This action might not be possible to undo. Are you sure you want to continue?

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

**Optical and Mechanical Behavior of the Optical Fiber Infrasound
**

Sensor

A thesis submitted in partial satisfaction of the

requirements for the degree

Master of Science

in

Earth Sciences

by

Scott DeWolf

Committee in charge:

Mark A. Zumberge, Chair

Duncan C. Agnew

Michael J. Buckingham

2009

Copyright

Scott DeWolf, 2009

All rights reserved.

The thesis of Scott DeWolf is approved, and it is accept-

able in quality and form for publication on microﬁlm and

electronically:

Chair

University of California, San Diego

2009

iii

DEDICATION

To Sarah and Delilah.

iv

EPIGRAPH

Only the mediocre are supremely conﬁdent of their ability.

— Sir Michael Atiyah

v

TABLE OF CONTENTS

Signature Page . . . . . . . . . . . . . . . . . . . . . . . . . . iii

Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

Epigraph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . vi

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . viii

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . x

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . xi

Abstract of the Thesis . . . . . . . . . . . . . . . . . . . . . . xii

Chapter 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Infrasound and Infrasound Instrumentation . . . . . . . . 1

1.2 The Optical Fiber Infrasound Sensor . . . . . . . . . . . 2

Chapter 2 Temperature-Induced Polarization Drift . . . . . . . . . . . . . 7

2.1 Observations of Polarization Drift . . . . . . . . . . . . . 8

2.2 The Polarization-Maintaining OFIS . . . . . . . . . . . . 9

2.2.1 Design, Construction and Deployment . . . . . . 9

2.2.2 Results: Ellipse Stability and Optical Phase Noise 10

2.3 Faraday Mirrors and the Michelson OFIS . . . . . . . . . 11

Chapter 3 Temperature-Induced Pressure Sensitivity Change . . . . . . . 20

3.1 Laboratory Observations . . . . . . . . . . . . . . . . . . 20

3.2 Temperature Dependence of the Elastic Modulus . . . . . 22

Chapter 4 Analytic Opto-Mechanical Model . . . . . . . . . . . . . . . . 28

4.1 Lam´e Equations for a Thick-Walled Cylinder . . . . . . . 28

4.2 Pre-Stressed, Double-Walled Thick Cylinder . . . . . . . 31

4.3 The Strain-Optic Eﬀect . . . . . . . . . . . . . . . . . . . 32

4.4 Single Fiber about a Thick Cylinder Subjected to Internal

Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.4.1 Theoretical Results . . . . . . . . . . . . . . . . . 34

4.4.2 Experimental Results . . . . . . . . . . . . . . . . 36

4.4.3 Suggestions for Future Work . . . . . . . . . . . . 36

Appendix A Optical Path Length Modulation and the Femtometer System 39

vi

Appendix B Derivation of Ellipse Parameters . . . . . . . . . . . . . . . . . 43

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

vii

LIST OF FIGURES

Figure 1.1: Typical conﬁguration for a high frequency 18 meter, 92-port

and a low frequency 70 meter, 144-port pipe array (shown to

scale). Each connects to a central microbarometer. . . . . . . . 5

Figure 1.2: The Mach-Zehnder OFIS schematic, along with an exposed

length shown next to a pencil for scale. Note that the OFIS is

wrapped in ﬁberglass insulation and inserted into a 0.1 meter

diameter corrugated drainage tube for both ground-lying and

burial deployments. . . . . . . . . . . . . . . . . . . . . . . . . 6

Figure 2.1: Two views of the same Poincar´e sphere, one rotated π in az-

imuth with respect to the other. The evolution of the SOP in

the single wrap arm is shown by circles, and the double wrap

arm by dark squares. . . . . . . . . . . . . . . . . . . . . . . . . 15

Figure 2.2: Three types of PM ﬁbers: PANDA, Bow-tie and elliptical-clad.

The center gray circle denotes the core of the ﬁber, whereas the

black shapes represent the stressing materials (not to scale).

The fast axis is along the horizontal dimension in this ﬁgure. . . 16

Figure 2.3: Four Camp Elliott OFISs, from left to right: Bow-tie, PANDA,

SMF-28, and a standard OFIS (not used for the present study,

however, was used for the experiment in section 2.1). Each

sensor was connected to an instrument box by a cylindrical PVC

adapter arm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Figure 2.4: Ellipse area probability density estimates for the PANDA, Bow-

tie, and SMF-28 OFISs, shown with a FM equipped Michelson

OFIS. Note that the SMF-28 OFIS consisted of a single wrap,

however, a standard OFIS ellipse behaves similarly with a non-

trivial probability of having an area at or near zero. . . . . . . . 17

Figure 2.5: The top time series shows the raw optical phase overlain with

the scaled null interferometer signal, while the lower shows their

diﬀerence along with a calibrated B&K 4193 precision infra-

sound microphone. Both signal amplitudes have been converted

to Pascals, but the PM OFIS signal appears slightly lower most

likely due to its directional response [34]. . . . . . . . . . . . . . 18

Figure 2.6: The Michelson OFIS schematic, showing that after traveling

the length of the sensor, the phase conjugate beam from both

arms are reﬂected back to the original splitter. It is at this

point that the interference fringes are formed and sent to the

photodetector. . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

viii

Figure 3.1: The 1.5 meter long OFIS heating and cooling chamber con-

nected to the thermoelectric temperature control system. This

unit was subsequently enclosed by insulation and mounted hor-

izontally for the calibration experiments and vertically for the

elastic modulus and linear expansion experiments. . . . . . . . 25

Figure 3.2: Schematic of the experimental setup for measuring the tem-

perature dependence of the pressure sensitivity. The two heat

exchangers (plumbed in series) submerged in water provided

suﬃcient thermal inertia to slow the cycle time to three hours.

Temperature data was taken from the thermistor located near

the center of the OFIS interior. . . . . . . . . . . . . . . . . . . 26

Figure 3.3: Results for one 24 hour period, or three-cycles of temperature

and sensitivity measurements. Each error bar corresponds to

the standard error within each 1

◦

C temperature bin, however,

the linear ﬁt was performed on the unbinned data. . . . . . . . 27

Figure 4.1: The left diagram shows a thick-walled cylinder with inner ra-

dius a and outer radius b, subjected to internal pressure p

a

and

external pressure p

b

. The right side shows the forces on the area

segment from the left diagram, both highlighted in gray. . . . . 38

Figure A.1: Schematic of the modulator box, showing the OFIS connections

on the left. While there are four connections to the OFIS, two

send light into the OFIS and receives the reﬂected light from

the FM found at the remaining two connections. Note that the

single and double wrap arms can be interchanged. . . . . . . . . 42

Figure A.2: Block diagram of the fringe modulation and Femtometer system

for estimating the optical phase diﬀerence between two OFIS

arms. The Femtometer output is the ﬁve ellipse parameters

(usually at a much lower sample rate) along with the (x, y)

inputs used to compute p. . . . . . . . . . . . . . . . . . . . . . 42

Figure B.1: Geometry of ellipse rotation, where η and ξ correspond to its

principle axes. . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

ix

LIST OF TABLES

Table 4.1: Material Parameters for 900 micron SMF-28 Fiber . . . . . . . . 35

x

ACKNOWLEDGEMENTS

I would like to thank Dr. Mark A. Zumberge for advising my research and

graduate academic career thus far. His examination of my experimental work in

the lab and in the ﬁeld has provided me many valuable insights on when and where

time is best spent, which will undoubtedly shape how I approach problems in the

future. I truly appreciate the ﬂexibility to reﬁne and re-run critical experiments

and to pursue the opto-mechanical model.

Thanks to Dr. Kristoﬀer T. Walker for countless insightful discussions and

comments throughout the course of this work, not to mention paying for many

of my lunches. Without his guidance on tasks ranging from the trivial to the

impossible, along with the authorship of essential scripts for handling CSS data, I

would have been substantially less productive.

I have come to understand that successful experimentation is, amongst

other less ﬂattering things, a combination of patience and experience. This has

been ampliﬁed in my case thanks to Dr. Glenn Sasagawa, who has always taken

the time to listen and provide useful suggestions for most of my laboratory work

to date.

Finally, thanks to Joel White for help with the winding and deployment

of the polarization maintaining OFISs, and for countless other small tasks that

accompany ﬁeld and laboratory work.

xi

ABSTRACT OF THE THESIS

Optical and Mechanical Behavior of the Optical Fiber Infrasound

Sensor

by

Scott DeWolf

Master of Science in Earth Sciences

University of California San Diego, 2009

Mark A. Zumberge, Chair

The Optical Fiber Infrasound Sensor (OFIS) is an interferometric pressure

transducer consisting of a pair of optical ﬁbers helically wrapped about a compliant

tube. While the OFIS has been successfully deployed for nearly a decade, its

performance has been plagued by downtime due to polarization fading (resulting

from no interference fringe intensity) and a nearly continuous change in its pressure

sensitivity. This thesis explores the use of very expensive polarization maintaining

ﬁber and inexpensive Faraday mirrors, both of which solve the polarization fading

issue. Laboratory measurements of the thermal behavior of the pressure sensitivity

are found to conﬁrm ﬁeld tests, but measurements of the temperature dependence

of the tube’s elastic modulus does not appear to adequately describe the observed

behavior. Therefore, a crude analytic mechanical and optical model is explored to

help investigate potential causes, however, the simple theory of compound thick

cylinders was found to be problematic as a realistic model for the OFIS.

xii

Chapter 1

Introduction

Sound can be deﬁned as pressure ﬂuctuations about some background level,

as in the case of human speech at atmospheric pressure. The frequency content of

sound can be broken into three categories roughly deﬁned by the range of human

hearing. This audible range is often taken to be 20-20,000 Hz, and sounds above

and below this threshold are termed ultrasound and infrasound, respectively. In

this thesis, the term infrasound is restricted to pressure ﬂuctuations below 20 Hz

with atmospheric pressure near the surface as the background level.

1.1 Infrasound and Infrasound Instrumentation

There are many examples of natural and anthropogenic sources of infra-

sound. While too numerous to describe individually, some naturally occurring

sources include earthquakes (seismic-to-acoustic coupling), auroras, meteors, tsu-

namis, volcanoes (Reference [4] provides a review of the preceding phenomena),

mountain-associated waves [7], atmospheric solitons [8], and convective storms

[16]. The most ubiquitous and commonly studied source in coastal areas are

microbaroms, which arise from the forcing of the lower atmosphere by nonlin-

ear wave-wave interactions at the ocean surface [28]. Investigations of man-made

sources range from shock waves from supersonic aircraft [12], wind turbines [22],

and rocket launches [11], to mining blasts [1] and nuclear explosions [9, 10]. It

is this last case that prompted the International Monitoring System (IMS) to in-

1

2

clude a global network of infrasound arrays along with seismic, hydroacoustic and

radionuclide sensors for monitoring the Comprehensive Nuclear Test Ban Treaty.

The most signiﬁcant limitation to measuring infrasound are pressure ﬂuc-

tuations due to turbulence, or “wind noise,” and while there exist several high

quality infrasound microphones, or microbarometers, instrument development is

largely focused on noise reduction technology. A comprehensive review of wind

noise physics, abatement schemes and their limitations can be found in Reference

[29], however, there appear to be two strategies that can be used individually or

in combination. The ﬁrst exploits the fact that these turbulent structures are

spatially incoherent by mechanically averaging the pressure over an area. This is

accomplished by connecting a microbarometer to an elaborate array of pipes whose

inlets are distributed over a wide area (Figure 1.1), and is the current method used

with IMS certiﬁed infrasound stations. There are an increasing number of studies

centered around the concept of screening the turbulence by enclosing a micro-

barometer with or without a pipe array in a porous fabric tent. The principle

drawbacks to either method are that they are expensive, labor intensive to build

and maintain, and have a large ecological footprint, exacerbated by the fact that

23 of 60 IMS infrasound stations are on, or proposed for, remote oceanic islands

[19].

1.2 The Optical Fiber Infrasound Sensor

Optical ﬁbers are well known for their ability to measure strain [6], tempera-

ture and pressure [20], which motivated the development of a fundamentally diﬀer-

ent microbarometer that came to be known as the Optical Fiber Infrasound Sensor

(OFIS) [34]. The OFIS consists of a compliant rubber tube helically wrapped with

a pair of optical ﬁbers, which form an equal-arm Mach-Zehnder interferometer.

Laser light, typically at 1,310 nanometer wavelength, enters one end of both the

double and single wrapping only to loop back where the other ends are combined

to form interference fringes (Figure 1.2). Since the ﬁbers are less elastic than the

tube, they act as local reinforcement such that the tube near the double wrapped

3

ﬁbers experiences less deformation than the single wrap for a given amount of

strain in the tube. Therefore, the OFIS measures diﬀerential strain in the tube as

a result of changes in the pressure ﬁeld. Finally, both path lengths are modulated

by a piezoelectric cylinder and the resulting interference fringe intensity is mon-

itored by a photodetector. The necessity of modulating the paths allows for the

discrimination between contractive and expansive fringes (negative and positive

pressures, respectively), which is described in Appendix A.

There are several unique characteristics of the OFIS that can be exploited

for measuring infrasound. Since the light experiences little attenuation in an op-

tical ﬁber, the OFIS can be made almost arbitrarily long. When deployed, the

OFIS is wrapped in ﬁberglass insulation and inserted into a perforated tube to

allow burial beneath a porous medium. Therefore the OFIS employs both wind

noise reduction technologies at once, but has the further advantage over mechanical

ﬁlters in that it averages the turbulence at the speed of light rather than the speed

of sound. The result is a lower limit on infrasonic noise in the 1 to 10 Hz band,

as evidenced by Reference [34], with a much smaller footprint than a conventional

pipe array. Given the ﬂexibility of the OFIS, it can be deployed in diﬀerent ge-

ometries, each with diﬀerent amplitude and frequency responses that depend upon

the direction of arrival. In the case of an array of linear OFISs, this has lead to

the development of novel beamforming techniques to estimate the elevation, back

azimuth, and phase velocity of infrasound signals [31].

Unfortunately, temperature has a severe impact on the performance of the

OFIS, and it is the purpose of this thesis to quantify what is happening and de-

scribe attempts to eliminate or mitigate these eﬀects. Chapter 2 is concerned with

the intensity of the interference fringes, which depends upon the polarization of

the laser light as it emerges from each arm. As the temperature of the OFIS

changes, the state of polarization in each of the two ﬁbers changes with respect to

the other. When the polarization states are orthogonal, the interference intensity

is zero causing the sensor to go oﬄine, and both the use of polarization maintaining

(PM) ﬁber and Faraday mirrors (FM) are investigated. Finally, chapter 3 examines

how temperature changes the sensitivity of the OFIS, which is quantiﬁed by the

4

optical phase diﬀerence per unit pressure. While temperature modestly changes

the elasticity of the tube, a crude analytic model is explored to understand possible

geometric eﬀects caused by thermal expansion.

5

Figure 1.1: Typical conﬁguration for a high frequency 18 meter, 92-port and a low frequency 70 meter, 144-port pipe array

(shown to scale). Each connects to a central microbarometer.

6

Figure 1.2: The Mach-Zehnder OFIS schematic, along with an exposed length shown next to a pencil for scale. Note that

the OFIS is wrapped in ﬁberglass insulation and inserted into a 0.1 meter diameter corrugated drainage tube for both

ground-lying and burial deployments.

Chapter 2

Temperature-Induced

Polarization Drift

The ﬁrst limitation of the OFIS concerns the size of the ellipse formed

by plotting the interference fringe signal versus its derivative. A least squares

ﬁt to this ellipse is performed by a digital signal processing system called the

Femtometer (Appendix A) to determine the optical phase diﬀerence in radians.

When the light from each arm of the OFIS is combined, the resulting interference

fringe intensity depends upon the relative phase of the two beams. However, the

intensity contrast between the constructive and destructive interference fringes

depends upon the relative state of polarization (SOP) of each arm with respect

to the other. For example, if both arms emit linearly polarized light where their

electric ﬁelds are both vertical, the resulting intensity will depend only on their

relative phase. However, if the electric ﬁeld of one arm is horizontal while the other

remains vertical then no interference can take place. Therefore, the ﬁtted ellipse

changes size with the relative SOP, which can lead to inaccurate phase estimates

when the SOP in the two arms are near orthogonal and the ellipse becomes very

small, taking the sensor oﬄine.

7

8

2.1 Observations of Polarization Drift

A common way to express the SOP of monochromatic light is in terms

of the Stokes parameters, which can be represented graphically as a point on a

sphere whose radius is determined by the intensity, known as the Poincar´e sphere.

Following the convention of Born and Wolf [5], the x and y components of an

electromagnetic wave propagating in the z direction are:

E

x

= a

1

cos (τ + δ

1

) (2.1)

E

y

= a

2

cos (τ + δ

2

) , (2.2)

where a

1,2

and δ

1,2

are the amplitudes and phases, and τ =

k · x − ωt, the wave

vector, spatial location, angular frequency, and time, respectively. The Stokes

parameters are deﬁned as:

s

0

= a

2

1

+ a

2

2

(2.3)

s

1

= a

2

1

−a

2

2

(2.4)

s

2

= 2a

1

a

2

cos (δ) (2.5)

s

3

= 2a

1

a

2

sin (δ) , (2.6)

where δ = δ

2

−δ

1

. It is straight forward to show that:

s

2

0

= s

2

1

+ s

2

2

+ s

2

3

. (2.7)

By eliminating τ from (2.1,2) and reducing the resulting quadratic to canonical

form, it can be shown that the SOP itself can be represented by an ellipse. Finally,

a stereographic projection of the SOP ellipse onto a sphere [23] yields:

s

1

= s

0

cos (2χ) cos (2ψ) (2.8)

s

2

= s

0

cos (2χ) sin (2ψ) (2.9)

s

3

= s

0

sin (2χ) . (2.10)

The results are the equations of a sphere of radius s

0

, with longitude 2ψ and

latitude 2χ. For example, (2ψ, 2χ) states of linearly polarized light fall along the

equator starting with horizontal at (0, 0) and vertical at (π, 0), whereas right and

9

left circular polarizations occur at the north and south poles, respectively. All the

states in between are generically referred to as elliptically polarized, with their

handedness determined by the upper and lower hemispheres.

Measuring the SOP of each OFIS arm was the ﬁrst step to determine what

could be done to solve the polarization drift problem. This was done with a

polarimeter, which typically operates by splitting the input beam into multiple

paths and passing each through a diﬀerent series of polarizers and retarders whose

output intensities are proportional to the Stokes parameters [17]. In this case, a

General Photonics POD-101A in-line polarimeter was used, whose accompanying

software was able to collect all four Stokes parameters at a minimum of 1,000

samples per second for a maximum of four hours. Since the interference ellipse

size changed most signiﬁcantly with temperature, and the steepest temperature

gradients occur at sunrise and sunset, SOP measurements were taken on each

arm of a 10 meter ground-lying OFIS from 4:30 to 8:30 PM local time. The

results (Figure 2.1) show that the SOP in the single wrap arm varies considerably,

especially in comparison to the double wrap arm. Since the SOP perturbations

are most likely caused by stress-induced birefringence imparted during thermal

expansion and contraction, and given that the OFIS functions based on the strain

diﬀerential between the single and double wrap ﬁbers, it follows that the SOP

variation would be greater for the single wrap.

2.2 The Polarization-Maintaining OFIS

2.2.1 Design, Construction and Deployment

Polarization can be maintained in an optical ﬁber by inducing a birefrin-

gence proﬁle in the plane perpendicular to the axis of the ﬁber, thus creating a

preferential orientation for the electric ﬁeld to propagate. In PM ﬁber (not to be

confused with polarizing ﬁber) this is accomplished by including a material with

a diﬀerent coeﬃcient of thermal expansion, such that when the glass is heated and

drawn into a ﬁber, thermal stress-induced birefringence occurs upon cooling. There

are three main types of PM ﬁber (Figure 2.2), each with diﬀering characteristics

10

such as birefringence contrast and uniformity over the length of the ﬁber. Only

Corning PANDA and Fibercore HB1500 Bow-tie types were used for the prototype

PM OFISs since PANDA is the most widely available and Bow-tie had the highest

birefringence contrast making it less likely to experience loss of PM characteristics

under helical winding.

Given that PM ﬁber was over one hundred times more expensive than the

typical Corning SMF-28 ﬁber, it was decided to deploy three unequal arm Mach-

Zehnder OFISs. Each OFIS consisted of two 30 meter, 900 micron jacketed ﬁbers

whose helical wraps were 5.5 cm apart, resulting in total lengths of approximately

18 meters. The availability of a NP Photonics “Rock” 1,533 nanometer wave-

length laser source with a coherence length of several hundred kilometers (spectral

linewidth of less than 1 kHz) allowed the second arm to be only 3 meters, which

was deployed in an attached instrument box. Finally, the sensors were wrapped

with ﬁberglass insulation and inserted into 0.1 meter diameter plastic perforated

tube.

The three sensors were deployed on the ground at the Camp Elliott Field

Station (Figure 2.3) located about 20 kilometers east of the SIO campus. Since

there was only enough hardware to run one sensor at a time, each was run indepen-

dently for several weeks. Temperature data was collected between the OFIS tube

and insulation and on the outside of the perforated tube with two Onset HOBO

sensors, which consist of a thermocouple and data logger. An OFIS calibration

system (contained in the gray box in the lower left-hand corner of Figure 2.3),

consisting of a commercial loudspeaker hermetically sealed to a plastic plate and

a calibrated Setra 265 pressure transducer, was connected to the internal volume

of each OFIS tube running a 0.1 Hz continuous wave at approximately 60 Pascals

(peak-to-peak).

2.2.2 Results: Ellipse Stability and Optical Phase Noise

The PM ﬁber provided very good ellipse stability, resulting in no downtime

due to polarization fading. This is quantiﬁed by the size of the interference el-

lipse, which can be calculated from the parameters determined by the Femtometer

11

system (Appendix B). Histograms of two days of data show that both PM OFIS

ellipse areas show some variability, but are not in danger of approaching zero. This

is in contrast to the SMF-28 OFIS whose ellipse area variance is over quadruple

that of the PM OFISs, and also has a non-zero probability of collapse. Hence, the

use of PM ﬁber does solve the polarization fading issue.

Shortly after the initial setup it became apparent that ﬂuctuations in the

laser wavelength caused by the unequal arms were contributing a signiﬁcant amount

of noise to the optical phase data. This was partially remedied by subtracting the

signal from a second unequal arm “null” interferometer consisting of a 30 meter

and 1 meter SMF-28 ﬁber housed in the adjacent building. Since the lengths of the

null interferometer arms were not identical to the OFISs, the null signal needed

to be scaled by a constant trivially determined by least squares. Figure 2.5 is an

example of this diﬀerencing on the Bow-tie OFIS, and shows a residual noise of

about 5 to 6 Hz not found in a Br¨ uel and Kjær (B&K) 4193 reference microphone

located 10 meters west of the OFISs. Investigating the causes of this residual noise

and/or constructing an identical null interferometer, in addition to other problems

such as the large amounts of scatter in the ﬁtted ellipse and oscillations in the el-

lipse area caused by the calibration signal were not investigated due to the results

of the following section.

2.3 Faraday Mirrors and the Michelson OFIS

Compensating for unwanted birefringence in the path of an interferometer

by creating a phase conjugate (or time reversed) beam was ﬁrst investigated by

Martinelli in 1989 [27], and successfully implemented shortly thereafter in a ﬁber

optic Michelson interferometer by Kersey [24]. This was done by using a Faraday

rotator followed by a mirror, which has come to be known as a Faraday rotator

mirror, or Faraday mirror (FM). To understand why this particular combination

works, one must be familiar with the Jones calculus for modeling optical compo-

nents.

The Jones calculus is a common way to mathematically express how fully

12

polarized light behaves as it passes through optical components such as polarizers,

rotators, and retarders or wave plates. Similar to Section 2.1, an electromagnetic

wave can be expressed as a complex vector [18]:

E =

E

x

E

y

=

a

1

e

i(τ+δ

1

)

a

2

e

i(τ+δ

2

)

(2.11)

For example, horizontally polarized light implies a

2

= 0, and normalization reduces

the Jones vector to:

E

h

=

1

0

**whereas right circularly polarized light implies that the component amplitudes are
**

equal (a

1

= a

2

), and the phase of E

y

is shifted by −π/2 with respect to E

x

(δ

1

= 0

and δ

2

= −π/2):

E

RCP

=

1

√

2

1

−i

.

Using this formalism, optical components are elementary matrices (A) that operate

on the Jones vectors (note that it is also common to use 4 × 4 Mueller matrices

to operate on a vector containing the Stokes parameters in the case of partially or

unpolarized light, but this added generality is not necessary in this context). The

incident beam is transformed into the output beam (denoted by a prime) :

E

′

= A

E, (2.12)

and can logically be extended to a system of n components where A = A

n

A

n−1

...A

2

A

1

.

An intuitive example of such a matrix is that of a horizontal linear polarizer:

A

HLP

=

1 0

0 0

,

13

which applied to the previous Jones vector:

E

′

= A

HLP

E

RCP

=

1

√

2

1 0

0 0

1

−i

=

1

0

=

E

h

,

where renormalization has eliminated the factor of 1/

√

2. Hecht [18] provides many

more examples of Jones matrices for common optical components.

To show how the Faraday mirror operates, consider modeling the optical

ﬁber by the Jones matrix for a general elliptical retarder (following the convention

of Kersey [24], but disregarding losses):

⇀

R =

1

d

a −b

∗

b a

∗

(2.13)

where a and b denote the complex birefringence of the ﬁber, d = det

⇀

R

, and the

harpoon indicates the forward propagating direction, while a beam propagating

through the same retarder in the reverse direction:

↼

R =

1

d

a −b

b

∗

a

∗

. (2.14)

The Faraday rotator imparts a π/4 rotation before and after the reﬂection, and

the total eﬀect of the rotator (A

ROT

(θ)) and mirror (A

M

) system is:

A

FM

= A

M

A

ROT

π

4

A

ROT

π

4

=

−1 0

0 1

cos

π

4

sin

π

4

−sin

π

4

cos

π

4

cos

π

4

sin

π

4

−sin

π

4

cos

π

4

=

0 −1

−1 0

(2.15)

14

Therefore, the Jones matrix for the ﬁber and FM is:

A =

↼

RA

FM

⇀

R =

0 −1

−1 0

, (2.16)

which is independent of the birefringence of the ﬁber. Qualitatively this implies

that the perturbations in the SOP by the ﬁber are “undone” by the reﬂected beam,

provided that such perturbations do not occur faster than the time of ﬂight of the

beam within the system.

To apply FM to the OFIS required the use of a Michelson rather than a

Mach-Zehnder interferometer (Figure 2.6). This was a trivial and very inexpensive

modiﬁcation, which simultaneously doubled the pressure sensitivity since the laser

is reﬂected back through the OFIS ﬁbers. Back reﬂections of the fringe signal to

the laser were eliminated by the a built-in isolator, or optical diode. However, in

the case where one laser is split into multiple paths to illuminate multiple OFISs,

there was concern that the backward reﬂecting fringe signal would also propagate

in the neighboring sensors thereby optically coupling each sensor. Fortunately the

use of ﬁber connectors with angled faces, or FC/APC connectors, are designed to

reduce such reﬂections, which resulted in no detectable fringe signal on adjacent

sensors. Finally, the FM equipped Michelson OFISs have worked very well (Figure

2.4), and are being used for all current and future sensors.

1

5

Figure 2.1: Two views of the same Poincar´e sphere, one rotated π in azimuth with respect to the other. The evolution of

the SOP in the single wrap arm is shown by circles, and the double wrap arm by dark squares.

16

Figure 2.2: Three types of PM ﬁbers: PANDA, Bow-tie and elliptical-clad. The

center gray circle denotes the core of the ﬁber, whereas the black shapes repre-

sent the stressing materials (not to scale). The fast axis is along the horizontal

dimension in this ﬁgure.

Figure 2.3: Four Camp Elliott OFISs, from left to right: Bow-tie, PANDA, SMF-

28, and a standard OFIS (not used for the present study, however, was used for

the experiment in section 2.1). Each sensor was connected to an instrument box

by a cylindrical PVC adapter arm.

17

Figure 2.4: Ellipse area probability density estimates for the PANDA, Bow-tie,

and SMF-28 OFISs, shown with a FM equipped Michelson OFIS. Note that the

SMF-28 OFIS consisted of a single wrap, however, a standard OFIS ellipse behaves

similarly with a non-trivial probability of having an area at or near zero.

18

Figure 2.5: The top time series shows the raw optical phase overlain with the

scaled null interferometer signal, while the lower shows their diﬀerence along with

a calibrated B&K 4193 precision infrasound microphone. Both signal amplitudes

have been converted to Pascals, but the PM OFIS signal appears slightly lower

most likely due to its directional response [34].

1

9

Figure 2.6: The Michelson OFIS schematic, showing that after traveling the length of the sensor, the phase conjugate

beam from both arms are reﬂected back to the original splitter. It is at this point that the interference fringes are formed

and sent to the photodetector.

Chapter 3

Temperature-Induced Pressure

Sensitivity Change

Pressure sensitivity of the OFIS is quantiﬁed by the optical phase change

in radians, per unit pressure, and normalized by the total length of the sensor

(rad· Pa

−1

· m

−1

). The implementation of a continuous calibration system revealed

that the pressure sensitivity of a 30 meter, ground lying OFIS changed by more

than a factor of two over a range of about 20 degrees Celsius [30]. Since the IMS

requires all microbarometer sensitivities vary less than ﬁve percent annually, it is

essential to understand and take steps to mitigate this behavior.

3.1 Laboratory Observations

To conﬁrm previous ﬁeld measurements, a 1.5 meter long heating and cool-

ing chamber was constructed from a 5 centimeter diameter copper pipe, helically

wrapped with 6 millimeter diameter copper tubing with a wrap spacing of 5 cen-

timeters that was periodically soldered to the larger pipe (Figure 3.1). This unit

was enclosed in 5 centimeters of foam insulation and end caps constructed to yield

a uniform temperature along its length. The temperature was controlled by pump-

ing a water and ethylene glycol mixture through the helical tubing using a Solid

State Cooling Systems 400 Watt ThermoCUBE thermoelectric temperature con-

trol system. This unit is capable of automatically cycling between two temperature

20

21

set points, but does not allow for the control of the time it takes to obtain a given

set point. Therefore, two Lytron CP15G05 aluminum heat exchangers submerged

in 20 liters of water were added to increase the thermal inertia of the system,

thereby reducing the cycle time. Finally, the temperature inside of the OFIS was

measured using a calibrated Fluke 80TK thermistor module, whose output voltage

was logged by the Femtometer system (Figure 3.2).

A calibration system was constructed to continuously determine the sensi-

tivity of the OFIS inside the heating and cooling chamber. This was a prototype

system for the unit deployed in the PM OFIS experiments, which also consisted

of a subwoofer glued to a back volume whose outlet was a standard barbed pipe

ﬁtting to accept ﬂexible tubing. The subwoofer was driven by a Tenma TG120

function generator and a Techron 5515 DC-coupled ampliﬁer, which ran a two sec-

ond sinusoid at a 2 Volt peak-to-peak amplitude. This was used to pressurize the

inner volume of the OFIS tube as well as a Setra 265 calibrated pressure trans-

ducer.

The pressure sensitivity of two 1.4 meter OFISs was measured continuously

over three, eight-hour heating and cooling cycles from approximately 5

◦

to 50

◦

Cel-

sius. After converting the Setra signal to Pascals, both the OFIS and Setra data

were narrowly bandpass ﬁltered about the calibration frequency and their enve-

lope functions obtained by computing the magnitude of the signal with its Hilbert

transform. Therefore, the instantaneous sensitivity was calculated by dividing the

Setra envelope by the OFIS envelope and the length of the sensor. To examine the

potential signiﬁcance of the environmental conditions under which the ﬁber was

wrapped about the silicone tube, one OFIS was wrapped at room temperature

(about 20

◦

C) while the other in a walk-in cooler at about 8

◦

C. While there were

no signiﬁcant diﬀerences in the nominal sensitivities or temperature dependences

between the warm and cold wrapped OFISs, both exhibited near linear behavior

(an example of which can be found in Figure 3.3). The average trend between all

cycles for both OFISs was found to be:

1

Φ(T)

dΦ

dT

= (10 ±1) ×10

−3 ◦

C

−1

,

22

relative to 20

◦

C. For comparison, consider the recent results from four, 30 meter

buried OFISs at the Camp Elliott Field station:

1

Φ(T)

dΦ

dT

= (30 ±7) ×10

−3 ◦

C

−1

.

These two results suggest that the temperature dependence of the pressure sen-

sitivity may also depend on the total length of the OFIS, however, more data is

required to establish a formal trend.

3.2 Temperature Dependence of the Elastic Mod-

ulus

Since Hooke’s law states that the stress in a linear elastic material is di-

rectly proportional to the strain and the material’s elastic (aka: Young’s) modulus,

it could be supposed that the observed changes in OFIS pressure sensitivity would

be inversely proportional to the temperature changes in the elastic modulus of the

silicone tubing. For example, at low temperatures one would expect the silicone

tube to become more rigid, corresponding to an increase in the elastic modulus,

requiring a greater amount of pressure to strain the tubing (as monitored by the

optical ﬁber) by the same amount, therefore lowering the sensitivity. It was this

reasoning that led to the measurement of the temperature dependence of the sili-

cone tubing’s elastic modulus.

The elastic modulus of the silicone tube can be easily measured as the ratio

of the tensile stress to tensile strain by hanging a length of the tubing from one end

and measuring the extension caused by a known mass suspended from the other

end. To show this, consider the following form of Hooke’s law:

σ = Eǫ,

where σ is the tensile stress, ǫ is the tensile strain, and E the elastic modulus.

Since the tensile stress is the force per unit of cross-sectional area (A) of the tube,

and (courtesy of Newton’s second law) the force is the applied mass (m) and the

23

acceleration due to local gravity (g):

mg

A

= Eǫ

m =

π (b

2

−a

2

) E

gL

∆L, (3.1)

where a and b are the inner and outer radii of the tubing, and the axial strain,

ǫ = ∆L/L, is simply the change in length over the total length. Therefore, the

slope of a linear least squares ﬁt to the applied mass versus change in length is

proportional to the elastic modulus.

The elastic modulus of the silicone tubing was measured for ten diﬀerent

temperatures from 5

◦

C to 50

◦

C using the same 1.5 meter long heating and cooling

chamber, but mounted vertically and without the added thermal inertia. Re-

peated heating and cooling measurements of the temperature inside and outside

the silicone tube revealed that the time needed for the temperature of the silicone

to equilibrate was approximately two hours: the e-folding time was found to be

τ = 1, 280 ± 220 seconds, therefore 5τ ≈ 6, 400 seconds, or about 1.8 hours. At

each temperature, a series of ten masses were used to stretch the silicone tube by

a distance measured with a traveling microscope. All measurements were repeated

four times: twice for increasing and decreasing temperature, respectively.

The results reveal a very modest change in the elastic modulus with tem-

perature. In order to account for the thermal changes in the length of the silicone

tube, the coeﬃcient of linear expansion was measured using the same setup, and

was found to be (relative to 20

◦

C):

1

L(T)

dL

dT

= (4.00 ±0.06) ×10

−4 ◦

C

−1

.

As shown earlier, the elastic modulus for each temperature was computed from

the slope of the applied mass versus extension. The temperature dependence of

the elastic modulus was very linear with a best ﬁt slope of:

1

E (T)

dE

dT

= (−2.06 ±0.08) ×10

−3 ◦

C

−1

,

relative to 3.48 ± 0.01MPa at 20

◦

C. This amounts to a 0.2% change per degree

Celsius compared to the 1.0% change shown in the laboratory measurements of

24

the previous section, and therefore further investigation is needed to understand

this discrepancy.

25

Figure 3.1: The 1.5 meter long OFIS heating and cooling chamber connected

to the thermoelectric temperature control system. This unit was subsequently

enclosed by insulation and mounted horizontally for the calibration experiments

and vertically for the elastic modulus and linear expansion experiments.

2

6

Figure 3.2: Schematic of the experimental setup for measuring the temperature dependence of the pressure sensitivity.

The two heat exchangers (plumbed in series) submerged in water provided suﬃcient thermal inertia to slow the cycle time

to three hours. Temperature data was taken from the thermistor located near the center of the OFIS interior.

2

7

Figure 3.3: Results for one 24 hour period, or three-cycles of temperature and sensitivity measurements. Each error bar

corresponds to the standard error within each 1

◦

C temperature bin, however, the linear ﬁt was performed on the unbinned

data.

Chapter 4

Analytic Opto-Mechanical Model

It was the inability for the temperature dependence of the elastic modulus

to explain the observed temperature dependent sensitivity that lead to exploring

a mechanical model for the OFIS. The ﬁrst step in this direction was to model the

OFIS as a single loop of ﬁber wrapped under tension about a silicone rubber tube

as a pre-stressed, double-walled thick cylinder, the basis of which is a well known

problem in elasticity dating back to Lam´e and Clapeyron in 1833 [26]. Derivations

similar to the one that follows can be found in nearly any introductory text on

elasticity or strength of materials.

4.1 Lam´e Equations for a Thick-Walled Cylinder

Assuming the cylindrical coordinate convention (r, θ, z), consider a thick-

walled cylinder of inner radius a and outer radius b subjected to both internal and

external pressure, p

a

and p

b

, respectively (Figure 4.1). Balancing the radial and

tangential (aka: “hoop”) forces for an element of unit thickness leads to:

(σ

r

+ dσ

r

) (r + dr) dθ = σ

r

rdθ + σ

θ

dθdr

where σ

r

and σ

θ

are the radial and tangential stresses, which in this case are

the only two non-zero elements of the stress tensor. By assuming that drdσ

r

≪

rσ

r

, rdσ

r

, σ

r

dr, and therefore disregarding drdσ

r

, canceling dθ, and dividing by dr,

28

29

the equation of equilibrium becomes:

r

dσ

r

dr

+ σ

r

−σ

θ

= 0. (4.1)

For an isotropic elastic material with Poisson’s ratio ν and elastic modulus E, the

stress-strain relationship can be expressed as:

ǫ

ij

=

1 + ν

E

σ

ij

−

ν

E

σ

αα

δ

ij

, (4.2)

where ǫ

ij

are the elements of the strain tensor, δ

ij

is the Kronecker delta, and the

repeated indices indicate summation. Without shear or axial stresses, the radial

and tangential strains are:

ǫ

r

=

1

E

(σ

r

−νσ

θ

) (4.3)

ǫ

θ

=

1

E

(σ

θ

−νσ

r

) . (4.4)

Rearranging (4.3,4), it follows that the two elements of the stress tensor are:

σ

r

=

E

1 −ν

2

(ǫ

r

+ νǫ

θ

) (4.5)

σ

θ

=

E

1 −ν

2

(ǫ

θ

+ νǫ

r

) . (4.6)

As proven by Housner and Vreeland [21], the two strain components in cylindrical

coordinates can be expressed in terms of the radial displacement u

r

:

ǫ

r

=

du

r

dr

(4.7)

ǫ

θ

=

u

r

r

. (4.8)

These results can be used to rewrite equations (4.5,6):

σ

r

=

E

1 −ν

2

du

r

dr

+ ν

u

r

r

(4.9)

σ

θ

=

E

1 −ν

2

u

r

r

+ ν

du

r

dr

. (4.10)

By diﬀerentiating (4.9) with respect to r:

dσ

r

dr

=

E

1 −ν

2

¸

d

2

u

r

dr

2

+ ν

1

r

du

r

dr

−

u

r

r

2

,

30

and using (4.9,10), the equation of equilibrium (4.1) can be expressed solely in

terms of the radial displacement u

r

:

d

dr

¸

1

r

d (ru

r

)

dr

= 0.

Integrating this separable equation twice yields:

u

r

(r) = C

1

r +

C

2

r

, (4.11)

where C

1

and C

2

are constants of integration. This is the general solution for the

radial displacement of a thick-walled cylinder, however, it must be re-written in

terms of stress to apply the boundary conditions (assuming the convention where

compressive stresses are negative):

σ

r

(r = a) = −p

a

(4.12)

σ

r

(r = b) = −p

b

. (4.13)

It may appear to be unnecessary to express the stress in terms of the radial dis-

placement only to solve for the stress, but it is required since σ

r

is a function of

u

r

and du

r

/dr. Deﬁning A =

E

1−ν

C

1

and B =

E

1+ν

C

2

, the general solutions for the

stresses are:

σ

r

(r) = A−

B

r

2

(4.14)

σ

θ

(r) = A +

B

r

2

. (4.15)

Application of the boundary conditions (4.12,13) determines A and B as:

A =

a

2

p

a

−b

2

p

b

b

2

−a

2

(4.16)

B =

a

2

b

2

(p

a

−p

b

)

b

2

−a

2

, (4.17)

and:

C

1

=

(1 −ν) (a

2

p

a

−b

2

p

b

)

E (b

2

−a

2

)

(4.18)

C

2

=

(1 + ν) a

2

b

2

(p

a

−p

b

)

E (b

2

−a

2

)

. (4.19)

31

Finally, the desired form for the radial displacement can be found with (4.11) and

(4.18,19):

u

r

(r) =

¸

(1 + ν) a

2

b

2

(p

a

−p

b

)

E (b

2

−a

2

)

1

r

+

¸

(1 −ν) (a

2

p

a

−b

2

p

b

)

E (b

2

−a

2

)

r, (4.20)

which is equation (10.46) of Faupel [14].

4.2 Pre-Stressed, Double-Walled Thick Cylinder

This section expands upon the previous one by adding a second cylinder

about the ﬁrst with an outer radius c, subjected to an external pressure p

c

, with

elastic modulus E

2

and Poisson’s ratio ν

2

. The radial displacements of the two

cylinders are:

u

r

1

(r) =

¸

(1 + ν

1

) a

2

b

2

(p

a

−p

b

)

E

1

(b

2

−a

2

)

1

r

+

¸

(1 −ν

1

) (a

2

p

a

−b

2

p

b

)

E

1

(b

2

−a

2

)

r,

for r ∈ [a, b] (4.21)

u

r

2

(r) =

¸

(1 + ν

2

) b

2

c

2

(p

b

−p

c

)

E

2

(c

2

−b

2

)

1

r

+

¸

(1 −ν

2

) (b

2

p

b

−c

2

p

c

)

E

2

(c

2

−b

2

)

r,

for r ∈ [b, c] . (4.22)

Since the second cylinder is force-ﬁt about the ﬁrst (this is analogous to either a

shrink-ﬁt or press-ﬁt construction), an “interference distance” δ can be deﬁned as:

δ = u

r

2

(r = b) −u

r

1

(r = b) . (4.23)

Note that the case of δ = 0 states that the radial displacement is continuous

across the interface. It is essential to understand that δ is the combination of the

shrinkage of the inner radius of the second cylinder (u

r

2

(b) > 0) and the outer

radius of the ﬁrst (u

r

1

(b) < 0). Texts on contact mechanics often use δ as the

“distance of mutual approach” [15], however, in this context δ is deﬁned as the

indentation. This leads to the constraint:

p

b

=

2E

1

bc

2

(b

2

−a

2

) p

c

+ E

2

(c

2

−b

2

) [2a

2

bp

a

+ δE

1

(b

2

−a

2

)]

b {E

1

(b

2

−a

2

) [b

2

+ c

2

+ ν

2

(c

2

−b

2

)] + E

2

(c

2

−b

2

) [a

2

+ b

2

−ν

1

(b

2

−a

2

)]}

.

(4.24)

32

(Note that setting p

a

= p

c

= 0 results in equations (10.51,2) of Faupel [14].)

However, to model the second cylinder as a loop of optical ﬁber the radial dis-

placement u

r

2

must be related to optical path length diﬀerence, which is governed

by the strain-optic eﬀect.

4.3 The Strain-Optic Eﬀect

The optical phase, φ, of light passing through an optical ﬁber can be written

as [6]:

φ = βL =

2πn

λ

L (4.25)

where β is called the propagation constant, n is the index of refraction of the ﬁber

core, λ is the wavelength of the light, and L is the length of the ﬁber. The change

in phase is then:

∆φ = β∆L + L∆β

= ∆φ

1

+ ∆φ

2

. (4.26)

The ﬁrst term represents the optical phase diﬀerence due the axial stretching of

the ﬁber:

∆φ

1

=

2πnL

λ

ǫ, (4.27)

where ǫ = ∆L/L is the axial strain in the ﬁber. The second term describes the

change in the index of refraction as well as the waveguide mode dispersion caused

by the reduction in diameter of the ﬁber core under axial stress. Making the

frequent assumption that the core diameter remains constant [6, 20, 32, 3], the

change in the index of refraction due to axial strain alone can be expressed as [3]:

∆φ

2

= −

πn

3

L

λ

[p

12

−ν

c

(p

11

+ p

12

)] ǫ, (4.28)

where ν

c

is the Poisson’s ratio of the ﬁber core and p

11

and p

12

are photoelastic

constants dependent upon the material used. Finally, the formula for the total

optical phase diﬀerence per unit strain can be formed from equations (4.32-34):

Ψ =

∆φ

ǫ

=

πnL

λ

¸

2 −n

2

[p

12

−ν

c

(p

11

+ p

12

)]

¸

. (4.29)

33

While this theory has been successfully applied innumerable times for un-

jacketed ﬁber, most relevantly in the case of a hydrophone whose design is similar

to the OFIS [32], it is not clear how jacketing aﬀects the stresses resolved on the

core. Therefore, an empirical measurement of the optical phase diﬀerence per

unit strain was performed on a 1.115 meter length of 900 micron jacketed Corning

SMF-28 ﬁber using a pair of ﬁber vices, a precision translation stage and a 1,310

nanometer wavelength laser. The result was found to be:

Ψ

empirical

= (6.111 ±0.009) ×10

6

rad · strain

−1

.

Using published values for the constants [20]:

n = 1.456

ν

c

= 0.17

p

11

= 0.121

p

12

= 0.270,

and equation (4.29), the theoretical value was calculated as:

Ψ

theoretical

= 6.107 ×10

6

rad · strain

−1

,

which is in agreement with the measured value.

4.4 Single Fiber about a Thick Cylinder Sub-

jected to Internal Pressure

In this section, the results of the previous two sections are used to model a

single loop of optical ﬁber wrapped under tension about a silicone rubber tube as

a pre-stressed, double-walled thick cylinder. The strain-optic behavior is included

to better model the outer layer of the cylinder as an optical ﬁber and to convert

its strain to an optical phase. After simplifying the previous results, an estimate

of the theoretical pressure sensitivity will be made and compared to experimental

data, followed by suggestions for future model improvements.

34

4.4.1 Theoretical Results

Assuming that the tube material is incompressible (ν

1

= 1/2), and dis-

carding the external pressure p

c

, the radial displacements (4.21,22) and interface

pressure (4.24) simplify to:

u

r

1

(r) =

¸

3a

2

b

2

(p

a

−p

b

)

2E

1

(b

2

−a

2

)

1

r

+

¸

(a

2

p

a

−b

2

p

b

)

2E

1

(b

2

−a

2

)

r, for r ∈ [a, b] (4.30)

u

r

2

(r) =

¸

(1 + ν

2

) b

2

c

2

p

b

E

2

(c

2

−b

2

)

1

r

+

¸

(1 −ν

2

) b

2

p

b

E

2

(c

2

−b

2

)

r, for r ∈ [b, c] (4.31)

p

b

=

2E

2

(c

2

−b

2

) [2a

2

bp

a

+ δE

1

(b

2

−a

2

)]

b {2E

1

(b

2

−a

2

) [b

2

+ c

2

+ ν

2

(c

2

−b

2

)] + E

2

(3a

2

+ b

2

) (c

2

−b

2

)}

.

(4.32)

The strain in the optical ﬁber at r = c simpliﬁes with the use of (4.8) and (4.31,32):

ǫ

θ

2

=

u

r

2

(r = c)

c

=

4b [2a

2

bp

a

+ δE

1

(b

2

−a

2

)]

2E

1

(b

2

−a

2

) [b

2

+ c

2

+ ν

2

(c

2

−b

2

)] + E

2

(3a

2

+ b

2

) (c

2

−b

2

)

,

(4.33)

and the sensitivity:

Φ =

ǫ

θ

2

p

a

Ψ

=

8π

2

nbc [2a

2

bp

a

+ δE

1

(b

2

−a

2

)] {2 −n

2

[p

12

−ν

c

(p

11

+ p

12

)]}

λp

a

{2E

1

(b

2

−a

2

) [b

2

+ c

2

+ ν

2

(c

2

−b

2

)] + E

2

(3a

2

+ b

2

) (c

2

−b

2

)}

,

(4.34)

where L = 2πc.

To calculate the pressure sensitivity of the single loop, the material param-

eters of the ﬁber must be known. As a ﬁrst estimate, the elastic modulus and

Poisson’s ratio of the jacketed ﬁber were determined by averaging the parameters

of the constitutive materials, weighted by their respective cross-sectional areas.

For example, the core and cladding of the ﬁber is 125 microns in diameter and

composed of fused silica. This is coated with UV cured polyacrylate to a total

diameter of 250 microns, which is coated again up to 900 microns with polyvinyl

35

Table 4.1: Material Parameters for 900 micron SMF-28 Fiber

Material Elastic Modulus E (GPa) Poisson’s Ratio ν

Fused Silica 71.7 0.17

Polyacrylate 3.2 0.35-0.42

PVC 2.5-4.1 0.33-0.46

chloride (PVC). Using published values for these materials in Table 4.1 [25, 20],

the average values were found to be:

E

2

= 4.6 ±0.7 GPa

ν

2

= 0.39 ±0.06.

The measured values of the elastic modulus of the silicone from the previous ex-

periments (taken at 20

◦

C) and the three radii were measured as:

E

1

= 3.48 ±0.01 MPa

a = 0.0091 ±0.0002 m

b = 0.0118 ±0.0001 m

c = 0.0126 ±0.0001 m.

The ﬁnal step to predict the sensitivity of the single loop from (4.34) is

to estimate the interference distance, or indentation δ. Assuming no pre-stressing

(δ = 0):

Φ

theoretical

(δ = 0) = (1.0 ±0.3) ×10

−3

rad · Pa

−1

.

The analytical evaluation of the indentation about an elastic cylinder due to an

elastic ring of circular cross-section appears to be an unsolved problem, except in

the case where the ring is rigid (ie: E

2

→∞) [2]. Unfortunately, the complexity of

its solution prevents it from being useful in this thesis, however, another estimate

of δ may be arrived at using the results of an elastic spherical indenter in an elastic

half-space [15]:

δ

3

=

1

R

3T

4E

∗

2

, (4.35)

where:

E

∗

=

1 −ν

2

1

E

1

+

1 −ν

2

2

E

2

−1

(4.36)

36

is the combined elastic modulus of the silicone tube and ﬁber, R = (c −b) /2 is

the radius of the ﬁber, and T is the indenter load, which is the tension of the ﬁber.

Using a value of T = 1.62 ±0.03 Newtons:

δ = (5.4 ±0.4) ×10

−4

m,

and the resulting sensitivity:

Φ

theoretical

δ = 5.4 ×10

−4

= 55 ±16 rad · Pa

−1

.

This illustrates the remarkable dependence the pressure sensitivity on δ and the

necessity for an accurate estimate, however, it may be useful as a free parameter.

4.4.2 Experimental Results

An experiment was conducted on a single loop of ﬁber about a silicone tube,

and its sensitivity measured with a similar calibration system to all the previous

experiments. The loop of ﬁber was wrapped at a tension of 1.62 ± 0.03 Newtons

and the PVC jacket welded over a length of 2-3 millimeters. After pressurizing the

interior of the tube at ﬁve diﬀerent amplitudes at a frequency of 0.5 Hz, the mag-

nitudes of the calibrated pressure transducer and optical phase were determined

using least squares. A linear ﬁt to the optical phase diﬀerence versus the applied

pressure resulted in a sensitivity of:

Φ

experimental

= (11.9 ±0.6) ×10

−3

rad · Pa

−1

.

While the experiment has been repeated on several other loops of varying tension

with similar results, it is clearly out of agreement with either of the above predicted

values. Inverting (4.34) to ﬁnd the indentation yields:

δ = (1.1 ±0.3) ×10

−7

m.

4.4.3 Suggestions for Future Work

It is diﬃcult to assess whether the pre-stressed, double-walled thick cylinder

can be useful to understand the temperature dependence of the pressure sensitivity

37

of the OFIS. The striking dependence on the interference distance, and the diﬃ-

culty in obtaining its estimate severely limits the predictive power of the model.

Inverting for δ could be useful for obtaining an initial estimate of the pressure sen-

sitivity that could be held constant to examine the eﬀects of temperature change

on the other model parameters. However, this approach would be highly problem-

atic since the formulations of δ depend upon these same parameters.

While there is some literature on helically wrapped composite tubes (Evans

[13] provides a review), most reduce the problem to a bulk cylinder whose material

properties are found by what amounts to an elaborate weighted-averaging scheme.

The other extreme is the highly complex contact problem as in Reference [2]. How-

ever, the strong dependence upon the contact conditions shown above indicate that

a realistic model for the OFIS should focus on the temperature dependence of the

contact strain ﬁeld, which could be addressed using a numerical method such as

ﬁnite diﬀerences or ﬁnite elements. In this way the composite nature of the ﬁber

could also be taken into consideration and better predict the strains resolved on

the ﬁber’s core.

3

8

Figure 4.1: The left diagram shows a thick-walled cylinder with inner radius a and outer radius b, subjected to internal

pressure p

a

and external pressure p

b

. The right side shows the forces on the area segment from the left diagram, both

highlighted in gray.

Appendix A

Optical Path Length Modulation

and the Femtometer System

While the primary focus of this thesis is the mechanical and optical behavior

of the OFIS as an interferometric pressure transducer, it is important to under-

stand how an estimate of the optical phase diﬀerence is calculated by measuring

the intensity of the interference fringes. This will explain why the ﬁber path length

is modulated, as mentioned in Chapter 1, and where interference ellipse originates,

since its area was used as a metric for assessing the uptime of the sensor in Chapter

2. Furthermore, it is essential to note that the following technique is not unique

to the OFIS, but can be applied to any Mach-Zehnder or Michelson type interfer-

ometer [33]. What follows is the description of the components needed to obtain

an estimate of optical phase for the current Michelson interferometer OFIS design

as shown in Figure 2.6.

As previously stated, there are two helical wraps of ﬁber such that there

are four connectors at the end of an OFIS, which interface with one end of a device

termed the modulator box (Figure A.1). At the other end of the modulator box is

the laser input for illuminating the OFIS, which is split equally among two paths

that wrap about a piezoelectric cylinder (piezo). A 156 kHz voltage signal supplied

to the piezo causes its diameter to change by a small amount, thereby stretching

the ﬁber a fraction of a wavelength. The piezo is split and wired such that one ﬁber

is strained π out of phase with the other. The modulated light is sent through both

39

40

arms where it is reﬂected oﬀ the Faraday mirrors and recombined at the splitter

to form interference fringes. Therefore, the purpose of the modulator box is to

supply the illumination to each OFIS arm, modulate the optical path length, and

return the modulated fringe signal.

A collection of electronics known as a “Hanada rack” (Figure A.2) bridges

the gap between the modulator box and the data acquisition system. The ﬁrst

module contains the oscillator and ampliﬁer to drive the piezo, while the last

consists of an ampliﬁed photodetector to receive the modulated fringe signal. In

between these two units is a phase-sensitive detector, or lock-in ampliﬁer. The

output of the photodetector, x, is fed into the lock-in ampliﬁer to demodulate the

fringe signal, producing an output, y, proportional to the spatial derivative of the

original fringe signal [33]. There is a secondary ampliﬁer module to adjust the x

and y voltage amplitudes and their DC oﬀsets, conditioning them for input to the

data acquisition system.

Recall that the point of modulating the optical path length is to discrim-

inate between contractive and expansive fringes that respectively correspond to

positive and negative external pressure acting on the OFIS. For example, if the

fringe intensity is a maximum, corresponding to perfect constructive interference,

it is impossible to determine if a subsequent decrease in intensity is the result of

the optical path length increasing or decreasing. However, the plot of a chang-

ing fringe signal versus its electronically generated derivative traces out an ellipse

whose instantaneous optical phase diﬀerence depends upon the (x, y) location on

the ellipse. Therefore, the next step is to obtain estimates of the ellipse location,

size, and shape to obtain an estimate of the optical phase.

The previously described components are essential for preparing the origi-

nal fringe signal for estimating the optical phase, but it is the data acquisition and

signal processing system called the Femtometer that yields the estimate. These

(x, y) data pairs are sampled at a rate of either 100 kHz or 50 kHz, and a user

determined number of pairs are stored in a circular buﬀer to obtain least squares

41

estimates of the constants in the following parametric equations [33]:

x = x

0

+ a sin (p + p

0

) (A.1)

y = y

0

+ b cos (p) , (A.2)

where x

0

and y

0

are DC oﬀsets on the (x, y) plane, a and b are the x and y ellipse

amplitudes, p

0

is the ellipse shape factor, and p is the optical phase diﬀerence. As

new data replaces the old in the buﬀer, the ellipse parameters are updated and

the optical phase is determined from the most recent (x, y) pair by solving both

equations (A1,2) for p and taking the arctangent of their ratio. It is important to

note that under extreme conditions, it is possible for the subsequent (x, y) point to

advance more than 2π around the ellipse despite the high sample rate. With the

previous limitation properly accounted for, the phase and ellipse parameter data

are decimated and archived following the CSS 3.0 convention.

42

Figure A.1: Schematic of the modulator box, showing the OFIS connections on

the left. While there are four connections to the OFIS, two send light into the

OFIS and receives the reﬂected light from the FM found at the remaining two

connections. Note that the single and double wrap arms can be interchanged.

Figure A.2: Block diagram of the fringe modulation and Femtometer system for

estimating the optical phase diﬀerence between two OFIS arms. The Femtometer

output is the ﬁve ellipse parameters (usually at a much lower sample rate) along

with the (x, y) inputs used to compute p.

Appendix B

Derivation of Ellipse Parameters

Standard formulas for ellipse properties such as area, eccentricity, foci, etc.

are usually given in terms of the semimajor and semiminor axes, therefore it is

beneﬁcial to ﬁnd how these are related to the ﬁve ellipse parameters (x

0

, y

0

, a, b

and p

0

) put out by the Femtometer. One approach is to reduce the parametric

equations (1,2) of Zumberge, et al. [33] to canonical form, which are:

x = x

0

+ a sin (p + p

0

) (B.1)

y = y

0

+ b cos (p) . (B.2)

It can immediately be assumed that x

0

= y

0

= 0 since the ellipse properties should

not depend on its position. Solving both equations for p and taking the cosine of

both sides:

y

b

= cos

arcsin

x

a

−p

0

. (B.3)

Splitting the argument of the cosine with cos (θ ±φ) = cos (θ) cos (φ)∓sin (θ) sin (φ)

and using the inverse identity cos [arcsin (θ)] =

√

1 −θ

2

, we arrive at:

y

b

=

1 −

x

a

2

cos (p

0

) +

x

a

sin (p

0

) . (B.4)

Isolating the square root, squaring both sides and collecting like terms results in a

quadratic in standard form:

Ax

2

+ Bxy + Cy

2

−D = 0, (B.5)

43

44

where:

A = a

−2

(B.6)

B = −

2

ab

sin (p

0

) (B.7)

C = b

−2

(B.8)

D = cos

2

(p

0

) . (B.9)

The next step is to rotate the quadratic to its principle axes by the transformation

(Figure B.1):

x = ξ cos (ψ) −η sin (ψ) (B.10)

y = ξ sin (ψ) −η cos (ψ) . (B.11)

Expanding and collecting like terms:

A

′

ξ

2

+ B

′

ξη + C

′

η

2

−D = 0, (B.12)

where:

A

′

= Acos

2

(ψ) + B cos (ψ) sin (ψ) + C sin

2

(ψ) (B.13)

B

′

= −2Acos (ψ) sin (ψ) + B

cos

2

(ψ) −sin

2

(ψ)

(B.14)

+2C cos (ψ) sin (ψ) (B.15)

C

′

= Asin

2

(ψ) −B cos (ψ) sin (ψ) + C cos

2

(ψ) (B.16)

and D remains unchanged from (B.9). For the rotation to arrive at the principle

axis, B

′

must be zero, leading to a deﬁnition of ψ:

tan (2ψ) =

B

A−C

=

2ab sin (p

0

)

a

2

−b

2

, (B.17)

with the aid of the trigonometric relationships:

cos

2

(ψ) −sin

2

(ψ) = cos (2ψ)

cos (ψ) sin (ψ) = sin (2ψ) .

Note that since tan (2ψ) ∈ (−∞, ∞), there always exists a ψ ∈

−

π

4

,

π

4

such that

B

′

= 0. Finally, by dividing by D we arrive at the canonical equation:

ξ

u

2

+

η

v

2

= 1. (B.18)

45

With two more inverse trigonometric relationships:

cos [arctan(θ)] =

1

√

1 + θ

2

sin [arctan(θ)] =

θ

√

1 + θ

2

.

and a fantastic amount of algebra we arrive at the semimajor and semiminor axes

in terms of the Femtometer parameters:

u =

1

2

a

2

+ b

2

+

a

4

+ b

4

−2a

2

b

2

cos (2p

0

)

(B.19)

v =

1

2

a

2

+ b

2

−

a

4

+ b

4

−2a

2

b

2

cos (2p

0

)

(B.20)

where a quick test shows that when p

0

= 0, u = a and v = b. Given that the area

of an ellipse is πuv, it is easy to show that:

Area = πab cos (p

0

) . (B.21)

Figure B.1: Geometry of ellipse rotation, where η and ξ correspond to its principle

axes.

Bibliography

[1] Stephen J. Arrowsmith, Michael A. H. Hedlin, Brian Stump, and Marie D.

Arrowsmith. Infrasonic Signals from Large Mining Explosions. Bulletin of the

Seismological Society of America, 98(2):768–777, 2008.

[2] A. Avei, A. Bulu, and A. Yapiei. Axisymmetric smooth contact for an elas-

tic isotropic inﬁnite hollow cylinder compressed by an outer rigid ring with

circular proﬁle. Acta Mechanica Sinica, 22:46–53, 2006.

[3] Chris S. Baldwin. Springer Handbook of Experimental Solid Mechanics, chap-

ter Optical Fiber Strain Gages, pages 347–370. Springer, Secaucus, New

Jersey, 2008.

[4] Henry E. Bass, Joydeep Bhattacharyya, Milton A. Garces, Michael Hedlin,

John V. Olson, and Robert L. Woodward. Infrasound. Acoustics Today,

2(1):9–19, 2006.

[5] Max Born and Emil Wolf. Principles of Optics. Pergamon, Headington Hill

Hall, Oxford, fourth edition, 1970.

[6] C. D. Butter and G. B. Hocker. Fiber optics strain gauge. Applied Optics,

17(18):2867–2869, 1978.

[7] George Chimonas. A Possible Source Mechanism for Mountain-Associated

Infrasound. Journal of the Atmospheric Sciences, 34(5):806–811, 1977.

[8] D. R. Christie, K. J. Muirhead, and A. L. Hales. On Solitary Waves in the

Atmosphere. Journal of the Atmospheric Sciences, 35(5):805–825, 1978.

[9] William L. Donn and Maurice Ewing. Atmospheric Waves from Nuclear Ex-

plosions. Journal of Geophysical Research, 67(5):1855–1866, 1962.

[10] William L. Donn and Maurice Ewing. Atmospheric Waves from Nuclear Ex-

plosions – Part II: The Soviet Test of 30 October 1961. Journal of the Atmo-

spheric Sciences, 19(3):264–273, 1962.

46

47

[11] William L. Donn, Eric Posmentier, Uri Fehr, and N. K. Balachandran. In-

frasound at Long Range from Saturn V, 1967. Science, 162(3858):1116–1120,

1968.

[12] William L. Donn and David Rind. Monitoring Stratospheric Winds with

Concorde-Generated Infrasound. Journal of Applied Meteorology, 18(7):945–

952, 7 1979.

[13] J. J. Evans and P. D. Wilcox. A structural model for high pressure helical wire-

wound thermoplastic hose. International Journal of Solids and Structures,

39:1307–1326, 2002.

[14] Joseph H. Faupel. Engineering Design: A Synthesis of Stress Analysis and

Materials Engineering. John Wiley and Sons, New York, New York, 1964.

[15] Anthony C. Fischer-Cripps. Introduction to Contact Mechanics. Springer,

New York, New York, second edition, 2007.

[16] T. M. Georges. Infrasound from Convective Storms: Examining the Evidence.

Reviews of Geophysics, 11(3):571–594, 1973.

[17] Dennis H. Goldstein. Polarized Light. CRC Press, New York, New York,

second edition, 2003.

[18] Eugene Hecht. Optics. Addison-Wesley, Reading, Massachusetts, second edi-

tion, 1989.

[19] Michael A. H. Hedlin, Jon Berger, and Frank L. Vernon. Surveying Infrasonic

Noise on Oceanic Islands. Pure and Applied Geophysics, 159(5):1127–1152,

2002.

[20] G. B. Hocker. Fiber-optic sensing of pressure and temperature. Applied Optics,

18(9):1445–1448, 1979.

[21] George W. Housner and Thad Vreeland Jr. The Analysis of Stress and De-

formation. Collier-Macmillan, London, 1966.

[22] Jrgen Jakobsen. Infrasound Emission from Wind Turbines. Journal of Low

Frequency Noise, Vibration and Active Control, 24(3):145–155, 2005.

[23] H. G. Jerrard. Transmission of Light through Birefringent and Optically

Active Media: the Poincar´e Sphere. Journal of the Optical Society of America,

44(8):634–640, 1954.

[24] A. D. Kersey, M.J. Marrone, and M. A. Davis. Polarisation-Insensitive Fibre

Optic Michelson Interferometer. Electronics Letters, 27(6):518–520, 1991.

48

[25] Dale O. Kipp. Plastic Material Data Sheets. MatWeb - Division of Automation

Creation, Inc., 2008.

[26] G. Lam´e and B.P.E. Clapeyron. M´emoire sur l’´equilibre int´erieur des corpes

solides homog`enes. Memoires par divers savans, 4, 1833.

[27] Mario Martinelli. A Universal Compensator for Polarization Changes Induced

by Birefringence on a Retracing Beam. Optics Communications, 72(6):341–

344, 1989.

[28] E. S. Posmentier. A Theory of Microbaroms. Geophysical Journal Interna-

tional, 13(5):487–501, 1967.

[29] Kristoﬀer T. Walker and Michael A. H. Hedlin. Infrasound Monitoring for

Atmospheric Studies, chapter A review of wind noise reduction methodologies,

pages 1–49. Springer, Secaucus, New Jersey, 2009.

[30] Kristoﬀer T. Walker, Mark A. Zumberge, and Matthew A. Dzieciuch. OFIS

Experiments at Camp Elliott: Paving the Way to Infrasonic Radar and a

Portable Infrasonic Calibrator. In Proceedings of the 29th Monitoring Research

Review: Ground-Based Nuclear Explosion Monitoring Technologies, 2007.

[31] Kristoﬀer T. Walker, Mark A. Zumberge, Michael A. H. Hedlin, and Peter M.

Shearer. Methods for determining infrasound phase velocity direction with

an array of line sensors. The Journal of the Acoustical Society of America,

124(4):2090–2099, 2008.

[32] Kai Yin, Min Zhang, Tianhuai Ding, Liwei Wang, Zhenguo Jing, and Yanbiao

Liao. An investigation of a ﬁber-optic air-backed mandrel hydrophone. Optics

Communications, 281:94–101, 2008.

[33] Mark A. Zumberge, Jonathan Berger, Matthew A. Dzieciuch, and Robert L.

Parker. Resolving Quadrature Fringes in Real Time. Applied Optics,

43(4):771–775, 2004.

[34] Mark A. Zumberge, Jonathan Berger, Michael A. H. Hedlin, Eric Husmann,

Scott Nooner, Richard Hilt, and Rudolf Widmer-Schnidrig. An optical ﬁber

infrasound sensor: A new lower limit on atmospheric pressure noise between

1 and 10 Hz. The Journal of the Acoustical Society of America, 113(5):2474–

2479, 2003.

Copyright Scott DeWolf, 2009 All rights reserved.

The thesis of Scott DeWolf is approved, and it is acceptable in quality and form for publication on microﬁlm and electronically:

Chair

University of California, San Diego

2009

iii

DEDICATION To Sarah and Delilah. iv .

EPIGRAPH Only the mediocre are supremely conﬁdent of their ability. — Sir Michael Atiyah v .

3 Faraday Mirrors and the Michelson OFIS . . . . .2. . . . . . .2 Pre-Stressed. . . . . . . . . . . . . . . . . . .1 Theoretical Results . 1 1 2 7 8 9 9 10 11 Chapter 2 Chapter 3 Temperature-Induced Pressure Sensitivity Change . . . . . . . . . 20 3. . . . . . . . . . 36 39 Chapter 4 Appendix A Optical Path Length Modulation and the Femtometer System vi . . . . . . . . . . . . . . . . . . . xii Chapter 1 Introduction . . . . . . . . . x xi Abstract of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . 33 4. Dedication . . . . . . . .2 The Polarization-Maintaining OFIS . . . . . . . . 2. . . . . . . . . . . . . . . . . . .1 Infrasound and Infrasound Instrumentation . . . . . . . . . . . . . viii List of Tables . . . . . . . . . . . . . . . . Temperature-Induced Polarization Drift . . . . . . . . . . . . . . Construction and Deployment . . . . . . . . . 20 3. . . . . . . . . . . . . . . . . 2. . . .1 Observations of Polarization Drift . . . . . . . . . . . . . Acknowledgements . . . .2 The Optical Fiber Infrasound Sensor . . . . . . .2 Experimental Results . . . . . . . . . . 36 4. . . . . . . . . . . . 1. . . . . . . . . . . . . . . .4. . . . . . . . . . . . . . . . . . 22 Analytic Opto-Mechanical Model . . . . . . . . . . 32 4.3 Suggestions for Future Work . . . . . . 2. . . 34 4. . . . . . . . . . . . . . . . . . . . . 2.1 Lam´ Equations for a Thick-Walled Cylinder . . .TABLE OF CONTENTS Signature Page . . . . . . . . . . . . . . . . . . . . 1. . . . . . .4 Single Fiber about a Thick Cylinder Subjected to Internal Pressure . .4. . . . . . 28 4.4. . . . . . . . . Table of Contents . . . . . . . . . . . . . .1 Laboratory Observations . . . . . . . . . . . 31 4. . . . . . . . . . . . . Double-Walled Thick Cylinder . . . . . . . . . . . . . . . .2 Results: Ellipse Stability and Optical Phase Noise 2. . .2 Temperature Dependence of the Elastic Modulus . . . .1 Design. . . . . . . . . . . . . . . iii iv v vi List of Figures . . . . . . . . . . .3 The Strain-Optic Eﬀect . . . 28 e 4. . . . . Epigraph . . . .2. . .

43 46 vii . . .Appendix B Derivation of Ellipse Parameters . . Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2: The Mach-Zehnder OFIS schematic. a standard OFIS ellipse behaves similarly with a nontrivial probability of having an area at or near zero. . . .4: Ellipse area probability density estimates for the PANDA. . . . . Each connects to a central microbarometer. . . . . PANDA. . from left to right: Bow-tie. . . . . . . .6: The Michelson OFIS schematic. . . .1: Two views of the same Poincar´ sphere. . . . was used for the experiment in section 2. 17 Figure 2. . . .1: Typical conﬁguration for a high frequency 18 meter. . . however. . The fast axis is along the horizontal dimension in this ﬁgure. It is at this point that the interference fringes are formed and sent to the photodetector. but the PM OFIS signal appears slightly lower most likely due to its directional response [34].1 meter diameter corrugated drainage tube for both ground-lying and burial deployments. . . . however. Bowtie. . 144-port pipe array (shown to scale).LIST OF FIGURES Figure 1. . along with an exposed length shown next to a pencil for scale. . . Each sensor was connected to an instrument box by a cylindrical PVC adapter arm. . . . Figure 1. . . . . . . 16 Figure 2. . Both signal amplitudes have been converted to Pascals.2: Three types of PM ﬁbers: PANDA. . . . . . 5 6 Figure 2. . . . . . . .1). . . . 19 viii . . . . . . . . . 18 Figure 2. The evolution of the SOP in the single wrap arm is shown by circles. 92-port and a low frequency 70 meter. . whereas the black shapes represent the stressing materials (not to scale). . Note that the SMF-28 OFIS consisted of a single wrap. . 16 Figure 2. Bow-tie and elliptical-clad. . . . . . . the phase conjugate beam from both arms are reﬂected back to the original splitter. . . . . . one rotated π in aze imuth with respect to the other. .3: Four Camp Elliott OFISs. The center gray circle denotes the core of the ﬁber. . . . . . . showing that after traveling the length of the sensor. . . . . . and SMF-28 OFISs. . . . . while the lower shows their diﬀerence along with a calibrated B&K 4193 precision infrasound microphone. SMF-28. . . . . . . . . . . . . and the double wrap arm by dark squares. . . . . shown with a FM equipped Michelson OFIS. . 15 Figure 2. . . . . . . . . . . . Note that the OFIS is wrapped in ﬁberglass insulation and inserted into a 0. . .5: The top time series shows the raw optical phase overlain with the scaled null interferometer signal. . and a standard OFIS (not used for the present study.

. two send light into the OFIS and receives the reﬂected light from the FM found at the remaining two connections. . . . . . The right side shows the forces on the area segment from the left diagram. . . . .1: The left diagram shows a thick-walled cylinder with inner radius a and outer radius b.2: Schematic of the experimental setup for measuring the temperature dependence of the pressure sensitivity. . . . . . . . or three-cycles of temperature and sensitivity measurements. The Femtometer output is the ﬁve ellipse parameters (usually at a much lower sample rate) along with the (x. . . . . where η and ξ correspond to its principle axes. This unit was subsequently enclosed by insulation and mounted horizontally for the calibration experiments and vertically for the elastic modulus and linear expansion experiments. . . . 26 Figure 3. . . Each error bar corresponds to the standard error within each 1◦ C temperature bin. . . . subjected to internal pressure pa and external pressure pb . . . .1: Schematic of the modulator box. . . . . the linear ﬁt was performed on the unbinned data. While there are four connections to the OFIS. . . . . . . . Figure A.3: Results for one 24 hour period. . . however.5 meter long OFIS heating and cooling chamber connected to the thermoelectric temperature control system. . . . Note that the single and double wrap arms can be interchanged. . Temperature data was taken from the thermistor located near the center of the OFIS interior. . . . . . . . 38 42 42 45 ix .Figure 3. . . y) inputs used to compute p.1: Geometry of ellipse rotation. showing the OFIS connections on the left. . . Figure B. . . . . . . The two heat exchangers (plumbed in series) submerged in water provided suﬃcient thermal inertia to slow the cycle time to three hours. 25 Figure 3. . . . . . . Figure A. both highlighted in gray. . . . . . .2: Block diagram of the fringe modulation and Femtometer system for estimating the optical phase diﬀerence between two OFIS arms. .1: The 1. . . . . . . . . . 27 Figure 4. . . . .

. . . . . . .1: Material Parameters for 900 micron SMF-28 Fiber . 35 x .LIST OF TABLES Table 4.

Kristoﬀer T. This has been ampliﬁed in my case thanks to Dr. and for countless other small tasks that accompany ﬁeld and laboratory work. amongst other less ﬂattering things. who has always taken the time to listen and provide useful suggestions for most of my laboratory work to date. Finally. thanks to Joel White for help with the winding and deployment of the polarization maintaining OFISs. along with the authorship of essential scripts for handling CSS data. a combination of patience and experience. I would have been substantially less productive. Glenn Sasagawa. which will undoubtedly shape how I approach problems in the future. not to mention paying for many of my lunches. I truly appreciate the ﬂexibility to reﬁne and re-run critical experiments and to pursue the opto-mechanical model.ACKNOWLEDGEMENTS I would like to thank Dr. Mark A. Thanks to Dr. Zumberge for advising my research and graduate academic career thus far. His examination of my experimental work in the lab and in the ﬁeld has provided me many valuable insights on when and where time is best spent. I have come to understand that successful experimentation is. Without his guidance on tasks ranging from the trivial to the impossible. xi . Walker for countless insightful discussions and comments throughout the course of this work.

Chair The Optical Fiber Infrasound Sensor (OFIS) is an interferometric pressure transducer consisting of a pair of optical ﬁbers helically wrapped about a compliant tube.ABSTRACT OF THE THESIS Optical and Mechanical Behavior of the Optical Fiber Infrasound Sensor by Scott DeWolf Master of Science in Earth Sciences University of California San Diego. both of which solve the polarization fading issue. Therefore. a crude analytic mechanical and optical model is explored to help investigate potential causes. Laboratory measurements of the thermal behavior of the pressure sensitivity are found to conﬁrm ﬁeld tests. but measurements of the temperature dependence of the tube’s elastic modulus does not appear to adequately describe the observed behavior. the simple theory of compound thick cylinders was found to be problematic as a realistic model for the OFIS. its performance has been plagued by downtime due to polarization fading (resulting from no interference fringe intensity) and a nearly continuous change in its pressure sensitivity. While the OFIS has been successfully deployed for nearly a decade. 2009 Mark A. xii . however. Zumberge. This thesis explores the use of very expensive polarization maintaining ﬁber and inexpensive Faraday mirrors.

The most ubiquitous and commonly studied source in coastal areas are microbaroms. mountain-associated waves [7]. and rocket launches [11]. to mining blasts [1] and nuclear explosions [9. atmospheric solitons [8].000 Hz. and convective storms [16]. which arise from the forcing of the lower atmosphere by nonlinear wave-wave interactions at the ocean surface [28]. respectively.1 Infrasound and Infrasound Instrumentation There are many examples of natural and anthropogenic sources of infra- sound. and sounds above and below this threshold are termed ultrasound and infrasound. 10]. 1.Chapter 1 Introduction Sound can be deﬁned as pressure ﬂuctuations about some background level. volcanoes (Reference [4] provides a review of the preceding phenomena). tsunamis. some naturally occurring sources include earthquakes (seismic-to-acoustic coupling). the term infrasound is restricted to pressure ﬂuctuations below 20 Hz with atmospheric pressure near the surface as the background level. as in the case of human speech at atmospheric pressure. meteors. The frequency content of sound can be broken into three categories roughly deﬁned by the range of human hearing. It is this last case that prompted the International Monitoring System (IMS) to in- 1 . wind turbines [22]. Investigations of man-made sources range from shock waves from supersonic aircraft [12]. This audible range is often taken to be 20-20. In this thesis. While too numerous to describe individually. auroras.

labor intensive to build and maintain. or “wind noise. they act as local reinforcement such that the tube near the double wrapped . typically at 1. exacerbated by the fact that 23 of 60 IMS infrasound stations are on. remote oceanic islands [19]. This is accomplished by connecting a microbarometer to an elaborate array of pipes whose inlets are distributed over a wide area (Figure 1. and have a large ecological footprint. abatement schemes and their limitations can be found in Reference [29]. Since the ﬁbers are less elastic than the tube. 1. There are an increasing number of studies centered around the concept of screening the turbulence by enclosing a microbarometer with or without a pipe array in a porous fabric tent. A comprehensive review of wind noise physics.2 The Optical Fiber Infrasound Sensor Optical ﬁbers are well known for their ability to measure strain [6]. The most signiﬁcant limitation to measuring infrasound are pressure ﬂuctuations due to turbulence. however.1).310 nanometer wavelength. tempera- ture and pressure [20]. The principle drawbacks to either method are that they are expensive. enters one end of both the double and single wrapping only to loop back where the other ends are combined to form interference fringes (Figure 1. or microbarometers. hydroacoustic and radionuclide sensors for monitoring the Comprehensive Nuclear Test Ban Treaty. Laser light. instrument development is largely focused on noise reduction technology. The OFIS consists of a compliant rubber tube helically wrapped with a pair of optical ﬁbers. The ﬁrst exploits the fact that these turbulent structures are spatially incoherent by mechanically averaging the pressure over an area. which form an equal-arm Mach-Zehnder interferometer. there appear to be two strategies that can be used individually or in combination. or proposed for.” and while there exist several high quality infrasound microphones. which motivated the development of a fundamentally diﬀerent microbarometer that came to be known as the Optical Fiber Infrasound Sensor (OFIS) [34]. and is the current method used with IMS certiﬁed infrasound stations.2).2 clude a global network of infrasound arrays along with seismic.

The result is a lower limit on infrasonic noise in the 1 to 10 Hz band. When the polarization states are orthogonal. There are several unique characteristics of the OFIS that can be exploited for measuring infrasound. Therefore. and phase velocity of infrasound signals [31]. both path lengths are modulated by a piezoelectric cylinder and the resulting interference fringe intensity is monitored by a photodetector. it can be deployed in diﬀerent geometries. the state of polarization in each of the two ﬁbers changes with respect to the other. the OFIS is wrapped in ﬁberglass insulation and inserted into a perforated tube to allow burial beneath a porous medium. In the case of an array of linear OFISs. Since the light experiences little attenuation in an optical ﬁber. Finally. each with diﬀerent amplitude and frequency responses that depend upon the direction of arrival. the OFIS measures diﬀerential strain in the tube as a result of changes in the pressure ﬁeld. As the temperature of the OFIS changes. as evidenced by Reference [34]. chapter 3 examines how temperature changes the sensitivity of the OFIS.3 ﬁbers experiences less deformation than the single wrap for a given amount of strain in the tube. temperature has a severe impact on the performance of the OFIS. Given the ﬂexibility of the OFIS. Therefore the OFIS employs both wind noise reduction technologies at once. which is quantiﬁed by the . which depends upon the polarization of the laser light as it emerges from each arm. The necessity of modulating the paths allows for the discrimination between contractive and expansive fringes (negative and positive pressures. Chapter 2 is concerned with the intensity of the interference fringes. and both the use of polarization maintaining (PM) ﬁber and Faraday mirrors (FM) are investigated. but has the further advantage over mechanical ﬁlters in that it averages the turbulence at the speed of light rather than the speed of sound. back azimuth. respectively). Finally. and it is the purpose of this thesis to quantify what is happening and describe attempts to eliminate or mitigate these eﬀects. with a much smaller footprint than a conventional pipe array. which is described in Appendix A. When deployed. Unfortunately. the OFIS can be made almost arbitrarily long. this has lead to the development of novel beamforming techniques to estimate the elevation. the interference intensity is zero causing the sensor to go oﬄine.

While temperature modestly changes the elasticity of the tube.4 optical phase diﬀerence per unit pressure. . a crude analytic model is explored to understand possible geometric eﬀects caused by thermal expansion.

Figure 1.1: Typical conﬁguration for a high frequency 18 meter, 92-port and a low frequency 70 meter, 144-port pipe array (shown to scale). Each connects to a central microbarometer.

5

Figure 1.2: The Mach-Zehnder OFIS schematic, along with an exposed length shown next to a pencil for scale. Note that the OFIS is wrapped in ﬁberglass insulation and inserted into a 0.1 meter diameter corrugated drainage tube for both ground-lying and burial deployments.

6

**Chapter 2 Temperature-Induced Polarization Drift
**

The ﬁrst limitation of the OFIS concerns the size of the ellipse formed by plotting the interference fringe signal versus its derivative. A least squares ﬁt to this ellipse is performed by a digital signal processing system called the Femtometer (Appendix A) to determine the optical phase diﬀerence in radians. When the light from each arm of the OFIS is combined, the resulting interference fringe intensity depends upon the relative phase of the two beams. However, the intensity contrast between the constructive and destructive interference fringes depends upon the relative state of polarization (SOP) of each arm with respect to the other. For example, if both arms emit linearly polarized light where their electric ﬁelds are both vertical, the resulting intensity will depend only on their relative phase. However, if the electric ﬁeld of one arm is horizontal while the other remains vertical then no interference can take place. Therefore, the ﬁtted ellipse changes size with the relative SOP, which can lead to inaccurate phase estimates when the SOP in the two arms are near orthogonal and the ellipse becomes very small, taking the sensor oﬄine.

7

known as the Poincar´ sphere. which can be represented graphically as a point on a sphere whose radius is determined by the intensity.8 2. respectively. whereas right and .2) and reducing the resulting quadratic to canonical form. with longitude 2ψ and latitude 2χ.1) (2. spatial location. and τ = k · x − ωt. angular frequency. and time. it can be shown that the SOP itself can be represented by an ellipse. It is straight forward to show that: s2 = s2 + s2 + s2 .6) where a1. For example. 2χ) states of linearly polarized light fall along the equator starting with horizontal at (0. The Stokes parameters are deﬁned as: s0 = a2 + a2 1 2 s1 = a2 − a2 1 2 s2 = 2a1 a2 cos (δ) s3 = 2a1 a2 sin (δ) . (2. where δ = δ2 − δ1 . the wave By eliminating τ from (2.9) (2.4) (2. (2ψ. 0) and vertical at (π. the x and y components of an electromagnetic wave propagating in the z direction are: Ex = a1 cos (τ + δ1 ) Ey = a2 cos (τ + δ2 ) .1 Observations of Polarization Drift A common way to express the SOP of monochromatic light is in terms of the Stokes parameters. e Following the convention of Born and Wolf [5].10) The results are the equations of a sphere of radius s0 .5) (2.1.3) (2.8) (2. a stereographic projection of the SOP ellipse onto a sphere [23] yields: s1 = s0 cos (2χ) cos (2ψ) s2 = s0 cos (2χ) sin (2ψ) s3 = s0 sin (2χ) . 0).2) vector. Finally. 0 1 2 3 (2.7) (2.2 and δ1. (2.2 are the amplitudes and phases.

2. a General Photonics POD-101A in-line polarimeter was used. In this case. All the states in between are generically referred to as elliptically polarized. such that when the glass is heated and drawn into a ﬁber.1 The Polarization-Maintaining OFIS Design. thus creating a preferential orientation for the electric ﬁeld to propagate. whose accompanying software was able to collect all four Stokes parameters at a minimum of 1. respectively. each with diﬀering characteristics . which typically operates by splitting the input beam into multiple paths and passing each through a diﬀerent series of polarizers and retarders whose output intensities are proportional to the Stokes parameters [17]. This was done with a polarimeter. The results (Figure 2.9 left circular polarizations occur at the north and south poles. Construction and Deployment Polarization can be maintained in an optical ﬁber by inducing a birefrin- gence proﬁle in the plane perpendicular to the axis of the ﬁber. Measuring the SOP of each OFIS arm was the ﬁrst step to determine what could be done to solve the polarization drift problem.2 2. Since the interference ellipse size changed most signiﬁcantly with temperature. with their handedness determined by the upper and lower hemispheres. Since the SOP perturbations are most likely caused by stress-induced birefringence imparted during thermal expansion and contraction.1) show that the SOP in the single wrap arm varies considerably. and given that the OFIS functions based on the strain diﬀerential between the single and double wrap ﬁbers. There are three main types of PM ﬁber (Figure 2. SOP measurements were taken on each arm of a 10 meter ground-lying OFIS from 4:30 to 8:30 PM local time. it follows that the SOP variation would be greater for the single wrap. In PM ﬁber (not to be confused with polarizing ﬁber) this is accomplished by including a material with a diﬀerent coeﬃcient of thermal expansion. and the steepest temperature gradients occur at sunrise and sunset.2).2.000 samples per second for a maximum of four hours. thermal stress-induced birefringence occurs upon cooling. especially in comparison to the double wrap arm.

each was run independently for several weeks. resulting in no downtime due to polarization fading.1 Hz continuous wave at approximately 60 Pascals (peak-to-peak). 2. was connected to the internal volume of each OFIS tube running a 0. The three sensors were deployed on the ground at the Camp Elliott Field Station (Figure 2.3). which can be calculated from the parameters determined by the Femtometer . Each OFIS consisted of two 30 meter.2. The availability of a NP Photonics “Rock” 1.1 meter diameter plastic perforated tube.2 Results: Ellipse Stability and Optical Phase Noise The PM ﬁber provided very good ellipse stability. Finally. consisting of a commercial loudspeaker hermetically sealed to a plastic plate and a calibrated Setra 265 pressure transducer. Temperature data was collected between the OFIS tube and insulation and on the outside of the perforated tube with two Onset HOBO sensors. the sensors were wrapped with ﬁberglass insulation and inserted into 0. An OFIS calibration system (contained in the gray box in the lower left-hand corner of Figure 2. it was decided to deploy three unequal arm MachZehnder OFISs. Since there was only enough hardware to run one sensor at a time.10 such as birefringence contrast and uniformity over the length of the ﬁber.533 nanometer wavelength laser source with a coherence length of several hundred kilometers (spectral linewidth of less than 1 kHz) allowed the second arm to be only 3 meters. 900 micron jacketed ﬁbers whose helical wraps were 5.5 cm apart.3) located about 20 kilometers east of the SIO campus. which was deployed in an attached instrument box. Only Corning PANDA and Fibercore HB1500 Bow-tie types were used for the prototype PM OFISs since PANDA is the most widely available and Bow-tie had the highest birefringence contrast making it less likely to experience loss of PM characteristics under helical winding. which consist of a thermocouple and data logger. resulting in total lengths of approximately 18 meters. Given that PM ﬁber was over one hundred times more expensive than the typical Corning SMF-28 ﬁber. This is quantiﬁed by the size of the interference ellipse.

Figure 2. in addition to other problems such as the large amounts of scatter in the ﬁtted ellipse and oscillations in the ellipse area caused by the calibration signal were not investigated due to the results of the following section. but are not in danger of approaching zero. and also has a non-zero probability of collapse. and shows a residual noise of about 5 to 6 Hz not found in a Br¨ el and Kjær (B&K) 4193 reference microphone u located 10 meters west of the OFISs. Investigating the causes of this residual noise and/or constructing an identical null interferometer. The Jones calculus is a common way to mathematically express how fully .3 Faraday Mirrors and the Michelson OFIS Compensating for unwanted birefringence in the path of an interferometer by creating a phase conjugate (or time reversed) beam was ﬁrst investigated by Martinelli in 1989 [27]. or Faraday mirror (FM).5 is an example of this diﬀerencing on the Bow-tie OFIS. which has come to be known as a Faraday rotator mirror. the null signal needed to be scaled by a constant trivially determined by least squares.11 system (Appendix B). 2. the use of PM ﬁber does solve the polarization fading issue. and successfully implemented shortly thereafter in a ﬁber optic Michelson interferometer by Kersey [24]. To understand why this particular combination works. one must be familiar with the Jones calculus for modeling optical components. Hence. Shortly after the initial setup it became apparent that ﬂuctuations in the laser wavelength caused by the unequal arms were contributing a signiﬁcant amount of noise to the optical phase data. This was done by using a Faraday rotator followed by a mirror. This is in contrast to the SMF-28 OFIS whose ellipse area variance is over quadruple that of the PM OFISs. Histograms of two days of data show that both PM OFIS ellipse areas show some variability. This was partially remedied by subtracting the signal from a second unequal arm “null” interferometer consisting of a 30 meter and 1 meter SMF-28 ﬁber housed in the adjacent building. Since the lengths of the null interferometer arms were not identical to the OFISs.

Similar to Section 2. The incident beam is transformed into the output beam (denoted by a prime) : E ′ = AE. An intuitive example of such a matrix is that of a horizontal linear polarizer: AHLP = 1 0 0 0 . rotators.12 polarized light behaves as it passes through optical components such as polarizers. (2. and retarders or wave plates. an electromagnetic wave can be expressed as a complex vector [18]: E= Ex Ey = a1 ei(τ +δ1 ) a2 ei(τ +δ2 ) (2.1. horizontally polarized light implies a2 = 0. and normalization reduces the Jones vector to: Eh = 1 0 whereas right circularly polarized light implies that the component amplitudes are equal (a1 = a2 ). .11) For example..12) to operate on a vector containing the Stokes parameters in the case of partially or and can logically be extended to a system of n components where A = An An−1 . Using this formalism. but this added generality is not necessary in this context).A2 A1 . and the phase of Ey is shifted by −π/2 with respect to Ex (δ1 = 0 and δ2 = −π/2): 1 ERCP = √ 2 1 −i .. optical components are elementary matrices (A) that operate on the Jones vectors (note that it is also common to use 4 × 4 Mueller matrices unpolarized light.

d = det ⇀ . and the R harpoon indicates the forward propagating direction. while a beam propagating through the same retarder in the reverse direction: ↼=1 R d b∗ a −b a∗ . consider modeling the optical ﬁber by the Jones matrix for a general elliptical retarder (following the convention of Kersey [24]. (2.13) where a and b denote the complex birefringence of the ﬁber. √ where renormalization has eliminated the factor of 1/ 2. and the total eﬀect of the rotator (AROT (θ)) and mirror (AM ) system is: AF M = AM AROT = −1 0 0 1 0 −1 −1 0 π AROT 4 cos − sin π 4 π 4 π 4 sin cos π 4 π 4 cos − sin π 4 π 4 sin cos π 4 π 4 = (2. Hecht [18] provides many more examples of Jones matrices for common optical components. To show how the Faraday mirror operates.13 which applied to the previous Jones vector: E ′ = AHLP ERCP 1 = √ 2 = 1 0 1 0 0 0 1 −i = Eh . but disregarding losses): ⇀=1 R d a −b∗ b a∗ (2.15) .14) The Faraday rotator imparts a π/4 rotation before and after the reﬂection.

the FM equipped Michelson OFISs have worked very well (Figure 2. To apply FM to the OFIS required the use of a Michelson rather than a Mach-Zehnder interferometer (Figure 2. Fortunately the use of ﬁber connectors with angled faces. Back reﬂections of the fringe signal to the laser were eliminated by the a built-in isolator. are designed to reduce such reﬂections.14 Therefore. which resulted in no detectable fringe signal on adjacent sensors. there was concern that the backward reﬂecting fringe signal would also propagate in the neighboring sensors thereby optically coupling each sensor. Finally. This was a trivial and very inexpensive modiﬁcation.16) which is independent of the birefringence of the ﬁber. . the Jones matrix for the ﬁber and FM is: A = ↼ FM⇀ = RA R 0 −1 −1 0 . in the case where one laser is split into multiple paths to illuminate multiple OFISs. (2. or optical diode. or FC/APC connectors.6). However. and are being used for all current and future sensors. which simultaneously doubled the pressure sensitivity since the laser is reﬂected back through the OFIS ﬁbers. provided that such perturbations do not occur faster than the time of ﬂight of the beam within the system.4). Qualitatively this implies that the perturbations in the SOP by the ﬁber are “undone” by the reﬂected beam.

one rotated π in azimuth with respect to the other. 15 .Figure 2. and the double wrap arm by dark squares. The evolution of e the SOP in the single wrap arm is shown by circles.1: Two views of the same Poincar´ sphere.

1). The center gray circle denotes the core of the ﬁber. and a standard OFIS (not used for the present study.2: Three types of PM ﬁbers: PANDA. Figure 2. The fast axis is along the horizontal dimension in this ﬁgure. however. PANDA. .16 Figure 2. was used for the experiment in section 2.3: Four Camp Elliott OFISs. whereas the black shapes represent the stressing materials (not to scale). from left to right: Bow-tie. Bow-tie and elliptical-clad. Each sensor was connected to an instrument box by a cylindrical PVC adapter arm. SMF28.

4: Ellipse area probability density estimates for the PANDA. Bow-tie.17 Figure 2. . shown with a FM equipped Michelson OFIS. Note that the SMF-28 OFIS consisted of a single wrap. however. and SMF-28 OFISs. a standard OFIS ellipse behaves similarly with a non-trivial probability of having an area at or near zero.

Both signal amplitudes have been converted to Pascals. but the PM OFIS signal appears slightly lower most likely due to its directional response [34]. .5: The top time series shows the raw optical phase overlain with the scaled null interferometer signal.18 Figure 2. while the lower shows their diﬀerence along with a calibrated B&K 4193 precision infrasound microphone.

6: The Michelson OFIS schematic. 19 . showing that after traveling the length of the sensor. the phase conjugate beam from both arms are reﬂected back to the original splitter. It is at this point that the interference fringes are formed and sent to the photodetector.Figure 2.

Chapter 3 Temperature-Induced Pressure Sensitivity Change Pressure sensitivity of the OFIS is quantiﬁed by the optical phase change in radians. Since the IMS requires all microbarometer sensitivities vary less than ﬁve percent annually. ground lying OFIS changed by more than a factor of two over a range of about 20 degrees Celsius [30]. (rad·P a−1 ·m−1 ).5 meter long heating and cool- ing chamber was constructed from a 5 centimeter diameter copper pipe. helically wrapped with 6 millimeter diameter copper tubing with a wrap spacing of 5 centimeters that was periodically soldered to the larger pipe (Figure 3. it is essential to understand and take steps to mitigate this behavior. This unit is capable of automatically cycling between two temperature 20 . This unit was enclosed in 5 centimeters of foam insulation and end caps constructed to yield a uniform temperature along its length. a 1.1 Laboratory Observations To conﬁrm previous ﬁeld measurements. The implementation of a continuous calibration system revealed 3. per unit pressure. and normalized by the total length of the sensor that the pressure sensitivity of a 30 meter. The temperature was controlled by pumping a water and ethylene glycol mixture through the helical tubing using a Solid State Cooling Systems 400 Watt ThermoCUBE thermoelectric temperature control system.1).

The average trend between all cycles for both OFISs was found to be: 1 dΦ = (10 ± 1) × 10−3 ◦ C−1 . Therefore. both exhibited near linear behavior (an example of which can be found in Figure 3. but does not allow for the control of the time it takes to obtain a given set point. the temperature inside of the OFIS was measured using a calibrated Fluke 80TK thermistor module. This was a prototype system for the unit deployed in the PM OFIS experiments. Therefore. the instantaneous sensitivity was calculated by dividing the Setra envelope by the OFIS envelope and the length of the sensor. The pressure sensitivity of two 1. Finally.3). two Lytron CP15G05 aluminum heat exchangers submerged in 20 liters of water were added to increase the thermal inertia of the system. After converting the Setra signal to Pascals. whose output voltage was logged by the Femtometer system (Figure 3. This was used to pressurize the inner volume of the OFIS tube as well as a Setra 265 calibrated pressure transducer. A calibration system was constructed to continuously determine the sensitivity of the OFIS inside the heating and cooling chamber.4 meter OFISs was measured continuously over three. which ran a two second sinusoid at a 2 Volt peak-to-peak amplitude. eight-hour heating and cooling cycles from approximately 5◦ to 50◦ Celsius. While there were no signiﬁcant diﬀerences in the nominal sensitivities or temperature dependences between the warm and cold wrapped OFISs. thereby reducing the cycle time.2). Φ (T ) dT . To examine the potential signiﬁcance of the environmental conditions under which the ﬁber was wrapped about the silicone tube. The subwoofer was driven by a Tenma TG120 function generator and a Techron 5515 DC-coupled ampliﬁer. both the OFIS and Setra data were narrowly bandpass ﬁltered about the calibration frequency and their envelope functions obtained by computing the magnitude of the signal with its Hilbert transform. which also consisted of a subwoofer glued to a back volume whose outlet was a standard barbed pipe ﬁtting to accept ﬂexible tubing. one OFIS was wrapped at room temperature (about 20◦ C) while the other in a walk-in cooler at about 8◦ C.21 set points.

22 relative to 20◦ C. and E the elastic modulus. however.2 Temperature Dependence of the Elastic Modulus Since Hooke’s law states that the stress in a linear elastic material is di- rectly proportional to the strain and the material’s elastic (aka: Young’s) modulus. ǫ is the tensile strain. To show this. consider the recent results from four. therefore lowering the sensitivity. 3. Φ (T ) dT These two results suggest that the temperature dependence of the pressure sensitivity may also depend on the total length of the OFIS. at low temperatures one would expect the silicone tube to become more rigid. more data is required to establish a formal trend. For example. For comparison. Since the tensile stress is the force per unit of cross-sectional area (A) of the tube. and (courtesy of Newton’s second law) the force is the applied mass (m) and the . where σ is the tensile stress. it could be supposed that the observed changes in OFIS pressure sensitivity would be inversely proportional to the temperature changes in the elastic modulus of the silicone tubing. The elastic modulus of the silicone tube can be easily measured as the ratio of the tensile stress to tensile strain by hanging a length of the tubing from one end and measuring the extension caused by a known mass suspended from the other end. 30 meter buried OFISs at the Camp Elliott Field station: 1 dΦ = (30 ± 7) × 10−3 ◦ C−1 . requiring a greater amount of pressure to strain the tubing (as monitored by the optical ﬁber) by the same amount. It was this reasoning that led to the measurement of the temperature dependence of the silicone tubing’s elastic modulus. consider the following form of Hooke’s law: σ = Eǫ. corresponding to an increase in the elastic modulus.

At each temperature. gL (3.0% change shown in the laboratory measurements of . but mounted vertically and without the added thermal inertia.06) × 10−4 ◦ C−1 . respectively. therefore 5τ ≈ 6. In order to account for the thermal changes in the length of the silicone tube.08) × 10−3 ◦ C−1 . 280 ± 220 seconds. the slope of a linear least squares ﬁt to the applied mass versus change in length is proportional to the elastic modulus. and was found to be (relative to 20◦ C): 1 dL = (4.8 hours. The results reveal a very modest change in the elastic modulus with temperature. and the axial strain. E (T ) dT relative to 3. The temperature dependence of the elastic modulus was very linear with a best ﬁt slope of: 1 dE = (−2. Repeated heating and cooling measurements of the temperature inside and outside the silicone tube revealed that the time needed for the temperature of the silicone to equilibrate was approximately two hours: the e-folding time was found to be τ = 1. a series of ten masses were used to stretch the silicone tube by a distance measured with a traveling microscope.00 ± 0. the elastic modulus for each temperature was computed from the slope of the applied mass versus extension. ǫ = ∆L/L. or about 1. Therefore.48 ± 0.01MP a at 20◦ C. is simply the change in length over the total length.5 meter long heating and cooling chamber.2% change per degree Celsius compared to the 1. The elastic modulus of the silicone tubing was measured for ten diﬀerent temperatures from 5◦ C to 50◦ C using the same 1. L (T ) dT As shown earlier.23 acceleration due to local gravity (g): mg = Eǫ A π (b2 − a2 ) E m = ∆L. This amounts to a 0. All measurements were repeated four times: twice for increasing and decreasing temperature. 400 seconds.1) where a and b are the inner and outer radii of the tubing. the coeﬃcient of linear expansion was measured using the same setup.06 ± 0.

and therefore further investigation is needed to understand this discrepancy. .24 the previous section.

25 Figure 3.1: The 1.5 meter long OFIS heating and cooling chamber connected to the thermoelectric temperature control system. . This unit was subsequently enclosed by insulation and mounted horizontally for the calibration experiments and vertically for the elastic modulus and linear expansion experiments.

Figure 3. The two heat exchangers (plumbed in series) submerged in water provided suﬃcient thermal inertia to slow the cycle time to three hours.2: Schematic of the experimental setup for measuring the temperature dependence of the pressure sensitivity. Temperature data was taken from the thermistor located near the center of the OFIS interior. 26 .

Each error bar corresponds to the standard error within each 1◦ C temperature bin. 27 . the linear ﬁt was performed on the unbinned data.Figure 3.3: Results for one 24 hour period. however. or three-cycles of temperature and sensitivity measurements.

the only two non-zero elements of the stress tensor. canceling dθ. pa and pb . and dividing by dr. Balancing the radial and tangential (aka: “hoop”) forces for an element of unit thickness leads to: (σr + dσr ) (r + dr) dθ = σr rdθ + σθ dθdr where σr and σθ are the radial and tangential stresses. which in this case are rσr . The ﬁrst step in this direction was to model the OFIS as a single loop of ﬁber wrapped under tension about a silicone rubber tube as a pre-stressed. double-walled thick cylinder. the basis of which is a well known problem in elasticity dating back to Lam´ and Clapeyron in 1833 [26]. consider a thick- walled cylinder of inner radius a and outer radius b subjected to both internal and external pressure.Chapter 4 Analytic Opto-Mechanical Model It was the inability for the temperature dependence of the elastic modulus to explain the observed temperature dependent sensitivity that lead to exploring a mechanical model for the OFIS. θ. Derivations e similar to the one that follows can be found in nearly any introductory text on elasticity or strength of materials. and therefore disregarding drdσr . z).1 Lam´ Equations for a Thick-Walled Cylinder e Assuming the cylindrical coordinate convention (r. respectively (Figure 4. 4. rdσr . By assuming that drdσr ≪ 28 .1). σr dr.

4) Rearranging (4. E E (4.4).7) (4.29 the equation of equilibrium becomes: r dσr + σr − σθ = 0.3) (4. it follows that the two elements of the stress tensor are: σr = σθ = E (ǫr + νǫθ ) 1 − ν2 (4. the stress-strain relationship can be expressed as: ǫij = 1+ν ν σij − σαα δij .1) For an isotropic elastic material with Poisson’s ratio ν and elastic modulus E. and the repeated indices indicate summation. .5) (4.9) with respect to r: dσr E d2 u r = +ν dr 1 − ν 2 dr 2 1 dur ur − 2 r dr r . dr (4. the radial and tangential strains are: ǫr = ǫθ 1 (σr − νσθ ) E 1 = (σθ − νσr ) . 1 − ν2 As proven by Housner and Vreeland [21].10) By diﬀerentiating (4.5. (4.9) (4.3. the two strain components in cylindrical coordinates can be expressed in terms of the radial displacement ur : ǫr = ǫθ dur dr ur . Without shear or axial stresses. = r (4. δij is the Kronecker delta.6) E (ǫθ + νǫr ) .8) These results can be used to rewrite equations (4. E (4.2) where ǫij are the elements of the strain tensor.6): σr = σθ = E 1 − ν2 E 1 − ν2 dur ur +ν dr r ur dur +ν r dr .

This is the general solution for the radial displacement of a thick-walled cylinder.30 and using (4. b2 − a2 (1 + ν) a2 b2 (pa − pb ) . r (4.1) can be expressed solely in terms of the radial displacement ur : d 1 d (rur ) = 0. dr r dr Integrating this separable equation twice yields: ur (r) = C1 r + C2 . E (b2 − a2 ) . but it is required since σr is a function of ur and dur /dr.15) E C 1−ν 1 and B = E C.16) (4.12.18) (4.11) where C1 and C2 are constants of integration.17) (4. however. it must be re-written in terms of stress to apply the boundary conditions (assuming the convention where compressive stresses are negative): σr (r = a) = −pa σr (r = b) = −pb . (4.13) It may appear to be unnecessary to express the stress in terms of the radial displacement only to solve for the stress.10). 1+ν 2 the general solutions for the a2 b2 (pa − pb ) .19) a2 pa − b2 pb b2 − a2 (4.13) determines A and B as: A = B = and: C1 = C2 = (1 − ν) (a2 pa − b2 pb ) E (b2 − a2 ) (4.9. Deﬁning A = stresses are: B r2 B σθ (r) = A + 2 . r σr (r) = A − Application of the boundary conditions (4. the equation of equilibrium (4.12) (4.14) (4.

however. The radial displacements of the two cylinders are: ur1 (r) = (1 − ν1 ) (a2 pa − b2 pb ) (1 + ν1 ) a2 b2 (pa − pb ) 1 r.2 Pre-Stressed. Texts on contact mechanics often use δ as the “distance of mutual approach” [15].46) of Faupel [14]. b {E1 (b2 − a2 ) [b2 + c2 + ν2 (c2 − b2 )] + E2 (c2 − b2 ) [a2 + b2 − ν1 (b2 − a2 )]} (4.18. + E2 (c2 − b2 ) r E2 (c2 − b2 ) (4.23) Note that the case of δ = 0 states that the radial displacement is continuous across the interface. subjected to an external pressure pc . c] .21) (1 + ν2 ) b2 c2 (pb − pc ) 1 (1 − ν2 ) (b2 pb − c2 pc ) r.11) and (4. the desired form for the radial displacement can be found with (4.20) which is equation (10. This leads to the constraint: pb = 2E1 bc2 (b2 − a2 ) pc + E2 (c2 − b2 ) [2a2 bpa + δE1 (b2 − a2 )] . an “interference distance” δ can be deﬁned as: δ = ur2 (r = b) − ur1 (r = b) .22) for r ∈ [b. with elastic modulus E2 and Poisson’s ratio ν2 . (4. in this context δ is deﬁned as the indentation. b] ur2 (r) = (4. Double-Walled Thick Cylinder This section expands upon the previous one by adding a second cylinder about the ﬁrst with an outer radius c. + E1 (b2 − a2 ) r E1 (b2 − a2 ) for r ∈ [a. It is essential to understand that δ is the combination of the shrinkage of the inner radius of the second cylinder (ur2 (b) > 0) and the outer radius of the ﬁrst (ur1 (b) < 0). + E (b2 − a2 ) r E (b2 − a2 ) (4. Since the second cylinder is force-ﬁt about the ﬁrst (this is analogous to either a shrink-ﬁt or press-ﬁt construction).24) .31 Finally.19): ur (r) = (1 + ν) a2 b2 (pa − pb ) 1 (1 − ν) (a2 pa − b2 pb ) r. 4.

The change in phase is then: ∆φ = β∆L + L∆β = ∆φ1 + ∆φ2 . Finally. (4.32-34): ∆φ Ψ = ǫ πnL 2 − n2 [p12 − νc (p11 + p12 )] .25) λ where β is called the propagation constant. 4. 3]. 32.51. λ is the wavelength of the light. φ.3 as [6]: The Strain-Optic Eﬀect The optical phase. the formula for the total optical phase diﬀerence per unit strain can be formed from equations (4. the change in the index of refraction due to axial strain alone can be expressed as [3]: πn3 L [p12 − νc (p11 + p12 )] ǫ. The second term describes the ∆φ1 = change in the index of refraction as well as the waveguide mode dispersion caused by the reduction in diameter of the ﬁber core under axial stress.28) λ where νc is the Poisson’s ratio of the ﬁber core and p11 and p12 are photoelastic ∆φ2 = − constants dependent upon the material used.26) The ﬁrst term represents the optical phase diﬀerence due the axial stretching of 2πnL ǫ. Making the frequent assumption that the core diameter remains constant [6. n is the index of refraction of the ﬁber φ = βL = core.27) λ where ǫ = ∆L/L is the axial strain in the ﬁber. and L is the length of the ﬁber. of light passing through an optical ﬁber can be written 2πn L (4.2) of Faupel [14].32 (Note that setting pa = pc = 0 results in equations (10.) However. to model the second cylinder as a loop of optical ﬁber the radial displacement ur2 must be related to optical path length diﬀerence. 20. (4.29) = λ the ﬁber: . (4. which is governed by the strain-optic eﬀect. (4.

270. Therefore. the theoretical value was calculated as: Ψtheoretical = 6.121 p12 = 0. followed by suggestions for future model improvements.107 × 106 rad · strain−1 .115 meter length of 900 micron jacketed Corning SMF-28 ﬁber using a pair of ﬁber vices.29). and equation (4. a precision translation stage and a 1. an empirical measurement of the optical phase diﬀerence per unit strain was performed on a 1.17 p11 = 0.4 Single Fiber about a Thick Cylinder Subjected to Internal Pressure In this section. which is in agreement with the measured value.009) × 106 rad · strain−1 .456 νc = 0. the results of the previous two sections are used to model a single loop of optical ﬁber wrapped under tension about a silicone rubber tube as a pre-stressed.111 ± 0. The result was found to be: Ψempirical = (6. 4. .310 nanometer wavelength laser.33 While this theory has been successfully applied innumerable times for unjacketed ﬁber. Using published values for the constants [20]: n = 1. most relevantly in the case of a hydrophone whose design is similar to the OFIS [32]. The strain-optic behavior is included to better model the outer layer of the cylinder as an optical ﬁber and to convert its strain to an optical phase. it is not clear how jacketing aﬀects the stresses resolved on the core. double-walled thick cylinder. an estimate of the theoretical pressure sensitivity will be made and compared to experimental data. After simplifying the previous results.

b] + 2E1 (b2 − a2 ) r 2E1 (b2 − a2 ) (1 + ν2 ) b2 c2 pb 1 (1 − ν2 ) b2 pb r. and dis- carding the external pressure pc .31. for r ∈ [a. For example.32) The strain in the optical ﬁber at r = c simpliﬁes with the use of (4. b {2E1 (b2 − a2 ) [b2 + c2 + ν2 (c2 − b2 )] + E2 (3a2 + b2 ) (c2 − b2 )} (4. To calculate the pressure sensitivity of the single loop.8) and (4.34 4.34) where L = 2πc.1 Theoretical Results Assuming that the tube material is incompressible (ν1 = 1/2). the material parameters of the ﬁber must be known.22) and interface pressure (4.30) (4.33) and the sensitivity: Φ = ǫθ2 Ψ pa 8π 2 nbc [2a2 bpa + δE1 (b2 − a2 )] {2 − n2 [p12 − νc (p11 + p12 )]} . which is coated again up to 900 microns with polyvinyl .4. for r ∈ [b.31) 2E2 (c2 − b2 ) [2a2 bpa + δE1 (b2 − a2 )] . 2E1 (b2 − a2 ) [b2 + c2 + ν2 (c2 − b2 )] + E2 (3a2 + b2 ) (c2 − b2 ) (4. This is coated with UV cured polyacrylate to a total diameter of 250 microns. weighted by their respective cross-sectional areas. c] + E2 (c2 − b2 ) r E2 (c2 − b2 ) (4. the radial displacements (4. As a ﬁrst estimate.24) simplify to: ur1 (r) = ur2 (r) = pb = (a2 pa − b2 pb ) 3a2 b2 (pa − pb ) 1 r. the elastic modulus and Poisson’s ratio of the jacketed ﬁber were determined by averaging the parameters of the constitutive materials.21.32): ǫθ2 = = ur2 (r = c) c 4b [2a2 bpa + δE1 (b2 − a2 )] . the core and cladding of the ﬁber is 125 microns in diameter and composed of fused silica. = λpa {2E1 (b2 − a2 ) [b2 + c2 + ν2 (c2 − b2 )] + E2 (3a2 + b2 ) (c2 − b2 )} (4.

06. 20]. the complexity of . the average values were found to be: E2 = 4.39 ± 0.48 ± 0.01 MP a a = 0. Using published values for these materials in Table 4.7 GP a ν2 = 0.5-4.0 ± 0.34) is to estimate the interference distance.33-0.0002 m b = 0. The ﬁnal step to predict the sensitivity of the single loop from (4.0091 ± 0.35-0.6 ± 0.0118 ± 0. or indentation δ. Assuming no pre-stressing (δ = 0): Φtheoretical (δ = 0) = (1.35) 2 2 1 − ν2 1 − ν1 + E1 E2 (4. −1 (4.1 [25.17 0. however. The analytical evaluation of the indentation about an elastic cylinder due to an elastic ring of circular cross-section appears to be an unsolved problem.3) × 10−3 rad · P a−1 .0001 m c = 0.35 Table 4.1 Poisson’s Ratio ν 0.7 3.2 2. Unfortunately.1: Material Parameters for 900 micron SMF-28 Fiber Material Fused Silica Polyacrylate PVC Elastic Modulus E (GP a) 71.46 chloride (PVC).36) . another estimate of δ may be arrived at using the results of an elastic spherical indenter in an elastic half-space [15]: δ3 = where: E∗ = 1 R 3T 4E ∗ 2 the case where the ring is rigid (ie: E2 → ∞) [2]. except in its solution prevents it from being useful in this thesis.0126 ± 0.42 0. The measured values of the elastic modulus of the silicone from the previous experiments (taken at 20◦ C) and the three radii were measured as: E1 = 3.0001 m.

03 Newtons and the PVC jacket welded over a length of 2-3 millimeters. however. The loop of ﬁber was wrapped at a tension of 1.4.4 ± 0. it may be useful as a free parameter.4) × 10−4 m. After pressurizing the interior of the tube at ﬁve diﬀerent amplitudes at a frequency of 0. 4.62 ± 0.62 ± 0. While the experiment has been repeated on several other loops of varying tension with similar results. R = (c − b) /2 is Using a value of T = 1.2 Experimental Results An experiment was conducted on a single loop of ﬁber about a silicone tube. and the resulting sensitivity: Φtheoretical δ = 5.5 Hz. and T is the indenter load. the radius of the ﬁber.9 ± 0.36 is the combined elastic modulus of the silicone tube and ﬁber.4 × 10−4 = 55 ± 16 rad · P a−1 .4. A linear ﬁt to the optical phase diﬀerence versus the applied pressure resulted in a sensitivity of: Φexperimental = (11.1 ± 0. This illustrates the remarkable dependence the pressure sensitivity on δ and the necessity for an accurate estimate. and its sensitivity measured with a similar calibration system to all the previous experiments. the magnitudes of the calibrated pressure transducer and optical phase were determined using least squares. double-walled thick cylinder can be useful to understand the temperature dependence of the pressure sensitivity . it is clearly out of agreement with either of the above predicted values.03 Newtons: δ = (5.34) to ﬁnd the indentation yields: δ = (1. which is the tension of the ﬁber. Inverting (4.3 Suggestions for Future Work It is diﬃcult to assess whether the pre-stressed. 4.6) × 10−3 rad · P a−1 .3) × 10−7 m.

37 of the OFIS. However. most reduce the problem to a bulk cylinder whose material properties are found by what amounts to an elaborate weighted-averaging scheme. . The other extreme is the highly complex contact problem as in Reference [2]. which could be addressed using a numerical method such as ﬁnite diﬀerences or ﬁnite elements. Inverting for δ could be useful for obtaining an initial estimate of the pressure sensitivity that could be held constant to examine the eﬀects of temperature change on the other model parameters. this approach would be highly problematic since the formulations of δ depend upon these same parameters. the strong dependence upon the contact conditions shown above indicate that a realistic model for the OFIS should focus on the temperature dependence of the contact strain ﬁeld. The striking dependence on the interference distance. and the diﬃculty in obtaining its estimate severely limits the predictive power of the model. However. While there is some literature on helically wrapped composite tubes (Evans [13] provides a review). In this way the composite nature of the ﬁber could also be taken into consideration and better predict the strains resolved on the ﬁber’s core.

both highlighted in gray.Figure 4. subjected to internal pressure pa and external pressure pb .1: The left diagram shows a thick-walled cylinder with inner radius a and outer radius b. The right side shows the forces on the area segment from the left diagram. 38 .

What follows is the description of the components needed to obtain an estimate of optical phase for the current Michelson interferometer OFIS design as shown in Figure 2. The modulated light is sent through both 39 . thereby stretching the ﬁber a fraction of a wavelength. as mentioned in Chapter 1. This will explain why the ﬁber path length is modulated.Appendix A Optical Path Length Modulation and the Femtometer System While the primary focus of this thesis is the mechanical and optical behavior of the OFIS as an interferometric pressure transducer. which interface with one end of a device termed the modulator box (Figure A. At the other end of the modulator box is the laser input for illuminating the OFIS. since its area was used as a metric for assessing the uptime of the sensor in Chapter 2. Furthermore. and where interference ellipse originates. it is essential to note that the following technique is not unique to the OFIS. but can be applied to any Mach-Zehnder or Michelson type interferometer [33]. A 156 kHz voltage signal supplied to the piezo causes its diameter to change by a small amount. As previously stated.6. The piezo is split and wired such that one ﬁber is strained π out of phase with the other. it is important to understand how an estimate of the optical phase diﬀerence is calculated by measuring the intensity of the interference fringes. there are two helical wraps of ﬁber such that there are four connectors at the end of an OFIS.1). which is split equally among two paths that wrap about a piezoelectric cylinder (piezo).

the purpose of the modulator box is to supply the illumination to each OFIS arm. y) data pairs are sampled at a rate of either 100 kHz or 50 kHz. However. and shape to obtain an estimate of the optical phase. and return the modulated fringe signal. modulate the optical path length. if the fringe intensity is a maximum. is fed into the lock-in ampliﬁer to demodulate the fringe signal. y) location on the ellipse. The ﬁrst module contains the oscillator and ampliﬁer to drive the piezo. A collection of electronics known as a “Hanada rack” (Figure A. x. conditioning them for input to the data acquisition system. producing an output.2) bridges the gap between the modulator box and the data acquisition system.40 arms where it is reﬂected oﬀ the Faraday mirrors and recombined at the splitter to form interference fringes. the plot of a changing fringe signal versus its electronically generated derivative traces out an ellipse whose instantaneous optical phase diﬀerence depends upon the (x. For example. while the last consists of an ampliﬁed photodetector to receive the modulated fringe signal. but it is the data acquisition and signal processing system called the Femtometer that yields the estimate. Recall that the point of modulating the optical path length is to discriminate between contractive and expansive fringes that respectively correspond to positive and negative external pressure acting on the OFIS. corresponding to perfect constructive interference. In between these two units is a phase-sensitive detector. The previously described components are essential for preparing the original fringe signal for estimating the optical phase. it is impossible to determine if a subsequent decrease in intensity is the result of the optical path length increasing or decreasing. or lock-in ampliﬁer. and a user determined number of pairs are stored in a circular buﬀer to obtain least squares . These (x. Therefore. size. the next step is to obtain estimates of the ellipse location. proportional to the spatial derivative of the original fringe signal [33]. The output of the photodetector. There is a secondary ampliﬁer module to adjust the x and y voltage amplitudes and their DC oﬀsets. Therefore. y.

41 estimates of the constants in the following parametric equations [33]: x = x0 + a sin (p + p0 ) y = y0 + b cos (p) . the ellipse parameters are updated and the optical phase is determined from the most recent (x.2) for p and taking the arctangent of their ratio. y) pair by solving both equations (A1. p0 is the ellipse shape factor. a and b are the x and y ellipse amplitudes. y) plane. it is possible for the subsequent (x. . the phase and ellipse parameter data are decimated and archived following the CSS 3. and p is the optical phase diﬀerence.1) (A.0 convention. As new data replaces the old in the buﬀer. (A. It is important to note that under extreme conditions.2) where x0 and y0 are DC oﬀsets on the (x. With the previous limitation properly accounted for. y) point to advance more than 2π around the ellipse despite the high sample rate.

42 Figure A. y) inputs used to compute p. While there are four connections to the OFIS. Figure A. Note that the single and double wrap arms can be interchanged.2: Block diagram of the fringe modulation and Femtometer system for estimating the optical phase diﬀerence between two OFIS arms. showing the OFIS connections on the left.1: Schematic of the modulator box. The Femtometer output is the ﬁve ellipse parameters (usually at a much lower sample rate) along with the (x. . two send light into the OFIS and receives the reﬂected light from the FM found at the remaining two connections.

b and p0 ) put out by the Femtometer. (B.4) Isolating the square root.3) Splitting the argument of the cosine with cos (θ ± φ) = cos (θ) cos (φ)∓sin (θ) sin (φ) √ and using the inverse identity cos [arcsin (θ)] = 1 − θ2 . a (B. therefore it is beneﬁcial to ﬁnd how these are related to the ﬁve ellipse parameters (x0 .5) 43 . Solving both equations for p and taking the cosine of both sides: y x − p0 . [33] to canonical form.Appendix B Derivation of Ellipse Parameters Standard formulas for ellipse properties such as area. foci. et al.1) (B. y0 . which are: x = x0 + a sin (p + p0 ) y = y0 + b cos (p) . One approach is to reduce the parametric equations (1. eccentricity. (B.2) of Zumberge. we arrive at: y = b 1− x a 2 cos (p0 ) + x sin (p0 ) . are usually given in terms of the semimajor and semiminor axes.2) It can immediately be assumed that x0 = y0 = 0 since the ellipse properties should not depend on its position. etc. a. squaring both sides and collecting like terms results in a quadratic in standard form: Ax2 + Bxy + Cy 2 − D = 0. = cos arcsin b a (B.

leading to a deﬁnition of ψ: tan (2ψ) = 2ab sin (p0 ) B = .17) (B. where: A′ = A cos2 (ψ) + B cos (ψ) sin (ψ) + C sin2 (ψ) B ′ = −2A cos (ψ) sin (ψ) + B cos2 (ψ) − sin2 (ψ) +2C cos (ψ) sin (ψ) C ′ = A sin2 (ψ) − B cos (ψ) sin (ψ) + C cos2 (ψ) axis.9) The next step is to rotate the quadratic to its principle axes by the transformation (Figure B.14) (B. (B.6) (B. there always exists a ψ ∈ − π .11) and D remains unchanged from (B. Expanding and collecting like terms: A′ ξ 2 + B ′ ξη + C ′ η 2 − D = 0. (B.18) . A−C a2 − b2 (B.16) (B.12) (B.7) (B.15) (B.1): x = ξ cos (ψ) − η sin (ψ) y = ξ sin (ψ) − η cos (ψ) . π such that 4 4 B ′ = 0.13) (B.10) (B. For the rotation to arrive at the principle with the aid of the trigonometric relationships: cos2 (ψ) − sin2 (ψ) = cos (2ψ) cos (ψ) sin (ψ) = sin (2ψ) . by dividing by D we arrive at the canonical equation: ξ u 2 + η v 2 = 1. Finally. ∞).8) (B.9). Note that since tan (2ψ) ∈ (−∞. B ′ must be zero.44 where: A = a−2 2 B = − sin (p0 ) ab −2 C = b D = cos2 (p0 ) .

where η and ξ correspond to its principle axes. Given that the area of an ellipse is πuv.45 With two more inverse trigonometric relationships: cos [arctan (θ)] = √ 1 1 + θ2 θ sin [arctan (θ)] = √ .20) where a quick test shows that when p0 = 0.1: Geometry of ellipse rotation.19) (B.21) Figure B. 1 + θ2 and a fantastic amount of algebra we arrive at the semimajor and semiminor axes in terms of the Femtometer parameters: u = v = 1 2 a + b2 + 2 1 2 a + b2 − 2 a4 + b4 − 2a2 b2 cos (2p0 ) a4 + b4 − 2a2 b2 cos (2p0 ) (B. u = a and v = b. it is easy to show that: Area = πab cos (p0 ) . . (B.

Springer. and A. 67(5):1855–1866. pages 347–370. Bulu. 35(5):805–825. Journal of the Atmospheric Sciences. 1962. 19(3):264–273. [4] Henry E. Journal of the Atmospheric Sciences. Journal of Geophysical Research. Oxford. 98(2):768–777. B. K. [8] D. 22:46–53. Infrasonic Signals from Large Mining Explosions. On Solitary Waves in the Atmosphere. Hocker. Baldwin. Milton A. 2008. 46 . Muirhead. D. Fiber optics strain gauge. 2008. and Robert L. and Marie D. Butter and G. Christie. [6] C. Headington Hill Hall.Bibliography [1] Stephen J. Secaucus. Axisymmetric smooth contact for an elastic isotropic inﬁnite hollow cylinder compressed by an outer rigid ring with circular proﬁle. Hales. John V. J. Bass. 34(5):806–811. 1978. Springer Handbook of Experimental Solid Mechanics. 17(18):2867–2869. Donn and Maurice Ewing. [7] George Chimonas. New Jersey. [3] Chris S. Woodward. [2] A. 2006. 2(1):9–19. Olson. R. 1977. 2006. Michael Hedlin. Arrowsmith. Joydeep Bhattacharyya. fourth edition. Hedlin. [10] William L. Garces. Avei. Acta Mechanica Sinica. Infrasound. Atmospheric Waves from Nuclear Explosions – Part II: The Soviet Test of 30 October 1961. Bulletin of the Seismological Society of America. Journal of the Atmospheric Sciences. [9] William L. Principles of Optics. Brian Stump. H. Applied Optics. chapter Optical Fiber Strain Gages. Atmospheric Waves from Nuclear Explosions. 1970. Arrowsmith. Yapiei. and A. Pergamon. [5] Max Born and Emil Wolf. Donn and Maurice Ewing. 1978. L. Michael A. A Possible Source Mechanism for Mountain-Associated Infrasound. A. Acoustics Today. 1962.

2005. Introduction to Contact Mechanics. B. Eric Posmentier. Housner and Thad Vreeland Jr. New York. John Wiley and Sons. 2003. [20] G. Balachandran. International Journal of Solids and Structures. Evans and P. CRC Press. 11(3):571–594. [21] George W. G. The Analysis of Stress and Deformation. Donn. [13] J. Massachusetts. Addison-Wesley. D. 1966. Wilcox. 1989. 1979. 1968. Journal of the Optical Society of America. e 44(8):634–640. 1954. [15] Anthony C. Fischer-Cripps. Hocker. Davis. 18(7):945– 952. 1967. Science. second edition. M. Vibration and Active Control. K. 162(3858):1116–1120. Goldstein. Journal of Low Frequency Noise. 39:1307–1326. [24] A. [16] T. Marrone. A structural model for high pressure helical wirewound thermoplastic hose. M.47 [11] William L. New York. [19] Michael A. Applied Optics. Surveying Infrasonic Noise on Oceanic Islands. 159(5):1127–1152. Engineering Design: A Synthesis of Stress Analysis and Materials Engineering. 1973. [22] Jrgen Jakobsen. and Frank L. London. 2002. J. 2002. New York. A. Reviews of Geophysics. 2007. [12] William L. 7 1979. H. 24(3):145–155. Donn and David Rind. Electronics Letters. 1991. New York. 18(9):1445–1448.J. [14] Joseph H. and M. [18] Eugene Hecht. Monitoring Stratospheric Winds with Concorde-Generated Infrasound. Pure and Applied Geophysics. Infrasound at Long Range from Saturn V. Polarized Light. [17] Dennis H. 27(6):518–520. New York. second edition. D. Hedlin. Polarisation-Insensitive Fibre Optic Michelson Interferometer. [23] H. and N. Transmission of Light through Birefringent and Optically Active Media: the Poincar´ Sphere. Reading. Vernon. Faupel. Kersey. Uri Fehr. Fiber-optic sensing of pressure and temperature. second edition. Springer. Jerrard. New York. Infrasound Emission from Wind Turbines. Georges. Infrasound from Convective Storms: Examining the Evidence. Journal of Applied Meteorology. Jon Berger. . 1964. Optics. Collier-Macmillan.

S. H. [30] Kristoﬀer T. Walker. Eric Husmann. A Theory of Microbaroms. Hedlin. Infrasound Monitoring for Atmospheric Studies. Optics Communications. [26] G. [29] Kristoﬀer T. Dzieciuch. Geophysical Journal International. A Universal Compensator for Polarization Changes Induced by Birefringence on a Retracing Beam.P. Mark A. Min Zhang. pages 1–49. The Journal of the Acoustical Society of America. chapter A review of wind noise reduction methodologies. 2003. 43(4):771–775. 4. Lam´ and B. An optical ﬁber infrasound sensor: A new lower limit on atmospheric pressure noise between 1 and 10 Hz. e [27] Mario Martinelli. 281:94–101. [33] Mark A. Richard Hilt. and Matthew A. 1833. Optics Communications. Posmentier. Zumberge. An investigation of a ﬁber-optic air-backed mandrel hydrophone. Zumberge. Applied Optics. H. 2008. New Jersey. and Robert L.. Walker and Michael A. Michael A. [34] Mark A. Zumberge. 72(6):341– 344. OFIS Experiments at Camp Elliott: Paving the Way to Infrasonic Radar and a Portable Infrasonic Calibrator. Dzieciuch. Jonathan Berger. [28] E. Zhenguo Jing. 113(5):2474– 2479. Walker. H. 2004. M´moire sur l’´quilibre int´rieur des corpes e e e e solides homog`nes. 2007. Hedlin.48 [25] Dale O. Secaucus. [32] Kai Yin. Memoires par divers savans. [31] Kristoﬀer T. Tianhuai Ding.E. . 124(4):2090–2099. and Rudolf Widmer-Schnidrig. 1989. 2008. Plastic Material Data Sheets. Mark A. Resolving Quadrature Fringes in Real Time. Hedlin. Parker. Zumberge. Clapeyron. 2009.Division of Automation Creation. MatWeb . and Peter M. Shearer. Liwei Wang. The Journal of the Acoustical Society of America. Springer. Methods for determining infrasound phase velocity direction with an array of line sensors. 2008. 1967. and Yanbiao Liao. 13(5):487–501. Matthew A. Kipp. Jonathan Berger. Inc. Scott Nooner. Michael A. In Proceedings of the 29th Monitoring Research Review: Ground-Based Nuclear Explosion Monitoring Technologies.

- thesis
- Numerical Analysis of the Warpage Problem in TSOP
- Asar Sample Prducts
- Articulos-Seleccionados_2014
- Theory of Elasticity
- ENG_2014061117130220
- circularly polarized antenna
- Exp-4 Torsion Test
- thesis
- thesis
- Elasticity - NetBadi.com
- Ch9 Settlement
- LRFD Chapter F. Beams and other flexural members
- Band Pass Fss
- GND Thesis Complete
- Champagne Hyperpol
- Index
- Dac tinh quang hoc cua thin film.pdf
- Yu - Free Flexural Vibration Analysis of Symmetric Honeycomb Panels GOOD
- MOM2Echap4B Anchor force
- D790 – 10.pdf
- Constitutive Laws for Geomaterials 1999 06 Papamichos v54n6
- thesis
- 00. Building Materials
- FE Analysis of a Box Beam.docx
- MR23
- Elemen Struktur Baja-I.overVIEW
- Introduction to Statics
- Numerics in geotechnics 2010.pdf
- Hardened properties of self-compacting concrete

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd