Transcript Locking improvements

Locking improvements
after the end of VSR1
Gabriele Vajente for the Locking Group
14th ILIAS WG1 meeting
Cascina – March 6th 2008
The end of the story…
October 1st 2007
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
2
The end of the story…
February 5th 2008
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
3
And now the full story…
Better diagonalization of driving
New sensing strategy
Modulation at 8 MHz
New filtering algorithms
Better speed performances
Much more flexible
Improved control filters
More low frequency gain
Less noise re-introduction
Improved noise subtraction
Alpha, beta, gamma techniques
Beam jitter subtraction
Reduced actuator noise
Locking with lateral coils
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
4
Virgo DOFs, sensing and driving
DARM = Differential long arm
CARM = Common long arm (not frequency)
PRCL = Power recycling cavity length
MICH = Short Michelson differential
PRCL
DARM,
CARM
MICH
ITF reflection
=
B2
Dark fringe
=
B1
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
BS pick-off
=
B5
5
Better diagonalization of MICH-PRCL
MICH correction is sent both to BS and PR
Measured the relative actuation gain in simple
configuration
PR – BS – WI cavity
Gained a better de-coupling of MICH and PRCL loops
WI
LOCKING
BS
PR
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
Reference
6
New sensing strategy
Main modulation 6 MHz
Additional modulation at 8 MHz, phase-locked
Not resonant in PRC or arm cavities
B2 beam (ITF reflection) demodulated at 8 MHz
Gives good signal for central cavity control
Old (VSR1) configuration
New sensing scheme
MICH
Controlled with
B5_ACq + B2_ACp_mix *
UGF @ 15 Hz
PRCL
Controlled with B2_ACp_mix *
UGF @ 40 Hz
Sidebands
Balanced with an offset on B2_3f_ACp
and/or B5_ACq
MICH
Controlled with B2_8MHz_ACp
UGF @ 15 Hz
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
PRCL
Controlled with B5_ACq
UGF @ 80 Hz
Sidebands
Balanced with an offset on MICH error
signal added in Gc
7
Performances of new sensing
8 MHz demodulated signal is much less noisy than 6
MHz one if used to control MICH
Large reduction of control noise at low frequency
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
8
New Global Control filtering algorithm
Just a “software change”
Reduce computation overhead
Allow to redefine control filters while locked
But a significant improvement in terms of
Robustness
Flexibility
Allowed a much faster development and test of new
control filters
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
9
New control filters / more LF gain
Largely increased loop gain below 100 mHz
No evidence of up-conversion of noise
MICH
PRCL
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
CARM
10
New DARM filter
DARM loop had low phase margin (13 degrees)
When cavity poles move (etalon) the DARM gain peaking changes a lot (3040 %)
This directly affects the dark fringe calibration
New filter for more margin: calibration changes by 2% at most
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
11
Low noise MICH filter
Developed a new filter, with more optimized high
frequency roll-off
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
12
Alpha, beta, gamma techniques
Auxiliary loop noises couples strongly to dark fringe
Noise subtraction techniques
The MICH, PRCL, CARM corrections are filtered and added to
DARM correction, to cancel noise
(B. Swinkels)
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
13
Performances
With alpha (MICH) we can attain a suppression of 1000
Less stringent requirements on
beta (PRCL) ~ 10
Gamma (CARM) ~ 100
ALPHA
Measurement and fit
Suppression predicted with
different filter orders
Suppression 1/1000
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
14
Beam jitter noise subtraction
MICH error signal (ITF reflection demod at 8 MHz) is
dominated by input beam jitter noise
Part of this is visible on BMS signals
It can be subtracted with a simple filter
Black: B2 8MHz
Red: Subtracted
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
15
Actuation noise
x4
UP
x4
LEFT
RIGHT
DOWN
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
Series
resistor on
L/R coils 4
times larger
than on U/D
16
Locking with Left + Right coils
Need to accurately measure
and balance actuation gains
If coils are not well tuned,
alpha filter changes
significantly over time and
during different locks
We suspect very small
alignment drifts
Solved developing a
technique to measure coil
gains with ITF in low noise
state
Now this is the standard
configuration
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
(V. Dattilo)
17
Effect of switching
central hall air
conditioning off
CALIBRATION LINES
CARM err
PRCL err
MICH err
Environmental noise
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
18
Conclusions
Lot of work after the end of VSR1
Improved locking robustness
Longitudinal noise is no more limiting at any frequency
Error signals are dominated by environmental noise
With AC off, locking is compliant with design
Developed many techniques to characterize the ITF
October 1st
Gabriele Vajente, 14th ILIAS WG1 meeting – Cascina
February 5th
19