Gyroscope and Its Applications

Your phone carries three MEMS gyroscopes that cost about a dollar and drift like drunkards — tens of degrees per hour of bias instability — yet fused with accelerometers, magnetometer, and GPS they stabilize your camera, orient your maps, and let a drone hover in gusting wind with control loops running on equation (12) a thousand times per second. One grade up, tactical FOGs and ring-laser gyros guide airliners between GPS fixes; the hemispherical resonator gyro — a vibrating quartz wineglass exploiting the same Coriolis physics as MEMS, but in a millimetre-perfect resonator — now flies on spacecraft with drift measured in ten-thousandths of a degree per hour and no wear-out mechanism, having replaced spinning rotors almost everywhere. The deep-physics end of the family is the quietest. Gravity Probe B's four gyroscopes — niobium-coated fused-quartz spheres round to forty atoms, electrostatically levitated and spun in vacuum, read out by SQUID magnetometers — orbited Earth from 2004 to 2005 to test whether spacetime itself precesses a gyroscope. General relativity predicted a geodetic drift of 6606 milliarcseconds per year from the curvature of space around Earth and a frame-dragging drift of 39 mas/yr from Earth's rotation dragging spacetime around with it. The mission measured both: $6602 \pm 18$ and $37 \pm 7$ mas/yr. The humble demand behind those error bars — a stray-torque drift below $10^{-11}$ degrees per hour, thirteen orders of magnitude quieter than your phone — is still equation (2), $\Omega_{\mathrm{drift}} = \tau/H$, pushed to its physical limit. Meanwhile large ring-laser gyros anchored in bedrock (the 4-metre ring at Wettzell) now track fluctuations in Earth's own rotation rate at the part-per-billion level — the gyroscope, invented to show the planet turns, has become the instrument that watches it stumble.

Spin a rotor up to speed and mount it in two nested gimbals, so that the frame can turn any way it likes without exerting torque on the rotor axis. Now rotate the frame. The rotor does not care. Its angular momentum is a vector fixed in inertial space, and with no torque path to it, $\mathrm{d}\vec{L}/\mathrm{d}t = 0$ — the axis keeps pointing at the same distant star while the vehicle around it pitches, rolls, and yaws. Read the gimbal angles and you have the vehicle's orientation, mechanically, with no external reference. This is gyroscopic rigidity, and it is nothing but angular momentum conservation dressed up as an instrument.

The complementary principle appears when the rotor is not free. Clamp one gimbal so the vehicle's turn is transmitted to the rotor as a forced precession $\vec{\Omega}$, and the rotor answers with the reaction torque $-\vec{\Omega}\times\vec{L}_s$ derived in the Gyroscopic Precession article. Measure that torque (or the spring deflection it causes) and you have a rate gyro — an instrument that outputs how fast the vehicle is turning. Amplify instead of measure and you have a control-moment gyro that steers a spacecraft. Rigidity gives you orientation; reaction gives you rotation rate and torque authority. Everything that follows — gyrocompass, MEMS chip, laser ring — is one of these two ideas wearing different hardware. One subtlety upgrades the whole subject from demo to instrument: a free gyro holds an inertial direction, and we do not live in an inertial frame. Point a perfect gyro at the horizon and walk away: to you it will appear to tilt and drift at up to Earth's rotation rate, $\omega_E = 7.292\times10^{-5}\ \mathrm{rad/s} = 15.04°/\mathrm{h}$, because the ground under it is turning. An "apparent drift" of fifteen degrees per hour is catastrophic for a compass and useless for navigation — until you realize the drift is Earth's rotation, at which point you can compensate it (directional gyro), integrate it away (INS), or harness it to find north (gyrocompass).

The master law and its instrument-grade consequences come first. About the rotor's support point,

where: $\vec{\tau}$ — net external torque; $\vec{L}$ — total angular momentum, dominated for a fast rotor by the spin term $\vec{L}_s \approx I_3\omega_3\hat{e}_3 \equiv H\hat{e}_3$. Engineers call $H$ the angular momentum capacity; it is the currency in which all gyroscopic performance is priced. A stray torque $\tau$ — bearing friction, mass unbalance, gimbal flexure — does not tip the axis; it precesses it at

where: $\Omega_{\mathrm{drift}}$ — the slow wander of the reference axis, the drift rate that defines an instrument's grade. Equation (2) is the fast-top precession law of the Gyroscopic Precession article read as an error budget: to build a better gyro, raise $H$ or starve the torque. A consumer MEMS part drifts by tens of degrees per hour; a marine gyrocompass by hundredths; Gravity Probe B, with electrostatically levitated quartz spheres in vacuum, reached below $10^{-11}$ degrees per hour — the same formula spanning thirteen orders of magnitude. Forcing the axis to turn at $\vec{\Omega}$ costs whoever does the forcing a torque, and the rotor pushes back:

where: the reaction is perpendicular to both the turn and the spin — derived and dissected in the Gyroscopic Precession article. Linear in $\Omega$ with the enormous gain $H$, equation (3) is why a modest gimbal motor tilting a 100-kg rotor can slew the entire ISS, and why a rate gyro's tiny spring deflection reads vehicle turn rate directly. For anything quantitative about orientation you need the kinematics: the rotor axis direction is tracked by the Euler angles $(\phi,\theta,\psi)$, with the angular velocity decomposing as

where: $\hat{n}$ — the line of nodes; the full construction, body-frame components and all, lives in the Angular Velocity in Terms of Euler Angles article. Feeding (4) into the symmetric-top Lagrangian yields two conserved momenta $p_\phi, p_\psi$ and the effective potential $V_{\mathrm{eff}}(\theta) = (p_\phi - p_\psi\cos\theta)^2/(2I_1\sin^2\theta) + MgR\cos\theta$ — the machinery derived in the Heavy Symmetric Top article. Two of its results matter constantly in gyro engineering. First, steady precession under a constant torque exists at two rates,

where: the slow branch $\dot\phi_- \to MgR/(I_3\omega_3)$ in the fast-spin limit is the operating point of every precession-based instrument. Second, any disturbance superposes nutation — a nodding of the axis at frequency and amplitude

where: the scalings say that spinning faster makes the wobble both faster and smaller — for instrument rotors at tens of thousands of rpm, nutation is a kilohertz shiver of microradian size, handled by damping and otherwise ignored (the Nutation article does the full analysis). And with no torque at all, the axis of a slightly-misaligned rotor traces the free precession cones of the Free Symmetric Top article at

where: $\omega_{\mathrm{body}}$ — the coning rate seen in the rotor frame. In an instrument this free coning is a defect (it smears the reference axis), which is why precision rotors are spun about their axis of maximum moment of inertia and fitted with nutation dampers.

Which formula answers which question

A free gyro drifts $15°/\mathrm{h}$ relative to the ground; the gyrocompass turns that bug into its sensing principle. Mount the rotor so its axis stays near the horizontal plane, and make it pendulous — hang a small mass below the rotor case, so that any tilt $\beta$ of the axis above the horizon produces a gravity torque $mgl\sin\beta$ about the horizontal. Now the feedback loop closes by itself. If the spin axis points east of true north by azimuth $\alpha$, Earth's rotation tilts the local horizon under it, so the axis — inertially fixed — appears to rise at a rate proportional to $\alpha$. The rising tilt engages the pendulous torque, and by equation (3) that torque precesses the axis in azimuth, back toward the meridian. Point west of north and the loop runs in reverse. True north is the only azimuth where nothing happens: the axis oscillates about the meridian and, with damping, settles on it.

The choice of an 84-minute period is not an accident of my example numbers — it is the Schuler period, $2\pi\sqrt{R_\oplus/g}$, the period of a pendulum as long as Earth's radius. An instrument tuned to it has a magic immunity: when the ship accelerates, the apparent gravity tilts, but a Schuler-tuned loop responds by tracking the accelerating vehicle's actual local vertical rather than being fooled — accelerations do not excite a north error. Every inertial navigation system carries the same 84-minute oscillation in its error dynamics for the same reason.

The spinning rotor has two solid-state successors, and both descend from equations you already own. The MEMS vibrating-structure gyroscope is a Coriolis machine. A silicon proof mass is driven to oscillate along $x$ with amplitude $x_0$ at frequency $\omega_d$, giving velocity amplitude $v_d = x_0\omega_d$. Rotate the chip about $z$ at $\Omega$, and in the chip's rotating frame the mass feels the Coriolis acceleration studied in the Coriolis Force article:

where: $a_C$ — peak Coriolis acceleration along the perpendicular sense axis $y$; $y_0$ — the resulting sense-mode amplitude when the sense resonance $\omega_s$ is matched to the drive and amplified by its quality factor $Q$; capacitive fingers convert $y_0$ to a signal proportional to $\Omega$. The numbers are humbling: for $x_0 = 5\ \mathrm{\mu m}$ at $20\ \mathrm{kHz}$, $v_d = 0.63\ \mathrm{m/s}$; a rotation of $1°/\mathrm{s}$ gives $a_C = 0.022\ \mathrm{m/s^2}$ and a bare (unamplified) sense deflection of $1.4$ picometres — smaller than an atom. Only resonant gain ($Q\sim10^3$ brings it to nanometres) and femtofarad capacitance readout make the phone-scale gyro possible. Two masses driven in antiphase and read differentially cancel linear vibration — the tuning-fork architecture:

The optical gyroscope removes even the vibration. Send two light beams around the same closed loop in opposite directions and rotate the loop at $\Omega$: the co-rotating beam chases a receding beamsplitter while the counter-rotating beam meets an advancing one, so they complete the circuit in different times.

1 · A fiber-optic gyro reads the planet (confidence builder)

A fiber-optic gyroscope uses $L_f = 1.0\ \mathrm{km}$ of fiber wound on a coil of diameter $8.0\ \mathrm{cm}$, operating at $\lambda = 1550\ \mathrm{nm}$. Its axis is aligned with Earth's rotation axis. What Sagnac phase does Earth's rotation produce, and what is the scale factor in phase per degree-per-hour?

Step 1. Coil geometry: area per turn $A = \pi(0.040)^2 = 5.03\times10^{-3}\ \mathrm{m^2}$; number of turns $N = L_f/(\pi d) = 1000/(\pi\times0.080) = 3979$.

Step 2. Earth's rate: $\Omega = \omega_E = 7.292\times10^{-5}\ \mathrm{rad/s}$ ($= 15.04°/\mathrm{h}$).

Step 3. Sagnac phase from equation (11): $\Delta\Phi = \dfrac{8\pi NA\,\Omega}{\lambda c} = \dfrac{8\pi \times 3979 \times 5.03\times10^{-3} \times 7.292\times10^{-5}}{1.55\times10^{-6}\times 2.998\times10^{8}} = \boxed{79\ \mathrm{\mu rad}}$.

Step 4. Scale factor: dividing by $15.04°/\mathrm{h}$ gives $\boxed{5.2\ \mathrm{\mu rad}\ \text{per deg/h}}$. Navigation-grade FOGs resolve phases of nanoradians, so Earth's rotation is a large signal for them — a stationary FOG gyrocompasses by measuring the direction-dependence of exactly this phase, something no consumer MEMS part can dream of.

2 · Designing a marine gyrocompass (exam level)

A marine gyrocompass rotor has $I_3 = 0.020\ \mathrm{kg\,m^2}$ spinning at $24{,}000$ rpm, and pendulosity giving a restoring torque coefficient $mgl = 1.5\ \mathrm{N\,m}$ per radian of tilt. At latitude $\lambda = 45°$: (a) find the north-seeking period; (b) the settled latitude-error tilt; (c) the drift rate a stray bearing torque of $10^{-6}\ \mathrm{N\,m}$ would cause if the same rotor were used as a free directional gyro.

Step 1. Angular momentum capacity: $\omega_3 = 2\pi \times 400 = 2513\ \mathrm{rad/s}$, so $H = I_3\omega_3 = 0.020\times2513 = 50.3\ \mathrm{kg\,m^2/s}$.

Step 2. North-seeking frequency from equation (8): $\omega_c = \sqrt{\dfrac{mgl\,\omega_E\cos\lambda}{H}} = \sqrt{\dfrac{1.5\times 7.292\times10^{-5}\times 0.707}{50.3}} = 1.24\times10^{-3}\ \mathrm{rad/s}$.

Step 3. Period: $T_c = 2\pi/\omega_c = 5065\ \mathrm{s} = \boxed{84\ \mathrm{min}}$ — the Schuler period, by design: the pendulosity was chosen so that ship accelerations cannot masquerade as north error. With damping, the compass settles on the meridian in two to three of these periods, which is why a ship's gyrocompass is started hours before departure.

Step 4. Latitude-error tilt from equation (9): $\beta^* = \dfrac{H\,\omega_E\sin\lambda}{mgl} = \dfrac{50.3\times 7.292\times10^{-5}\times 0.707}{1.5} = 1.73\times10^{-3}\ \mathrm{rad} = \boxed{0.10°}$ — a small, latitude-dependent tilt the readout must subtract.

Step 5. Free-gyro drift from equation (2): $\Omega_{\mathrm{drift}} = \tau/H = 10^{-6}/50.3 = 1.99\times10^{-8}\ \mathrm{rad/s} = \boxed{4.1\times10^{-3}\ °/\mathrm{h}}$. A micronewton-metre of bearing friction — a fingerprint's worth — already costs $0.1°$ of heading per day. This single number explains a century of air bearings, magnetic suspensions, and finally the flight from mechanical rotors to optical loops with no bearings at all.

References. Morin, Introduction to Classical Mechanics, ch. 9 (Angular Momentum, Part II) · Goldstein, Poole & Safko, Classical Mechanics, 3rd ed., ch. 5 · Landau & Lifshitz, Mechanics, 3rd ed., §35–37 · Taylor, Classical Mechanics, ch. 10

Provenance. Equations and limits follow Goldstein (Classical Mechanics, ch. 5), Landau–Lifshitz (Mechanics §35–37), Morin (ch. 9), and Taylor (ch. 10). Verifier should confirm: the steady-precession quadratic and its two roots; the $\Omega=MgR/I_3\omega_3$ fast-top limit; the effective-potential reduction with conserved $p_\phi,p_\psi$; the sleeping-top stability threshold $\omega_3^2>4I_1MgR/I_3^2$; and the torque-free precession rates. No empirical constants are quoted beyond standard $g$. Additional results on this page: the gyrocompass dynamics $\omega_c^2 = mgl\,\omega_E\cos\lambda/H$ and latitude error $\beta^* = H\omega_E\sin\lambda/(mgl)$ (Wrigley/Anschütz standard theory; cf. Klein–Sommerfeld vol. IV); the Sagnac results $\Delta t = 4A\Omega/c^2$, $\Delta\Phi = 8\pi NA\Omega/(\lambda c)$; MEMS Coriolis readout $a_C = 2\Omega v$. Worked numbers recomputed from these formulas: FOG 1 km/8 cm/1550 nm → $79\ \mathrm{\mu rad}$ at Earth rate ($5.2\ \mathrm{\mu rad}$ per deg/h); gyrocompass $H = 50.3\ \mathrm{kg\,m^2/s}$, $mgl = 1.5\ \mathrm{N\,m}$, $\lambda = 45°$ → $T_c = 84$ min, $\beta^* = 0.10°$; free-gyro drift $10^{-6}\ \mathrm{N\,m}/50.3 = 4.1\times10^{-3}\,°/\mathrm{h}$; Earth rate $15.04°/\mathrm{h}$; GP-B values 6606/39 mas/yr predicted, $6602\pm18$/$37\pm7$ measured (Everitt et al. 2011).

Builds on: Angular Momentum of a Rigid Body · Principal Axes and Principal Moments of Inertia · Euler Angles · Euler's Equations of Motion · Classification of Rigid Bodies: Spherical, Symmetric, and Asymmetric Tops. Leads to: Angular Velocity in Terms of Euler Angles · Rotational Kinetic Energy · Inertia Ellipsoid.