Chapter 14 — Key Takeaways (Spacecraft Attitude Dynamics)
A one-page reference. Reread before an exam, or before you size an attitude control system.
The one idea
Attitude is orientation, not position. A rigid spacecraft has 6 degrees of freedom: 3 translational (the orbit, driven by force) and 3 rotational (the attitude, driven by torque). They are separate problems with separate hardware. Almost every subsystem — power, comms, science, thermal, burns — is a pointing requirement in disguise, so losing attitude control usually loses the mission.
Three ways to represent a rotation
| Representation | Numbers / constraints | Verdict |
|---|---|---|
| Euler angles (3–2–1 yaw–pitch–roll) | 3 / 0 | Intuitive, but gimbal lock at pitch $=90^\circ$: two axes align, a DOF is lost, rate equations blow up ($\div\cos\theta$). |
| Direction cosine matrix (DCM) | 9 / 6 | Orthonormal $R_{B/N}$, no singularity, but heavy (9 numbers) and needs re-orthonormalizing. |
| Quaternion (Hamilton, scalar-first) | 4 / 1 | No gimbal lock, minimal redundancy, numerically robust — the working representation of flight software. |
Conventions (match Appendix D): passive DCM, $\mathbf{v}_B = R_{B/N}\mathbf{v}_N$, with $+\sin\theta$ in the upper-right of $R_z$. Quaternion $\mathbf{q}=(q_0,q_1,q_2,q_3)$, $q_0=\cos(\theta/2)$, $(q_1,q_2,q_3)=\hat{\mathbf{e}}\sin(\theta/2)$, unit norm. Euler's rotation theorem: any orientation is one rotation $\theta$ about one axis $\hat{\mathbf{e}}$ — the basis of the quaternion.
Rigid-body dynamics (why attitude changes)
$$\mathbf{I}\,\dot{\boldsymbol{\omega}} + \boldsymbol{\omega}\times(\mathbf{I}\,\boldsymbol{\omega}) = \mathbf{M}$$
| Symbol | Meaning | Units |
|---|---|---|
| $\boldsymbol{\omega}$ | angular velocity (turn rate) | rad/s |
| $\mathbf{I}$ | inertia tensor (rotational "mass") | kg·m² |
| $\mathbf{M}$ | external torque (moment) | N·m |
| $\mathbf{H}=\mathbf{I}\boldsymbol{\omega}$ | angular momentum | N·m·s (kg·m²/s) |
The coupling term $\boldsymbol{\omega}\times(\mathbf{I}\boldsymbol{\omega})$ makes torque-free bodies tumble and the intermediate axis unstable (tennis-racket theorem).
The four environmental torques
| Torque | Scaling with distance | Where it dominates | Formula / note |
|---|---|---|---|
| Aerodynamic | falls fast with altitude ($\propto\rho$) | very low LEO | $F_{\text{drag}}\cdot d$, drag at offset center of pressure |
| Gravity-gradient | $\propto 3\mu/r^3 = 3n^2$ | LEO | $M_{gg}=\dfrac{3\mu}{2r^3}\lvert I_{\max}-I_{\min}\rvert\sin2\theta$ |
| Magnetic | $\propto 1/r^3$ | LEO | $\mathbf{M}=\mathbf{m}\times\mathbf{B}$ |
| Solar radiation | $\approx$ altitude-independent | GEO and beyond | $P_\odot A(1+\rho)\cdot d$, $P_\odot\approx4.5\times10^{-6}\,\text{N/m}^2$ |
All are tiny ($\sim10^{-7}$–$10^{-4}$ N·m) but relentless — with no damping in vacuum they integrate. Gravity-gradient stiffness $=3n^2$ ties directly to Chapter 8's mean motion $n=\sqrt{\mu/a^3}$.
Determination — knowing where you point
Key geometric fact: one direction measurement fixes only 2 of 3 DOF; you need two non-parallel vectors (or one star tracker) for full 3-axis attitude.
| Sensor | Measures | Accuracy | Notes |
|---|---|---|---|
| Star tracker | full 3-axis attitude | few arcsec | best; blinded by Sun/Moon/Earth limb; solves "lost in space" |
| Sun sensor | Sun direction (2-axis) | 0.1–1° | cheap, safe-mode anchor |
| Earth/horizon sensor | nadir direction | ~0.1° | for nadir pointers |
| Magnetometer | field vector (2-axis) | 1–5° | LEO only; also drives magnetorquers |
| Rate gyro | angular velocity | drifts | integrate to propagate; drifts → must be reset |
Standard architecture: fuse drifting gyro + absolute star tracker via a Kalman filter (Ch. 13).
Control — changing where you point
| Actuator | Internal/External | Torque | Cost / limit |
|---|---|---|---|
| Reaction wheel | internal | small, precise | no propellant, but saturates |
| Control moment gyro (CMG) | internal | large (amplified) | complex; gimbal singularities |
| Thruster | external | large | burns propellant (mass, finite life); works anywhere |
| Magnetorquer | external | weak, 2-axis | no propellant; LEO only; can't torque along $\mathbf{B}$ |
Momentum dumping (desaturation): $\dot{\mathbf{H}}_{\text{total}} = \mathbf{M}_{\text{external}}$. Internal actuators only redistribute momentum; an external torque (thrusters or magnetorquers) must remove what disturbances pile into the wheels before they saturate. Every reaction-wheel spacecraft needs an external actuator too.
Spin vs. three-axis; pointing budget
| Spin-stabilized | Three-axis | |
|---|---|---|
| Steadiness | passive gyroscopic stiffness | active control (wheels/CMGs) |
| Pointing | spin axis only | any direction, precise |
| Complexity/mass | low | high |
| Rule | spin about max-inertia axis (major-axis rule; Explorer 1) | full sensor + actuator suite |
Pointing budget: separate accuracy (control error), knowledge (determination error), jitter (short-term wobble). Combine independent errors by root-sum-square: $\sigma=\sqrt{\sum\sigma_i^2}$. Attack the largest (squared) term first.
Decision aid — "which do I use?"
| Situation | Choice |
|---|---|
| Flight-software attitude state | quaternion (no singularity, compact) |
| Rotate a vector between frames | DCM ($\mathbf{v}_B=R_{B/N}\mathbf{v}_N$) |
| Human display / pilot intuition | Euler angles (watch for gimbal lock) |
| Precise steerable pointing | three-axis + reaction wheels |
| Cheap, robust, one-axis pointing | spin (about max-inertia axis) |
| Dump wheel momentum in LEO | magnetorquers (free) |
| Dump momentum in deep space | thrusters (finite propellant) |
| Best absolute attitude knowledge | star tracker (+ gyro for rate) |
Common pitfalls
| Pitfall | Reality |
|---|---|
| "Good orbit ⇒ good pointing." | Independent problems; a satellite can orbit perfectly while tumbling. |
| "Weightless ⇒ no torques." | Free fall removes weight, not torque; four disturbances act unopposed. |
| "A gyro tells you which way you point." | It measures rate; integrating it drifts — reset with a star tracker. |
| "Reaction wheels need no propellant, so no thrusters needed." | Wheels saturate; dumping needs an external torque. |
| "Three angles are enough, so use them." | They gimbal-lock; flight software uses quaternions. |
| "Spin a pencil about its long axis." | Violates the major-axis rule; energy dissipation → flat spin (Explorer 1). |
Numbers worth remembering
- Disturbance torques: $\sim10^{-4}\ \text{N·m}$ (small sat), spanning $10^{-7}$–$10^{-3}$ N·m.
- Solar pressure near Earth: $P_\odot\approx4.5\times10^{-6}\ \text{N/m}^2$.
- Star trackers: a few arcsec; $1^\circ = 3600\ \text{arcsec}$.
- Gravity-gradient bracket $3\mu/r^3 = 3n^2$ (Ch. 8 mean motion).
- Pointing requirements: comsat $\sim0.05$–$0.1^\circ$; imager $\sim0.01$–$0.05^\circ$; great telescope $\ll1\ \text{arcsec}$.
Mission / astrotools additions this chapter
- MDR: added the attitude section — pointing requirement (accuracy/knowledge/jitter), spin vs. three-axis choice, and the actuator + sensor suite with a momentum-dumping method.
attitude.py:quat_multiply(q1, q2)(Hamilton, scalar-first) anddcm_from_quat(q)(passive DCM, matches Appendix D). Self-test: $\mathbf{q}_z(90^\circ)\otimes\mathbf{q}_x(90^\circ)=(0.5,0.5,0.5,0.5)$, a $120^\circ$ turn about $(1,1,1)$, whose DCM is the clean axis-cycling permutation.