Craig, Georgia
On Friday we pico'd AS_B to offload some of the OMC ASC onto OM1 and OM2. This afternoon we tried switching the OMC ASC from the QPD alignment to the dither alignment again.
It seemed like the dither and QPD alignments were similar in PIT and a bit off in YAW. First I walked the QPD_A and QPD_B PIT offsets to try to get the components of the dither alignment error signals (OMC_ASC_P1_I and OMC_ASC_P2_I) closer to 0. The two offsets needed to be walked as just increasing the QPD_A offset would saturate the OMC SUS in PIT. While walking the offsets I found a small change improved Kappa C. Offsets are tabulated below:
| QPD_A_PIT OFFSET | QPD_B_PIT_OFFSET | |
| Starting offset | -0.368 | -0.15 |
| Best Kappa C | -0.338 | -0.18 |
| reduced dither error signal | -0.158 | 0.1 |
We tried switching the OMC ASC half way to dither, things seemed ok, so we switched all the way to dither. After a minute or so it was clear that the control signals were going off into space, the error signals were getting further from 0, and our kappa_c was tanking. Looking at the error signals in the first attached screenshot the POS_X and POS_Y loops seem normal but the ANG_X and ANG_Y error signals and control signals grow over time. I also tried switching to dither loops and swapping the sign of the master gain (from 0.02 to -0.02) but this was bad for all 4 loops (second screenshot).
The last time the OMC was successfully put on dither alignment there was a large alignment shift between the two alignments (see first attachment of alog 58970), and a bit of a shift in kappa_c, but it was a lot gentler than what we saw today.
To facilitate this transition in the future, I wrote a OMC ASC DOF comparison script.
It reconstructs the OMC ASC DOFs you would get from the dithers vs the QPDs using the input matrices, and plots the ASDs.
At low frequencies the gains are overall matching.
It seems that the dithers are sensitive to some actual displacement peaks seen by the QPDs, so these error signals are not total nonsense.
We hope to upgrade it slightly tomorrow to do the passive TFs so we can properly switch from QPDs to dithers.
This transition shouldn't be hard but for some reason it has been for us.
It's in /ligo/gitcommon/labutils/reconstruct_omc_asc_dofs, and also https://git.ligo.org/aligo_commissioning/labutils/-/tree/master/reconstruct_omc_asc_dofs.
First, it might be useful to do beacon-type QPD instead of DC QPD using one of four calibration lines between 15.6 and 17.6Hz (or any other DARM lines visible in QPD_SUM).
QPDs will observe the centering of the field that carries the DARM information, ignoring everything else. We can do it for ASC-OMC_A and B as well as ASC-AS_C. See the cartoon representing the idea.
If we use one of four calibration lines between 15.6 and 17.6Hz, since they are only 0.5-0.8Hz apart from each other it will be very slow, but probably that's fine.
Second, deacon dither infrastructure could be used instead of simple dither (alog 45825) but it will take a long time to tune, at first to see if the deacon oscillator could be stably sustained using the current hardware w/o bad side effect (deacon was never tested in aLIGO era as far as I remember) and then to see if it's usable for ASC.
Update: Evan and Nancy briefly tried Deacon infrastructure (alog 64632) and it only somewhat half-worked while the IFO was in high dynamic range state. Evan asks if there's enough oomph for ESD in low dynamic range mode.
