Download as pdf or txt
Download as pdf or txt
You are on page 1of 57

Linear control

Summer Semester 2018

TU Munich 11th June 2018

roberto.lampariello@dlr.de
Contents

We introduce automatic control and PID control as a linear control


method.

The attitude of a satellite defines its orientation in space. Attitude


control mainly consists of a stabilization control problem, to account for
orbital disturbances.

The standard attitude control problem is linear and can be treated with
the powerful linear control theory. We introduce here some of its basic
principles in the time domain. This provides a base for the nonlinear
control problems found in robotics.

Folie 2
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Generalities on automatic control

Aim is to design a control system to obtain a desired system response


Process is component to be controlled, controller that which controls it
Automatic: the control is accomplished without human intervention

Folie 3
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
The process

The process outputs are the variables (states) to be controlled


The process inputs are the variables (actuator actions) that are
manipulated by the controller

Folie 4
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
The controller
The controller outputs are the variables that manipulate the process, i.e.
the actuator actions (forces and torques).
The controller inputs include the desired values of the process output
variables, and may also include the actual values of the process
variables (feedback), to provide an actual error of the process variables.
The objective of the controller is to make the actual values of the process
output variables equal to the desired values, via the actuator actions.

Open loop
control
(no feedback)

Closed loop
control
(with feedback)

Folie 5
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
The control law
The controller acts by means of its control law, which determines the
characteristics of the desired system response. It is the mathematical
relation between the inputs and the outputs of the controller.
Important features that derive from it are:
Response speed - responsiveness of the controller to an error induced
by a disturbance
Accuracy
dynamic overshoot
steady-state error
Stability
Robustness to
errors in model
non-linear and variable process dynamics
effects of unmodelled disturbances
Folie 6
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
System mathematical model

Why do we need a system model?


It is the first step in the feedback control system design
We need models to predict the effects of control actions - in dynamical
system the effects of actions do not occur immediately
Certain control laws are based on a mathematical model of the system
dynamics (see Lesson on Robot Control)

Definition of mathematical model:


Mathematical relationship that relate the output of a system to its input
It should be understood that no mathematical model of a physical system
is exact
We generally strive to develop a model that is adequate for the problem
at hand without making the model overly complex

Folie 7
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Classical (linear) PID control

Folie 8
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Classical PID control (linear control) y(t)
Recall the mass-damper-spring system from L. 3.
Let’s say that we want to control the position y(t) of the
mass m. To do this we use a motor which applies a force F
to mass m.

The equation of motion for this system is given by


F = m𝑦𝑦̈
To achieve a position ydes, we command the motor to apply
𝑡𝑡
a force F = −K p 𝑒𝑒 − K d 𝑒𝑒̇ − K i ∫0 𝑒𝑒 𝑡𝑡 𝑑𝑑𝑑𝑑 y(t)
where 𝑒𝑒 = 𝑦𝑦 𝑡𝑡 − 𝑦𝑦𝑑𝑑𝑑𝑑𝑑𝑑 is the difference between the
desired position 𝑦𝑦𝑑𝑑𝑑𝑑𝑑𝑑 and measured position 𝑦𝑦(𝑡𝑡). This is the
general form of the PID controller, with its three terms:
Proportional term K p 𝑒𝑒 mot F
motor
Derivative term K d 𝑒𝑒̇ or
𝑡𝑡
Integral term K i ∫0 𝑒𝑒 𝑡𝑡 𝑑𝑑𝑑𝑑
We can also write the equation of motion in terms of the error state, as
𝑡𝑡
m𝑒𝑒̈ + K d 𝑒𝑒̇ + K p 𝑒𝑒 + K i ∫0 𝑒𝑒 𝑡𝑡 𝑑𝑑𝑑𝑑 = 0
Folie 9
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Classical PID control (linear control) Kp=1.6
Kp=1.1
Proportional gain Kp
Term K p e proportional to error Kp=0.5
If Kp is too high, system can become unstable ydes
If Kp too low, too small response to
disturbances Kd=0.5 K =2
d
Derivative gain Kd
Term K d e proportional to error derivative
Kd=1
Improves settling time and stability
Slows down transient ydes
Sensitive to noise due to derivative
Integral gain Ki Ki=2
t

Term K i e(τ ) dτ proportional to magnitude
0
of error and its duration
Eliminates steady-state error Ki=1
Ki=0.5 ydes
If Ki too high, overshoot

Folie 10
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Classical PID control (linear control)
The control law is then of the form:
t
u (t ) = K p e + K i ∫ e( τ )dτ + K d e
0

where u is the control variable (actuator action) and 𝑒𝑒(𝑡𝑡) = 𝑦𝑦𝑑𝑑𝑑𝑑𝑑𝑑 (𝑡𝑡) − 𝑦𝑦(𝑡𝑡)

PID stands for: u(t)


P (Proportional) ydes(t)
I (Integral)
D (Derivative)

Each term of the control law has an associated gain: K p , K i , K d . Each of these factors is
determined through tuning. Some tuning methods include Ziegler-Nichols, Damped-
Oscillation, Cohen-Coon and Minimum Error Integral methods.

In modern industry PID structures control more that 95% of closed-loop industrial
processes. The majority of onboard attitude control loops belong to the standard family
of PID loops.
Folie 11
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Attitude control

Folie 12
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Motivation for attitude control

Communication - antenna pointing


Power - solar panel pointing
Payload - camera/telescope/robot pointing
Orbit maneuvers - thrust in right direction
Synchronization maneuvers - thrust to give right rotational motion

Stabilization: maintain a given fixed attitude

Slew maneuvers: reach a new given attitude

Folie 13
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Causes of attitude disturbances

Gravity gradient

Air drag S/C body


Solar pressure
Centre of
mass
Magnetic forces
Centre of
Structural dynamics pressure
Micrometeoroids impacts Solar panel

Space debris impacts


Sun light
Internal forces (robot, or drag
solar appendages)

Folie 14
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Single rigid body assumption
We limit ourselves to considering the attitude control problem for a “normal-
sized” spacecraft, which can be modelled as a single rigid body.

When is the single rigid body assumption not valid?


The structural flexibility of a spacecraft, e,g. of solar appendages, with an
oscillatory dynamics under mechanical and thermal loads, can be seen
as a source of small disturbances to the attitude. Due to the high-
precision pointing requirements which some satellites have, this may turn
out to be non-negligible.
A robotic spacecraft (free-flying robot), which is effectively a multibody
system: the dynamic coupling introduces a large disturbance on the
carrying satellite.
For “large” spacecraft (e.g., ISS) flexibility is intrinsic to the main
structure.

Folie 15
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Active vs. passive attitude control

The control problem may be solved via a passive approach or an active


approach, depending on the mission requirements:
Active attitude control: on-board actuators are used in conjunction with
the on-board computer and suitable control laws. These are generally
based on the theory of automatic control.
Passive attitude control: natural torques, such as the gravity gradient
or magnetic torques or spin stabilization as seen in L.3, are used to the
advantage of the mission. These are very attractive methods since they
require little engineering effort and actuation energy (if at all). However,
they do not allow a high pointing precision.

Folie 16
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
The onboard control system of a spacecraft

The tasks and functions of an automatic onboard system are the following:

Preparation and execution of maneuvers and control of trajectory and


attitude (guidance, navigation and control – GNC)
Sequencing of phases, GNC modes or maneuvers and scheduling of
equipment for such modes (mission and vehicle management – MVM)
Detection and recovery from system and equipment failures and from
critical state vector deviations (failure detection, isolation and recovery -
FDIR)
Data exchange with the ground control center

Folie 17
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2016 | Orbit dynamics
Guidance, navigation and control
The tasks and functions of the GNC are:

The navigation function consists of a Kalman filter, which processes


the various information of attitude (gyros, updated by Sun, Earth and/or
star sensors) and position sensors (GPS, RGPS) and propagates the
spacecraft state in position and attitude by using the dynamic model and
the information of the actual thruster command.
The guidance function defines the set values for the nominal evolution
of the spacecraft state, i.e. the references for the control of the position,
velocities, attitude and angular rates at each point in time.
The control function produces the force and torque commands
necessary to achieve the desired corrections in attitude and position and
to ensure stability of the vehicle.

Folie 18
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2016 | Orbit dynamics
Typical closed loop control for one of the six DOFs

Sensors GNC functions Actuators


θdes
guidance
rendezvous
sensors τdes thrusters

θis navigation actuator reaction


τis
GPS receiver -+ controller
filter management wheels

attitude θ̂ is momentum
sensors gyros

measurement plant
of environment dynamics
spacecraft & disturbances control
state as forces
seen by dynamic
sensors Spacecraft dynamics disturbances
and environment

Controller computes τdes based on difference θ des - θ̂ is Source: Fehse


Folie 19
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2016 | Orbit dynamics
Basic linearized attitude control equations
Generally speaking, the spacecraft attitude dynamics equations are three
second-order, coupled nonlinear equations. Automatic control theory does
not provide exact analytical solutions and design procedures for such
dynamic plants.
However, linearization of the equations allows to use standard automatic
control techniques. Using Euler angles, and for small Euler angle rotations,
the following approximation for the differential kinematics holds:
p ≈ ψ , q ≈ ∆ , r ≈ φ
The linearized dynamics, assuming a diagonal inertia matrix, then follows as:
 = τ , I φ = τ
I xψ = τ x , I y ∆ y z z
which are three single-axis (decoupled) second-order dynamic equations.

To control such a system control torques are necessary, which are function
of the attitude error. These control torques equations can be written using
the classical PID control method.

Folie 20
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Spacecraft attitude control loop

A simplified block diagram of a typical spacecraft attitude control loop:

If the switch is open there is an open-loop system where the controller


response is independent of the actual output
With the switch closed we have a feedback system

Folie 21
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
PD control for spacecraft attitude
Recalling the linearized attitude dynamics equations in S.16, considering
only one degree of rotation, θ M, and adding a reaction wheel, we obtain
𝐼𝐼𝜃𝜃̈M + ℎ̇ = 𝑁𝑁𝐷𝐷
where 𝐼𝐼 is the satellite inertia about the given rotation axis, ℎ is the
angular momentum of the reaction wheel and 𝑁𝑁𝐷𝐷 any external torque.
Note that ℎ̇ provides an actuation torque, which we choose as our control
variable, NC .

Folie 22
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
PD control for satellite attitude
A common control law for spacecraft is the PD control law:

( )
N C = −h = − K d θM − θREF − K p (θ M − θ REF ) = − K dθM − K p (θ M − θ REF )
where θREF is constant. Substituting in the equation of motion we then get
the error dynamics of the controlled system as

I θ + K dθ + K pθ = N D

with error θ =θM – θREF. This is a forced mass-damper-spring system!

In other words, our choice of the control law allows us to model the error
dynamics as a mass-damper-spring system. We can then easily suitably
defined parameters (stiffness, damping, stability, transient, etc.) to obtain
the desired system response.

Folie 23
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Analysis of PD control for spacecraft attitude
Such a control law has different advantages:
it is easy to understand intuitively, since the error dynamics behaves
like a mass-damper-spring system, which is easy to imagine
it is easy to derive a mathematical theory for such a system - linear,
constant coefficients - and therefore easily determine its behavior for
different disturbances
In fact, recalling that: ω0 = k m ζ = c 2 mk
we have: ω0 = K p I ζ = K d 2 IK p
We can choose the desired system response by suitably defining the
control gains Kp and Kd.
To analyze the system response we introduce a step function of size ND
as the disturbance input. The time response is then given by
θ (t) =
ND
Kp
[ (
1- 1- ζ2 )−1 2
exp(− ζω0t ) sin (ω1t +ψ ) ]
where ω1 = ω0 1− ζ 2 ψ = arctan (1 - ζ 2 ) [ 12
]
ζ = arccos(ζ )
Folie 24
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Analysis of PD control for satellite attitude
Example for ND /Kp = 1
ρ=ζ

UNDERSHOOT

Folie 25
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Analysis of PD control for satellite attitude

With 2 / 2 ≤ ζ ≤ 1 there is only 1 overshoot and no undershoots.


At ζ = 1 the system is critically damped, however the system response is
slow.

The system performance can be expressed in terms of the acceptable


steady-state error, which is the difference between the desired output and
the actual output after the system has settled:

lim ND
θ steady −state-error = θ (t ) = radians
t →∞ Kp

How could this steady-state error be eliminated? We could include an


Integrator term in the control law.

Folie 26
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Example of attitude controller for slewing (closed loop)


τ Plant γ, θ rw

τ − control variable
γ − controlled attitude angle
θ − reaction wheel position
Actuation via High Torque Wheels (BIROS, DLR): challenge in
structural vibration suppression due to large rotational rates of
wheels combined with small misalignment of CoM w.r.t. rotation axis.
Folie 27
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Attitude control: passive techniques
Gravity gradient
τ M

Nadir (downward) pointing only F1


CM
2-axis only M
R R1
F= µ M / R2, therefore F1 < F2 F2

For Cubesats need long extendable booms


R2

Earth
Lumped-mass satellite model

Folie 28
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Attitude control: passive techniques
Passive magnetic
Aligned with geomagnetic field
2-axis only
Simple
τmt = mm x B

Geomagnetic field Relation of spacecraft magnetic


moment mm to geomagnetic field B
Folie 29
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Sensors for attitude control
While the measurement of a spacecraft orbit is based on ground systems
(ground station antennas) and space components (transponders, receivers),
the determination of the spacecraft attitude requires dedicated sensors
onboard.

Usual on-board attitude sensors are

Sun sensor - absolute attitude measurement


Earth sensor - ” ”
Magnetometers - ” ”
Star sensor - ” ”
Gyroscope - absolute angular rate measurement
Visual camera - relative attitude measurement
LIDAR - relative attitude measurement

Folie 30
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Satellite Sensors

Folie 31
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Actuators for attitude control
Usual attitude actuators are:

that exert external torque on spacecraft:


Thrusters (or gas jets) - require fuel
Gravity gradient - no fuel or energy required
Magnetic coils - no fuel required

that exert internal torque on spacecraft:


Reaction wheels - no fuel required
(or rotors – see Lesson 2)
Control Momentum Gyros - no fuel required

Note: internal actuators such as reaction wheels work on the principle of


conservation of angular momentum. As such, to compensate for the
effect of the external orbital torques, they have to constantly
accelerate. This eventually leads to their saturation. Desaturation is
regularly necessary via the thrusters.
Folie 32
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Satellite Actuators

Folie 33
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Actuators for attitude control

Add figure 15-19 in Wertz

Folie 34
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Actuators for attitude control

Source: Wertz

Folie 35
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Sensors and actuators for attitude control
(not part of exam material)

Folie 36
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Sun Sensor
The most simple sun sensor consists basically of a solar cell where
the output current is proportional to the cosine of the incident angle:

Folie 37
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Sun Sensor
Sun sensors providing a digital output signal are usually made of an array of
solar cells of different sizes and locations, masked by a slit:

Folie 38
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Earth Sensor
Earth sensors are typically used on geostationary satellites and sense the
warm infrared radiation of the Earth disc in front of the cold space.

Compared to the Sun (0.5o) the Earth disc is much larger (17.4o from GEO)
and it is more difficult to measure the vector pointing to the centre.

Spinning spacecrafts (e.g. Meteosat) use horizon scanners made of 2 pencil-


beam bolometers (thermometers to measure infrared radiation).

The Earth direction is determined as a function of the spin rate ω, the times
between horizon crossings Δt1 and Δt2 and the angle 2α between the two
bolometer beams. ω

ω∆t1

ω

ω∆t2

Folie 39
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Magnetometers
Magnetometers are used on the largest part of Earth orbiting spacecrafts.

They are reliable, have low mass and low power requirements.

They are insensitive towards radiation and temperature gradient and have no
moving parts.

Since the Earth magnetic field decreases with distance by 1/r3, the accuracy
also decreases with increasing altitude.

Today however even the Galileo spacecrafts will be equipped with


magnetometers.

Three magnetometers are required to


determine the Earth’s magnetic field vector.

Folie 40
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Star sensor

Star sensors are the most precise attitude sensors and hence the most
important ones during nominal operations.

Operating a star sensor is based on loading a star catalogue into the onboard
computer (usually the AOCS processor).

For each tracked star its direction is measured within the star sensor
reference frame fixed to the sensor optical head.

Folie 41
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Star sensor

A star sensor usually consists of

• Optical system (lens + stray light shading)


• Detector for starlight (e.g. CCD array)
• Electronics and signal processing hardware

Folie 42
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Star sensor: example Cryosat
Star tracker accommodation and Sun exclusion zones:

Folie 43
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Star sensor: example Cryosat
Operational star tracker selection (predicted and uplinked by control
centre)

Folie 44
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Gyroscopes
Rate gyros don’t provide an absolute attitude like the sensors mentioned
before, but rather measure spacecraft angular rates.

They require an initial attitude reference, from which numerical integration of


the rates yields the attitude change information.

Because of the integration process, gyros are characterized by a bias,


typically of 1 deg./hour.

Mechanical gyros have the drawback of


rotating mechanical parts which are subject
to degradation.

Most gyros today use a different


measurement principle, based on
interferometric effects between
laser signals (fibre-optic gyros,
ring- laser gyros).
Folie 45
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Gyroscopes
The output of a mechanical rate gyro is generated by measuring the rotation
of the gimbal about the output axis:

Spinning disk Output gimbal
(rotor) rotation axis

θ measured about Ô
Gimbal

Spring at joint


Spin axis
of rotor

Frame
Î Input axis attached to satellite
Folie 46
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Gyroscopes
The relation between the angular rate about the input axis Î and the angular
displacement θ about the output axisÔ can be derived via the total angular
momentum of the system composed of the gimbal and the rotor, expressed as

Γ = I 0θ O
ˆ + LSˆ (1)
where L = LS ˆ is the angular momentum of the rotor, I0 is the moment of inertia
of the gimbal system about the output axis.
Given the angular velocity vector as ω = ωIIˆ + ωO O
ˆ + ω Sˆ , we can write the
S
equations of motion as (see Lesson 5)
dΓ 2 dΓ
= + ω×Γ = τ (2)
dt dt
Furthermore, the torques acting at the joints of the gimbal are given as

τ O = −( Kθ + Dθ) Ô (3)

to represent the spring restoring force and a damping friction force.

Folie 47
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Gyroscopes
Substituting (1) in (2) we obtain
2
d ( LSˆ + Iθ O
ˆ)
+ ω × ( LSˆ + Iθ O
ˆ)=τ
dt
With (3) and for the component along the Ô axis, we obtain

I 0θ + Dθ + Kθ − ωI L = 0
Coupling term
The steady-state solution is then
gyroscopic effect:
ωI L
θ=
K The rotation of the angular
momentum vector Γ,
where the sensor output θ is proportional to the directed along Ŝ , by ω I ,
satellite angular rate ω I. directed along Î , gives rise
to a rotation directed along
Ô !
Folie 48
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Satellite Sensors

Folie 49
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Reaction wheels
Torquers with the capability of momentum storage such as reaction
wheels and momentum wheels are internal torquers.
Momentum wheels operate at high rpm and provide spin
stability. The delivered torque is in the range of 0.01 – 1 Nm.
Even if there is no torque demand by the attitude controller the
wheel is kept at a certain idle speed in order not to have any
stiction effects.
Reaction wheels rotation speed is varied to trade angular
momentum with the spacecraft for attitude control
Also here 3 orthogonal wheels vary all
components of angular momentum about the
spacecraft body axes.
A 4th wheel at an oblique angle provides
redundancy and can replace any of the other
wheels.

Folie 50
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Reaction wheels
The angular momentum exchange can be well understood when writing the
expression for the angular momentum of the system satellite plus reaction
wheels (1D):
H sat ω sat + H rw θ rw = K ⇒ θ rw = H -1rw (K - H sat ω sat )
where K is an arbitrary constant representing the total angular momentum
of the system. Note that this is constant, with respect to the torques acting
between the satellite and the reaction wheels, which are internal.
From the above equation, we can compute the necessary reaction wheel
 for a desired satellite angular velocity ω sat .
angular velocity θ rw
In the presence of an external torque τ, we can write the equations of
motion as
(
d H sat ω sat + H rw θ rw

)
dt
( )
If we let d H rw θ rw dt = τ, we ensure that d (H sat ω sat ) dt = 0 , which is the
goal of the attitude controller, given that ω sat (0) = 0, i.e. ω
 sat = 0 .
Note however, that this implies that θ rw ≠ 0 , leading to rw saturation.
Folie 51
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Reaction wheels
After a certain time of operations a reaction wheel will reach its limit
of angular velocity and is not able to further provide torques in this
direction.
The reaction wheel must be decelerated
This is called desaturation or momentum dumping.
The produced deceleration torque during this procedure must be
compensated by other actuators, e.g. magnetic torquers or thruster.

Venus Express
example:
desaturation
every 24 hrs

Folie 52
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Control Momentum Gyros
Following the discourse on the gyroscopes, it is evident that we can
consider inverting the roles of such a device, i.e. generating torques
by commanding a gimbal rotation and thereby changing the spin axis
orientation. The resulting torque on the satellite is approximately
given by: τ I = ωO L = θ L
The output torque for a 100 kg mass can arrive to 0(1000) Nm.
Due to the expense and weight, CMGs are only used on large
spacecraft.
Due to their high mechanical stress, a redundant
number are usually mounted on a given spacecraft,
e.g. six for the Hubble telescope, and 4 on the
International Space Station (see figure).

Folie 53
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Magnetic Torquerods

Torqrods are used extensively in the attitude control of spacecraft. They are
designed to generate controllable magnetic dipole moments.

They are used for active damping in gravity gradient attitude-stabilized


spacecraft. They are also used for unloading of momentum exchange
devices.

Torquerods are useful because they are substitutes for consumables (such as
fuel for thrusters), thus reducing weight.

At 400 km altitude, the maximum achievable torque is 2.56 x 10-3 Nm. At


geostationary altitude it is only 10.46 x 10-6 Nm.

Control systems using torquerods must usually include a magnetometer for


estimating the earth’s magnetic field.

Folie 54
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Thrusters
Thrusters produce forces by expelling propellant in the opposite direction.

Hot gas thrusters produce forces by the acceleration of


propellant molecules where the energy comes from either a
chemical reaction or a thermodynamic expansion. Thrust levels
can be in the range of 0.1 – 500 N.

Cold gas thrusters expel fuel molecules (typically Nitrogen)


where the energy comes from previous compression of the fuel.
Achievable thrust is typically in the range of 5 N.

Ion thrusters accelerate ionized molecules electrodynamically


where the energy eventually comes from solar cells. Achievable
thrust is typically in the range of 5 – 20 x 10-3 N.

Folie 55
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Thrusters

Folie 56
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Literature

Wertz: “Spacecraft Attitude Determination and Control”, Reidel


Publishing Company, 1980.
Hughes, “Spacecraft Attitude Dynamics”, 2004.
Sidi, M., “Spacecraft Dynamics and Control”, 2006.

Folie 57
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control

You might also like