in ASC which were later patched up. We then tried staying at 75W on Friday night but we ran into a 14Hz instability in PRCL. This is due to a signifcant loss in the PRCL gain which we can counteract with increasing the PRCL digital gain (as the guardian does). However we believe this isn't fixing a long running underlying issue whilst powering up that the 9MHz build up that TCS or PRC alignment changes don't fix.

Whilst we can fix the PRCL loop with a digital gain easily, if the sidebands are getting distorted this will eventually degrade our ASC sensing - which we suspect is happening. We see offsets developing in various REFL wavefront signals and the ASC sensing is changing in gains and cross couplings. These seem to get worse going up above 60W. When we replaced ITMY after O3 which gave us back a lot more 9MHz we have been able to power-up and control ASC better, perhaps we are running into the same issue now as we increase power again and further degrade the 9MHz.

ITM ring heaters - not as scary as we were assuming

The only actuator we have yet to really experiment with is the ITM ring heaters. In theory these should be able to correct for the thermal lensing and curvature deformations from self heating and bring back some of the 9 and carrier recycling gains. We haven't touched these for a few months after moving ITMY up which turned into a week long locking disaster.

So the first thing we tried on Friday night was to switch ITMY RH back on to 0.44W - common with ITMX. This wasn't that big of a issue as we saw in Feb. After changing it late Friday by Saturday late morning after an initial alignment we could lock DRMI fine, in fact it seemed to be locking even quicker but this could be mircoseism was low. The buildups weren't great but I put 0.5W of central CO2Y on at the locking stage and this seemed to fix it. We did not have any CARM issues and could get up to 70W easily. We were seeing fast locklosses after an hour, suspect a PI but couldn't see any in 64kHz, didn't check the 512kHz. There was no clear benefit to having ITMY on at equal levels and it seemed to actually make the 9 buildups slightly worse. We also tried both ITM RH off but that seems to make ASC unhappy at 0.46Hz after reaching 70W for 20-30 minutes.

Spot positions and alignment - RF18 vs PRCL gain

We have been trying to determine if the 9 buildups are being hurt by alignment issues or thermal issues. Given our large test mass A2L gains we are suspicious of excess clipping in the PRC (BS and PR3). It is clear in MOVE_SPOTS that RF18 gets worse. We lose about 20% of our PRCL gain here which happens when ETMX yaw moves. This reduction in PRCL gain is anti-correllated with the PRG which goes up but at a lesser rate compared to how much 9MHz we lose.

Another thought is that we've had some issues recently with yaw ASC sensing and noise, perhaps this yaw spot move is clipping the 9 too much in the horiztonal direction which then gets worse as we thermalise as spot sizes change shape in the PRC.

POPAIR RF18 also did not corellate with the PRCL gain during some spot moves. Moving the spots on just ETMX but keeping the rest centered dropped RF18 by 50% but the PRCL gain did not change. We suspect this might mean POPAIR is clipping in this particular spot state. At our nominal spot positions this does not seem to be happening though and RF18 roughly follows PRCL gain but not entirely. We end up losing more PRCL gain than the RF18 signal suggests.

On Sunday evening we tried picking some new spot positions which make a trade off in PRCL optical gain and PRG and power up to 60W. During the power up we actually see PRCL gain increasing for a change, but PRG was down to 40. Changing CO2 in this state had a small effect on PRCL gain but overall more of either annular or central made it worse.

ITM ring heater pulses at 25W

Given we couldn't hold high power lock for more than an hour we decided to stay at 25W with the spots centered and see how the ring heaters affect the 9MHz. Pulsing the ring heaters with +2W for 20minutes produced a reasonable size transient in the spherical power which could be clearly seen in various channels. This transient lasted about 2.5 hours. We could reduce that to 10 minutes for faster tests in future. About 20 minutes after the pulse you start to see the effects.

This small common ITM RH pulse reduced the PRCL gain by 40%, with a spherical power change in the HWS of about 16uD (normally on power-up we get +100uD of self heating). This seems excessive but needs more modelling to understand better. Pulsing the ITMs improved the PRCL gain at 25W. The magnitude of the PRCL gain loss with such a small common RH change was not expected.

Up to 81W input then back to 78W

On Sunday we tried a few more ring heater and CO2 tests and still found we could not get the PRCL gain back up. We had ITM X and Y RH at zero. To get this back to nominal (ITMX at 0.44W and ITMY at 0W) we tried the inverse ring heater to speed up the process which worked fine. However, user error on my part misread the instructions and now we are at 0.88W (Whoops). Going up to 0.88W seemed to help our PRCL gain build up a bit but reduced PRG, this has also led to worse frequency noise. As the IFO is still locked I will leave it for now, but we can drop it back to 0.44W Monday night and it should be ready for after maintenace day.

From there we just powered up from 60W to 81W with some ASC changes inbetween (68519). Staying at 81W input does not seem possible currently due to 0.4Hz oscillations. We kept stepping down to 78W where it seems ok, 430kW in the arms. We had better than this last week with the ITMX RH at 0.48W and the ETM RH at 1W so that could be brought back.

We think we saw some PI ringing up around 27 and 36kHz but in preparation I increased the ETMX ringheaters to 1.4W each segment to reduce the arm separation frequencies closer to 5.2kHz. Those modes reduced after doing this.

Undocumented ASC changes

DSOFTY = 20 was 5 (not in the guardian)