Two Ways to Move Torque Across a Gap

Once a motor is chosen, the next decision is how its output reaches the thing that has to move. In most product-scale mechanisms the choice comes down to a toothed belt on two pulleys, or a train of meshing gears. Both transmit torque and change speed. They fail differently, cost differently, and sound different, and picking wrong shows up either in the unit cost or in the warranty rate.

A timing belt wraps a reinforced elastomer loop around two toothed pulleys. Steel or aramid cords carry the tension, the tooth profile prevents slip, and the whole assembly tolerates shaft misalignment and axial distance changes. GT2, GT3, HTD, and the older T and AT profiles cover most small mechanisms.

A gear train transmits torque by direct tooth contact, either between two parallel-shaft spur or helical gears, or through worm, bevel, or planetary arrangements. Contact is rigid, the ratio is exact, and the shafts must be located precisely relative to one another.

When a Belt Is the Right Call

Belts win more often than engineers trained on industrial machinery expect, particularly in consumer and light commercial products.

  • Long center distance. Moving torque 6 to 24 in (150 to 600 mm) needs one belt or a five-gear idler train. The belt is cheaper, lighter, and quieter.
  • Noise matters. A properly tensioned belt runs 5 to 15 dBA below an equivalent spur gear set. For anything used near people at home or in an office, this alone often decides it.
  • Loose housing tolerances. Belts tolerate a few tenths of a millimeter of center distance error and a degree of angular misalignment. Gears do not, and holding gear centers in a molded plastic housing is expensive.
  • Shock loads and jam protection. A belt stretches slightly and, if you size it deliberately, skips teeth instead of stripping a gear or stalling a motor into a burned winding.
  • No lubrication allowed. Food contact, medical, and dusty environments all favor a dry running belt over a greased gear mesh that attracts contamination.
  • Low volume. Belts and pulleys are catalog parts. Custom molded gears need tooling, so at 500 or 2,000 units a year the belt is unambiguously cheaper.

When Gears Win

Gears earn their extra precision requirements in four situations.

High torque in a small package. A planetary gearhead delivers ratios of 5:1 to 100:1 in a cylinder the diameter of the motor. No belt arrangement is close on power density.

Positional accuracy under load. Belts have compliance. Under a reversing load, a belt-driven axis winds up and unwinds, and that lost motion is worse than gear backlash on anything demanding repeatability at the output. Anti-backlash and preloaded gear sets push this further.

Non-parallel shafts or self-locking behavior. Bevel gears turn a corner. A worm drive gives a very high ratio in one stage and, at low lead angles, will not back-drive, which means a lifting mechanism holds position with the motor unpowered. That single property frequently decides the architecture.

High volume and long life. Above roughly 20,000 units a year, molded plastic gears at $0.15 to $0.60 each beat belt-plus-two-pulleys at $3 to $8 per assembly, and there is no consumable to replace.

How the Decision Actually Gets Made

Work in this order and the answer usually falls out before you finish.

Start with the torque and speed at the output, then work back through the ratio to what the motor must supply, using the method in how to calculate the motor torque your product needs. Include acceleration torque, not just steady-state load, since a fast-indexing mechanism spends most of its torque budget on inertia.

Next, check whether the load must be held when unpowered. If yes, either a worm gear, a brake, or a self-locking lead screw enters the design and the belt option narrows considerably.

Then look at the layout. If the driven shaft is far away, offset, or moving, favor a belt. If it is adjacent and coaxial, favor gears.

Then check the acoustic and cost targets against annual volume. Then confirm the motor type is compatible with the plan, since the stepper, servo, and brushed DC options in stepper vs servo vs brushed DC each pair naturally with different transmissions. Hybrid arrangements are legitimate: a planetary gearhead on the motor for ratio, then a belt to carry that output across the machine, is one of the most common architectures in production equipment.

Mistakes That Repeat

  • Sizing the belt on torque alone. Belt capacity depends on the number of teeth in mesh and on belt width. A 6 mm GT2 belt on a 16-tooth pulley carries a fraction of what the same belt carries on a 40-tooth pulley. Use the manufacturer's tables rather than a stress calculation.
  • Ignoring minimum pulley diameter. Wrapping a belt tighter than its rated minimum destroys the tension cords through bending fatigue, and the failure appears after months, not on the bench.
  • No tension adjustment. Belts need a tensioner, a slotted motor mount, or an idler. Fixed centers with no adjustment means either a loose belt that skips or an over-tight belt that eats bearings, a load path detailed in bearing selection for a new product.
  • Plastic gears sized without heat in mind. Acetal and nylon gears fail by tooth root fatigue and by thermal softening at high duty cycles. Compare the two materials in nylon vs acetal, and derate hard for continuous running.
  • Forgetting the side load on the motor shaft. Belt tension is a radial load on the motor bearing. Small stepper and DC motors have modest radial ratings, and exceeding them quietly shortens motor life.
  • Testing only a few hundred cycles. Belt tooth wear, gear pitting, and pulley flange abrasion appear between 10,000 and 100,000 cycles. Both technologies pass a demo. Only one of them may pass the life test, which is the pattern described in material fatigue in product design.

A related layout question, when the motion is linear rather than rotary, is whether a belt is driving a carriage on rails at all; the supporting hardware options are compared in linear motion guides.

Get the Transmission Sized Before the Housing Is Drawn

Projects House sizes drive trains against the real duty cycle, not the nameplate, and locks the layout before the enclosure geometry makes the decision for you. Send your load, speed, cycle count, and noise target through our contact form.