Skip to content

Chapter 7: Attitude Determination and Control Systems

Key Concepts in This Chapter

  • Reference frames: ECI, ECEF, orbital (RSW), and body frames

  • Attitude representations: Euler angles, DCM, quaternions, and Modified Rodrigues Parameters

  • Kinematic and dynamic equations of motion for a rigid spacecraft

  • Environmental disturbance torques: solar radiation pressure, gravity gradient, aerodynamic, and magnetic

  • Attitude sensors: star trackers, sun sensors, magnetometers, gyroscopes, and GPS

  • Attitude determination algorithms: TRIAD, QUEST, and Extended Kalman Filter

  • Attitude actuators: reaction wheels, CMGs, magnetorquers, and thrusters

  • Control laws: PD controller, LQR, and sliding mode control

7.1 Introduction to Attitude Determination and Control

The Attitude Determination and Control System (ADCS) is the spacecraft subsystem responsible for measuring the spacecraft’s orientation in space and actively controlling it to the commanded state. The attitude of a spacecraft—its three-dimensional orientation relative to some reference frame—determines fundamentally important quantities: whether the solar arrays face the Sun, whether the antennas point toward the ground, whether the imaging instruments view the intended targets, and whether the thrusters fire in the correct direction.

ADCS design spans an enormous range of complexity, from passively spin-stabilised or gravity-gradient-stabilised spacecraft with no active control at all, to highly agile Earth observation satellites with arc-second pointing accuracy maintained by reaction wheel systems and precision star trackers, to interplanetary probes that must point radio antennas with milli-degree accuracy across billions of kilometres. The unifying mathematical framework for all these applications is the subject of this chapter.

7.2 Reference Frames

Attitude is always defined relative to a reference frame. The major frames used in spacecraft engineering are:

7.2.1 Earth-Centred Inertial Frame (ECI)

The ECI frame is a (quasi-)inertial frame with its origin at the Earth’s centre. The fundamental realisation is the International Celestial Reference Frame (ICRF), implemented through the J2000.0 epoch:

  • \(\hat{x}_I\): Points toward the vernal equinox (direction of the Sun on the first day of spring) at epoch J2000.0 (1 January 2000, 12:00 TT)

  • \(\hat{z}_I\): Points toward the North Celestial Pole (NCP), aligned with the Earth’s mean rotation axis at J2000.0

  • \(\hat{y}_I = \hat{z}_I \times \hat{x}_I\): Completes the right-handed triad

The ECI frame does not rotate with the Earth; it is approximately inertial over mission timescales (precession of the equinoxes causes slow drift, corrected through proper motion updates). Spacecraft positions and velocities in orbital mechanics are primarily expressed in ECI (Vallado 2013).

7.2.2 Earth-Centred Earth-Fixed Frame (ECEF)

The ECEF frame rotates with the Earth, with \(\hat{x}_E\) pointing toward the prime meridian at the equator, \(\hat{z}_E\) toward the geographic North Pole. Its relationship to ECI involves a rotation by the Greenwich Apparent Sidereal Time (GAST) angle \(\theta_{GAST}\) about the \(\hat{z}\) axis, plus small corrections for polar wander and UT1-UTC offset. Ground station coordinates, coverage analysis, and geolocation computations use ECEF.

7.2.3 Orbital (RSW) Frame

The Local Vertical Local Horizontal (LVLH) or RSW frame (also called the Hill frame) is defined relative to the spacecraft’s orbital position:

  • \(\hat{r}\) (Radial): Points from Earth’s centre outward through the spacecraft (nadir direction is \(-\hat{r}\))

  • \(\hat{s}\) (Along-track): In the orbital plane, perpendicular to \(\hat{r}\), in the direction of spacecraft velocity for circular orbits

  • \(\hat{w}\) (Cross-track): Normal to the orbital plane, \(\hat{w} = \hat{r} \times \hat{s}\) (for prograde orbit, \(\hat{w}\) points north)

For a nadir-pointing Earth observation satellite, the nominal attitude has the payload boresight along \(-\hat{r}\). Deviations from nadir pointing are the Euler angles roll, pitch, and yaw in the RSW frame.

7.2.4 Body Frame

The body frame is fixed to the spacecraft structure, typically aligned with the spacecraft’s principal axes of inertia. Attitude sensors, actuators, and payload boresights are defined in the body frame. The ADCS task is to drive the rotation from the body frame to the desired reference frame (e.g., the orbital frame for nadir pointing) to zero (or to a commanded target).

7.3 Attitude Representations

7.3.1 Euler Angles

Euler angles describe a sequence of three elemental rotations. The 3-2-1 (or 321) sequence, corresponding to yaw-pitch-roll, is most common in aerospace:

  1. Rotate by \(\psi\) (yaw) about \(\hat{z}\)

  2. Rotate by \(\theta\) (pitch) about the new \(\hat{y}'\)

  3. Rotate by \(\phi\) (roll) about the new \(\hat{x}”\)

The composite rotation matrix (Direction Cosine Matrix, DCM) is:

\[C = R_1(\phi) R_2(\theta) R_3(\psi)\tag{7.1}\]

where:

\[R_1(\phi) = \begin{bmatrix} 1 & 0 & 0 \\ 0 & \cos\phi & \sin\phi \\ 0 & -\sin\phi & \cos\phi \end{bmatrix}, \quad R_2(\theta) = \begin{bmatrix} \cos\theta & 0 & -\sin\theta \\ 0 & 1 & 0 \\ \sin\theta & 0 & \cos\theta \end{bmatrix}\]

Euler angles suffer from the gimbal lock singularity: when \(\theta = \pm 90\degree{}\) for the 321 sequence, the first and third rotations become about the same axis and the representation is degenerate. This limits the use of Euler angles in attitude control software for highly manoeuvrable spacecraft.

7.3.2 Direction Cosine Matrix

The DCM \(C \in \mathbb{R}^{3\times3}\) is a proper orthogonal matrix (\(C^T C = I\), \(\det(C) = +1\)) with 9 elements satisfying 6 orthogonality constraints, leaving 3 independent parameters encoding the attitude. The DCM transforms a vector \(\mathbf{v}_A\) expressed in frame \(A\) to its representation \(\mathbf{v}_B\) in frame \(B\):

\[\mathbf{v}_B = C_{BA} \mathbf{v}_A\]

The kinematic differential equation for the DCM is:

\[\dot{C} = -[\boldsymbol{\omega}]_\times C\tag{7.4}\]

where \([\boldsymbol{\omega}]_\times\) is the skew-symmetric cross-product matrix of the angular velocity vector \(\boldsymbol{\omega}\):

\[[\boldsymbol{\omega}]_\times = \begin{bmatrix} 0 & -\omega_3 & \omega_2 \\ \omega_3 & 0 & -\omega_1 \\ -\omega_2 & \omega_1 & 0 \end{bmatrix}\]

7.3.3 Quaternions

The unit quaternion (also called Euler parameter or rotation quaternion) provides a singularity-free attitude representation with only one constraint (unit norm). By Euler’s theorem, any attitude can be represented as a single rotation by angle \(\Phi\) about a unit vector \(\hat{\mathbf{e}}\) (the Euler axis). The quaternion is defined as:

\[\mathbf{q} = \begin{bmatrix} \boldsymbol{\varepsilon} \\ \eta \end{bmatrix} = \begin{bmatrix} \hat{\mathbf{e}} \sin(\Phi/2) \\ \cos(\Phi/2) \end{bmatrix} = \begin{bmatrix} q_1 \\ q_2 \\ q_3 \\ q_4 \end{bmatrix}\tag{7.6}\]

with the constraint \(\|\mathbf{q}\| = q_1^2 + q_2^2 + q_3^2 + q_4^2 = 1\).

The quaternion kinematic equation is:

\[\dot{\mathbf{q}} = \frac{1}{2} \Xi(\mathbf{q}) \boldsymbol{\omega}\tag{7.7}\]

where the \(4\times3\) matrix \(\Xi(\mathbf{q})\) is:

\[\Xi(\mathbf{q}) = \begin{bmatrix} \eta I_3 + [\boldsymbol{\varepsilon}]_\times \\ -\boldsymbol{\varepsilon}^T \end{bmatrix}\]

Quaternion multiplication for composing two rotations \(\mathbf{q}_1\) followed by \(\mathbf{q}_2\) gives:

\[\mathbf{q}_2 \otimes \mathbf{q}_1 = \begin{bmatrix} \eta_2 \boldsymbol{\varepsilon}_1 + \eta_1 \boldsymbol{\varepsilon}_2 - \boldsymbol{\varepsilon}_2 \times \boldsymbol{\varepsilon}_1 \\ \eta_1 \eta_2 - \boldsymbol{\varepsilon}_1 \cdot \boldsymbol{\varepsilon}_2 \end{bmatrix}\tag{7.9}\]

The inverse (conjugate) quaternion \(\mathbf{q}^{-1} = [-\boldsymbol{\varepsilon}; \eta]\) represents the reverse rotation. The attitude error quaternion between current attitude \(\mathbf{q}\) and commanded attitude \(\mathbf{q}_c\) is \(\delta\mathbf{q} = \mathbf{q}_c^{-1} \otimes \mathbf{q}\).

7.3.4 Modified Rodrigues Parameters

The Modified Rodrigues Parameters (MRP) offer a minimal three-parameter attitude representation:

\[\boldsymbol{\sigma} = \frac{\boldsymbol{\varepsilon}}{1 + \eta} = \hat{\mathbf{e}} \tan(\Phi/4)\tag{7.10}\]

MRPs have a singularity at \(\Phi = 360\degree{}\) (a full rotation), which can be avoided in practice by switching between the “shadow” MRPs \(\boldsymbol{\sigma}^s = -\boldsymbol{\sigma}/\|\boldsymbol{\sigma}\|^2\) when \(\|\boldsymbol{\sigma}\| > 1\). MRPs are differentiable everywhere (except the singularity) and are useful in optimal control and estimation formulations.

7.4 Equations of Motion

7.4.1 Kinematic Equations

The kinematic equations relate the time derivative of the attitude representation to the angular velocity \(\boldsymbol{\omega}\) of the body frame relative to the inertial frame, expressed in the body frame. For quaternions, Equation (7.7) applies. For the DCM, Equation (7.4) applies.

7.4.2 Euler’s Rotational Equations of Motion

For a rigid body, Euler’s rotational equations of motion relate the angular momentum of the spacecraft to the applied external torques. Written in the body frame:

\[I \dot{\boldsymbol{\omega}} + \boldsymbol{\omega} \times (I \boldsymbol{\omega}) = \mathbf{T}_{\mathrm{ext}}\tag{7.11}\]

where \(I \in \mathbb{R}^{3\times3}\) is the spacecraft inertia tensor (symmetric, positive definite), \(\boldsymbol{\omega}\) is the angular velocity in body frame (rad/s), and \(\mathbf{T}_{\mathrm{ext}}\) is the total external torque vector including control torques and environmental disturbances.

When the spacecraft carries rotating wheels (reaction wheels or CMGs), the wheel angular momentum \(\mathbf{H}_w\) must be included:

\[I \dot{\boldsymbol{\omega}} = -\boldsymbol{\omega} \times (I\boldsymbol{\omega} + \mathbf{H}_w) + \mathbf{T}_{\mathrm{ext}} - \dot{\mathbf{H}}_w\tag{7.12}\]

The control torque applied by a reaction wheel is \(\mathbf{T}_c = -\dot{\mathbf{H}}_w\) (the reaction to the torque applied to the wheel).

7.5 Environmental Disturbance Torques

7.5.1 Solar Radiation Pressure Torque

Solar radiation pressure (SRP) exerts a force on any surface exposed to sunlight. The SRP force on a flat surface depends on whether the surface is perfectly absorbing (black), specularly reflecting, or diffusely reflecting. For a spacecraft with solar radiation pressure force \(\mathbf{F}_{\mathrm{srp}}\) not acting through the centre of mass:

\[\mathbf{T}_{\mathrm{srp}} = \mathbf{d}_{\mathrm{cpa}} \times \mathbf{F}_{\mathrm{srp}}\tag{7.13}\]

where \(\mathbf{d}_{\mathrm{cpa}}\) is the vector from the centre of mass to the centre of pressure. The SRP pressure is \(P_{\mathrm{srp}} = S/c = 1361/2.998\times10^8 = 4.56 \times 10^{-6}\) N/m\(^2\) at 1 AU. For a total absorbing plate of area \(A\) perpendicular to the Sun, \(F = P \cdot A\). For a reflective surface with reflectivity \(\rho\), \(F = P \cdot A \cdot (1 + \rho)\). Asymmetric solar array deployment, unequal surface properties, or offset centres of mass versus centres of pressure create residual SRP torques that the ADCS must overcome.

7.5.2 Gravity Gradient Torque

The non-uniform gravitational field of the Earth exerts differential forces on extended bodies, producing a gravity gradient torque that tends to align the spacecraft’s long axis with the local vertical. For a spacecraft with principal moments of inertia \(I_x, I_y, I_z\) (with \(I_z\) along the orbit normal), the gravity gradient torque components in the orbital frame are approximately:

\[T_{gg,x} = 3\omega_o^2 (I_z - I_y) \sin\phi \cos\phi \approx 3\omega_o^2 (I_z - I_y) \phi\tag{7.14}\]
\[T_{gg,y} = \frac{3\mu}{r^3}(I_x - I_z)\sin\theta\cos\theta \approx \frac{3\mu}{r^3}(I_x - I_z)\theta\tag{7.15}\]

where \(\omega_o = \sqrt{\mu/r^3}\) is the mean orbital rate. The gravity gradient effect can be exploited for passive attitude stabilisation in LEO by deliberately designing an elongated spacecraft with \(I_z \ll I_x, I_y\).

7.5.3 Aerodynamic Torque

In low Earth orbit below approximately 600 km, residual atmospheric drag produces torques analogous to those from SRP when the centre of atmospheric pressure does not coincide with the centre of mass. The aerodynamic torque is:

\[T_{\mathrm{aero}} = \frac{1}{2} \rho v^2 C_D A \cdot d_{cp}\tag{7.16}\]

where \(\rho\) is the atmospheric density (strongly dependent on altitude and solar activity, from roughly \(10^{-12}\) kg/m\(^3\) at 400 km to roughly \(10^{-14}\) kg/m\(^3\) at 800 km, varying by an order of magnitude or more with solar activity, as given by empirical models such as NRLMSISE-00), \(v\) is the orbital velocity, \(C_D\) is the drag coefficient (typically 2.2–2.5 for spacecraft in free molecular flow), \(A\) is the cross-sectional area, and \(d_{cp}\) is the moment arm. Aerodynamic torques dominate the disturbance environment for spacecraft below 500 km.

7.5.4 Residual Magnetic Dipole Torque

Spacecraft carry residual magnetic dipole moments from current-carrying conductors, magnetised materials, and hysteresis losses. The resulting torque is:

\[\mathbf{T}_{\mathrm{mag}} = \mathbf{m}_{\mathrm{res}} \times \mathbf{B}\tag{7.17}\]

where \(\mathbf{m}_{\mathrm{res}}\) is the residual magnetic dipole moment (A\(\cdot\)m\(^2\)) and \(\mathbf{B}\) is the geomagnetic field vector. This torque is cyclic (varies with the changing direction of \(\mathbf{B}\) along the orbit) and typically has a zero-mean over an orbit, but peak values can reach about \(3\times10^{-5}\) N\(\cdot\)m for a 1 A\(\cdot\)m\(^2\) residual dipole at LEO field strengths of \(\sim 3\times10^{-5}\) T.

7.6 Attitude Sensors

7.6.1 Star Trackers

The star tracker (or star camera) is the most accurate attitude sensor in operational use. It captures images of the star field using a CCD or CMOS focal plane array and determines the spacecraft’s three-axis attitude by pattern matching against an on-board star catalogue. Modern star trackers achieve attitude determination accuracy of 1–10 arcseconds per axis (1\(\sigma\)) with an update rate of 1–10 Hz.

The lost-in-space (LIS) problem—identifying star patterns without prior knowledge of attitude—is solved by algorithms such as:

TRIAD: Uses two non-collinear measured unit vectors (in body frame) and their known inertial counterparts to compute the DCM directly. Deterministic and fast but uses only two measurements.

QUEST (Quaternion Estimator): Minimises Wahba’s loss function:

\[L(C) = \frac{1}{2} \sum_{i=1}^N w_i \|\mathbf{b}_i - C \mathbf{r}_i\|^2\tag{7.18}\]

where \(\mathbf{b}_i\) are measured unit vectors in body frame, \(\mathbf{r}_i\) are corresponding known reference vectors in inertial frame, and \(w_i\) are weights. QUEST reformulates this as an eigenvalue problem and yields the optimal attitude quaternion.

FOAM (Fast Optimal Attitude Matrix): A computationally efficient variant using SVD.

Star trackers are vulnerable to Solar or lunar blinding when bright objects enter or approach the field of view, and to centroid errors caused by radiation-induced detector hot pixels. Radiation-hardened designs and anti-blinding baffles address these issues.

7.6.2 Sun Sensors

Sun sensors measure the direction of the Sun vector in the body frame. Two types are used:

Coarse Sun Sensors (CSS): Simple photodiode arrays on each spacecraft face. The ratio of illumination levels on adjacent diodes gives the Sun vector direction with accuracy of \(\pm 1--5\degree{}\). Extremely robust and used in safe/acquisition modes when the attitude is highly uncertain.

Fine Sun Sensors (FSS): Two-dimensional quadrant or lateral effect photodiodes, or digital sun sensors with slit apertures, achieving accuracies of \(\pm 0.01--0.1\degree{}\) over a field of view of \(\pm 30--60\degree{}\). Used in nominal operations for solar array pointing verification and as a backup to the star tracker.

7.6.3 Magnetometers

Three-axis fluxgate magnetometers measure the geomagnetic field vector \(\mathbf{B}\) in the body frame to accuracies of 1–100 nT. Comparing the measured \(\mathbf{B}\) with the IGRF (International Geomagnetic Reference Field) model prediction gives attitude information, but with low accuracy (typically \(\pm 1--5\degree{}\)) because the IGRF model itself has errors and the geomagnetic field direction varies slowly. Magnetometers are most useful in LEO and as a sensor for magnetorquer control loops; above 1000 km altitude the field weakens and measurement errors become more significant.

7.6.4 Earth Horizon Sensors

Infrared horizon sensors detect the boundary between the warm Earth (\(\sim 255\) K mean) and cold space (\(\sim 3\) K) in the thermal infrared (\(14--16~\mu\)m CO\(_2\) emission band). Scanning or static two-axis sensors determine the nadir vector direction with accuracy of \(\pm 0.05--0.1\degree{}\) in LEO. Horizon sensors were the primary attitude sensor for earlier spacecraft generations; star trackers have largely replaced them for accuracy-critical applications but they remain in use as robust backup sensors.

7.6.5 Gyroscopes

Gyroscopes measure angular velocity and, by integration, can propagate attitude estimates between absolute sensor updates. Four principal types are used in space:

Table 7.1. Comparison of gyroscope types for space applications.
Type ARW Bias Instability Mass Applications
(/\(\sqrt{\mathrm{h}}\)) (/h) (kg)
MEMS (ADXRS, IMU) 0.1–1.0 1–100 0.01–0.1 CubeSats, attitude rate sensing
Fiber Optic (FOG) 0.001–0.1 0.01–1.0 0.2–2 LEO, medium accuracy
Ring Laser (RLG) 0.001–0.01 0.001–0.1 0.5–3 High accuracy, legacy systems
Hemispherical Resonator (HRG) 0.0001–0.001 0.0001–0.01 0.5–2 Long-life precision missions

The HRG (Northrop Grumman) uses the Coriolis effect in a fused quartz hemispherical shell vibrating at its fundamental mode. With no moving parts and solid-state operation, it offers unmatched lifetime (> 20 years demonstrated) and is the sensor of choice for high-reliability long-duration missions.

Angle Random Walk (ARW) characterises high-frequency noise in the angular rate measurement; it causes attitude estimation error to grow as \(\sqrt{t}\) during gyro-only propagation. Bias instability characterises the low-frequency drift of the gyro output; it dominates attitude error over long time intervals between absolute sensor updates.

7.7 Attitude Determination Algorithms

7.7.1 Extended Kalman Filter for Attitude Estimation

The Extended Kalman Filter (EKF) is the standard algorithm for fusing gyroscope measurements with periodic absolute attitude measurements from star trackers or other sensors. The state vector for attitude estimation is:

\[\mathbf{x} = \begin{bmatrix} \delta\boldsymbol{\theta} \\ \boldsymbol{\beta} \end{bmatrix} \in \mathbb{R}^6\tag{7.19}\]

where \(\delta\boldsymbol{\theta} \in \mathbb{R}^3\) is the small-angle attitude error (Gibbs vector or MRP of error quaternion), and \(\boldsymbol{\beta} \in \mathbb{R}^3\) is the gyro bias estimate vector. The propagation (predict) step uses the gyro measurement \(\tilde{\boldsymbol{\omega}}\):

\[\dot{\hat{\mathbf{q}}} = \frac{1}{2}\Xi(\hat{\mathbf{q}})(\tilde{\boldsymbol{\omega}} - \hat{\boldsymbol{\beta}})\tag{7.20}\]
\[\dot{\hat{\boldsymbol{\beta}}} = \mathbf{0}\tag{7.21}\]

The state error covariance \(P\) is propagated via:

\[\dot{P} = F P + P F^T + G Q G^T\tag{7.22}\]

where \(F\) is the linearised system matrix, \(Q\) is the process noise covariance (characterised by the gyro ARW and bias random walk), and \(G\) maps noise into the state.

When a star tracker measurement \(\mathbf{q}_{\mathrm{meas}}\) arrives, the measurement update (correct) step is:

\[\delta\boldsymbol{\theta}_{\mathrm{meas}} = 2 \boldsymbol{\varepsilon}(\hat{\mathbf{q}}^{-1} \otimes \mathbf{q}_{\mathrm{meas}})\tag{7.23}\]
\[K = P H^T (H P H^T + R)^{-1}\tag{7.24}\]
\[\hat{\mathbf{x}}^+ = \hat{\mathbf{x}}^- + K (\delta\boldsymbol{\theta}_{\mathrm{meas}} - H \hat{\mathbf{x}}^-)\tag{7.25}\]
\[P^+ = (I - KH) P^-\tag{7.26}\]

where \(R\) is the measurement noise covariance and \(H = [I_3 \; \mathbf{0}_{3\times3}]\) (star tracker measures attitude error, not bias). The attitude estimate \(\hat{\mathbf{q}}\) is then updated by rotating by the estimated error \(\delta\boldsymbol{\theta}\).

7.8 Attitude Actuators

7.8.1 Reaction Wheels

A reaction wheel (RW) is a motor-driven flywheel mounted in the spacecraft body. By accelerating or decelerating the flywheel, the ADCS exchanges angular momentum with the spacecraft body, generating a control torque. The torque provided by a single reaction wheel is:

\[T_{rw} = \dot{h}_w = I_w \dot{\Omega}_w\tag{7.27}\]

where \(I_w\) is the wheel moment of inertia and \(\Omega_w\) is the wheel spin speed. The angular momentum stored in the wheel is \(h_w = I_w \Omega_w\).

For three-axis attitude control, at least three reaction wheels (one per axis) are required, though most spacecraft use four wheels in a pyramidal or skewed arrangement to provide redundancy against single-wheel failure. Standard wheel speed range is \(\pm 3000\)–\(6000\) RPM; momentum capacity ranges from 0.4 Nms for a small LEO satellite to 100 Nms for a large GEO platform.

Figure 7.1. Angular momentum envelope for a four-reaction-wheel system in a pyramidal configuration. Each wheel can store momentum \(h_w\); the combined momentum is the vector sum. When any wheel approaches its saturation speed \(\Omega_{\max}\), momentum must be dumped using an external torque source (magnetorquers or thrusters).

Momentum desaturation: Over time, reaction wheels accumulate angular momentum from residual environmental disturbance torques that have a mean value. The wheels gradually spin up (or down) toward their speed limits. Momentum must be periodically unloaded using external torque sources: magnetorquers exploit the Earth’s magnetic field (usable in LEO but not in GEO), or thrusters fire in pairs to generate a net torque without translational acceleration.

Micro-vibrations: Reaction wheel imbalance (both static and dynamic) generates harmonic vibrations at the wheel spin frequency and harmonics. These micro-vibrations propagate through the spacecraft structure and can limit the pointing accuracy of sensitive optical instruments. Vibration isolation mounts (passive: elastomeric bushings; active: piezoelectric actuators in hexapod arrangements) attenuate micro-vibrations. The SOFIA airborne observatory and the Hubble Space Telescope have employed wheel isolation systems for this reason.

7.8.2 Control Moment Gyroscopes

A Control Moment Gyroscope (CMG) is a device in which a spinning flywheel is mounted in gimbals. The flywheel spins at constant high speed, and the output torque is generated by rotating (gimballing) the spinning wheel. The gyroscopic effect amplifies the gimbal input torque into a much larger output torque on the spacecraft.

For a single-gimbal CMG (SGCMG), the output torque is:

\[\mathbf{T}_{\mathrm{cmg}} = h_0 \dot{\delta} \hat{\mathbf{t}}\tag{7.28}\]

where \(h_0\) is the constant flywheel angular momentum magnitude, \(\dot{\delta}\) is the gimbal rate, and \(\hat{\mathbf{t}}\) is the torque direction (perpendicular to both the spin axis and gimbal axis). CMGs can provide torques 10–100 times greater than reaction wheels of equivalent mass, making them attractive for agile Earth observation satellites (e.g., Pleiades, WorldView series).

The fundamental challenge of CMG arrays is the singularity problem: for certain combinations of gimbal angles, the output torque vectors of all CMGs become coplanar and no torque can be produced in the orthogonal direction. Singularity avoidance and escape algorithms are an active research area.

7.8.3 Magnetorquers

Magnetorquers (also called torque coils or magnetic torque rods) generate a magnetic dipole moment \(\mathbf{m}\) by passing current through a coil or rod (ferromagnetic core wound with wire), which interacts with the Earth’s geomagnetic field to produce a torque:

\[\mathbf{T} = \mathbf{m} \times \mathbf{B}\tag{7.29}\]

The dipole moment is \(\mathbf{m} = N I A \hat{\mathbf{n}}\) where \(N\) is the number of turns, \(I\) is the current, \(A\) is the coil area, and \(\hat{\mathbf{n}}\) is the coil normal. The fundamental limitation of magnetorquers is that the torque is always perpendicular to \(\mathbf{B}\): full three-axis control is not possible at any given instant. Over time (one orbit or fraction thereof), the varying direction of \(\mathbf{B}\) allows three-axis control authority. Magnetorquers are widely used for momentum desaturation and detumbling in LEO satellites, where field strengths of \(20\)–\(60~\mu\)T are available.

7.8.4 Thruster-Based Attitude Control

Chemical thrusters provide attitude control torques in several operational modes: large-angle slews during LEOP, emergency de-tumbling, momentum desaturation (especially in GEO where magnetorquers cannot be used), and reorientation manoeuvres during orbit maintenance burns. Thrust is typically provided in on/off pulsed mode; the minimum impulse bit (the smallest controllable impulse) is an important parameter for fine attitude control. Plume impingement on solar arrays, antennas, and sensors is a critical concern in thruster placement.

7.9 Control Laws

7.9.1 PD Controller

The classical three-axis PD (Proportional-Derivative) attitude controller is:

\[\mathbf{T}_c = -K_p \delta\boldsymbol{\theta} - K_d \boldsymbol{\omega}\tag{7.30}\]

where \(\delta\boldsymbol{\theta}\) is the attitude error vector (small angle representation), \(\boldsymbol{\omega}\) is the angular velocity error (actual minus commanded), \(K_p\) is the proportional gain matrix, and \(K_d\) is the derivative gain matrix. For a symmetric spacecraft with diagonal inertia matrix \(I = \mathrm{diag}(I_x, I_y, I_z)\), the gains are decoupled: \(K_{p,i} = I_i \omega_{n,i}^2\) and \(K_{d,i} = 2 \zeta_i \omega_{n,i} I_i\) where \(\omega_{n,i}\) is the natural frequency and \(\zeta_i\) is the damping ratio of each axis. Critical damping is achieved at \(\zeta = 1\).

The PD controller eliminates constant disturbances only if an integrating term (PID) is added, or if integral action is implemented through a reaction wheel bias (momentum-bias system). For disturbances that are constant in the body frame, the PD controller will exhibit a steady-state pointing error.

7.9.2 LQR Optimal Control

The Linear Quadratic Regulator (LQR) provides an optimal control law minimising the cost functional:

\[J = \int_0^\infty \left( \mathbf{x}^T Q \mathbf{x} + \mathbf{u}^T R \mathbf{u} \right) dt\tag{7.31}\]

where \(\mathbf{x}\) is the state vector (attitude error and angular velocity), \(\mathbf{u}\) is the control input (wheel torques), \(Q \geq 0\) is the state weighting matrix, and \(R > 0\) is the control effort weighting matrix. The optimal control is:

\[\mathbf{u}^* = -K_{\mathrm{LQR}} \mathbf{x}\tag{7.32}\]

where \(K_{\mathrm{LQR}} = R^{-1} B^T P_\infty\) and \(P_\infty\) is the solution to the algebraic Riccati equation:

\[P A + A^T P - P B R^{-1} B^T P + Q = 0\tag{7.33}\]

LQR provides a structured trade-off between pointing accuracy (weight \(Q\)) and actuator usage (weight \(R\)), and guarantees stability margins of at least 6 dB gain margin and \(\pm 60\degree{}\) phase margin for the linear system.

7.9.3 Nonlinear and Robust Control

For highly manoeuvrable spacecraft or in the presence of significant inertia uncertainty, nonlinear control laws are preferred. Sliding mode control (SMC) is robust to parameter uncertainties and external disturbances up to a known bound. The sliding surface is defined as:

\[\mathbf{s} = \dot{\boldsymbol{\varepsilon}} + \lambda \boldsymbol{\varepsilon}\tag{7.34}\]

and the control law drives the system onto the surface \(\mathbf{s} = 0\) and keeps it there, regardless of bounded disturbances and inertia uncertainty. The chattering phenomenon (high-frequency switching) in classical SMC is mitigated by using boundary layer methods or higher-order sliding mode approaches.

7.10 Pointing Modes and Mission Operations

Modern spacecraft operate in multiple pointing modes depending on mission phase and science/communications objectives:

Sun acquisition mode: Used during LEOP and after contingencies. The spacecraft manoeuvres to a Sun-pointing attitude using coarse sun sensors and magnetometers/gyros, to ensure solar array power. Once Sun is acquired, the spacecraft can proceed to safe mode or nominal mode checkout.

Nadir pointing (Earth observation mode): The payload boresight is maintained along the nadir direction (\(-\hat{r}\)), with the spacecraft yaw compensated for the Sun angle to maintain solar array illumination. This requires continuous attitude maintenance against orbital mechanics perturbations. Tracking a ground target (spotlight mode) requires attitude manoeuvres at the orbital rate modified by the target geometry.

Inertial pointing (science mode): The spacecraft is pointed at a fixed inertial direction (star, planet, deep field), with the orbital motion causing the Earth to drift through the field of view. Gyroscopes and star trackers provide attitude hold.

Spin stabilisation: Early spacecraft and some modern missions (e.g., spin-stabilised kick motors, some planetary probes) use body spin for gyroscopic stiffness. The angular momentum vector is fixed in inertial space, providing passive attitude stability. Three-axis control is not available in pure spin stabilisation but can be augmented by precession manoeuvres.

7.11 ADCS Block Diagram and System Architecture

Figure 7.2. ADCS functional block diagram. Sensor data flows to the ADCS computer, which performs attitude determination (EKF fusing gyro and absolute sensor data), orbit determination, and control law computation. Actuator commands are distributed to reaction wheels, magnetorquers, and thrusters. The ADCS computer also manages wheel desaturation logic and mode transitions.

7.11.1 Sensor Suite Comparison

Table 7.2. Comparison of attitude sensor specifications for space applications.
Sensor Accuracy Update Rate Mass (kg) Typical Use
Star Tracker 1–10 arcsec 1–10 Hz 0.3–2.0 Precision 3-axis
Fine Sun Sensor 0.01–0.1 10–100 Hz 0.05–0.3 Sun vector, array pointing
Coarse Sun Sensor 1–5 \(>\)100 Hz 0.01–0.05 Acquisition, safe mode
Earth Horizon Sensor 0.05–0.1 1–10 Hz 0.3–1.5 Nadir vector, backup
Magnetometer (fluxgate) \(\pm 1\)–100 nT 1–100 Hz 0.05–0.5 Coarse attitude, magnetorquer control, LEO
MEMS Gyroscope 0.1–1/\(\sqrt{\mathrm{h}}\) ARW 100–1000 Hz 0.01–0.2 Rate sensing, CubeSats
HRG 0.0001/\(\sqrt{\mathrm{h}}\) ARW 200–400 Hz 0.5–2.0 Precision, long-life
GPS (single) 3–10 m position 1–10 Hz 0.1–0.5 Orbit determination
GPS (multi-antenna) 0.1–1 attitude 1–10 Hz 0.5–2.0 Attitude, baselines

7.12 Micro-Vibration and Jitter Analysis

For optical imaging satellites with resolution requirements of sub-meter at altitude, micro-vibration from rotating machinery (reaction wheels, cryocoolers, compressors) can be the dominant error source limiting image quality. The metric is the line-of-sight (LOS) jitter expressed as an angle (arcseconds RMS) in the focal plane.

Reaction wheel imbalance produces disturbances at the wheel spin frequency \(f_w\) and harmonics \(n f_w\). The structural response of the spacecraft amplifies disturbances near modal frequencies. The LOS jitter is computed through a transfer function from wheel force/torque to LOS angle, integrated over all wheel speeds and frequencies.

Passive isolation mounts (such as the Moog CSA hexapod isolator) provide 6-DOF isolation, attenuating disturbances above the corner frequency (typically 10–40 Hz) by 20–40 dB/decade. Active isolation using accelerometer-based control further extends the isolation bandwidth.

7.13 Summary

The ADCS constitutes one of the most mathematically rich subsystems of the spacecraft. From the reference frames that define the context for attitude measurement, through the singularity-free quaternion representation of attitude and its kinematic equations, to the Euler rigid-body equations of motion coupling attitude dynamics to external torques—the theoretical foundation provides the tools for analysis. Environmental disturbances (SRP, gravity gradient, aerodynamic, magnetic) drive the actuator sizing. Sensor technologies spanning coarse photodiodes to precision star trackers and HRG gyroscopes provide the measurements for the EKF attitude estimator. Reaction wheels provide precision torque; magnetorquers enable efficient momentum management in LEO; CMGs provide high agility. The PD and LQR control laws, with their known stability properties, complete the closed-loop system. Chapter 8 addresses the propulsion system, which interfaces closely with the ADCS through orbit maintenance manoeuvres and angular momentum desaturation.