5: Vibration Isolation (Passive) : Peter Saulson Syracuse University

5: Vibration isolation (passive)
Peter Saulson
Syracuse University
25 December 2013
ICTS Winter School on Experimental

Gravitational-wave Physics
My lectures during this School

1.
1
2.
3
3.
4.
5
5.
6.
7
7.
8.
9.
10.
Overview of gravitational waves and sources

Interactions of waves and detectors
Sh t noise
Shot
i and
d radiation
di ti pressure noise
i
Theory of linear systems
Vib ti isolation
Vibration
i l ti (passive)
(
i )
Thermal noise
O i off F
Optics
Fabry-Perot
b P
cavities
ii
Feedback control systems
Description off LIGO and
d other
h current detectors
d
Future detectors in space
25 December 2013

Outline
1.
1
2.
3.
4.
Displacement
Di
l
t noise
i goall
Seismic noise
Seismic isolators in theory and practice
Limits of performance
25 December 2013

Two classes of noise

I the
In
th nextt two
t
mornings,
i
we will
ill ttalk
lk about
b t
two kinds of noise that move an
i t f
interferometers
t mirrors:
i
seismic noise
thermal
h
l noise
This is in distinction to readout noise like shot

noise, which only affects our ability to see
where the mirrors are.
Radiation pressure noise, shot noises quantummechanical twin, was our first example of
di l
displacement
t noise.
i
25 December 2013

Large mechanical noise

How large?
H
l
?
Seismic: xrms ~ 1 m.
Thermal
mirrors CM: ~ 3 x 10-12 m.
mirrors surface: ~ 3 x 10-16 m.
25 December 2013

LIGOss sensitivity goal

LIGO
Earlier, I loosely gave initial LIGOs
Earlier
LIGO s sensitivity
as h ~ 10-21. What did I mean?
We want the standard deviation of strain measurements
averaged over the 10 msec duration of, say, a signal
from a supernova
p
or black hole ringdown
g
to be 10-21.
Lets convert this spec to power spectrum language
150 Hz
21 2
h ( f )df 10
.
This means we want a noise amplitude spectral density
near 100 Hz of
h( f 100 Hz) 10 22 / Hz .
50 Hz
25 December 2013

Displacement noise goal

Whatt spectrum
Wh
t
off di
displacement
l
t noise
i iis
consistent with this goal?
Well have four key mirrors (two in each arm to
make the Fabry-Perot cavities, see later
lecture.)
p
noises are incoherent, they
y
If their displacement
add in quadrature. The net result is
2
hdisp ( f ) x( f )).
L
Thus, we need x(f) = 2 10-19 m/Hz1/2.

25 December 2013

How strong is seismic noise?

Amplitude spectrum of seismic noise above 10 Hz is
typically
10-9 m/Hz1/2*(10
(10 Hz/f)2, quite diverse below 10 Hz.
Note basic pattern: Strong at low frequencies, weak at high
frequencies.
At our target frequency of 100 Hz,

x(f) = 10-11 m/Hz1/2,
19 m/Hz
f ffrom x(f)
far
(f) = 2 10-19
/H 1/2.
Seismic noise is serious!

W need
We
d 8 orders
d
off magnitude
it d off isolation
i l ti att 100 H
Hz.
25 December 2013

Seismic noise spectrum
25 December 2013

Our test masses need to be free

ffree to
t move iin response tto gravitational
it ti
l
wave
free from large disturbances from the
environment
25 December 2013

10
Strategy to deal with

seismic noise
Seismic noise is so large
large, we need multiple
strategies for dealing with it.
1st, as much isolation as we can build.
build
This will partly solve the problem, but will leave
strong noise at low frequency.
2nd, build an instrument that can be sensitive at

high frequencies (where we can make seismic
noise small), while tolerating the remaining
strong
g noise at low frequencies.
q
This means ensuring linearity, and making
thoughtful use of servo control.
25 December 2013

11
SHO as filter
A mass on a spring
i makes
k a good
d iisolator.
l t
Frequency response goes like 1/f2 above resonance.
25 December 2013

12
Pendulum as isolator
One such
O
h SHO iis b
built
ilt iinto
t our plans
l
already
l d
each test mass must be suspended as a
pendulum,
d l
to
t allow
ll
it tto respond
d ffreely
l tto th
the
gravitational wave.
IIt h
has a resonant ffrequency near 1 Hz.
H
Thus, we should multiply the input spectrum
b (1 H
by
Hz/f)
/f)2 to
t fi
find
d output
t t spectrum
t
(i
(i.e., motion
ti
of mirror.)
25 December 2013

13
Core Optics
i t ll ti and
installation
d alignment
li
t
25 December 2013

14
Optic Suspension
10 kg Fused Silica
25 cm diameter
10 cm thick
Magnet / coils control mirrors

Suspending gives 1/f2 isolation above ~1 Hz.
Low loss music wire minimizes thermal noise
25 December 2013

magnet
15
Pendulum helps, but only makes seismic

noise good enough at f > 1 kHz
25 December 2013

16
Multiple stages of isolation

for better filtering
One SHO makes
O
k a good
d filt
filter, b
butt nott good
d
enough.
If we make a chain of N oscillators, we have a
coupled system with N resonances, above
which the frequency response is
2N
f0
G ( f ) .
f
Just need to build enough stages of isolation.

isolation
25 December 2013

17
Add 3 stages of 3 Hz isolation,

LIGO can work down to 30 Hz
25 December 2013

18
Need for multi-dof isolation

We need good vertical (and transverse horizontal)
isolation as well.
Why?
If not, some of the large transverse motion would be
converted into optic-axis motion by asymmetries in the
pendulum.
Interferometer has some sensitivity to vertical motion,
because arms cant
can t be level everywhere
everywhere.
The earth is curved!
So, we need q
quasi-isotropic
p isolation. Such isolators are
called mass-spring stacks. They were invented by
Joseph Weber.
25 December 2013

19
25 December 2013

20
Real designs
25 December 2013

21
25 December 2013

22
Internal modes as limits to

performance
Note
N
t that
th t LIGO stacks
t k ffrequency response
levels off above ~ 30 Hz, at a level of
about
b t 10-66 (about
( b t 120
120 dB.)
dB )
Internal degrees of freedom in springs and
masses have resonant frequencies there.
E.o.m. isnt well approximated by the pointmass model in that regime.
25 December 2013

23
Virgos Free falling test masses
25 December 2013

24
GEO600 Reaction Pendulum
25 December 2013

25
aLIGO suspension
25 December 2013

26
aLIGO has more isolation

(see Ranas talk on active SEI)
Stage 2
S
Actuation Plane
Optical Table
Suspension
Point
25 December 2013
+
L
+
+ Z
+ X+
182
V Y
cm
+ (X
,Y
)
T [mm]
(+200 +426
(+200,
+426.1)
1)
ITM
((+200
200 , -369.8)
369.8)
ETM

27
Newtonian gravitational noise

Fluctuating density of the ground near the test masses
causes gravitational force noise.
It is like coupling the test masses to the ground through
a spring of resonant frequency
f grav
1
2
G earth .
This is a low frequency (below 10-4 Hz), but is only like

g
one isolation stage.
Net effect is un-isolatable noise that is stronger than
seismic noise below about 3 Hz.
(Perhaps can measure seismic field, model this noise,
and subtract )
25 December 2013

28
The case for space

The best seismic isolators are still low-pass
low pass filters,
filters
admitting all seismic noise below a few Hz.
Even if you work hard on direct isolation
isolation, there will
always be the Newtonian noise.
N.B. There are some ideas for measuring
g the seismic noise at
many places around each mass, computing the Newtonian
noise, and subtracting it.
Why not get away from it all, in space?

A good
d idea.
id A little
littl more about
b t proposed
d space
missions on Friday afternoon.
25 December 2013

29

5: Vibration Isolation (Passive) : Peter Saulson Syracuse University

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

5: Vibration Isolation (Passive) : Peter Saulson Syracuse University

Uploaded by

Copyright:

Available Formats

5: Vibration isolation (passive)

ICTS Winter School on Experimental

My lectures during this School

Overview of gravitational waves and sources

ICTS Winter School on Experimental

ICTS Winter School on Experimental

Two classes of noise

This is in distinction to readout noise like shot

ICTS Winter School on Experimental

Large mechanical noise

ICTS Winter School on Experimental

LIGOss sensitivity goal

ICTS Winter School on Experimental

Displacement noise goal

Thus, we need x(f) = 2 10-19 m/Hz1/2.

ICTS Winter School on Experimental

How strong is seismic noise?

At our target frequency of 100 Hz,

Seismic noise is serious!

ICTS Winter School on Experimental

Seismic noise spectrum

ICTS Winter School on Experimental

Our test masses need to be free

ICTS Winter School on Experimental

Strategy to deal with

2nd, build an instrument that can be sensitive at

ICTS Winter School on Experimental

ICTS Winter School on Experimental

ICTS Winter School on Experimental

ICTS Winter School on Experimental

Magnet / coils control mirrors

ICTS Winter School on Experimental

Pendulum helps, but only makes seismic

ICTS Winter School on Experimental

Multiple stages of isolation

Just need to build enough stages of isolation.

ICTS Winter School on Experimental

Add 3 stages of 3 Hz isolation,

ICTS Winter School on Experimental

Need for multi-dof isolation

ICTS Winter School on Experimental

ICTS Winter School on Experimental

ICTS Winter School on Experimental

ICTS Winter School on Experimental

Internal modes as limits to

ICTS Winter School on Experimental

Virgos Free falling test masses

ICTS Winter School on Experimental

GEO600 Reaction Pendulum

ICTS Winter School on Experimental

ICTS Winter School on Experimental

aLIGO has more isolation

ICTS Winter School on Experimental

Newtonian gravitational noise

This is a low frequency (below 10-4 Hz), but is only like

ICTS Winter School on Experimental

The case for space

Why not get away from it all, in space?

ICTS Winter School on Experimental

You might also like