Star trackers and gyros measure which way a satellite faces; reaction wheels spin one way to turn the satellite the other way, with no fuel used.
These three parts do most of the fine pointing on modern satellites. Two of them sense. One of them moves.
A star tracker is a small camera with a computer. It photographs the sky and matches the pattern of stars against a built-in catalogue. From that match it works out exactly which way the satellite faces. NASA says most modern units know their pointing to within a few arcseconds.
They struggle when the satellite spins fast: NASA notes that cheaper units track up to about 0.3° per second and better ones up to about 3° per second.
A gyroscope measures how fast the satellite is rotating. Between star-tracker updates, the computer adds up the gyro readings to keep track of attitude. NASA notes that fibre-optic gyros generally perform better than tiny MEMS gyros, which are lighter but more sensitive to radiation.
ISRO's GSAT-19 (2017) flew a fibre-optic gyro as one of its new technologies.
A reaction wheel is a heavy disc spun by an electric motor. Speed the wheel up one way and the satellite turns the other way. This is conservation of angular momentum, the same reason a person on a spinning chair turns when they swing a heavy book. Three wheels at right angles give control in all directions; NASA notes a fourth is often added as a spare.
There is a limit. If outside forces keep pushing the satellite the same way, a wheel keeps speeding up until it reaches top speed — it is saturated. It then has to be slowed down by magnetic torquers or thrusters. ISRO's GSAT-19 uses exactly this trio: momentum wheels, magnetic torquers and thrusters.
Worked example (illustrative numbers): speeding up a wheel with a moment of inertia of 0.1 kg·m² by about 330 rpm turns a 2,000 kg·m² satellite at 0.1° per second the other way. Slowing the wheel stops the turn — no propellant used.
Pick an Earth-observation satellite and watch its ground track: to image a target below, wheels must keep its camera locked downward on every lap. Open the ISRO satellite tracker
Reaction wheels obey conservation of angular momentum. With no outside torque, the total angular momentum of satellite plus wheel stays constant:
I_sat × ω_sat + I_wheel × ω_wheel = constant
So a change in wheel speed Δω_wheel produces a satellite rate
ω_sat = −(I_wheel ÷ I_sat) × Δω_wheel
Worked example (illustrative numbers). A satellite with I_sat = 2,000 kg·m² must turn at 0.1°/s = 0.001745 rad/s. It needs angular momentum H = 2,000 × 0.001745 ≈ 3.5 N·m·s. A wheel with I_wheel = 0.1 kg·m² must change speed by 3.5 ÷ 0.1 = 35 rad/s, about 330 rpm. Stopping the turn means slowing the wheel back down.
Saturation: a steady disturbance torque τ_d (from solar pressure, gravity gradient or air drag) adds momentum τ_d × t to the wheel. If a wheel can store H_max, it saturates after t = H_max ÷ τ_d. Desaturation — NASA calls it momentum dumping — applies an outside torque with magnetic torquers or thrusters while the wheel spins down.
Star trackers identify stars from the angles between them, then compute the rotation from the catalogue frame to the camera frame. Between star-tracker fixes, gyro rates are integrated to carry the attitude forward. NASA notes that gyro bias must be estimated onboard, so each new star-tracker fix is also used to correct the gyro bias.
Conservation of angular momentum: when the wheel's spin changes one way, the satellite body rotates the opposite way.
A wheel saturates when steady outside torques have pushed it to its maximum speed. It is desaturated by using magnetic torquers or thrusters to apply an external torque while the wheel slows down.
It photographs stars and matches their pattern against an onboard star catalogue, which gives the camera's pointing direction to within a few arcseconds.