mechanical Q factor measurement

[YoshinoriFujii]

Todo

• measure the mechanical Q-factors for each resonant mode.

  • [x] w/o damping controls
  • [x] w/ damping controls

• First, Do finish this work at ETMX.

[YoshinoriFujii]

Notes

• For a check with FFT: /users/VIS/TypeA/general/qfactor_measurement/ETMX/wo_control/mode39/MNY2TMY_hpfiltered_0.5Hz/exe_plot_asd.py

• If you want to get filtered timeseries: /users/VIS/TypeA/general/qfactor_measurement/get_timeseries_filtered.py

[YoshinoriFujii]

Log at 2019.10.11

• Since I had no idea what are the exact frequency of the mode 56 and mode 58, I tried to measure the mechanical transfer function with swept sine. However, based on the measurement, it seemed the suspension did not have those resonances at 6.0Hz.
  • [ ] Measure TF more concretely.

[YoshinoriFujii]

• The resonant frequency of #50 was identified by looking at a MNL-excited spectra:
  • In the plot (following), 2.15 Hz and 2.48Hz were the dominant ones.
  • see /users/VIS/TypeA/general/qfactor_measurement/ETMX/wo_control/mode49/timeseries/figures
  • Before filtering:
  • After filtering with a band-pass from 1.9Hz to 3.0Hz:

[YoshinoriFujii]

Note 0n 2019.11.1

• applying bandpass in python
  • ref: https://scipy-cookbook.readthedocs.io/items/ButterworthBandpass.html
  • test files: /users/VIS/TypeA/general/qfactor_measurement/bk/tools_bk/bp_filter_bk.py
  • its outputs: users/VIS/TypeA/general/qfactor_measurement/ETMX/wo_control/mode63_Oct11th/timeseries/test/

[YoshinoriFujii]

Note on 2019.11.9

• how to get the time-series from KAGRA main server:
  • see https://www.evernote.com/l/AoOJksKhCtNEiZ5ym7p1npInYc1dG4eBWEg/

[YoshinoriFujii]

Note 2019.11.12

• It was observed that one resonant peak at 9.69 Hz had longer te compared to the requirement: about 70sec.

• This peak looks appeared in the TF of MNVexc to MNV/MNPoplev/TMPoplev, and thus it might be useful to close the following loops:

  • MNVps to MNV
  • TMPoplev to MNV

• [x] Just try

---

• FYI, the transfer functions:

---

• [X] This was damped with a BP-comb filter using TMPoplev and MNV actuator.
  • see https://docs.google.com/spreadsheets/d/1VoEfaNRhGDV501i5V56EScmZ8ik1PI-ZFldY0ue5di4/edit#gid=0
  • see also https://docs.google.com/spreadsheets/d/1roQkUBhXvQ-Y1gkTwliK7Gg17kzqKy_00ZoyPeD7AzA/edit#gid=0

[YoshinoriFujii]

Note on 2019.11.29

• It was found that another mode had longer te than the requirement, i.e., 60sec.
  • mode#43, 1.5Hz (T): ~200sec.
  • I would write that: "This can be damped with a BP-comb filter '(MNP to MNR etc). So far, this is not issued for the lock-acquisition, and thus nt much attention was paid."

---

• Some unexpected modes are found.

  • Based on a model which does not include the heat-link and its VIS.

  • 0.14Hz in Yaw, te was about 167sec

  • This peak was seen in BFY2BFY loop (with I-loops were closed), and the BFY loop (red-colored one) should work for this:

    • However, around this peak there was so small coherence (it is natural though..)

  • How this should be treated:

  • [ ] say that "only for this mode, the requirement is not met. It would be true that this mode would be hardly excited and also its te is like 3 min., not the ridiculously long one, and thus, for now it is okay. Actually this mode was found in BF_LVDT signal (in Yaw), hwever, this peak was not found at the TM stage. Further improvement should be done though. For lock acquisition for now should be okay."

    • [ ] いや、でもBF揺れたら最終的にTMも揺れるよね。どこで運動量保存してるの？によるけど、どこで？ - [ ] Or, measure it again with larger amplitude and see the result.

  • [ ] Actually what is this mode? where comes from?

[YoshinoriFujii]

Result

• resonant frequency vs. 1/e decay time constant:

---

• mode number in a model vs. 1/e decay time constant:
  • This model does NOT have the information of heat links and its VIS.
  • A few (it looks 3.) unknown modes were found and thus I just put #76 for one of it.
  • The others were NOT plotted in the bellow.