Fiber Optic Gyro Technology Trends- A Honeywell Perspective

MA2

(Invited)

9:00am-9:30am

Fiber Optic Gyro Technology Trends

-

A

Honeywell Perspective

S.J. Sanders,

L.K.

Strandjord,

D.

Mead

Honeywell Space Systems, MS M22C5

21111

N

19fh Ave, Phx, AZ, 85036-1111

Phone 602-436-6604; Fax 602-436-2821

I. Introduction

Development of Fiber Optic Gyroscopes (FOGs) at Honeywell began in the mid-1980s’

with tactical-grade devices employing first-generation components such as piezoelectric

phase modulators and superluminescent diode sources.

The operational mode of these

gyros was the open-loop configuration, i.e., without rotation-nulling feedback. Typical

sensor performance was 1 “/hr bias instability,

0.1

0/hr1’2 angle random walk

(ARW)

and

2500 ppm scale factor

(SF)

inaccuracy. More than 9000 of these gyros have been sold to

commercial aircraft makers such as Domier, Embraer, and Boeing over the last decade.

Over this same time frame, Honeywell has continually worked to enhance the

performance

of

closed-loop FOGs by optimizing the optical and electronic design,

refining FOG noise and error models, advancing depolarized gyro technology, and

making use of improved components from the optical telecommunications industry. This

15-year effort has resulted in roughly lOOOx improvement in FOG performance from the

original tactical-grade devices: better than 0.0003”/hr bias stability, 0.00008”/hr”2

ARW,

and 0.5-ppm scale factor inaccuracy have been achieved. Honeywell’s FOG emphasis

today is primarily on these high-performance devices for a variety

of

applications (land,

air, sea and space). This paper presents the evolution of

FOG

development and

performance capabilities at Honeywell.

A

timeline of key milestones is given, and the

technology developments and resulting performance improvements are outlined. Data

from Honeywell’s strategic-grade, high-precision FOGs (HPFOG) is included.

11.

Major FOG Technologies and Timeline at Honeywell

Honeywell has developed two major architectures of the closed-loop FOG, each

architecture offering advantages for particular applications.

The two architectures are:

(a) the PM gyro, in which the coil is wound

of

polarization-maintaining fiber. The PM

coil minimizes polarization errors and maximizes light throughput, resulting

in

optimum bias stability and ultra low noise. Honeywell’s highest precision gyros are

based on the PM FOG architecture.

(b) the twin-depolarized gyro, in which the coil is wound of non-PM, single-mode

(SM)

fiber, and Lyot depolarizers are placed at each end of the Sagnac loop to eliminate the

problem of signal fading due to polarization wander

[

11; thus the coil can be wound

with standard

SM

fiber, offering significant cost savings compared to relatively

expensive PM fiber.

SM

FOG coils also offer lower magnetic sensitivity than PM

coils [2] and SM fiber generally exhibits lower radiation-induced darkening [3],

making depolarized gyros better suited for certain application environments.

Honeywell’s navigation-grade

FOGs

are based on the depolarized architecture, and

better than navigation-grade performance has been achieved with this architecture,

paving the way for its use in higher-precision applications.

0-7803-7289-1/02/$17.0oO2002

IEEE

5

A timeline of Honeywell’s development of these FOG architectures is given below:

0

1986: Tactical grade,

PM

“AHRS” gyro begins development

0

1989: Navigation-grade PM gyro effort begins

0

1991

:

First AHRS

gyros

ship to commercial aircraft makers

1993

:

Navigation-grade depolarized gyro effort begins

0

1994: Mini PM FOG nav-grade: 0.02”/hr bias, 0.0015”/hr”2 ARW, 20 ppm

SF

0

1994: High-precision PM gyro begins:

0.001

“/hr bias, 0.0002°/hr’/2 ARW achieved

0

1995-’97: Improved HPFOG: 0.00023”/hr bias, 0.0001 8”/hr’/’ ARW, 0.3ppm SF

0

1996: Miniature depolarized FOG achieves nav-grade performance

0

1998: Development of high-precision depolarized gyro begins

0

2000: HPFOG ARW reduced below 0.00009”/hr1/2

As can be seen, Honeywell FOG performance has dramatically improved over the last

decade, and emphasis has shifted from tactical-grade to high precision, strategic-grade

applications. The rest

of

the paper will concentrate on these high-precision

gyros.

111.

High-precision

FOG

Figure 1 shows a block diagram of the Honeywell high-precision FOG. The salient

features of the design are

a

2 to 4-km, PM fiber coil for maximum bias stability

0

a high-power fiber light source (FLS) for excellent wavelength stability and low noise

0

a feedback loop to reduce relative intensity noise (RIN) of the FLS output intensity,

(e.g., via lithium niobate intensity modulator

[4])

0

a “dual ramp” feedback scheme [5] maximizing SF linearity and dynamic range

0

error suppression modulation to maximize bias stability [6]

Figure

2

gives

a

root-Allan variance or “cluster” curve

of

typical output data for an

HPFOG. From this plot, the gyro bias instability and

ARW

may be estimated. For this

run, the

ARW

is about

79

pdeg/hr’’2. This noise result can be compared to a detailed

model of FOG noise developed at Honeywell over the last decade. The model, which has

been validated with abundant data from many high-precision FOGS, includes all the

major

ARW

contributors, including shot noise, Johnson noise,

RIN,

and thermal phase

noise, aliasing, and environmental effects such as radiation. In this case, the model

predicts agreement to within 2% of the measured value. This modeling capability is

a

key tool in

gyro

optimization for many applications, enabling quantitative design trades

between

ARW

and other performance parameters. Finally, we note that industry-

standard gyro analysis tools such as Autofit corroborate the low

ARW

estimates for these

gyros: for instance, the Autofit estimate for the data in Fig.

2

was

79.5

ydegihr”’.

As for bias instability or flicker noise, the cluster curve of this 58-hour data set does not

depart from a white noise

(ARW) signature until about three hours of integration time.

Even with this departure, the curve remains within the 3-sigma error bars for the cluster

estimation process out to 20 hours of integration time. Nevertheless, a worst-case

estimate for the bias instability coefficient,

Bflicker,

may be obtained from the cluster value

OB

=

0.00005

“/hr

at three hours:

Bflicker

=

0-x[n/21n(2)]”~

x

0.000075 “/hr. Autofit

6