On Tuesday between 8am and 9am LT, I had added Acl code into SDB1_LC
to be able to substitute the B5 QD2 signals used for drift control
with the error signal of the SDB2 B1p QD1 galvo loop. I could at that
time successfully lock the bench drift control, after having changed
the sign of the error signal, and use DC centering instead of 2f
centering for the error signal.
Doing the same in LN3 has failed twice (once on Tuesday and once on
Thursday, each time at the beginning of the commissioning shift).
I went then to DRMI 1F, where one can move SDB1 significantly while
still having a 112MHz 2f signal. After a few tries I have realized I
have used the H B1p QD1 signal twice both for horizontal and vertical
bench control. After changing that and adjusting the vertical sign I
was able to close the drift control of SDB1 using B1p QD1. I have
disabled floating set points to avoid potential complication from the
set point automation.
The first lock acquisition failed (at CARM to MC), but the second went
to CARM NULL 1F without problems, and without any manual commands for
the SDB1 LC.
Unlocked while opening the OMC slow shutter, but before it actually
opened. The correction on SDB1 TY were high at +/-30V before that, for
at least a minute. Reducing the horizontal signal gain by a factor
3. Most of the RMS was at ~5Hz, which could be the zero phase crossing
of the SDB1 TY open loop TF. On B5 QD2 there was a significantly
different offset compared to the normal lock acquisition, so to lock
the OMC actually a way of adding offset in the SDB1 control in this
configuration needs to be found.
Adding offsets on B1_DARM_TX/TY_offset of the order of 1 or 2 works to
correct the offset in the SDB1 alignment. Adding static galvo
correction using B1p_QD1_H/V_offset is a better option, as that keeps
the beam centered on the quadrant.
To complete the alignment scheme in this configuration one would need
to turn back on the galvo on B1p QD1 with a low bandwidth once the OMC
drift control based on the DARM line is enabled. Today doing it by hand by
adjusting the static galvo correction.
17:13 UTC (3min) LN2 reference time
continuing to LN3 and adjusting the B1p QD1 galvo offset to follow the
misalignment of SR
17:31 UTC (30min) LN3 reference time
Figure 1 compares the in-loop control in LN3 with B1p QD1 112MHz
signal in loop (blue) and with normal B5 QD2 in loop (purple). The
sensing noise is clearly higher using B1p QD1 112MHz by up to factor
10. But it is not coherent with the sensitivity, and doesn't increase
the h(t) spectrum.
Around 18:08 UTC Restored control to using B5 QD2 while the
interferometer was locked. It did a large step in the alignment, as the
OMC drift control had to learn the difference in offset. The B5 QD2 offset
is set for the aligned SR condition, and not the SR misaligned. After
a few minutes the SDB1 alignment came back to the starting point.
Figure 2 at 18:26 UTC I have change the SR TY set point from 175 to
190, so that the DCP pole measured by Hrec moves back to the 180-185Hz
that was measured before the DARM notch filter were changed. I haven't
changed the automation so this will be reset at the next relock.
Renable floating set point one minute after going to Science, prompted by re-reading the logbook entry before posting.
I will add an SR TX offset after one hour of Science mode.
