Figure 1. Today I have tried a large exploration of the PRCL offset to see if that can increase the sideband gain. I also have adjusted the MICH offset to keep the sideband balanced on B1p, as a bad interference on sidebands will mean a lower gain. Even for large offsets (60) the impact on the sideband gain has been small, maybe a 5% increase, and not very clear. For large offset the power of the carrier started to decrease by 10%, which makes sense as for a strongly detuned PRCL the recycling of the carrier will not be as good. The time series starts just after a relock, and I have tried to cut out of the picture most of the transient at the beginning of the lock.

There are a couple of other things that improve slightly for a large PRCL offset, the DCP might be increasing from 420Hz to 430Hz, and the B1p power decreasing by 10-20%. But at the same time the 56MHz sideband becomes very misbalanced on B4, with 40% more power on the LSB than the USB. In fact the relative power between the 56MHz LSB and USB on B4 is the only quantity that looks very well correlated with the PRCL offset. 

This quite different from measurements done last year where an offset of PRCL of ~10 could change the sideband power by a few percent https://logbook.virgo-gw.eu/virgo/?r=64109.  However, that measurement was done in LN3, and the sideband gain was 30% higher to start with for the 6MHz sideband. 

It doesn't look like the PRCL offset is going to be a magical solution to solve the problem that the sideband gain in CARM NULL / LN is much smaller than expected based on defects in the mirrors + CP defects + point absorbers (VIR-0870A-22). It might still be worth check what happens in LN2 or LN3, to see if the behavior has changed compared to last year.