It dawned on me recently that with the scratched view ports for CHETA, the delay of the CHETA roll out could have a subtle saving grace.
The intended use for IR1 is to heat the ITM with the same power as the absorbed power, so that the ITMs would remain in more or less the same thermal state at all times, making DRMI locking fast and reliable, and removing some of the impact on SRC detuning changing over time as the ITM lenses would evolve less than in O4.
It appears that in the CHETA Design Document, the effects on mode spacing and therefore Parametric Instabilities (PIs) was completely overlooked. Shame, no PI expert in LIGO was consulted, I'm even an author on the cited document in the design document and how we used CHETA like actuators at UWA to mitigate PI.
Very simplistically, rolling out CHETA only for the ITMs (as was the plan for the upcoming IR1) would have two impacts on how the optical mode spacing evolves:
- Mode spacings would start closer to the final, thermalised, frequencies.
- They would evolve with the same time constant (only ETM heating) as in O4, to the thermalised state.
- But! Because they start closer, the mode spacing would evolve slower.
In O4 LHO's QUAD Radii of curvature's operating point has them stuck between two lock-losing PIs, and they haven't managed to demonstrate reliable damping of either. This lead to several periods in O4 where their lock stretches would be repeatedly cut short at around 4 hours.
The implementation of CHETA on ITMs only would have put them closer to their 10.4 kHz PIs as a starting condition, and because the mode spacing would be changing more slowly over time, they would experience the frequency matched condition of PI for longer periods of time. Thus:
- LHO would lose locks more reliably.
- LHO would have shorted locks.
I could spend time modelling this but its blindingly obvious once you notice.
CHETA on both ETMs and ITMs would theoretically allow you to start with essentially thermalised mode spacing, allowing you to more reliably side-step PIs.
This interesting, though I am certainly no PI expert. Naively, I am not as alarmed for the 10.4kHz PIs for IR1 because the YARM TMS is already below it in its thermalized state, and I hope that we can change ring heaters successfully before IR1 to match the XARM to that for frequency noise reasons. I understand that this has additional implications on the higher frequency PIs and I have no intuition if this move is the right direction for those (maybe you do).
But just a quick order estimate, the absorptions in the ITMs are estimated to be about double what the ETMs are, and the coupling of beam heating to ETM ROC change is about 75% of that for the ITM, so overall I might estimate 3-4 times less thermalization transient when we have ITM CHETAs (so maybe a total change in LG10 spacing of 80-100Hz or so). I think if we set the ring heaters correctly, we can stay away from the 10.2 and 10.4kHz acoustic modes, but maybe more modeling should be done.
The ring heater settings in O4 for H1 were too low to compensate for heating from absorbed laser power. For IR1 we would like to increase the ring heater powers to bring our thermal state in lock closer to the cold state, where we believe that our PRG and mode matching to the OMC are better. We still hope to install Cheta and that having cheta will reduce the difficulties that we have seen in lock acquistion with changes in ring heater settings.
This will certainly mean there are changes to the PI situation, but if we are sucsesfull we should keep the optical modes both below where the accoustic modes are.