SIESTA for Virgo locking experience

L. BarsottiUniversity of Pisa – INFN Pisa

on behalf of the

Virgo Locking Group

Cascina, March 16th 2004 Simulation Workshop

OutlinesOutlines

Commissioning of the first 3–km cavity

Recombined mode

Full Virgo

Other activities in parallel

North Cavity Optical SchemeNorth Cavity Optical Scheme

B1p

T=8%

T=50 ppmT=12%

6 W

B5

B7

PR, WI, WE mirrors misalignedWE

WI

NENIBS

PR

Commissioning of the North CavityCommissioning of the North Cavity

Feedback characterization:

• optical gain

• open loop transfer function

Analysis of the lock algorithm efficiency

• linearized error signal

• no linearized error signal

Comparison with real data (C1, C2 runs)

Real suspensions, real actuators, real photodiodes, computational delays included in the simulation

North Cavity Control SchemeNorth Cavity Control Scheme

B1pT=8%

B7

NENIBSPR

Hz

| Gain

|

frequency

B7_DCB1p_demod

Lock Acquisition

Linearized error signal:

1/m 102.4 8Optical Gain:Optical Gain: Measured

Simulated

Transfer Function Open LoopTransfer Function Open Loop

simulatedmeasured

Gain

Phase

M

G

zErrzCorr

noise

Lock Algorithm Efficiency Lock Algorithm Efficiency

Lock almost always acquired at the first trial

C1 run data : Several lock events collected locking and delocking the cavity

linearized error signal

Lock Algorithm Efficiency Lock Algorithm Efficiency

Failed locking attempt v ~ 12.5

m/s

8 m/s: maximum velocity

for the lock acquisition

success

10

33

m

Fv

λαBv

BΔt

MAXMAX

MAX

2

2

1

sμm

sμm

Constraints on the velocity according to the theory:

Gain due to the linearization:

~ 10

Lock Algorithm EfficiencyLock Algorithm Efficiency

With velocity lower than 10 m/s lock at the first attempt

With velocity higher than 10 m/s lock at the second attempt

Lock failed

Sweep at 12 m/s :

Lock event

Lock Algorithm Efficiency Lock Algorithm Efficiency

Failed locking attempts

not linearized error signalC1 data

Simulation

SIESTA link to real time controlSIESTA link to real time control

SIESTA

Control signalsPhotodiodes signals

Algorithms running in the global control

SIESTA link to real time controlSIESTA link to real time control

Control signalsPhotodiodes signals

Algorithms running in the global control

VIRGO

Recombined Optical SchemeRecombined Optical Scheme

B1

T=8% B5

B7

B8

B2

WE

NENI

WI

BSPR

PR mirror misaligned

Recombined modeRecombined mode

2 Steps locking strategy:

•sensing matrix

•procedure to find experimentally the algorithm parameters from simple optical systems

3 Steps locking strategy

•sensitivity curve

•comparison with real data

Linear locking

Reconbined Reconbined 2 Steps2 Steps Control Scheme Control Scheme

B1

B5

B7

B8

B2

north cavity controlled with B5

west cavity and michelson controlled at

the sime time

hLength_mic

tLength_wes

B2_quad

B1p_quad

10

2π1

Theorical optical matrix:

hLength_mic

tLength_wes

B2_quad

B1p_quad

10.08

0.041

Optical matrix measured by Siesta:

Michelson and West cavity controlled with the symmetric (B2_quad) and the antysimmetric signal (B1p_quad)

Sensing matrix

Locking simulation – North cavity

Locking

Locking simulation – Mich & WestPowers Lengths

Triggers

Corrections

B7_demod

B1p_demod

B2

North arm

West arm

B5

B8_demod

switch from B1p to B1 after the lock acquisition

Recombined Recombined 3Steps 3Steps Control SchemeControl Scheme

Lock acquisition -Lock acquisition - simulationsimulation

“Simple” simulation:

real suspensions and actuators

Lock acquisition -Lock acquisition - simulationsimulation

First lock acquisition27th February

Locking event At 3.25 am

Sensitivity - Sensitivity - simulationsimulation

Improvement:

real photodiodes (electronic noise, shot noise)

SensitivitySensitivity

Simulated

Measured

Switch to the linear locking state

west

nord

mich

4.3e34.3e31.8e2

2.6e42.6e40

10384

d1p_quad

d2_phase

d2_quad

Optical matrix:

d2_quad

d2_phase

d1p_quad

MICH

CARM

DARM

d1p_quad

d2_phase

d2_quad

4-1.16e5-1.9e0

4-1.16e-5-1.9e0

002-1.19e

west

nord

mich

Inverse optical matrix:

⊗

B1p_quadB2_quad

North arm

West arm

B2_phase

Linear Locking Linear Locking Control SchemeControl Scheme

Linear lock of the recombined

Simulation

Full Virgo Optical SchemeFull Virgo Optical Scheme

B1

B5

B7

B8

B2

WE

NENI

WI

BSPR

Multi–states approach (LIGO scheme)

Dynamical inversion of the optical matrix

Lock acquisition of full Virgo Lock acquisition of full Virgo

Lock acquisition of full VirgoLock acquisition of full Virgo

Something more…Something more…

Modal simulationModal simulation

Longitudinal local control optimizationLongitudinal local control optimization

Spikes removalSpikes removal

Modal simulation

High order modes (n + m ≤ 5 )

• compromise with the computational time 1 sec @ 20

kHz ⇒ 45 sec

Check with other codes in progress

0.113

misalignment of 2 rad in y of the curve mirror

Something more…Something more…

Modal simulationModal simulation

Longitudinal local control optimizationLongitudinal local control optimization

Spikes removalSpikes removal

Optimization of the z damping loop – I

10 sec

zCorr zMirrorm

Hz

Unity gain @ 0.65 Hz

measured

Open loop transfer function

Damping time sec

Optimization of the z damping loop – II

simulated

Open loop transfer function

Critical damping @

1.45 Hz

Hz

mV

zCorr zMirror

2 sec

Optimization of the z damping loop – III

measured after the optimization

mV

~ 2 sec

zCorr zMirror

Guadagno open loop

Hz

Critical damping @

1.45 Hz

Something more…Something more…

Modal simulationModal simulation

Longitudinal local control optimizationLongitudinal local control optimization

Spikes removalSpikes removal

Spikes removal Spikes removal

Spikes removalSpikes removal

Rearrange the algo:

Error signal derivative window integrator window

Other activity: Hierarchical controlHierarchical control

marionetta

reference mass

mirror

z

Control from the reference mass

Control from the marionetta

Transfer function betweeen force on steering filter and z movement of the

mirror

preliminary results

wwwcascina.virgo.infn.it/collmeetings/presentations/Mar2004/Fiori_11Mar04_MarioLockSim.ppt

ConclusionsConclusions

Siesta: fundamental tool for locking studies

Link to the real time control system

Work in parallel with other groups to improve the simulation (suspensions, alignment)

Noise analysis