[Vaishali, Kiwamu, Sheila(remotely)]
WP 6497
Sheila suggested us doing an interesting test today during the commissioning time window. The measurement we did is a coherence measurement between the bullseye sensor and DARM while changing the PR3 spot position.
The result suggests that a part of the broadband lump in 200 Hz-1 kHz is due to beam size jitter with a caveat that we couldn't fully reproduce the broadband lump.
[Background]
Back in pre-O2 commissioning, we had unidentified broadband noise in 200 Hz - 1 kHz (e.g. alog 30752) which was found to be a function of the PR3 spot position. One hypothesis was beam size jitter somehow coupling to OMC DCPDs.
[The measurements and results]
We tested three different PR3 spot positions, all of which is characterized by POP_A QPD readouts as follows.
Our intention was to reproduce what Sheila observed in this past November (31628) with a hope to reproduce the broadband lump. However, we didn't get drastic increase in the frequency band -- we only obtained a 18% increase at around 400 Hz. See the second attachment for the increased DARM noise. Despite different alignment, configurations (B) and (C) gave almost identical DARM noise. Increased noise below 30 Hz is presumably due to mistuned A2L couplings, as we have moved the spot position.
The first attached figure shows the coherence between the bullseye sensor (sorry for any confusion, but PIT = beam size jitter, YAW = horizontal pointing jitter). The dashed lines are the ones from configuration (A). Notice that they are almost at the same level as the previous measurement (31628). The solid lines are the ones from configuration (B). Those from configuration (C) are not shown as they are almost identical to configuration (B). As we have moved the PR3 spot position, the coupling of beam size jitter increased while the horizontal jitter coupling reduced. The coherence for beam size jitter became 0.05-ish above 300 Hz corresponding to a fractional contribution of 22 % to the DARM amplitude spectral density. Also, at the same time, the power recycling gain improved which is consistent with what Sheila observed.
These results indicate that broadband noise is more or less reproducible, and also, a part of broadband noise is due to beam size jitter.
Later, we tried further exploring the parameter space of the PR3 spot position to see if we can further elevate noise in 200 Hz - 1 kHz. This lead to a lockloss presumably due to too much misalignment on PRM when we were at (POP_A_PIT, POP_A_YAW) = (-0.3, +0.6). We couldn't find a location where the broadband lump becomes more prominent.
[A minor change in bullseye signals]
Besides, we made minor modifications on the bullseye settings.
Another caveat:
Later, Keita pointed out that we shouldn't have put a low pass in the sum channel because it spoils the dynamic cancelation of intensity noise. However, if this theory holds, the implementation of the low pass shouldn't decrease the coherence between PIT and SUM as it allows for intensity noise to contaminate both channels in a same way. But we saw a significant decrease in the coherence by a factor of 100 or so. We will get back to this point in the next week and assess what is going on.
With single precision floating point representation and two poles at 30 mHz anything above ~30 Hz is probably bit noise.
A follow up on the beam size jitter measurement.
So far the data seems still consistent with the hypothesis that excess noise in 200 - 1000 Hz is due to beam size jitter which happens to be coherent with intensity noise at the output of the high power oscillator (HPO).
The first attachment is a screenshot of plots showing the coherence of DARM with some relevant intensity noise channels. The data is from configuration (C), see the above entry for details. The below is a list of remarks on the plots.