Gears, belts, chains, and transmissions

In brief

Shift a bicycle into a low gear before a steep hill and pedaling suddenly gets easier, though slower for each turn of the pedals; shift up on a flat stretch and the reverse happens. Nothing about your legs changed, and nothing about the road changed. What changed is the ratio between two toothed wheels, the front chainring and the rear cog, connected by a chain. That single ratio is the entire idea behind every gear, belt, and chain transmission ever built: connect two rotating wheels so their rims agree at the point of contact, and you are forced into a fixed trade between how fast something turns and how much twisting force, or torque, it can deliver. Choosing that ratio correctly is the difference between a vehicle that cannot start moving on a hill and one that cannot reach a useful speed.

The full treatment

First look: why a low gear feels easier

On a bicycle, shifting to a low gear puts a small chainring in front and a large cog behind. The pedals must turn many times to spin the rear wheel once, so your feet move faster relative to the ground speed gained, but each push of your legs is amplified into a larger turning force at the wheel. This is exactly the lever trade in a different costume: rotating levers connected in a fixed ratio, one favoring effort, sacrificing speed, and one favoring speed, sacrificing effort.

Building the idea: the no-slip contact condition

Consider two circular wheels, radius r1 (the driver, receiving the input) and radius r2 (the driven wheel, delivering the output), touching at a single point, or connected by a belt or chain so that this point behaves as if they touch. Assume no slipping at that contact: the two wheels' rims must move at the same linear speed at the point where they meet, otherwise one surface would be scraping past the other. The rim speed of a wheel spinning at angular speed omega (radians per second) at radius r is v equals omega times r. Setting the two rim speeds equal, omega1 times r1 equals omega2 times r2, gives the entire kinematic result at once: omega2 divided by omega1 equals r1 divided by r2. The output spins slower than the input exactly in proportion to how much larger its radius is.

For gears, teeth are cut at a fixed, uniform spacing around each wheel, so the number of teeth on a gear is directly proportional to its radius. This lets the same relationship be written in a form that can be read straight off a parts catalog, without ever measuring a radius: gear ratio equals number of teeth on the driven gear divided by number of teeth on the driver gear, and the output angular speed equals the input angular speed divided by that gear ratio.

Building the idea: torque follows from conserved power

Speed is only half the story; the other half is torque, and it follows from a separate, equally basic constraint: energy conservation. In an ideal, lossless transmission, the mechanical power flowing in must equal the power flowing out, since no power is created or destroyed at an ideal contact. Power delivered by a rotating shaft equals torque times angular speed, so torque_in times omega_in equals torque_out times omega_out. Rearranging, torque_out divided by torque_in equals omega_in divided by omega_out, which by the earlier result equals the gear ratio itself. Slowing the output down by a factor of the gear ratio multiplies its torque by that same factor. This is the entire content of a transmission in one sentence: it cannot change power, only redistribute a fixed amount of it between speed and torque, in a ratio fixed by geometry.

Real transmissions lose a small fraction of that power to friction at each meshing contact, at bearings, and, for belts, to internal flexing of the belt material, so real output torque is the ideal value multiplied by an efficiency factor less than one, typically 0.95 to 0.99 for a single well-made gear stage.

Three ways to build the contact: gears, belts, and chains

Gears enforce the no-slip condition through interlocking teeth: contact is positive, meaning the ratio holds exactly regardless of load, up to the point the teeth themselves break. Belts, particularly smooth flat or V-shaped belts, rely instead on friction between the belt and the pulley surface to transmit torque without teeth; this makes them quiet and tolerant of misalignment, but it means the connection can slip if the demanded torque exceeds what friction can sustain, a real limit rather than a rare accident. Chains and toothed (synchronous) belts split the difference: a chain's links engage sprocket teeth positively, like a gear, so it does not slip, while still offering a belt's flexibility to span a distance and route around obstacles that two gears in direct contact could not.

Lineage

Gearing is ancient: the Antikythera mechanism, a Greek device from roughly the second century BCE, used dozens of bronze gears to model astronomical cycles, and geared devices independently appear in early Chinese engineering, including the south-pointing chariot. Belt drives and chain drives matured through the water-wheel and steam-era factory, where a single power source drove long lines of shafting through belts running the length of a mill. The formal engineering treatment of gear-tooth geometry, ensuring the no-slip contact condition holds smoothly through an entire revolution rather than only at one instant, was worked out through the involute tooth profile in the eighteenth and nineteenth centuries and remains the geometry cut into nearly every gear made today, standardized by bodies such as AGMA and documented in texts including Shigley's Mechanical Engineering Design and Norton's Design of Machinery.

The strongest case for it

Gear transmissions deliver very high efficiency, typically ninety-five to ninety-nine percent per stage, and, because their contact is positive rather than frictional, they hold their ratio exactly regardless of load, which is why precision instruments and robotics rely on gear trains wherever exact, repeatable motion matters. The speed-torque trade-off predicted by the power-conservation argument is confirmed continuously, at every scale from wristwatch gear trains to ship propulsion drives, which is part of why gear, belt, and chain design is standardized rather than treated as a matter of trial and error.

The strongest case against it

The no-slip assumption is an idealization that fails in specific, well-known ways. Friction belts are deliberately capable of slipping past a torque threshold, which is sometimes a designed-in safety feature, protecting a motor from a jammed load, but is a real failure mode whenever the required torque is underestimated. Chains and gears both suffer real wear: a chain's pins and bushings wear and the chain effectively stretches, or elongates, over its service life, changing its engagement with the sprocket teeth and eventually causing skipping; gear teeth wear and can suffer fatigue failure at the tooth root under repeated loading. Manufacturing tolerance and gear backlash, the small gap left between mating teeth so they do not jam, mean the output shaft's angular position lags the input's by a small, variable amount, which matters for precision positioning even though it barely matters for delivering power. Multi-stage transmissions compound their per-stage efficiency losses, so a gearbox with several reductions in series can lose a noticeably larger fraction of power than any single stage suggests. A common misconception is expecting a single fixed gear ratio to serve both a demanding start and a high top speed well; the two requirements pull the ratio in opposite directions, which is exactly why multi-speed transmissions, or continuously variable transmissions that trade some efficiency for a continuously adjustable ratio, exist at all.

Where it stands now

The kinematic and power-conservation reasoning behind gear, belt, and chain ratios is settled engineering, unchanged in its fundamentals for well over a century and standardized internationally. Active engineering effort continues in materials, tooth-profile optimization for noise and efficiency, and continuously variable transmission designs, but none of it revises the basic contact-condition and power-conservation logic developed here.

Test yourself

A small electric motor delivers its rated torque at 200 radians per second but stalls, delivering roughly three times that torque, at zero speed. The vehicle you are building needs to climb a steep ramp from a standing start, which requires at least five times the motor's rated torque at the wheel, and afterward needs to cruise at a wheel speed of 40 radians per second. Using the relationships that output speed equals input speed divided by the gear ratio and output torque equals input torque times the gear ratio (times an efficiency factor you may take as 0.95), determine whether a single fixed gear ratio can satisfy both the hill-start torque requirement and the cruising speed requirement, and if not, propose how many distinct ratios you would need and roughly what they should be.