Angular Momentum & Gyroscopes · Guide · glidepathscience.org
GuideAngular Momentum & Gyroscopes
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round. Part 1 prepares the Intro track and Part 2 the Advanced track, about fifteen minutes each. Each section ends with a worked example that is not one of the sim’s own problems. Throughout, z points up, counterclockwise seen from +z counts as positive, and g = 9.81 m/s².
Part 1 · The essentials
Math you’ll need: algebra, right-triangle trigonometry and the right-hand rule; calculus only in a box marked “With calculus.”
Spin is a vector
Linear momentum tells you how hard it is to stop something that is moving and which way it is going. A spinning wheel, a planet and a figure skater carry a second kind of momentum, one that a push through their center cannot change: angular momentum. Before we can measure it, we need a direction for a spin, because a spin, like a velocity, has one.
Every point of a turning wheel moves in a different direction, so no direction of motion can stand for the whole wheel. What every point shares is the axle. So the spin’s vector, the angular velocity $\vec\omega$, lies along the axis of rotation, and its length is the rate of turning in radians per second. Which way along the axis it points is fixed by the right-hand rule: curl the fingers of your right hand the way the body turns, and your thumb points along $\vec\omega$. A wheel turning counterclockwise seen from above has $\vec\omega$ pointing up; turning clockwise, down. The angular momentum of a body spinning about a symmetry axis, $\vec L$, points the same way.
The commonest mistake is to point the spin along the motion, around the rim; but the rim’s points move in every direction. The same rule makes Earth’s angular velocity point out through the North Pole, toward the North Star, since Earth turns toward the east: counterclockwise seen from above that pole.
The angular momentum of a moving body
Angular momentum is always measured about a point, and a body need not go around that point to have some. Picture a puck sliding past a post: its motion has a turning sense about the post, even though its path is straight. The angular momentum of a particle of momentum $\vec p$ at position $\vec r$ from the point O is the cross product
$\vec l = \vec r \times \vec p$
Its size is $l = rp\sin\theta$, with θ the angle between $\vec r$ and $\vec p$, and its direction is perpendicular to both, by the right-hand rule. There is a neater way to read the size: $r\sin\theta$ is the perpendicular distance from O to the line along which the body moves, the lever arm $r_\perp$, so
$l = r_\perp p$
That form shows at once why a body moving in a straight line with no force on it keeps its angular momentum. As it moves on, r grows, but sin θ shrinks in exact step, because $r\sin\theta$ is the distance to a line that does not move. Equal areas in equal times, Kepler’s second law, is the same fact for a planet: the Sun’s pull points at the Sun, so it can change neither the planet’s angular momentum about the Sun nor the rate at which its line from the Sun sweeps out area.
A rigid body: L = Iω
For a body turning about a fixed axis, each bit of mass m at distance r from the axis moves at $r\omega$ on a circle, with lever arm r, so it carries $mr\cdot r\omega$ about the axis. Adding them up gives the angular momentum about the axis,
$L_{axis} = \left(\sum mr^2\right)\omega = I\omega$
where $I = \sum mr^2$ is the body’s moment of inertia about that axis, the quantity Rotational Dynamics & Rolling computes by shape: $MR^2$ for a hoop, $\tfrac12 MR^2$ for a solid disk, and so on. $L_{axis}$ is the part of $\vec L$ along the axis; for a body turning about a symmetry axis it is all of $\vec L$. The equation is the rotational twin of $p = mv$: $I$ plays the part of the mass and ω the part of the velocity. It only works with ω in radians per second; a speed in rpm must be multiplied by $2\pi/60$ first, and one in revolutions per second by $2\pi$. A bicycle wheel’s mass is nearly all in its rim, so it is close to a hoop; using a disk’s ½ there halves the answer.
Angular momentum, like linear momentum, changes only when something acts from outside. For a spin, that something is a torque.
Torque changes angular momentum
Newton’s second law in its momentum form, $\sum\vec F = \Delta\vec p/\Delta t$, has a rotational partner: the net outside torque about a point is the rate at which the angular momentum about that point changes.
$\sum\vec\tau = \Delta\vec L/\Delta t$
Torque and angular momentum must be taken about the same fixed point (or about the center of mass, which also works). When the torque is steady, the change it makes is the torque times the time, the angular impulse:
$\tau\,\Delta t = \Delta L$
A wheel with no torque on it turns forever; it does not need a torque to keep spinning, any more than a puck needs a force to keep sliding. A torque changes the spin. A brake pad’s friction force f at the rim of radius R has a torque fR against the spin, and the wheel stops when that torque has removed all of its L.
Conservation, and the energy of a spin
If no outside torque acts, $\Delta\vec L$ = 0: the angular momentum keeps both its size and its direction. For a body whose moment of inertia can change, a skater pulling in their arms or a diver tucking, keeping $L = I\omega$ means
$I_i\omega_i = I_f\omega_f$
Bring the mass closer to the axis and I falls, so ω must rise. The ice barely twists a skater, and in the air nothing twists a diver about their center of mass, since gravity acts at that center.
The kinetic energy of the spin is $K_R = \tfrac12 I\omega^2$, which, with $L = I\omega$, can be written
$K_R = L^2/2I$
With L fixed, a smaller I means more kinetic energy, by the factor $I_i/I_f$. Where does it come from? The arms tend to fly outward as the body turns, and pulling them in against that tendency takes work, which the muscles do: chemical energy becomes kinetic energy. So “energy is conserved” is true for the whole story, but the kinetic energy alone is not conserved, and the usual mistake, keeping $\tfrac12 I\omega^2$ the same, gives a spin that grows only as the square root of $I_i/I_f$.
Catches and flips
Conservation also governs bodies that grab one another. When a spinning body is caught by, or dropped onto, another on the same axis, friction between them acts until they turn as one. Those friction torques are equal and opposite, inside the system, so the total angular momentum is kept:
$I_1\omega_1 + I_2\omega_2 = (I_1 + I_2)\,\omega_f$
This is a rotational sticking collision, and like two carts that stick, it keeps momentum but not kinetic energy: the rubbing turns kinetic energy into heat. A student at rest on a turntable who grips a spinning wheel ends up turning slowly, with most of the wheel’s energy gone into their hands.
A flip moves angular momentum between bodies without any rubbing. A student at rest on a frictionless turntable holds a wheel spinning with $\vec L_w$ up, in frictionless bearings, so that nothing can change the wheel’s own spin. Turning its axle upside down reverses the wheel’s $L_z$; since the total $L_z$ must stay $+L_w$, the student must take up $+2L_w$ and turn the way the wheel first spun, at
$\Omega = 2L_w/I_s$
The wheel’s own moment of inertia does not join the student’s, because the free bearings let the wheel keep its spin whatever the stool does. Tilting the axle only part way, by α, leaves the wheel with $L_w\cos\alpha$ along the vertical, and the student takes up $L_w(1 - \cos\alpha)$.
Why a gyroscope does not fall
A fast spinner has a large angular momentum, and that makes it hard to turn. Mounted in frictionless gimbals, rings that let its frame turn freely around it, a rotor feels no torque at all, so its $\vec L$ and its axle keep pointing the same way in space however the frame is turned. Pilots call this rigidity in space; the attitude and heading indicators are built on it.
Now rest one end of a spinning gyroscope’s axle on a pivot, the axle level, and let go. Gravity pulls down at the center of mass, a distance r from the pivot, so about the pivot there is a torque of size $rMg$, horizontal and at right angles to the axle. By $\Delta\vec L = \vec\tau\,\Delta t$, the change in $\vec L$ points along that torque: sideways, not down. A torque at right angles to $\vec L$ turns it without changing its size, so the axle swings slowly around the pivot. That steady turning is precession. Its direction is the direction of $\vec\tau = \vec r \times \vec W$; reversing the spin, or moving the weight to the other side of the pivot, reverses it.
In a short time Δt the horizontal $\vec L$ swings through a small angle $\Delta\phi = \Delta L/L = \tau\,\Delta t/L$, so the precession rate is
$\omega_P = \tau/L = rMg/(I\omega)$ $T_P = 2\pi/\omega_P$
This counts only the spin’s angular momentum, so it holds while the precession is slow, $\omega_P \ll \omega$, and the precession’s own small angular momentum can be left out. Spin faster, and it precesses more slowly. Nothing here cancels gravity: the center of mass does not fall because the pivot pushes up with a force equal to the weight, and the spin only decides what the torque does with $\vec L$. A rotor that is not spinning has no $\vec L$ to turn sideways, so the same torque simply tips it over.
Part 2 · Going further
Math you’ll need: cross products, derivatives and one integral; differential equations in outline.
Angular momentum about any point
A body that both moves and spins has angular momentum of two kinds about a point O. Splitting each particle’s position into the center of mass’s position plus its place relative to the center, the cross terms vanish and
$\vec L_O = \vec R_{cm} \times M\vec V_{cm} + \vec L_{cm}$
The first term, the orbital part, depends on where O is; the second, the spin about the center of mass, is the same about every point. So the same wheel can have different angular momenta about different points, and the two parts must be added as vectors: if the orbital part turns one way and the spin the other, they subtract. In the plane, $L_z = \pm MVd + I\omega$, with d the distance from O to the center’s line of motion.
Impacts about a pivot, and the sweet spot
When something strikes a body that turns on a fixed pivot, a hinged door or a hanging rod, the pivot pushes during the impact. That push changes linear momentum, so $p$ is not kept. But the pivot’s force acts at the pivot, so its torque about the pivot is zero, and angular momentum about the pivot is kept. For a point mass m at speed v striking a body square on at distance d from the pivot, and sticking,
$mvd = (I_h + md^2)\,\omega$
where $I_h$ is the body’s moment about the pivot ($\tfrac13 Mw^2$ for a uniform door of width w). The pivot’s impulse is then whatever closes the momentum books: $J = p_{after} - p_{before}$, with $p_{after} = (M\tfrac{w}{2} + md)\,\omega$.
There is one place where the pivot has nothing to do. If the hit lands beyond the sweet spot, the body turns so fast that its near end would swing backward, so the pivot pushes it forward, along the hit; if the hit lands nearer the pivot, the near end would be carried forward, and the pivot holds it back. In between, at the center of percussion, the two conditions agree with no pivot impulse: $(Mw/2 + md)\,d = I_h + md^2$, which gives exactly $d = 2w/3$ for a uniform slab or rod, whatever the striking mass. A bat or a tennis racket has the same sweet spot, where a hit sends no jolt into the hands.
The work of pulling in
Part 1 found the skater’s extra kinetic energy from $K = L^2/2I$. The same number comes from the force. A mass m held at radius r on a body turning at ω needs a centripetal pull $m\omega^2 r$ inward, and pulling it in by dr does work $m\omega^2 r\,|dr|$, with ω rising as it comes in. For two dumbbells,
$W = \int 2m\omega(r)^2 r\,|dr| = \tfrac12 L^2\left(1/I_f - 1/I_i\right)$
The integral and the energy difference agree exactly, because ω(r) follows from L. In the turning frame of the stool the story looks different but ends the same way: a dumbbell moving inward feels a Coriolis force forward, along its motion around the axis, and the hands holding it are pushed forward by it, so the student speeds up. Holding ω at its first value, the usual shortcut, underestimates the work.
The flip, more closely
Part 1’s flip assumes frictionless bearings: the wheel’s spin seen from the ground never changes, so the student takes up the whole change in its $L_z$, $\Omega = 2L_w/I_s$. If a motor instead held the wheel at a fixed speed relative to the student’s hands, the wheel would be carried round with the student, and its own moment would join theirs: $\Omega = 2L_w/(I_s + I_w)$.
A part tilt adds a smaller effect: the tilted wheel is also carried round the vertical across its own axle, where the bearings do pass torque, with $\tfrac12 MR^2\sin^2\alpha$ of a hoop’s moment about a diameter counting about the vertical:
$\Omega = L_w(1 - \cos\alpha)/(I_s + \tfrac12 MR^2\sin^2\alpha)$
This sim, like Part 1’s formula, leaves that term out: it is a few percent, and zero for a full flip, where sin α = 0.
When L does not point along ω
For a body turning about a symmetry axis, $\vec L = I\vec\omega$ points along the axis. In general it need not. Each particle contributes $m\,\vec r \times (\vec\omega \times \vec r)$, and the sum is a linear function of $\vec\omega$ written with the inertia tensor, a 3 × 3 array of moments and products of inertia:
$\vec L = \mathbf{I}\vec\omega$
Every body has three perpendicular principal axes along which $\mathbf{I}$ acts like a single number, so that $\vec L$ lies along $\vec\omega$. About any other axis, $\vec L$ points off the axle and turns with the body. Since $\vec L$ then changes, even a perfectly steady spin needs a torque from the bearings, $\vec\tau = \vec\omega \times \vec L$, turning with the shaft. For two masses on a rod tilted θ from its axle, each a from the center, $\vec L$ is perpendicular to the rod and $\tau = 2ma^2\omega^2\sin\theta\cos\theta$. That rotating torque is what shakes a car with an unbalanced wheel; balancing weights put a principal axis back on the axle. A wheel can balance on a static stand (its center of mass on the axle) and still shake: that is dynamic imbalance.
With no torque at all, a free body still need not spin steadily. Euler’s equations, $I_1\dot\omega_1 = (I_2 - I_3)\,\omega_2\omega_3$ and its two partners in the principal axes, show that a spin about the axis of largest or smallest moment is stable, but one about the intermediate axis is not: a small wobble grows at the rate $\omega_2\sqrt{(I_2 - I_1)(I_3 - I_2)/(I_1 I_3)}$, and the body flips over every couple of seconds while its $\vec L$ stays fixed in space. A book or a phone tossed spinning about the axis across its cover shows it; cosmonaut Vladimir Dzhanibekov saw a wing nut do it in mid-air on Salyut 7 in 1985 (the Dzhanibekov effect). And a body that loses energy while keeping L, by flexing, ends up spinning about its axis of largest moment, where $L^2/2I$ is least: Explorer 1, set spinning about its long axis in 1958, was soon tumbling end over end.
A spacecraft turns with nothing to push on the same way: a reaction wheel inside it spun up one way turns the craft the other, keeping the total L at zero, $I_{sat}\Omega \approx -I_w\omega_w$.
The heavy top
A top, or a gyroscope on a pivot, is a symmetric body spinning at $\omega_3$ about its axle, with its center of mass r from the pivot, its moment $I_3$ about the axle and $I_1$ across it about the pivot. Tilted θ from upright, gravity’s torque is $Mgr\sin\theta$, but the part of $\vec L$ that swings around is also only its horizontal part, $L\sin\theta$, so the tilt cancels and a fast top precesses at
$\Omega \approx Mgr/(I_3\omega_3)$
whatever its tilt. The exact steady precession rates, keeping the small angular momentum of the precession itself, are the roots of
$I_1\Omega^2\cos\theta - I_3\omega_3\Omega + Mgr = 0$
The slow root is the gyroscopic rate above, a little higher; the fast root, near $I_3\omega_3/(I_1\cos\theta)$, is a quick precession rarely seen. Real roots exist only while $(I_3\omega_3)^2 \ge 4I_1 Mgr\cos\theta$, which sets the second result: a top spinning upright, a sleeping top, stays up only while
$I_3\omega_3 > 2\sqrt{I_1 Mgr}$
As friction at its tip slows it through that spin, a sleeping top begins to wobble and falls.
Finally, steady precession needs the right start. Let go from rest with its axle level, a gyroscope first falls a little, and that fall is turned sideways into the beginning of the precession, so the free end traces a line of small cusps: it dips and nods, or nutates. For a fast top the nods come at about
$f_n = I_3\omega_3/(2\pi I_1)$
and the first dip is about $\Delta\theta = 2I_1 Mgr\sin\theta_0/(I_3\omega_3)^2$ radians. Friction damps the nods, leaving steady precession with the axle slightly low.
Gyroscopes at work: instruments and the Earth
Two principles run a cockpit’s gyro instruments, and both are this Guide’s. The heading indicator and the attitude indicator use rigidity in space: a rotor in gimbals keeps its axle’s direction while the airplane turns, pitches and rolls around it. The turn coordinator uses precession. Its rotor sits in a gimbal held by springs. When the airplane turns at rate Ω, the case turns the rotor’s $\vec L$ with it, which takes a torque ΩL; the gimbal tilts until the springs supply it, $k\theta = \Omega L$, so the tilt measures the turn rate. A standard-rate turn, 3°/s, takes two minutes for a full circle. If the rotor runs down, L falls and the instrument reads low.
A free gyro keeps its direction in space, not on the Earth. The Earth turns under it once a sidereal day, at $\Omega_E$ = 15.04°/h, and at latitude λ only the vertical part of that turn, $\Omega_E\sin\lambda$, swings the local north around a level axle. So a heading indicator drifts against the compass, by that much and more from bearing friction, and pilots reset it to the magnetic compass about every fifteen minutes.
The Earth is itself a gyroscope. Its spin makes it bulge at the equator, and the Sun and the Moon pull harder on the near side of the bulge than on the far side. Because Earth’s axis is tilted, that pull has a torque on the bulge at right angles to $\vec L$, and Earth’s axis precesses around the axis of its orbit, the opposite way to its spin. From the Sun alone, $\omega_P = (3n^2/2\omega_E)\,\tfrac{C - A}{C}\cos\varepsilon$, with n the orbital rate and $(C - A)/C$ the bulge’s size; the Moon’s pull is about twice the Sun’s, and together they turn the axis once in about 26,000 years. The pole star changes: Vega was it about 14,000 years ago and will be again.
A propeller is a rotor too: when a tailwheel airplane raises its tail on the takeoff roll, the pitch-down turns the propeller’s $\vec L$ downward, and the airframe feels the reaction as a yaw. A bicycle, though, mostly stays up by steering into its fall: Kooijman and colleagues built one whose wheels’ spins cancel, and it still balanced itself.
Reference
Ten common mistakes
Part 1
- ω along the motion. Spin points along the axis, by the right-hand rule, not around the rim.
- No angular momentum for straight-line motion. About a point its line misses, $l = r_\perp p$.
- rpm in $L = I\omega$. ω must be in rad/s: multiply rpm by $2\pi/60$.
- A torque needed to keep spinning. A torque changes L; with none, a spin goes on.
- Kinetic energy conserved when I changes. L is kept; $K_R = L^2/2I$ grows by the work done.
- The wheel’s own I in the flip. A wheel free in its bearings keeps its spin; the student takes $2L_w/I_s$.
- The spin holding the gyroscope up. The pivot holds it up; the spin turns the torque into precession.
Part 2
- Linear momentum kept at a hinge. The hinge pushes; only L about the hinge is kept.
- L always along ω. Only about a principal axis; otherwise $\vec L = \mathbf{I}\vec\omega$ is off the axle.
- Precession that depends on tilt. For a fast top the tilt cancels: $\Omega \approx Mgr/(I_3\omega_3)$.
The equation sheet
| Part 1 | |
|---|---|
| A particle | $l = r_\perp p = rp\sin\theta$ |
| A rigid body | $L = I\omega$ about the axis; $I = MR^2$ (hoop), $\tfrac12 MR^2$ (disk) |
| Torque | $\tau = \Delta L/\Delta t$ |
| Conservation | $I_i\omega_i = I_f\omega_f$ |
| Energy | $K = \tfrac12 I\omega^2 = L^2/2I$ |
| A flipped wheel | $\Omega = 2L_w/I_s$; a tilt: $\Delta L_z = L(1 - \cos\alpha)$ |
| Precession | $\omega_P = rMg/(I\omega)$ (slow, $\omega_P \ll \omega$); $T_P = 2\pi/\omega_P$ |
| Part 2 | |
|---|---|
| About a point | $L_O = R_{cm} \times MV_{cm} + L_{cm}$ |
| About a hinge | $mvd = (I_h + md^2)\omega$ |
| Pulling in | $W = \tfrac12 L^2(1/I_f - 1/I_i)$ |
| Tensor | $\vec L = \mathbf{I}\vec\omega$; $\tau = \omega \times L$, so $\tau = 2ma^2\omega^2\sin\theta\cos\theta$ |
| Heavy top | $I_1\Omega^2\cos\theta - I_3\omega_3\Omega + Mgr = 0$, and $\Omega \approx Mgr/(I_3\omega_3)$ |
| Sleeping, nodding | $I_3\omega_3 > 2\sqrt{I_1 Mgr}$; $f_n = I_3\omega_3/(2\pi I_1)$; $\Delta\theta = 2I_1 Mgr\sin\theta_0/(I_3\omega_3)^2$ |
| Instruments | drift $= \Omega_E\sin\lambda$; $k\theta = \Omega L$ |
l, L angular momentum (kg·m²/s), $r_\perp$ the lever arm, I moment of inertia (kg·m²), ω angular velocity (rad/s), τ torque (N·m), α a wheel’s tilt from vertical, $I_h$ a body’s moment about its hinge, d where a hit lands, $\mathbf{I}$ the inertia tensor, $I_3$ and $I_1$ a top’s moments about and across its axle (about the pivot), r its pivot-to-center-of-mass distance, θ its tilt from upright, $\Omega_E$ = 15.04°/h, λ the latitude, k a gimbal spring’s stiffness.
Sources and further reading
- OpenStax, University Physics Volume 1, by Samuel J. Ling, Jeff Sanny and William Moebs, §11.2–11.4, an open textbook (CC BY-NC-SA 4.0). The Walkthrough follows their order and notation.
- Federal Aviation Administration, Pilot’s Handbook of Aeronautical Knowledge (FAA-H-8083-25), the chapters on flight instruments and the aerodynamics of flight, and the Instrument Flying Handbook (FAA-H-8083-15).
- D. Kleppner and R. Kolenkow, An Introduction to Mechanics (2nd edition, Cambridge University Press, 2014), the chapters on rigid-body motion and angular momentum; J. R. Taylor, Classical Mechanics (University Science Books, 2005), the chapter on rotational motion of rigid bodies; H. Goldstein, C. Poole and J. Safko, Classical Mechanics (3rd edition, Addison-Wesley, 2002), the heavy symmetric top.
- J. D. G. Kooijman, J. P. Meijaard, J. M. Papadopoulos, A. Ruina and A. L. Schwab, “A bicycle can be self-stable without gyroscopic or caster effects,” Science 332, 339 (2011).
- This sim’s Field Manual (H): the model, its engines and what it leaves out.
Angular Momentum & Gyroscopes v1.3 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org