Reports 1-1 of 1 Clear search Modify search
fabian.arellano - 12:26 Sunday 26 May 2019 (8979) Print this report
Sensor noise injected by the control system in the observation frequency band.

The picture attached shows the sensor noise fedback by the control system.

  • The current control system does not fulfill the requirement above 10 Hz, which is where the interferometer observation band begins.
  • From the results of ther prototype test in Mitaka the OSEM noise contibution was expected to be high. As far as I understand, the strategy would be to turn off the local OSEM and oplev control and switch to the WFS (waveform sensing) of the interferometer for feedback.
  • The GAS filter LVDT contributuon is far too high. As far as I understand we aim to have those control loops closed during observation mode.
  • The amount of coupling from the optic vertical to optic longitudinal is 1%.
  • In the simulation I used the filters which are currently used in the suspension.
  • I used the measured noise of the LVDTs and OSEMs.
  • I didn't include oplev noise.
  • I didn't include geophone noise. The effect of inertial damping is not included yet.

Comments about GAS control filters:

  • In the prototype the control gain at high frequencies was supressed by using Butterworth filters with cut-off frequencies below 5 Hz, where the electromagnetic coupling from the actuator to the LVDT begins. See pages 100, 101 and 172 in Takanori's thesis.
  • For instance, In the case of SRM F0 (entry 7500 and 7513 for BF) the coupling begins about 2.5 Hz with the keystone locked and about 5 Hz when unlocked.

Reference: entry 8866.

Images attached to this report
Comments to this report:
fabian.arellano - 12:26 Monday 26 August 2019 (10120) Print this report

In case you don't see the pictures attached to this report please click here: 10120.

In the commented entry it was reported that the DC control noise for the GAS filters was too high. I rolled-off the filters using a high-order Butterworth filters (12th order at 4 Hz) and achived an amount of noise below the requirement. See pictures. However, more analysis is necessary:

  • Stability should be checked.
  • The order of the filter seems too high, people have suggested me to use Chebyshev or Elliptical low pass filters instead.
  • The measured sensitivity of displacement sensors are also shown.
Images attached to this comment
fabian.arellano - 13:11 Friday 22 November 2019 (11698) Print this report

In case you don't see the attached plots please click here: 11698.

Instead of using a Butterworth filter I used an elliptic filter of the 4th order with band pass ripple of 1 dB and attenuation of 40 dB in th stop band. Still, in order to reach reduce the necessary amount of noise the square of the filter was necessary:

  • The first plot shows the GAS filter control noise for all GAS stages without broadband damping filter, only the DC control filter is on.
  • The second plot shows the control noise with the F0 damping filter on together with the  DC control filter on for all GAS stages.

The last plot indicates we will need to roll off the F0 damping filter.

I checked the stability for the BF using the filters and suspension models uploded in Foton. The unity gain frequency is 25 mHz with a phase at that frequency of -93 degrees.

Images attached to this comment
Search Help