For background, I attempted to push a new calibration on 7/3 to account for the change in the SRCL offset that we made on 6/26, but it failed due to the broadband PCAL measurement showing a larger uncertainty that we had beforehand (see 85529). Since then, we have been running with the same calibration we have had since 6/10, which has a low error (~3%), but is based on a model that know to be incorrect. Namely, the model created and pushed on 6/10 has a small, positive spring, and we believe now that DARM has no spring to at least 10 Hz. We are especially confused because we expected the model change to be focused around the 10-30 Hz region, since this is the band where we expect significant change due to the SRCL offset, but the measurement shows large, >5%, error at 100 Hz.

I have made a series of plots comparing a variety of PCAL broadband measurements from different points since 6/10, measuring PCAL with GDS CALIB STRAIN and CAL DELTA L.

Plot 1 shows CALIB STRAIN/PCAL and DELTA L/PCAL on 6/11 after we pushed a new calibration modeled with a positive spring. The calibration at this point was very good; the calibration line uncertainties showed error of 3% or less. However, this plot is already showing something a bit confusing- a difference in CAL DELTA L and GDS CALIB STRAIN, where GDS CALIB STRAIN has a higher uncertainty around 70-200 Hz. We believe the application of the kappas should further reduce the uncertainty of GDS CALIB STRAIN.

Plot 2 shows CALIB STRAIN/PCAL and DELTA L/PCAL on 6/26 after changed the SRCL offset. The calibration report generated from that day indicates that the sensing function is flatter with the adjusted SRCL offset. Because the calibration still expects a spring, we were not surprised to see that the low frequency uncertainty changed.

Plot 3 shows CALIB STRAIN/PCAL and DELTA L/PCAL on 7/3 after we pushed a new calibration which was supposed to account for the flatter sensing function. However, we saw that the uncertainty increased at 100 Hz, which we did not expect. This measurement was run slightly early during the "TDCF burn in" so it may not have been an accurate look at the effect of the new calibration.

Plot 4 shows CALIB STRAIN/PCAL and DELTA L/PCAL on 7/3 after we pushed a new calibration, and then were only relocked for 10 minutes. The uncertainty was even larger than the previous uncertainty measurement. We were also very confused that CAL DELTA L changed significantly compared to plot 3. We're not sure if the kappas were significantly different from 1 to also cause problems in GDS CALIB STRAIN when applied.