Gabriele, Sheila, Evan
Today we looked at how changing the 45 MHz modulation index changes the appearance of broadband excess noise in the DCPDs. As one should expect, this coupling goes like the square of the modulation index. With our current index, the excess of the sum over the null is about 8 %.
To be a bit more quantitative:
-
We started with the nominal EOM power (23.2 dBm), which should correspond to a modulation index of 0.28 rad, as was computed in LHO#15674. Around 1.3 kHz (a clean part of the DCPD spectrum) the DCPD sum is 8.3×10−8 mA/Hz1/2, and the coherence with the ac EOM driver readback is 0.11. That means the RFAM-induced contribution to the DCPD sum is 2.8×10−8 mA/Hz1/2, compared to the null contribution of 7.7×10−8 mA/Hz1/2. These two contributions add in quadrature to produce 8.2×10−8 mA/Hz1/2, which would seem to account for most of the excess.
-
Then we reduced the EOM power by 3 dB, and compensated accordingly in LSC POP45 and some of the ASC diodes (REFL A/B 45 and AS A 45). (It was not clear to us how one should compensate the AS 36 diodes.) Then again at 1.3 kHz, the DCPD sum is 7.9×10−8 mA/Hz1/2, and the coherence with the ac EOM driver readback is 0.030. Then the RFAM-induced contribution is reduced to 1.4×10−8 mA/Hz1/2 (a factor of two lower than before), consistent with the notion that this coupling goes like Γ2 (as was anticipated; see LHO#20199).
We tried to reduce the modulation index further, but by 4 dB it seems the interferometer will not stay locked for longer than a minute or so. While changing the modulation index we can see that the uncontrolled quadratures of the AS 36 WFS change, perhaps indicating that the error signals for the BS and SRM loops are also changing in an undesirable way.
The attachment shows the DCPD streams, along with the EOM driver RFAM readback and the IOP channel of ASC AS C segment 1. The solid traces are taken at the nominal modulation index, and the dashed traces are taken with a modulation index that is 3 dB lower.