Virgo Logbook

Tonight we tried to understand the B4 6MHz Q signal for MICH, and how to manage the handoff. Unfortunately we're even more confused by the B4 6MHz Q channel than before.

The alignment is much more stable than last night, following the fix to the WI local controls. We can remain locked at points arbitrarily close to dark fringe for tens of minutes.
We tried the MICH handoff a few times, again tuning the offset and gain to match the DC signal. We found we couldn't do this without handing DARM off to B1p, and the offset tuning is still mostly guesswork. But, it can be done.
The offset of B4 6MHz Q is a mystery. In Figure 1, we show data from last night, which indicates the offset *grows* as we approach dark fringe. Notice that the 12MHz signal also grows, so maybe it's a fixed offset, and the magnitude is growing as the sideband buildup increases?
There is a relative offset between DARM from B7B8 and DARM from B1p. Switching to B1p does something strange to the beams: B1p gets a clear TEM10 mode, mostly in TYl, and the arm powers drop, but in an asymmetric way. We can re-introduce an offset on DARM after the handoff and partially recover the beam shape on B1p, but not the arm power.
We're very confused by the demod phase used by SSFS for B4 6MHz. There is a hard-coded offset between the demod phase in SPRB_Photodiodes and the SSFS process, but we're not sure if this offset is tuned correctly - maybe it has changed after reboots of the SSFS box? The B4 6MHz I-phase signal is not suppressed to zero like we expect it to be (and it appears to change DC values from lock to lock). This is something to understand in the morning, perhaps we have to wait until appropriate signals are available from the SSFS process.
We measured the MICH and PRCL loops with swept-sine measurements, see Figures 2 and 3. Things look good; these measurements were taken at 98% dark fringe, MICH on DC, PRCL on B2 8MHz. The gps times are kept in /olusers/virgorun/ISC/TFs, look for the MICH and PRCL files from 16Feb. Tomorrow we will examine the B4 6MHz Q phase signal at these times and see if the transfer function between MICH,PRCL and that signal has the expected shape. We saw, during MICH noise injections, a lot of coupling into PRCL, so maybe our sensors are so cross-coupled that the strange shape of B4 6MHz Q is due to PRCL loop interactions.
Also it might be interesting to measure these loops after DARM is on B1p. We've seen the loop gains drop significantly after the DARM transition, possibly this is correlated to the weird transition of the B1p beam and arm powers.
Starting at 2:07:30 UTC and lasting for ~50 minutes we collected data at 98% dark fringe and DARM on B1p. DARM in this state has a lot of noise features, and we'd expect (hope) it would just be sensing noise. Might be worth looking for coherences with ENV channels. The 1Hz comb shows up everywhere; this needs to be fixed.

Other things:

After handing DARM off to B1p, we notice a lot of motion around 2-3Hz, so we designed a second boost filter for the DARM loop. When we tried to enagage this filter we found that the loop gain at low frequency was much, much larger than we expected. See Figure 4, where the purple trace is with the old boost filter, the gold trace is the new boost filter, and the blue trace is the two of them together. The blue trace is much larger than the product of the purple and gold! We think that the addition of this filter exceeds the allocation of pole/zero pairs that we have hard-coded into the DARM loop filter, and some low-frequency features are being dropped in the filter computation. The biggest consumer of pole/zero pairs for DARM are the many notches for the mains lines. We dropped these notches from the filter selection, and the boost worked as expected. (Also: the Reload Filters button is not working as we expect, filters are not updated in real time when we save & reload? Do we need to re-set the filter for the new parameters to take effect?)
Thinking the strange B4 6MHz offset might be due to a PRCL detuning, we scanned a PRCL offset, but found no evidence that an offset different from zero is useful.
We should measure the PRCL length to make sure the sidebands are resonant in the cavity.
We've looked at using the 119MHz signals for MICH, but these don't pick up the MICH line with good signal-to-noise. Maybe we could bring back the 56MHz, to have an alternative locking signal that isn't coupled to the SSFS.
Some of our locklosses are closing the shutter on B1p PD2. If a shutter is closed by a bright flash, does the ensure_PD_enabled() function not reopen the shutter? It looks like the shutter needs to be rearmed, maybe this is not included in the function.
The MICH --> PR_CORR signal was ringing up the PR violin mode at 219Hz; this was notched in the PRCL loop but not in MICH. We've notched it. It's a good sign that we have locks long enough to ring up wire resonances.