Jeff K., TVo, Jenne, Hang
We finally managed to go through the low noise ASC state (by hand) with the current 30 W pre-loading. However, compromises had to be made to keep the loop stable. For now it seemed that the most fragile ones are CHARD YAW, and somehow whenever it grew unstable, an instability in DHARD PIT would be triggered as well at the same frequency.
============================
Things happened:
1). We measured the PRC1/PRC2/CHARD YAW to REFL_9/45_A/B_I at 20 W. While dithering PRC1/PRC2, we saw those signals showing up strongly in the CHARD YAW error point even we already put notch filters. On the other hand, we barely saw the CHARD YAW excitation in its own error signal though the line was already high in DARM after running the A2L script once. Thus it seemed to indicate that a large cross coupling in the current YAW sensing matrix.
The response we measured at 20 W was (caution: the CHARD YAW had low coherence and would need to be re-measured in the future.)
| PRC1 YAW | PRC2 YAW | CHARD YAW | |
| RF9_A_I | 0.41 | 0.19 | 0.67 |
| RF45_A_I | -0.45 | -0.37 | 0 |
| RF9_B_I | 1 | 0.57 | 0.54 |
| RF45_B_I | 0.8 | 1 | 1. |
which can be psuedo inverted to get an input matrix. Currently, we only use the ones for CH Y:
| RF9_A_I | RF45_A_I | RF9_B_I | RF45_B_I | |
| 1.67 | 1.4 | -0.4 | 0.43 | CHARD YAW |
Using this matrix we could lock the CHARD YAW at 2 W without any issue. However, by changing the matrix we still could not reduce the CHARD YAW gain to the level we had last week.
2). For the CHARD YAW loop we leave its DC gain at 1.5, corresponding to ~ 4 Hz UGF. For the blending setup the A path (DC signal from the REFL WFS) has a gain of 1.3 and the B path (AC signal from the TR QPDs) has a gain of 0.77. Also it seemed that we could not engage the original LP filter (FM2) due to lack of phase margin. Instead we made a less aggressive one (FM3).
We tried once to change the blending frequency down to 0.15 Hz where we should have more control and thus more tolerance to imperfections in the blending cross over. But we did not succeed. As soon as we engage the filter we immediately lost lock. Future investigations needed.
3). We also saw that somehow the DHARD PIT oscillated together with CHARD YAW (strange...) As soon as the CH YAW started oscillate the DH PIT would grew unstable and vice versa. So far we had to compromise by having a high BW of 5 Hz to keep both DH P and CH Y stable. Due to this high BW we could not engage FM1 (ELP10b) but had to use a less aggressive filter FM4 ELP17.
4). The DH YAW and CH PIT seem to behave well as last week.