1LIGO-G020147-00-M

Initial LIGO Seismic Isolation System Upgrade

Dennis Coyne

LIGO Seminar

March 29, 2002

2LIGO-G020147-00-M

Problem

Ground motion at LLO with the initial LIGO seismic isolation system makes it impossible to hold the interferometers locked reliably during the day» Steady-state ambient noise is higher due to anthropomorphic sources» Transients, particularly from logging

Wind induced seismic noise at LHO:» exceeds locking threshold at ~25 mph, or 4% of the time» Expect up-conversion is a problem at significantly lower wind speeds & a large

fraction of the time Upgrade is required to allow both reliable locking and to allow

better noise performance while locked» Need 90% duty cycle & lock durations > 40 hours» Need to reduce noise in the control band (< 40 Hz) to permit a smaller

suspension actuator authority & lower noise » Suppression in the 1-3 Hz band is most important due to excitation of the

lower stack modes (Q ~ 30)

3LIGO-G020147-00-M

Typical Day/Night Seismic Noise Levels in the 1-3 Hz Band

4LIGO-G020147-00-M

Integral Histogram of the Peak Ground Velocity

5LIGO-G020147-00-M

Up-conversion Example

6LIGO-G020147-00-M

Initial Vibration Isolation Systems

HAM Chamber BSC Chamber

» Reduce in-band seismic motion by 4 - 6 orders of magnitude» Little or no attenuation below 10Hz; amplification at stack mode resonances» Large range actuation for initial alignment and drift compensation» Quiet actuation to correct for Earth tides and microseism at 0.15 Hz during

observation

7LIGO-G020147-00-M

Seismic Isolation – Springs and Masses

damped springcross section

8LIGO-G020147-00-M

Seismic System Performance

102

100

10-2

10-4

10-6

10-8

10-10

Horizontal

Vertical

10-6

HAM stackin air

BSC stackin vacuum

9LIGO-G020147-00-M

Active External Pre-Isolation (EPI)& Active Internal Damping (AID)

Matrix &Control Law

SupportStructureDynamics

Kaman Inductiveposition sensors 4-5 Hz

L4C Geophones0.5 to 3-4 Hz

STS-2 SeismometerSensor Correction

(on PSL Table)

GS-13 Seismometer(floor mounted)

FeedForward

~40-50 Hz UGF

Force

Local Loop BlendFrequency ~3-4 Hz

LSC Control Offload

PER ISOLATION PLATFORM

ETM only

Microseismic @~0.2 Hz for ETMs only

Passive StackDynamics

LVDTposition sensor

Matrix &Control Law

Force

~10-15 Hz UGF

ACTIVE INTERNAL DAMPING (AID)

EXTERNAL PRE-ISOLATION (EPI)

10LIGO-G020147-00-M

Alternate Approaches

# Approach Description/comments Options Isolation Stack Damping

1a Hydraulic actuator Y ?

1b

External pre-isolation 6 DOF isolation with co-located sensing & actuation at the base of the passive stack; feedback & feedforward control to be explored including use of OSEM sensing EM actuator Y ?

2 Internal active damping Co-located sensing and actuation on the internal optics table (e.g. LVDT and voice coil) to sense & damp from support structure to optics table. The addition of inertial sensing on the optics table may permit isolation.

Voice Coil or EM linear motor

LVDT or geophone

? Y

3 Existing fine actuators Longitudinal & yaw velocity feedback with co-located geophones. Being pursued as an interim measure.

Y N

4 COTS isolation systems Piezo isolation systems like Stacis; minus-k compact low frequency spring, etc. which can perform the external pre-isolation task.

Various Y unlikely

5 SAS-like Implementation A hybrid passive/active “soft” alternative approach to the stiff external pre-isolation approach.

Y N

6 Tuned Mass Dampers With existing payload mass limits the optimum reduction in stack mode resonance is ~4. This does not meet requirements and requires in-vacuum hardware

Viscous fluid, eletro-restrictive or eddy current

N Y

7 Multiple pendulum or longer period suspensions

Too invasive, too large a schedule & cost impact; not clearly a solution either

Y N

8 Cooled suspension coil drive electronics with larger dynamic range

Does not preclude increased noise due to bi-linear coupling mechanisms & large amplitude of real motion; might be a last ditch effort after other measures are taken

Y N

9 Short across 1 layer of the HAM Stack

Compromise the better-than-needed high freq. HAM isolation performance; shift stack modes; not clear this works; seems wrong to compromise performance

N Y

10 Replace some or all springs with lower Q springs

Too invasive & marginal improvement in Q without complete replacement

N Y

11 Add eddy current damping between stages

Too invasive & marginal improvement in Q without the addition of many components

11LIGO-G020147-00-M

EPI Requirements

Net Noise at the base of the stack:» Maintain present drift stability: 1 month, 10 microns pk-pk» Alignment precision for lock: 100 seconds, 1 micron pk-pk» Microseismic peak: 0.16 Hz, 4e-7 m/√Hz» Integrated rms level similar to Hanford at night & consistent with the

technology: 1 Hz 1e-9 m/√Hz 10 Hz 4e-10 m/√Hz

» Suspension vertical bounce mode: 15 Hz, 2e-10 m/√Hz» No degradation in current in-band isolation:

30 Hz 6e-11 m/√Hz > 50 Hz 2e-11 m/√Hz

Dynamic Range: 10mm p-p static alignment, 300 microns p-p

12LIGO-G020147-00-M

AID Requirements

Noise contribution in the GW band must be < 1/10 of the Science Requirements Document (SRD), or at the optics table:» BSC:

Horizontal: 1.2e-13 m/√Hz at 20 Hz, 2.5e-17 m/√Hz at > 50 Hz Vertical: 5e-13 m/√Hz at 20 Hz, 1e-16 m/√Hz at > 50 Hz

» HAM: Horizontal: 4e-12 m/√Hz at 20 Hz, 2.5e-16 m/√Hz at > 50 Hz Vertical: 2e-11 m/√Hz at 20 Hz, 1e-15 m/√Hz at > 50 Hz

Damping to Q~3 on at least the 1.2 and 2.1 Hz BSC modes, with no ‘spillover’ in excess of 1.5 x total rms

13LIGO-G020147-00-M

2 DOF Pre-Isolation with the existingFine Actuation System (FAS)

Proof-of-principle test for external seismic isolation layer. Existing fine actuators in end (and mid)station,driven in

pairs,can move stack base in beam direction and in yaw. GS-13 seismometers placed on crossbeams above FAS Provide inertial error signals for 2,1-DOF SISO servos. dSpace signal processing board and software in PC allows

rapid controller/compensation development and provides GUI control panel

Resonant gain added at two troublesome stack modes (1.2 and 2.1 Hz), allowing factor of a few more gain there without destabilizing overall servo.

This resulted in about a factor of 7 decrease in motion seen by the test mass at the stack modes.

14LIGO-G020147-00-M

2 DOF Pre-Isolation with the existingFine Actuation System (FAS)

GS-13 geophone

PZT fine Actuator & Flexure assembly

15LIGO-G020147-00-M

2 DOF Pre-Isolation with the existingFine Actuation System (FAS)

16LIGO-G020147-00-M

2 DOF Pre-Isolation with the existingFine Actuation System (FAS)

17LIGO-G020147-00-M

Planned Initial Detector Modifications

HAM

BSC

18LIGO-G020147-00-M

Spring/Actuator Assembly with theElectro-Magnetic Actuator

Horizontal electromagnetic actuator

Vertical electromagnetic actuator

Seismic support pier

Point of attachment to the seismic external crossbeams

Horizontal and Vertical geophones

Seismic stack static load carrying springs

19LIGO-G020147-00-M

The Quiet Hydraulic Actuator

Current Design Current Design ImplementedOn LIGO Pier

Bellows

ActuationPlate

ServoValve

DisplacementSensor

20LIGO-G020147-00-M

Differential Bellows for Quiet Actuator

1) Pump

2) Differential Flapper Valve

3) Bellows Supply

4) Differential Bellows

5) Actuation Plate

21LIGO-G020147-00-M

Conditioning a Pressure Source

Servo Valve

CompensatorPressureSensor

Pump

Reservoir

Accumulator

Active and Passive Suppressionof Pressure Fluctuations

22LIGO-G020147-00-M

Pressure Noise at the Actuator

100

101

10-4

10-2

Freq (Hz)

pres

sure

AS

D (

psi/

Hz)

no accumulator with accumulator accumulator and servo control

23LIGO-G020147-00-M

The Test Platform at StanfordVertical Actuator

Displacement Sensor

Seismometer(Geotech S-13)

Seismometer(Streckeisen STS-2)

800 lb Test Mass

Horizontal Actuator

Vertical Actuator

24LIGO-G020147-00-M

Sensor Correction

zg

zm - zg

STS2 Filter

Compensator

+

ExternalAlignmentInput

+

25LIGO-G020147-00-M

Vertical Isolation

26LIGO-G020147-00-M

Horizontal Isolation

27LIGO-G020147-00-M

Active Internal DampingSystem Layout

28LIGO-G020147-00-M

Active Internal Damping (AID)Prototyping

HAM PrototypeLVDT (left) & Constant force Actuator horizontal doublet

29LIGO-G020147-00-M

LASTI Full ScalePrototype Testing

Stand-alone subsystem testing is underway for each subsystem

The AID & HEPI subsystems will be tested on a BSC isolation stack/chamber at the LASTI facility (MIT) starting in June/July

The MEPI subsystem will be tested at the same time on a LASTI HAM stack/chamber

Initially all controls will be performed with D-Space controllers before integrating the systems into the LIGO Epics Supervisory Control & DAQ systems

30LIGO-G020147-00-M

Summary

Testing and analyses to date all look promising An interim solution which should enable LLO to lock

reliably between the Science 1 and Science 2 runs is being installed (2 DOF pre-isolation with the fine actuation system)

Seismic retrofit with an active pre-isolation system and an active internal damping system is planned for after the Science 2 run, in Jan 2003

A Design Review will be held in 2 weeks» April 12, 9:00 am PT

» Documentation is nearly ready for release

31LIGO-G020147-00-M

Functional Description of the System

~10-6 m rms above 0.1Hz~10-8 m rms above 1Hz, ~4 10-10 m/√Hz at 10 Hz

~10-7 m rms (0.1Hz), ~10-8 m rms (1Hz), ~4 10-10 m/√Hz

10-14 m rms differential, 10-19 m/√Hz (10 Hz)

~10-7 m rms (0.1Hz), ~10-11 m rms (1Hz), 2 10-13 m/√Hz (10 Hz)

32LIGO-G020147-00-M

Motivation for an External Stage

• Isolation– From the micro-seismic peak to 10 Hz

• Alignment– Seasonal Temperature Changes

• Control Reallocation– Reduce control effort / noise from inner stages

• High Impedance Support– Inner stages react against stiff, damped

foundation

33LIGO-G020147-00-M

Performance Requirements• Range of Motion

– Mechanical Adjustment: 5 mm.– Active Control: +/- 1 mm.

• Response – Initial Response: 1 mm. in 10 sec.– Bandwidth: .1 - 10 Hz.

• Resolution and Noise– Reduce the ground motion by 10 from 0.1 to

3 Hz

34LIGO-G020147-00-M

New Pumping Stand

•Build new system withimproved low freq range, and capacity to drive 8 actuators.

•Allow Stanford and LASTI to order identical stations (easier debugging).

ReservoirMotor and pump

Granite block(isolation)

35LIGO-G020147-00-M

New Pump Station is Progressing

Motor

Pump

Motor controlle

r

Bypass valve

Pressure sensor

Accumulator

Filter

36LIGO-G020147-00-M

Candidate Actuators

Hydraulic

Forc

eV

eloc

ity

Disp

lace

men

tSt

ictio

n

Mec

hani

cal

Noi

seHys

tere

sis

Stiff

ness

High Med

High

LowMedLow LowLow

High

High

High

Low

Low High High High Low

HighLow Low LowLow

High High LowLow HighPiezo orMagnetostriction

Linear Motor

Ball Screw

High

37LIGO-G020147-00-M

Equations of Motion

Valve

Volumes

Flow

Mass

m2t

zd

d

2 p 2 p 1 A k z z g

D

tp 1 Vd

d

Q 1 A

tzd

d tz g

dd

tp 2 Vd

d

Q 2 A

tzd

d tz g

dd

p s p 1

RQ 1

p 1 p R

R 10

p s p 2

RQ 2

p 2 p R

R 20

R 1R

1 Cv i

R 2R

1 Cv i