It is likely that the increase in noise level is just a consequence of higher power and maybe different modulation depth for the 56MHz.
A coupling mechanism could be that if the 56MHz modulation depth is noisy, then the laser power stabilization that acts on the total power (carrier + sidebands), will translate that noisy side-band into a noisy carrier.
The coupling factor should be proportional to the square of the modulation depth (i.e. the relative power of the sideband to the carrier), so the easiest check is to change the 56MHz modulation depth in LOW NOISE 3. A change in the modulation depth of the 8MHz could also be tried (6MHz is probably to small modulation depth to matter).
It is possible also to make a direct measurement of the 56MHz modulation noise by locking on the 56MHz as was done back in May 2017, but that takes a bit more time, and then the result has to be modeled to translate into a noise projection.
I don't think this is the explanation of our 1/f^2.5 noise, as when I try to do a projection quickly in the noise budget a 1/f^2.5 RIN noise becomes ~1/f^5.5 in the sensitivity. But there might some other coupling mechanism than the one going through the total power stabilization, so changing 56MHz and 8MHz modulation depth can be an interesting experiment if there is no other leads on the 1/f^2.5 noise.
The modulation amplitude was raised from 7 dBm (O2) to 12 dBm (Post-O2 Commissioning) on Aug 27 (see entry 42544), as usual between Dark Fringe and LOW_NOISE_1; at the time it resulted in some improvement, today we made the test to reduce it once in LOW_NOISE_3 in order to check it. In the attached plots there is a zoom of both Hrec and DARM in the band of interest, with the following timestamps ( + 180 s) and colormap:
| 56MHz Mod Ampl | UTC Time | Color |
|---|---|---|
| 12 dBm | 16:15:40 | Red |
| 11 dBm | 16:20:00 | Light Green |
| 10 dBm | 16:27:00 | Black |
| 9 dBm | 16:32:30 | Cyan |
| 8 dBm | 16:38:00 | Purple |
| 7 dBm | 16:46:00 | Blue |
Unfortunately no big change was observed and the pattern is not related to the decrease of the modulation amplitude; the only remark is that at 11 dBm the noise seems quite better; to be checked if this was just a coincidence or if maybe our modulator is less burdened by a slightly lower modulation amplitude.
