Reports 1-1 of 1 Clear search Modify search
MIF (ITF Control)
kentaro.komori - 3:31 Sunday 13 April 2025 (33385) Print this report
Design of new filter for better PRCL feedback

Abstract:

I designed a new filter to improve PRCL control.
The unity gain frequency (UGF) will shift from 20 Hz to 40 Hz, and the phase margin at the UGF will increase from 30 degrees to 45 degrees.

Detail:

Although the reason for the low coherence between DARM and PRCL is not understood well, it is likely that PRCL displacement dominates the DARM sensitivity in the 10–30 Hz range, as discussed in klog:33357.
This is primarily because the current PRCL feedback filter provides a limited phase margin of approximately 30 deg around the UGF at 20 Hz, which leads to an enhancement of PRCL displacement in this frequency band.
As shown in the bottom right panel of Fig.1, which displays the closed-loop transfer function 1/(1 + G), the displacement at 30 Hz is amplified by roughly 9 dB.

Given that the PRCL displacement can now be measured clearly up to 50 Hz, due to increased optical power compared to O4a, it is feasible to increase the UGF without injecting significant sensing noise at high frequencies.
Based on this, I have designed a new PRCL feedback filter.

I temporarily updated FM1 in PRCL2 (named "temp"), which previously contained a simple gain(10 dB) filter.
The new filter consists of a straightforward zero-pole configuration: zpk([1; 2; 2; 10], [0; 0.05; 0.05; 100], 1, n) zpk([], [1000], 1, n) gain(111 dB).
To avoid displacement enhancement near the UGF, no sharp filters were introduced to suppress high-frequency feedback of the sensing noise.

Figure 2 compares the current (blue) and proposed (red) PRCL open-loop transfer functions.
The new UGF and phase margin will be 40 Hz and 45 degrees, respectively.
This modification is expected to yield an improvement of approximately 5–10 dB in DARM sensitivity in the 10–30 Hz band.

Next, I will produce a projection plot showing the expected PRCL contribution with the new filter and confirm that feedback of the sensing noise will not become a limiting factor for the DARM sensitivity.

Images attached to this report
Comments to this report:
kentaro.komori - 2:37 Tuesday 15 April 2025 (33419) Print this report

Abstract:

I designed new PRCL filters that can also be used during lock acquisition.
By updating the guardian and performing the same actions in the PRCL1 module instead of PRCL2, the lock acquisition should proceed smoothly, and the new final filters can be applied.

Details:

To develop appropriate PRCL filters compatible with the lock acquisition process, I investigated which filters are enabled or disabled at each stage.

  • FM5 is enabled from the beginning.

  • During PRMI 3F locking, FM9 is enabled and the gain is ramped up to -1.0.

  • After the IR handover, during the transition to 1F, FM3 and FM7 are turned on.

  • In the observation state, FM7 is turned off and FM2 is enabled.

These operations are currently handled within the PRCL2 module.
I have now implemented new filters in the same filter modules (FMs), corresponding to each IFO state, but within the PRCL1 module.

I compared the open-loop transfer functions under two conditions:

  • During PRMI 3F locking (FM9 on, gain applied), shown in Fig.1.

  • During the transition to 1F (FM3 and FM7 on), shown in Fig.2.

In both figures, the red line represents the new filter configuration, and the blue line represents the current one.
As discussed in the original post, the filters in the observation state (FM7 off, FM2 on) will reduce the PRCL coupling.

Since the new filters produce relatively similar transfer functions at all stages, the lock acquisition and observation should work well if the guardian is modified to use the PRCL1 module instead of PRCL2.

Images attached to this comment
takaaki.yokozawa - 9:44 Tuesday 15 April 2025 (33425) Print this report
I implemented to VERTEX guardian.
But when PRMI 3f locked with new filter, 1 Hz oscillation started and lock loss happened.

Now VERTEX guardian back to original filter
Images attached to this comment
Search Help
×

Warning

×