lee.mcculler@LIGO.ORG - posted 15:04, Friday 15 November 2019 (53280)
Speculative source of CLF related noise
Following the discussion on the ISC call, there was a question of mechanisms to generate or propagate 3.125MHz noise out of the interferometer that the AS port has over vacuum. I remember that the themal noise of bulk modes in the optics can reach up to rather high frequencies. While the arms are the usual place to look, they aren't a great lead for the 3.125MHz since that won't resonate, but MICH sensing is still likely plenty sensitive to see bulk modes of the ITMs, in fact with 4kW in MICH it is more sensitive that the 2kW in the Holometer experiment interferometers. Those were 2kW power recycled Michelson interferometers (no arm or signal recycling cavities). Take a look at the AS port RF spectrum of that instrument.
The green is the shot noise. The red is the ASPD cross correlation over ~100 Hours, ~400Hz bin-width. Note that there are plenty of spots ABOVE shot noise and and almost everywhere the noise is above vacuum (even if below shot noise).
The peaks are separated by some 230kHz, which is longitudinal-mode frequency separation of the 0.5"-thick optics used in the Holometer. See section 6.5 of Class. Quantum Grav. 34 065005 for more details related to this plot.
So the question is if a similar mechnism could explain noise at 3.125MHz coming out of the AS port in the LIGO IFO. The MICH degree of freedom still has ~1e-18m/rtHz sensitivity at the AS port. Actually, it may be decently better than that, since the SRCL cavity is resonant above the cavity pole (antiresonant below, for signal extraction), I believe it should enhance the MICH sensitivity at the CLF frequency (haven't checked yet though).
What is the mode density of the test masses? Maybe a higher CLF frequency could get above them. The magnificantly high Q's of the ITMs means the lines will be narrower and higher (more resolvable), potentially enough to offset the 700x larger mass of the testmasses vs 2" optics shown above. The larger optic size also means they will be much more dense as well. I'll have to do a more accurate comparison of the sensitivities. Perhaps we are unlucky and right in a forest of modes (we should see it in the LO error spectrum if there are unexplained lines are the right spacing for the test-mass modes). With 20cm thick ITMs, the longitudinal bulk mode is
14.750kHz and the CLF 3.125MHz modulo 14.750kHz is -2kHz (so the 211th longitudinal mode is at 3.11225MHz), which means that the CLF is actually pretty far and must be in the forest of mixed longitudinal+HOM frequencies. Since mixed-mode frequencies add in quadrature, the mode density is quite high at MHz, as seen in the attached plot. While the OMC cleans up the mechanical->optical HOMs, the fact they show so readily in other fringe-offset Michelsons indicates there is likely enough overlap with the carrier 00 that the OMC won't fully clean them out.
If these lines are there and large enough, they could also inject effective phase noise, which should diminish with higher CLF power injection as the bulk-thermal noise lines are overpowered in the RF3 demod.
This is a bit of a crazy idea, but maybe not unreasonable to look into further. We are likely looking for a broadband noise that this wouldn't explain unless the mode density is truly high, but given that the CLF related noise is only shown to be a few percent on range, it could be a set of weak narrow lines. Long integrations/cross correlations at different CLF powers might resolve changing line features. Moving the CLF may also change things (minding where the arm notch lands).
Images attached to this report
