When the Hubble Space Telescope finishes observing one galaxy and turns toward another, nothing fires into space. There is no puff of hydrazine and no small rocket hidden behind the solar arrays. The manoeuvre begins with electric motors changing the speeds of metal wheels inside the observatory.

That is not merely a fuel-saving technique. Hubble was built without thrusters for attitude control, in part because exhaust products could contaminate its field of view. Since the telescope was deployed in 1990, routine pointing has depended on two linked systems: reaction wheels that turn the spacecraft, and magnetic torquer bars that keep those wheels from accumulating too much momentum.

A telescope turns by speeding up something inside it

NASA’s description of Hubble’s pointing control lists four reaction-wheel assemblies. Each weighs about 100 pounds, or 45 kilograms, and is roughly two feet across. An electric motor can accelerate a wheel or slow it down.

The operating principle is conservation of angular momentum. If a wheel gains angular momentum in one direction, the rest of the spacecraft must respond in the other direction. Accelerating the appropriate combination of wheels starts Hubble rotating; changing their speeds again brakes the telescope as it approaches the commanded attitude.

The wheels are not propellers and they do not push against the air. They work precisely because the wheel and spacecraft form a coupled mechanical system. The wheel turns one way inside the structure while the roughly 12-tonne observatory turns the other way around their shared centre of mass.

Four wheels solve a three-axis problem

A free spacecraft can rotate about three axes, conventionally called roll, pitch and yaw. Three independently controlled reaction wheels are sufficient for three-axis control. Hubble carries four so that the wheel set has redundancy, with their orientations allowing the control computer to combine their torques into the movement required.

The resulting rotation is intentionally slow. According to NASA’s mission-operations guide, turning Hubble through 90 degrees takes about 14 minutes. NASA compares its fastest movement across the sky to the minute hand of a clock. A slow slew limits the forces imposed on the long telescope and gives the control system time to arrive at the next target accurately.

Stopping at the approximate direction is only part of the operation. Mission planners must also allow time for the observatory to identify guide stars and settle into the stable pointing needed for an exposure. The Space Telescope Science Institute’s pointing primer describes a slew limit of about six degrees per minute and a separate guide-star acquisition period. Time spent moving and acquiring guide stars is one reason closely spaced observations can be more efficient than repeated large turns.

The gyroscopes sense motion; they do not make it

Reaction wheels are often confused with Hubble’s gyroscopes because both contain rotating components and both belong to the pointing-control system. Their jobs are different. Gyroscopes are sensors. They report the telescope’s rotational rate and help the computer determine how its attitude is changing. Reaction wheels are actuators. They apply the torque that makes the change happen.

Other sensors contribute different kinds of knowledge. Sun sensors establish a coarse reference, magnetometers measure the local magnetic field, star trackers recognise star patterns, and the Fine Guidance Sensors lock onto guide stars for the precision needed during science observations. NASA’s Hubble overview puts the observatory’s pointing accuracy at 0.007 arcsecond, which is comparable to holding a laser beam on a small coin hundreds of kilometres away.

This distinction matters when reading reports about gyro failures. SpaceDaily has previously covered Hubble’s move to single-gyroscope operations. That change reduced the sensor configuration used to establish and maintain attitude; it did not mean that one gyro had taken over the physical work of turning the observatory.

Reaction wheels can fill up with momentum

In an idealised system with no outside torque, a reaction wheel could speed up for a slew and later give that momentum back. Hubble’s environment is not ideal. Even several hundred kilometres above Earth, traces of atmosphere exert drag. Sunlight produces radiation pressure, and gravity can apply small orientation-dependent torques across an extended spacecraft.

The pointing system continually corrects for these disturbances. Every correction slightly changes the momentum stored in the wheels. If the environmental torque has a persistent bias, one or more wheels gradually move toward their maximum or minimum useful speed. Eventually a wheel would reach saturation, leaving too little speed margin to absorb the next disturbance or execute the next command.

A closed spacecraft cannot solve that problem simply by spinning another internal part. Moving momentum among the wheels changes its distribution, but the total remains inside the vehicle. To unload the wheels, Hubble needs to exert torque against something outside itself.

Magnetic bars provide an external purchase

That outside connection is Earth’s magnetic field. Hubble has four magnetic torquer bars, positioned at 90-degree intervals around the spacecraft. Each is an iron rod about eight feet, or 2.4 metres, long and wrapped in coils of wire. Sending current through a coil gives the bar a controllable magnetic field.

The interaction between that field and Earth’s field produces torque on Hubble. While the torquer bars provide the external torque, the control system can reduce the reaction-wheel speeds and move them back toward a useful operating range without allowing the telescope to tumble. Engineers call the process momentum unloading or desaturation.

The phrase “bleed off momentum” is convenient, but the momentum is not destroyed. Hubble exchanges angular momentum with Earth through the magnetic interaction. The available torque depends on the strength and direction of the local field, so it is weaker and less direct than a rocket firing. It is nevertheless enough for routine wheel management in low Earth orbit.

Why Hubble does not use thrusters to point

Many spacecraft combine reaction wheels with small thrusters. When their wheels approach saturation, brief firings supply the external torque needed to unload them. That method is effective, but it consumes a finite fluid and releases exhaust close to the vehicle.

For Hubble, contamination was a central concern. A film deposited on optical surfaces could scatter light or reduce instrument performance, while material drifting through the viewing direction could interfere with observations. Magnetic unloading offered a clean alternative because it ejects no mass. Solar-array electricity powers both the wheel motors and the torquer coils.

This design removes one familiar lifetime limit: Hubble cannot run out of attitude-control propellant. It does not remove mechanical limits. Wheel bearings, motors, sensors and electronics can wear or fail, and astronauts replaced a reaction-wheel assembly during the 2002 servicing mission. NASA’s life-extension work has included ways to operate with fewer working reaction wheels if that becomes necessary.

Propellant-free steering does not mean propulsion

Reaction wheels change where Hubble points; they do not change the path of its centre of mass around Earth. Magnetic torquers likewise manage attitude rather than providing a sustained orbital boost. Hubble cannot raise its own orbit or replace the altitude gradually lost to atmospheric drag. During the servicing era, visiting space shuttles performed reboosts.

The engineering achievement is narrower and more durable. For more than three decades, a telescope the size of a bus has crossed the sky without spending fuel on each turn. Its wheels trade momentum with the spacecraft, its magnetic bars trade accumulated momentum with Earth, and its sensors tell the system when the required attitude has been reached. The result is a steering architecture whose ordinary operation is almost invisible, even though nearly every Hubble observation begins with it.