I increased REFL PD power from 5.5 mW to 16.7 mW while keeping CARM UGF around 50 kHz (fig1).
Figure 2 shows DARM spectrum when the REFL power was 5.5 mW (blue) and 16.7 mW (red).
At high frequency region, there is no significant reduction of the noise, which implies the current shot noise level doesn't contaminated by CARM shot noise.
Note:
I keep the REFL power at 16.7mW until lockloss happened.
After the lockloss, HWP angle will automatically go back to the nominal one and REFL power will be 5.5mW from the next lock.
To measure the spectrum I used the time from 11:02:00 UTC for 5.5mW lock and 11:12:00 UTC for 16.7mW lock.