Attitude Determination and Control

(ADCS)
16.684 Space Systems Product Development
Spring 2001
Olivier L. de Weck
Department of Aeronautics and Astronautics
Massachusetts Institute of Technology

ADCS Motivation

Motivation

In order to point and slew optical

systems, spacecraft attitude control
provides coarse pointing while
optics control provides fine
pointing

Spacecraft Slew Maneuvers

Spacecraft Control

Spacecraft Stabilization
Spin Stabilization
Gravity Gradient
Three-Axis Control
Formation Flight
Actuators
Reaction Wheel Assemblies
(RWAs)
Control Moment Gyros
(CMGs)
Magnetic Torque Rods
Thrusters

Sensors: GPS, star trackers, limb

sensors, rate gyros, inertial
measurement units
Control Laws

Euler Angles
Quaternions
Key Question:
What are the pointing
requirements for satellite ?

NEED expendable propellant:

On-board fuel often determines life
Failing gyros are critical (e.g. HST)

Outline

Definitions and Terminology

Coordinate Systems and Mathematical Attitude Representations
Rigid Body Dynamics
Disturbance Torques in Space
Passive Attitude Control Schemes
Actuators
Sensors
Active Attitude Control Concepts
ADCS Performance and Stability Measures
Estimation and Filtering in Attitude Determination
Maneuvers
Other System Consideration, Control/Structure interaction
Technological Trends and Advanced Concepts

Opening Remarks

Nearly all ADCS Design and Performance can be viewed in

terms of RIGID BODY dynamics
Typically a Major spacecraft system
For large, light-weight structures with low fundamental
frequencies the flexibility needs to be taken into account
ADCS requirements often drive overall S/C design
Components are cumbersome, massive and power-consuming
Field-of-View requirements and specific orientation are key
Design, analysis and testing are typically the most
challenging of all subsystems with the exception of payload
design
Need a true systems orientation to be successful at
designing and implementing an ADCS

Terminology
ATTITUDE :

Orientation of a defined spacecraft body coordinate

system with respect to a defined external frame (GCI,HCI)

ATTITUDE DETERMINATION: Real-Time

or Post-Facto knowledge,
within a given tolerance, of the spacecraft attitude
Maintenance of a desired, specified attitude
within a given tolerance

ATTITUDE CONTROL:

Low Frequency spacecraft misalignment;

usually the intended topic of attitude control

ATTITUDE ERROR:

High Frequency spacecraft misalignment;

usually ignored by ADCS; reduced by good design or fine
pointing/optical control.

ATTITUDE JITTER:

Pointing Control Definitions

target
estimate
c

a
true

k
s

target
true
estimate
a
s
k
c

desired pointing direction

actual pointing direction (mean)
estimate of true (instantaneous)
pointing accuracy (long-term)
stability (peak-peak motion)
knowledge error
control error

a = pointing accuracy = attitude error

s = stability = attitude jitter
Source:
G. Mosier
NASA GSFC

Attitude Coordinate Systems

(North Celestial Pole)
^
Z
GCI: Geocentric Inertial Coordinates
Cross product
Geometry: Celestial Sphere

^ ^ ^
Y=ZxX

dihedral

th
g
n
e
l
Arc

D
VERNAL
EQUINOX

^
X

D: Right Ascension

G : Declination

Inertial Coordinate
System

^
Y

X and Y are
in the plane of the ecliptic

Attitude Description Notations

{} = Coordinate system
*
P = Vector
A*
P = Position vector w.r.t. { A}

Z A
Pz

A*

Py
Px

Px
A*
P = Py
P
z

YA

X A

1 0 0
Unit vectors of { A} = X A YA Z A = 0 1 0

0 0 1

Describe the orientation of a body:

(1) Attach a coordinate system to the body
(2) Describe a coordinate system relative to an
inertial reference frame

Rotation Matrix
Z A

{ A} = Reference coordinate system

Jefferson
Memorial Y
Z B

{ B} = Body coordinate system

YA

X A

Z A

X A X B

0
1
A

B R = 0 cos
0 sin

Special properties of rotation matrices:

(1) Orthogonal:

Jefferson
MemorialY
Z

A A
A
A
R
=
X
ZB
B
B YB

X B

Rotation matrix from {B} to {A}

RT R = I , RT = R 1

YA

- sin

cos
0

(2) Orthonormal:
R =1

(3) Not commutative

A B
B A
B R C R C R BR

Euler Angles (1)

Euler angles describe a sequence of three rotations about different
axes in order to align one coord. system with a second coord. system.
Rotate about Z A by
Z A

Z B

Rotate about YB by
Z B

Z C

YB

X A

YA

X B

cos - sin
A

cos
B R = sin
0
0

X B

Z D
YB

Z C

YC

X C

0
0

Rotate about X C by

cos
B

CR= 0
- sin

YC
X C

0
1
0

A
A B C
D R= B R C R D R

sin
0

cos

YD

X D

0
0
1
C

D R = 0 cos - sin
0 sin cos

Euler Angles (2)

Zi (parallel to r)
about Yi
Concept used in rotational
about X
kinematics to describe body
Yaw
orientation w.r.t. inertial frame about Zb
Roll
Sequence of three angles and
Body
Xi
prescription for rotating one
CM
(parallel
reference frame into another
to v)
Pitch
Can be defined as a transformation
(r x v direction)
matrix body/inertial as shown: TB/I
Yi
Euler angles are non-unique and
r
nadir
exact sequence is critical

Note:

Goal: Describe kinematics of body-fixed

frame with respect to rotating local vertical

TB/1I = TI / B = TBT/ I

cos
Transformation
-sin
from Body to T
=
B/I

Inertial frame:

(Pitch, Roll, Yaw) = (TI\)

Euler Angles

sin
cos

0 1
0
0 cos 0 -sin
0 0 cos sin 0
1
0
0
0
1 0 -sin cos sin 0 cos

 

 



YAW

ROLL

PITCH

Quaternions

*
q = A vector describes the

Main problem computationally is

q1
axis of rotation.
q q*
the existence of a singularity

2
Q = = q4 = A scalar describes the
Problem can be avoided by an
q3 q4
amount of rotation.

application of Eulers theorem:
q
4
A

K
Z A
EULERS THEOREM
Jefferson Memorial

The Orientation of a body is uniquely

specified by a vector giving the direction
of a body axis and a scalar specifying a
rotation angle about the axis.

Definition introduces a redundant

fourth element, which eliminates
the singularity.
This is the quaternion concept
Quaternions have no intuitively
interpretable meaning to the human
mind, but are computationally
convenient

Z B

X A
A: Inertial
B: Body

k x
A
K = k y
k
z

YB

YA
X B

q1 = k x sin
2

q2 = k y sin
2

q3 = k z sin
2

q4 = cos
2

Quaternion Demo (MATLAB)

Comparison of Attitude Descriptions

Method

Euler
Angles

Direction
Cosines

Angular
Velocity Z

Quaternions

Pluses

If given ,,
then a unique
orientation is
defined

Orientation
defines a
unique dir-cos
matrix R

Vector
properties,
commutes w.r.t
addition

Minuses

Given orient
then Euler
non-unique
Singularity

6 constraints
must be met,
non-intuitive

Integration w.r.t Not Intuitive

time does not
Need transforms
give orientation
Needs transform

Best for
analytical and
ACS design work

Must store
initial condition

Computationally
robust
Ideal for digital
control implement

Best for
digital control
implementation

Rigid Body Kinematics

Z

Time Derivatives:
(non-inertial)

Body
CM
r

^
k

^
K

Z = Angular velocity
of Body Frame

^i

Inertial
Frame

^j
Rotating
Body Frame
Y

J^
^
I

X
Applied to
position vector r:

r = R+

BASIC RULE:
Position

r = R +  BODY +

 INERTIAL =  BODY +
Expressed in
the Inertial Frame

Rate

 + 


 +
2

+
r=R
BODY
BODY
Inertial
relative accel
accel of CM
w.r.t. CM

coriolis

angular
accel

centripetal

Acceleration

Angular Momentum (I)

.
rn
Angular Momentum

H total =

r i mi ri
i =1

mn

Z
System in
motion relative
to Inertial Frame

.
ri

mi

rn
ri
r1

m1
.
r1

Y
If we assume that
X
(a)
(b)
Then :

Origin of Rotating Frame in Body CM

Fixed Position Vectors ri in Body Frame
(Rigid Body)
Angular Momentum Decomposition

H total =

mi R R
i =1




n

ANGULAR MOMENTUM
OF TOTAL MASS W.R.T
INERTIAL ORIGIN

mi i  i

i =1 




H BODY
BODY ANGULAR
MOMENTUM ABOUT
CENTER OFMASS

Collection of point
masses mi at ri

Note that Ui is
measured in the
inertial frame

Angular Momentum (II)

For a RIGID BODY
we can write:

 i =

 i ,BODY





+ i = i

RELATIVE
MOTION IN BODY

And we are able to write:

H = I

RIIGID BODY, CM COORDINATES

H and Z are resolved in BODY FRAME

The vector of angular momentum in the body frame is the product

of the 3x3 Inertia matrix and the 3x1 vector of angular velocities.
Inertia Matrix
Properties:

I11
I = I 21
I 31

I12
I 22
I 32

Real Symmetric ; 3x3 Tensor ; coordinate dependent

n

I13
I 23
I 33

I12 = I 21 = mi i 2 i1

I13 = I 31 = mi i1 i 3

I 23 = I 32 = mi i 2 i 3

I11 = mi i22 + i23

i =1
n

I 22 = mi i21 + i23
i =1
n

I 33 = mi i21 + i22
i =1

i =1
n

i =1
n

i =1

Kinetic Energy and Euler Equations

Kinetic
Energy

Etotal

2 1 n
1 n
= mi R + mi  i2
2 i =1
2 i =1







E-TRANS

E-ROT

1
1
EROT = H = T I
2
2

For a RIGID BODY, CM Coordinates

with Z resolved in body axis frame

H = T I

Sum of external and internal torques

In a BODY-FIXED, PRINCIPAL AXES CM FRAME:

H 1 = I11 = T1 + ( I 22 I 33 ) 23
H 2 = I 2 2 = T2 + ( I 33 I11 ) 31
H = I  = T + ( I I )
3

3 3

11

22

1 2

Euler Equations

No general solution exists.

Particular solutions exist for
simple torques. Computer
simulation usually required.

Torque Free Solutions of Eulers Eq.

TORQUE-FREE
CASE:

An important special case is the torque-free motion of a (nearly)

symmetric body spinning primarily about its symmetry axis

x , y << z =

By these assumptions:

The components of angular velocity

then become:

And the Euler equations become:

x (t ) = xo cos n t
y (t ) = yo cos n t

The Zn is defined as the natural

or nutation frequency of the body:
Z

 x =

n2 = K x K y 2
Z

I xx I yy

H
Z

I zz I yy
I xx





Kx

I zz I xx
y
 y =
I yy





e
Con

Iz < Ix = I y

y
Bod

ne
Co
dy
Bo

Sp
ac
eC
on
e

Ky

Q : nutation
angle

Space
Cone

Iz > Ix = I y

 z = 0
H and Z never align
unless spun about
a principal axis !

Spin Stabilized Spacecraft

UTILIZED TO STABILIZE SPINNERS

Zb

Perfect Cylinder

I xx = I yy

Antenna
despun at
1 RPO

I zz

mL
2
= + R
4 3

mR 2
=
2

Yb
Xb

DUAL SPIN

:
BODY

Two bodies rotating at different rates

about a common axis
Behaves like simple spinner, but part
is despun (antennas, sensors)
requires torquers (jets, magnets) for
momentum control and nutation
dampers for stability
allows relaxation of major axis rule

Disturbance Torques
Assessment of expected disturbance torques is an essential part
of rigorous spacecraft attitude control design
Typical Disturbances

Gravity Gradient: Tidal Force due to 1/r2 gravitational field variation

for long, extended bodies (e.g. Space Shuttle, Tethered vehicles)
Aerodynamic Drag: Weathervane Effect due to an offset between the
CM and the drag center of Pressure (CP). Only a factor in LEO.
Magnetic Torques: Induced by residual magnetic moment. Model the
spacecraft as a magnetic dipole. Only within magnetosphere.
Solar Radiation: Torques induced by CM and solar CP offset. Can
compensate with differential reflectivity or reaction wheels.
Mass Expulsion: Torques induced by leaks or jettisoned objects
Internal: On-board Equipment (machinery, wheels, cryocoolers, pumps
etc). No net effect, but internal momentum exchange affects attitude.

Gravity Gradient
n = / a 3 = ORBITAL RATE

1) Local vertical
2) 0 for symmetric spacecraft

Gravity Gradient:

3) proportional to

r = sin

Zb

^r

T = 3n 2 r I r

Gravity Gradient
Torques

In Body Frame

1/r3

Small
angle
approximation

T
Xb

sin 1 sin sin [

2

- sin T

1]

Earth

Resulting torque in BODY FRAME:

Typical Values:
I=1000 kgm2
n=0.001 s-1
T= 6.7 x 10-5 Nm/deg

( I zz I yy )
T 3n 2 ( I zz I xx )

Pitch Libration freq.:

lib = n

3 ( I xx I zz )
I yy

Aerodynamic Torque
T = r Fa

r = Vector from body CM

to Aerodynamic CP

1
Fa = V 2 SCD
2
Aerodynamic
Drag Coefficient

Fa = Aerodynamic Drag Vector

in Body coordinates

1 CD 2

Typically in this Range for

Free Molecular Flow

S = Frontal projected Area

V = Orbital Velocity

Typical Values:
Cd = 2.0
S = 5 m2
r = 0.1 m
r = 4 x 10-12 kg/m3
T = 1.2 x 10-4 Nm

Notes
(1) r varies with Attitude
(2) U varies by factor of 5-10 at
a given altitude
(3) CD is uncertain by 50 %

U = Atmospheric Density
2 x 10-9 kg/m3 (150 km)
3 x 10-10 kg/m3 (200 km)
7 x 10-11 kg/m3 (250 km)
4 x 10-12 kg/m3 (400 km)

Exponential Density Model

Magnetic Torque

T = M B
M = Spacecraft residual dipole
in AMPERE-TURN-m2 (SI)
or POLE-CM (CGS)
M = is due to current loops and
residual magnetization, and will
be on the order of 100 POLE-CM
or more for small spacecraft.

Typical Values:
B= 3 x 10-5 TESLA
M = 0.1 Atm2
T = 3 x 10-6 Nm

B = Earth magnetic field vector in

spacecraft coordinates (BODY FRAME)
in TESLA (SI) or Gauss (CGS) units.
B varies as 1/r3, with its direction
along local magnetic field lines.
Conversions:
1 Atm2 = 1000 POLE-CM , 1 TESLA = 104 Gauss

B ~ 0.3 Gauss
at 200 km orbit

Solar Radiation Torque

T = rFs
Fs = (1 + K ) Ps S

Ps = I s / c

r = Vector from Body CM

to optical Center-of-Pressure (CP)
Fs = Solar Radiation pressure in
BODY FRAME coordinates
K = Reflectivity , 0 < K <1
S = Frontal Area

I s = 1400 W/m 2 @ 1 A.U.

Notes:

(a) Torque is always to sun line

(b) Independent of position or
velocity as long as in sunlight
Typical Values:
K = 0.5
S =5 m2
r =0.1 m
T = 3.5 x 10-6 Nm

Is = Solar constant, depends on

heliocentric altitude
SUN
Significant for
spacecraft
with large
frontal area
(e.g. NGST)

Mass Expulsion and Internal Torques

T = rF

Mass Expulsion Torque:

Notes:
(1)

May be deliberate (Jets, Gas venting) or accidental (Leaks)

(2)

Wide Range of r, F possible; torques can dominate others

(3)

Also due to jettisoning of parts (covers, cannisters)

Internal Torque:
Notes:
(1)

(2)

Momentum exchange between moving parts

has no effect on System H, but will affect
attitude control loops
Typically due to antenna, solar array, scanner
motion or to deployable booms and appendages

Disturbance Torque for CDIO

' offset
Expect residual
gravity torque to be
largest disturbance

r
mg
Pivot Point
Air Bearing

Body
CM

Air
Bearing

ground
Important
to balance
precisely ! Initial Assumption:

T = r mg 0.001 100 9.81 1 [Nm]

Passive Attitude Control (1)

Passive control techniques take advantage of basic physical
principles and/or naturally occurring forces by designing
the spacecraft so as to enhance the effect of one force,
while reducing the effect of others.
SPIN STABILIZED

H = T = rF

Requires Stable Inertia Ratio: Iz > Iy =Ix

Requires Nutation damper: Eddy Current, Ball-inTube, Viscous Ring, Active Damping
Requires Torquers to control precession (spin axis
drift) magnetically or with jets
Inertially oriented
Precession:

dH H

H=

dt
t

H rF t
'H

H = 2 H sin
H = I
2

Large Z
=
gyroscopic
stability

rF t rF

=
t
H
I

'T
H

T = rF
F into page

Passive Attitude Control (2)

GRAVITY GRADIENT

Requires stable Inertias: Iz << Ix, Iy

Requires Libration Damper: Eddy Current,
Hysteresis Rods
Requires no Torquers
Earth oriented
No Yaw Stability (can add momentum wheel)

Gravity Gradient with Momentum wheel:

rd
wa
for

DUAL SPIN with GG

torque providing
momentum control

BODY rotates at
one RPO (rev per orbit)

O.N.
do

nadir

wn

Wheel spins
at rate :

Gravity Gradient Configuration

with momentum wheel for
yaw stability

Active Attitude Control

Active Control Systems directly sense spacecraft attitude
and supply a torque command to alter it as required. This
is the basic concept of feedback control.

Reaction Wheels most common actuator

Fast; continuous feedback control
Moving Parts
Internal Torque only; external still
required for momentum dumping
Relatively high power, weight, cost
Control logic simple for independent axes
(can get complicated with redundancy)
Typical Reaction (Momentum) Wheel Data:

Operating Range: 0 +/- 6000 RPM

Angular Momentum @ 2000 RPM:
1.3 Nms
Angular Momentum @ 6000 RPM:
4.0 Nms
Reaction Torque: 0.020 - 0.3 Nm

Actuators: Reaction Wheels

One creates torques on a spacecraft by creating equal but opposite

torques on Reaction Wheels (flywheels on motors).

For three-axes of torque, three wheels are necessary. Usually use four
wheels for redundancy (use wheel speed biasing equation)
If external torques exist, wheels will angularly accelerate to counteract
these torques. They will eventually reach an RPM limit (~3000-6000
RPM) at which time they must be desaturated.
Static & dynamic imbalances can induce vibrations (mount on isolators)
Usually operate around some nominal spin rate to avoid stiction effects.
Ithaco RWAs
(www.ithaco.com
/products.html)
Waterfall plot:

Needs to be carefully balanced !

Actuators: Magnetic Torquers

Magnetic Torquers

Can be used

Often used for Low Earth Orbit

(LEO) satellites
Useful for initial acquisition
maneuvers
Commonly use for momentum
desaturation (dumping) in
reaction wheel systems
May cause harmful influence on
star trackers

for attitude control

to de-saturate reaction wheels

Torque Rods and Coils

Torque rods are long helical coils

Use current to generate magnetic
field
This field will try to align with the
Earths magnetic field, thereby
creating a torque on the spacecraft
Can also be used to sense attitude
as well as orbital location

ACS Actuators: Jets / Thrusters

Thrusters / Jets

Thrust can be used to control

attitude but at the cost of
consuming fuel
Calculate required fuel using
Rocket Equation
Advances in micro-propulsion
make this approach more feasible.
Typically want Isp > 1000 sec

Use consumables such as Cold Gas

(Freon, N2) or Hydrazine (N2H4)
Must be ON/OFF operated;
proportional control usually not
feasible: pulse width modulation
(PWM)
Redundancy usually required, makes
the system more complex and
expensive
Fast, powerful
Often introduces attitude/translation
coupling
Standard equipment on manned
spacecraft
May be used to unload accumulated
angular momentum on reaction-wheel
controlled spacecraft.

ACS Sensors: GPS and Magnetometers

Global Positioning System (GPS)

Currently 27 Satellites
12hr Orbits
Accurate Ephemeris
Accurate Timing

Stand-Alone
100m
DGPS
5m
Carrier-smoothed DGPS 1-2m

Magnetometers

Measure components Bx, By, Bz of

ambient magnetic field B
Sensitive to field from spacecraft
(electronics), mounted on boom
Get attitude information by
comparing measured B to modeled B
Tilted dipole model of earths field:

Bnorth
3 C
B = 6378 0
east r
Bdown km 2S

S C
S S 29900

S
C 1900
2C C 2C S 5530

Where: C=cos , S=sin, =latitude, =longitude

Units: nTesla
+Z

Me

flux
lines

+Y
+X

ACS Sensors: Rate Gyros and IMUs

Rate Gyros (Gyroscopes)

Measure the angular rate of a

spacecraft relative to inertial space
Need at least three. Usually use
more for redundancy.
Can integrate to get angle.
However,
DC bias errors in electronics
will cause the output of the
integrator to ramp and
eventually saturate (drift)
Thus, need inertial update

Inertial Measurement Unit (IMU)

Mechanical gyros
(accurate, heavy)
Ring Laser (RLG)
MEMS-gyros

Courtesy of Silicon Sensing Systems, Ltd. Used with permission.

Integrated unit with sensors,

mounting hardware,electronics and
software
measure rotation of spacecraft with
rate gyros
measure translation of spacecraft
with accelerometers
often mounted on gimbaled
platform (fixed in inertial space)
Performance 1: gyro drift rate
(range: 0 .003 deg/hr to 1 deg/hr)
Performance 2: linearity (range: 1
to 5E-06 g/g^2 over range 20-60 g
Typically frequently updated with
external measurement (Star
Trackers, Sun sensors) via a
Kalman Filter

ACS Sensor Performance Summary

Reference

Typical
Accuracy

Remarks

Sun

1 min

Simple, reliable, low

cost, not always visible

Earth

0.1 deg

Magnetic Field

1 deg

Stars

0.001 deg

Inertial Space

0.01 deg/hour

Orbit dependent;
usually requires scan;
relatively expensive
Economical; orbit
dependent; low altitude
only; low accuracy
Heavy, complex,
expensive, most
accurate
Rate only; good short
term reference; can be
heavy, power, cost

CDIO Attitude Sensing

Will not be able to
use/afford STAR TRACKERS !
From where do we get
an attitude estimate
for inertial updates ?

Potential Solution:
Electronic Compass,
Magnetometer and
Tilt Sensor Module
Specifications:
Heading accuracy: +/- 1.0 deg RMS @ +/- 20 deg tilt
Resolution 0.1 deg, repeatability: +/- 0.3 deg
Tilt accuracy: +/- 0.4 deg, Resolution 0.3 deg
Sampling rate: 1-30 Hz

Problem: Accuracy insufficient to meet requirements alone,

will need FINE POINTING mode

Spacecraft Attitude Schemes

Spin Stabilized Satellites

Spin the satellite to give it

gyroscopic stability in inertial
space
Body mount the solar arrays to
guarantee partial illumination by
sun at all times
EX: early communication
satellites, stabilization for orbit
changes
Torques are applied to precess the
angular momentum vector

Gravity Gradient Stabilization

Three-Axis Stabilization

De-Spun Stages

Some sensor and antenna systems

require inertial or Earth referenced
pointing
Place on de-spun stage
EX: Galileo instrument platform

Long satellites will tend to point

towards Earth since closer portion
feels slightly more gravitational
force.
Good for Earth-referenced pointing
EX: Shuttle gravity gradient mode
minimizes ACS thruster firings

For inertial or Earth-referenced

pointing
Requires active control
EX: Modern communications
satellites, International Space
Station, MIR, Hubble Space
Telescope

ADCS Performance Comparison

Method

Typical Accuracy

Remarks

Spin Stabilized

0.1 deg

Gravity Gradient

1-3 deg

Passive, simple; single axis

inertial, low cost, need slip
rings
Passive, simple; central
body oriented; low cost

Jets

0.1 deg

Consumables required, fast;

high cost

Magnetic

1 deg

Near Earth; slow ; low

weight, low cost

Reaction Wheels

0.01 deg

Internal torque; requires

other momentum control;
high power, cost

3-axis stabilized, active control most common choice for precision spacecraft

ACS Block Diagram (1)

T
desired
attitude

+
-

'T
error
signal

Spacecraft
Control
Actuators

Tc
gain

Ta
Actual
Pointing
Direction

Attitude Measurement

Feedback Control Concept:

T = K
c

Correction
torque = gain x error

Force or torque is proportional to deflection. This

is the equation, which governs a simple linear
or rotational spring system. If the spacecraft
responds quickly we can estimate the required
gain and system bandwidth.

Gain and Bandwidth

Assume control saturation half-width sat at torque command Tsat, then

Tsat
K
sat

hence

K

+ sat 0
I

Recall the oscillator frequency of a

simple linear, torsional spring:

K
I

[rad/sec]

I = moment
of inertia

This natural frequency is approximately

equal to the system bandwidth. Also,

f =
2

[Hz]

1 2
= =
f

Is approximately the system time constant W.

Note: we can choose any two of the set:

, sat ,

EXAMPLE:

sat = 102

[rad]

Tsat = 10 [Nm]
I = 1000

[kgm 2 ]

K = 1000

=1

[Nm/rad]
[rad/sec]

f = 0.16

[Hz]

= 6.3 [sec]

Feedback Control Example

Pitch Control with a single reaction wheel
Rigid Body
Dynamics

I = Tw + Text = I  = H

Wheel
Dynamics

 + ) = T = h
J (
w

Feedback
Law, Choose


Tw = K p K

r
, ,

 + (K r / I ) + (K p / I ) = 0

s 2 + (K r / I ) s + K p / I = 0
s + 2 s + = 0
2

Im

Stabilize
RIGID
BODY

Rate
feedback

Position
feedback

Then:

Y
D
BO

Re

Laplace Transform
Characteristic Equation
Nat. frequency

= Kp / I

damping

=K r / 2 K p I

Jet Control Example (1)

l

Introduce control torque Tc via

force couple from jet thrust:

Tc

SWITCH
LINE

Only three possible values for Tc :

l
.

I = T c

Chatter due to minimum

on-time of jets.
START

Problem

Fl

c
T = 0
Fl

Can stabilize (drive T to zero)

by feedback law:

T c = Fl sgn + 
Where

T c = Fl
T = Fl
c

PHASE PLANE

On/Off
Control
only

sgn ( x ) =

x
x

)
prediction
term

W = time constant

Jet Control Example (2)

Chatter leads to a
limit cycle, quickly
wasting fuel
.
T
SL

Solution:
Eliminate Chatter by Dead Zone ; with Hysteresis:

T c = Fl

TC

+
-

E1 E2

T c = Fl
= + 

Is 2

1+ s

PHASE PLANE
At Switch Line:

+  = 0

Low Frequency Limit Cycle

Mostly Coasting
Low Fuel
. Usage
T and T bounded

Results in the following motion:

.

max

DEAD ZONE

max

ACS Block Diagram (2)

In the REAL WORLD things are somewhat more complicated:
control
error
target

estimate

dynamic
disturbances

Controller

Estimator

true

Spacecraft

Sensors

accuracy + stability

knowledge error

sensor noise,
misalignment

Spacecraft not a RIGID body, sensor , actuator & avionics dynamics

Digital implementation: work in the z-domain
Time delay (lag) introduced by digital controller
A/D and D/A conversions take time and introduce errors: 8-bit, 12-bit,
16-bit electronics, sensor noise present (e.g rate gyro @ DC)
Filtering and estimation of attitude, never get q directly

Attitude Determination

Attitude Determination (AD) is the process of of deriving estimates

of spacecraft attitude from (sensor) measurement data. Exact
determination is NOT POSSIBLE, always have some error.
Single Axis AD: Determine orientation of a single spacecraft axis
in space (usually spin axis)
Three Axis AD: Complete Orientation; single axis (Euler axis,
when using Quaternions) plus rotation about that axis
KALMAN
Constant

1
raw
gyro rate
2
inertial
update

Switch
Kalman
Logical
NOT

Fixed
Gain

Example:
Attitude
Estimator
for NEXUS

Wc comp rates
2
filtered/corrected
rate
estimated
1
quaternion
Switch1

Single-Axis Attitude Determination

true
solution

Utilizes sensors that yield an arc^

Z
Locus of
length measurement between
possible S/C
sensor boresight and known
attitude from
reference point (e.g. sun, nadir)
sun cone angle
measurement
Requires at least two independent
with error band
measurements and a scheme to
choose between the true and false sun
solution
Total lack of a priori estimate
requires three measurements
Cone angles only are measured, not
^
X
full 3-component vectors. The
reference (e.g. sun, earth) vectors
are known in the reference frame,
false
Earth
but only partially so in the body
solution
nadir
frame.

a priori
estimate

^
Y
Locus of
possible attitudes
from earth cone
with error band

Three-Axis Attitude Determination

Need two vectors (u,v) measured in

the spacecraft frame and known in
reference frame (e.g. star position
on the celestial sphere)
Generally there is redundant data
available; can extend the
calculations on this chart to include
a least-squares estimate for the
attitude
Do generally not need to know
absolute values

Define:

i = u / u
j = (u v ) / u v
k = i j
Want Attitude Matrix T:

jB kB = T iR jR kR
i
B







u,v

So:

T = MN 1

Note: N must be non-singular (= full rank)

Effects of Flexibility (Spinners)

The previous solutions for Eulers equations were only valid for
a RIGID BODY. When flexibility exists, energy dissipation will occur.

H = I

CONSTANT
Conservation of
Angular Momentum

EROT

1 T
= I
2
DECREASING

Spin goes to maximum

I and minimum Z

CONCLUSION: Stable Spin is

only possible about the axis of
maximum inertia.
Classical Example:

EXPLORER 1

initial
spin
axis

energy dissipation

Controls/Structure Interaction
Spacecraft

Sensor

Flexibility

Cant always neglect flexible modes (solar

arrays, sunshield)
Sensor on flexible structure, modes introduce
phase loss
Feedback signal corrupted by flexible
deflections; can become unstable
Increasingly more important as spacecraft
become larger and pointing goals become tighter

Loop Gain Function: Nichols Plot (NGST)

Flexible
modes

Stable
no encirclements
of critical point

NM axis 1 to NM axis 1

200

Gain
[dB]

-200
-2000

-1500

-1000

-500

Phase [deg]

500

1000

Other System Considerations (1)

Need on-board COMPUTER

Ground Processing

Increasing need for on-board performance and autonomy

Typical performance (somewhat outdated: early 1990s)
35 pounds, 15 Watts, 200K words, 100 Kflops/sec, CMOS
Rapidly expanding technology in real-time space-based computing
Nowadays get smaller computers, rad-hard, more MIPS
Software development and testing, e.g. SIMULINK Real Time Workshop,
compilation from development environment MATLAB C, C++ to target
processor is getting easier every year. Increased attention on software.
Typical ground tasks: Data Formatting, control functions, data analysis
Dont neglect; can be a large program element (operations)

Testing

Design must be such that it can be tested

Several levels of tests: (1) benchtop/component level, (2) environmental
testing (vibration,thermal, vacuum), (3) ACS tests: air bearing, hybrid
simulation with part hardware, part simulated

Other System Considerations (2)

Maneuvers

Attitude/Translation Coupling

Typically: Attitude and Position Hold,Tracking/Slewing, SAFE mode

Initial Acquisition maneuvers frequently required
Impacts control logic, operations, software
Sometimes constrains system design
Maneuver design must consider other systems, I.e.: solar arrays pointed
towards sun, radiators pointed toward space, antennas toward Earth
 'v from thrusters can affect attitude
(2) Attitude thrusters can perturb the orbit

Simulation

CM
l
T

(1)

F1 (2)

F
T

'F

Numerical integration of dynamic equations of motion

Very useful for predicting and verifying attitude performance
F1 = F2
Can also be used as surrogate data generator
Hybrid simulation: use some or all of actual hardware, digitally simulate
the spacecraft dynamics (plant)
A/D
can be expensive, but save money later in the program
H/W
sim
D/A

Future Trends in ACS Design

Lower Cost

Integration of GPS (LEO)

Standardized Spacecraft, Modularity

Smaller spacecraft, smaller Inertias
Technological progress: laser gyros, MEMS, magnetic wheel bearings
Greater on-board autonomy
Simpler spacecraft design
Allows spacecraft to perform on-board navigation; functions independently
from ground station control
Potential use for attitude sensing (large spacecraft only)

Very large, evolving systems

Space station ACS requirements change with each added module/phase

Large spacecraft up to 1km under study (e.g. TPF Able kilotruss)
Attitude control increasingly dominated by controls/structure interaction
Spacecraft shape sensing/distributed sensors and actuators

Advanced ACS concepts

Formation Flying in Space

Visible Earth Imager using

a Distributed Satellite System

Exploit natural orbital dynamics to

synthesize sparse aperture arrays
using formation flying
Hills equations exhibit closed freeorbit ellipse solutions

No V required for collector

spacecraft
Only need V to hold combiner
spacecraft at paraboloids focus



x 2yn
3n 2 x = a x


= ay
y + 2xn

Hype rbola (Foci)

+ n 2z

z

x/R o (Ze nith Nadir)

Circula r P a ra boloid
0.5

Optima l Focus (p/R o =2.2076)

0

Ellips e

-0.5

-1
-1.5

P roje cte d Circle

-1 -0.5

0.5

z/R o (Cros s axis )

1.5

1.5

0.5

-0.5

-1

-1.5

y/R o(ve locity ve ctor)

TPF

= az

ACS Model of NGST (large, flexible S/C)

PID bandwidth is 0.025 Hz

NGST
ACS
Design

Command 3
Rate

Command
Position

3rd order LP elliptic filters for

flexible mode gain suppression
Wheel model includes non-linearities
and imbalance disturbances

KF Flag

FEM

3
72

72 DOF

ACS Rate
Matrix

Estimated Structural Filters

Inertia
PID
Attitude
Tensor
Controllers
Determination

gyro

Qt true attitude
Qt prop

Kalman Filter blends 10 Hz IRU and

2 Hz ST data to provide optimal attitude
estimate; option exists to disable the KF
and inject white noise, with amplitude given
by steady-state KF covariance into the
controller position channel

6x1
Forces &
Torques

Large (200m 2 ) deployable

sunshield protects from sun,
earth and moon IR radiation(ISS)

4
Wt true rate

Wheels

Deployable
secondary
Mirror (SM)

Spacecraft support module

SSM (attitude control,
communications, power,
data handling)

cold side

Open telescope (no

external baffling) OTA
Science
allows passive
Instruments
cooling to ~50K
Beryllium
(ISIM)
Isolation truss
Primary mirror (PM)

arm side

Attitude Jitter and Image Stability

total LOS error at target
is the RSS of these terms
pure LOS error from
uncompensated high-frequency
disturbances plus guider NEA

*
Guide Star

Source: G. Mosier
NASA GSFC

*
Target

roll about boresight produces

image rotation (roll axis shown
to be the camera boresight)

Guider

FSM rotation while guiding on a

star at one field point produces
image smear at all other field points

Camera

Important to assess impact of attitude jitter (stability) on image

quality. Can compensate with fine pointing system. Use a
guider camera as sensor and a 2-axis FSM as actuator.

Rule of thumb:
Pointing Jitter
RMS LOS < FWHM/10

E.g. HST: RMS LOS = 0.007 arc-seconds

References

James French: AIAA Short Course: Spacecraft Systems Design and

Engineering, Washington D.C.,1995
Prof. Walter Hollister: 16.851 Satellite Engineering Course Notes,
Fall 1997
James R. Wertz and Wiley J. Larson: Space Mission Analysis and
Design, Second Edition, Space Technology Series, Space Technology
Library, Microcosm Inc, Kluwer Academic Publishers