[Elenna, Georgia, Craig, Varun, Sheila, Keita, Jenne, Camilla, Ryan]
Today we did some work on powering up past 50W, similar to what was done last Friday. And, similar to last Friday, we aren't able to increase power much beyond 55W.
We ran sensing matrix measurements as well as open loop gain measurements at different powers, and see similarly strange behavior to what the team saw last Friday.
It seems like the relative phase between MICH and SRCL in POP_A_RF45 is a nice sensible almost-90-deg when we're down at 25W, but then it increases to 135 degrees at 56 W. I suspect that it's just increasing more and more as we increase power. But, I don't undersand why it's changing. In a perfect IFO, the relative phase wouldn't really change as we change power.
We noticed that the SRCL signal in POP_A_RF45 was not so close to being fully in the I-phase (where we want it such that we're not contaminating the MICH signal with SRCL), although it had been back in April 2021. While at 55W, Elenna stepped the POP_A_RF45 demod phase by -1 degree per step, until we had SRCL lined up in I-phase (a total of -28 degrees). Along the way, Elenna measured the MICH open loop gain, and twice compensated the digital MICH1 gain to keep the actual MICH loop gain up at the place it was at 25W (In the end she increased from a gain of 2.8 to 3.6).
After we were happy at 55W, we stepped up 1W in input power to 56W, and saw the dramatic MICH gain decrease that had been seen on Friday. Our sensing matrix phases didn't really change, but the magnitude of everything in the POP45 diode decreased. This is particularly confusing since POP90 doesn't look like it's changing much, and the POPDC and arm powers all look like they're increasing, so we're not losing carrier power at the level we'd need to explain our loss in gain.
We start to suspect that perhaps we need to lower our modulation depths further, in case we're starting to saturate some RF electronics.
Earlier, Georgia moved the ITMY A2L gains to improve the buildups a little more (pit didn't do much, so that's left at P2L of 0, but yaw is now at Y2L of -1.1). We had last set the spot positions at 25W input, so know that its possible there's a better location. We wondered if moving the spots on the ITMs would change the relative phase of the degrees of freedom in POP45, but it seems to have very little effect.
I attach our whole chronology of sensing matrix radar plots, just so we can look at them later - the suffix is meant to hint at the IFO config at the various times. The two to compare are the first (25W measurement, original POP45 phase) and the last one (56W measurement, after rephasing). The magnitudes of these phasors are not normalized by the laser power, although they do take into account the MICH gain increase from 2.8 to 3.6 (PRCL and SRCL gains unchanged between these plots). So, we're seeing that there is not as much gain increase in MICH and SRCL as we'd expect, having more than doubled the power incident. Also, we see that the relative phase between MICH and SRCL is much larger at higher power.
Note that we have now successfully relocked with the new POP45 phasing, so I have accepted it in SDF.
We also looked at the phase of various PDs while injecting a frequency line at 702 Hz, and see that there is some room for improvement on phasing of our REFL 9 PDs. But, we didn't go out onto the floor to change them.
The attached screenshot shows the injection into LSC AO2 (after having enabled the Common Mode Board excitation), and REFL9 is in the top right. This plot only has REFL_A_9, but REFL_B_9 is basically the same (which is good that they're the same).
In the last lock of the evening we managed to sit at 60W input power for ~18 mins. We passed the GPS time milestone of 13333333333 while locked at 60W 8)
To get there we:
- Lowered the RF9 modulation depth by 6dB while we were sitting at 50W so that LSC-POP_A_RF9 didn't saturate at 0dBm
- Stepped up the power to 55W, then to 60W in 1W steps, measuring and adjusting the MICH gain as we went (see first attached ndscope)
In the end the POP_A_RF45_DEMOD_RFMON tanked to -40dBm, and the MICH gain had been pushed up to 10 (started at 2.8). The RF levels on POP and REFL are kind of confusing during powerup, we are definitely saturating LSC-REFL_A_RF45 when we're above 57W. second attachment shows the RF levels for the full lock duration.
While we were sitting at 60W we saw high frequency noise reminiscent of whistles.
Attaching the power trends from the 60 W lock (teal in first two PDFs) No real surprises here, similar story to before (higher input power -> lower PRG, POP18 bad and getting worse with high power.) REFL LF seems to be rising quickly, and is not really correlated with a increased PRG. This indicates increased carrier losses in CARM. Also attaching the new and improved LSC sensing matrix dart plots which includes all TFs taken during measurement time. This shows the uncertainty in our phasing/magnitude. Third plot is 50W measurement, Fourth plot is 60W measurement. This shows how the optical gain of the MICH and SRCL loops are decreasing rapidly as we go up in power (50% and 75% respectively).
The lockloss from 60W was messy. There was a 14.5Hz oscillation in almost all LSC and ASC loops. I must have mixed up the times - according to the lockloss tool we lost lock before the 1333333333 party.