Conor, Chiara

After finding a discrepancy in the low-frequency calibration in earlier work, we found the cause and re-enabled the 'correct' low-frequency calibration for CAL_PRCL. This created a new 'feature', due to Foton, at high frequencies, see below.

To undo this change, simply disable the output of the M1 CAL filter module.

The filter module:

H1CALCS_CS_SUM_PRCL_PRM_M1

is responsible for the M1 drive part of the PRCL calibration. This is required for correct calibration below the M3-->M1 hand-off, which is about 0.7Hz. The filter output is currently 'off'. We will update this filter in two ways before enabling it.

1. Adding a module for "M1 and M2 Lock".

In the SUS model, the PRCL control signal passes through three suspension locking filters, M1Lock, M2Lock, and M3Lock (see block diagram below). The Calibration model, on the other hand, has these filters in parallel. To correct for that, we are installing a module which has the combined filters and gain settings. Currently M1Lock is just unity, and M2Lock is unity with a gain setting of +10.

This will replace the module (not engaged) named: BS270_430     which contained:     ellip("BandStop",4,1,40,270,430)gain(1.12202)

2. AC-coupling.

We suspect that the reason this module output is disengaged is to prevent saturating numerical precision with large offsets. To ensure there is no problem with that, we will AC-couple the signal. For aesthetic reasons, we are using an 8mHz Butterworth high-pass filter, which should approximately match the in-built STS-2 AC-coupling frequency.

We enabled H1CALCS_CS_SUM_PRCL_PRM_M1 filter bank with the two new filters.

-------------------

High-frequency feature

After implementing the calibration change, there was a roll-up of noise starting at about 3kHz in CAL_PRCL. Some investigation reveals that this is due to the extreme roll-off in the calibration filters encountering some numerical error in foton. The Bode plot shows a discontinuity at ~2.5kHz (the corner frequency for the new feature), and when we empirically toyed with the calibration filters, we could move the feature around, but not remove it.

For now, we will leave the M1 Calibration filter on, and accept that the calibration is incorrect at high frequencies. It might be possible to remove this feature with an updated suspension calibration filter, especially if it is lower-order, or has a flatter spectrum at high frequencies.

Attachment 2 shows the Bode Plot (and zoom of the troubled region).
Attachment 3 shows the ASD of CAL_PRCL before (red) and after (blue, reference trace) the calibration change.