Blog · 2026-08-15 · Linear
Rotary motors are easy to buy; straight-line motion is what you actually need. The five mechanisms that convert one to the other differ by an order of magnitude in force, speed and price — and one property, back-driving, quietly decides most designs.
Lifting a gantry, extending a gripper, driving a Z axis, pushing a syringe: all the same underlying problem, and all solved by picking a rotary-to-linear converter. The candidates are lead screws, ball screws, belt drives, rack and pinion, and integrated linear actuators. Rather than listing specifications, it's more useful to understand the two axes that separate them — force versus speed, and whether the mechanism holds position when unpowered — because those two decide almost every real design.
Every one of these mechanisms obeys the same conservation: force times distance in equals force times distance out, minus losses. A mechanism that multiplies force must divide speed by the same factor. A lead screw moving 2 mm per motor revolution multiplies torque enormously and moves slowly; a belt moving 40 mm per revolution does the reverse. There is no configuration that gives both, which is why the first question in any linear design is: am I short of force, or short of speed?
linear speed = rev/s × travel per revolution
force ≈ 2π × torque × efficiency ÷ travel per revolution
Notice efficiency appearing in the force equation but not the speed one. That's where the mechanisms genuinely differ, and it's a bigger factor than most people expect.
A threaded rod turning inside a nut. The hobby standard is the T8 trapezoidal screw with a brass or POM nut, and it dominates 3D printer Z axes for a specific reason: it self-locks. Friction in the thread is high enough that a load cannot back-drive the screw, so the axis holds its position with the motor completely unpowered — no brake, no holding current, no drift.
That holding ability is the same friction that makes lead screws inefficient: typically 25–40%, so most of the motor's work becomes heat in the nut. Speed is limited too, both by the fine lead and by "whip" — a long thin screw spun fast bends into a whirling arc and becomes violently unstable. The Lead Screw Calculator works out the torque and tells you whether a given screw and lead combination self-locks.
Choose a lead screw for vertical axes, presses, grippers and anything where holding a load without power matters more than speed. Choose a fine lead (T8×2) for force and resolution, a coarse one (T8×8) for speed — remembering that the coarse multi-start versions generally do not self-lock.
Replace sliding thread contact with recirculating ball bearings and efficiency jumps to around 90%. The consequences are all good except two: far less motor torque needed for the same force, minimal heat, much longer life, and very low backlash in preloaded versions — which makes ball screws the standard for CNC machines where positional accuracy under cutting loads is everything.
The two costs are money and back-driving. A ball screw and nut costs several times a lead screw, and because it's efficient it will happily be driven backwards by its load: a vertical ball-screw axis sinks under gravity the moment power is removed. That's not a defect, it's the inverse of the lead screw's friction bargain, and it means vertical ball-screw axes need a motor brake or counterbalance. For hobby robots, ball screws are usually overkill unless precision under load is the actual requirement.
A toothed belt with the carriage clamped to it, driven by a pulley. Travel per revolution is large — a 20-tooth GT2 pulley moves 40 mm per turn, twenty times a T8×2 screw — so belts are fast, and their cost barely increases with length, making them the obvious choice for long axes. Efficiency is high, they're quiet, and there's no whip limit.
What you give up is force and rigidity. The same twenty-fold speed advantage is a twenty-fold force disadvantage from the same motor, and belts stretch slightly under load, adding compliance that shows up as position error under varying force. Belts also back-drive completely freely, so a vertical belt axis needs a counterweight or brake. Use belts for horizontal axes, fast motion and long travel — sized with the Belt & Pulley Calculator and tensioned as described in the tensioning guide.
Rack and pinion — a gear running along a toothed bar — gives unlimited travel with no stretch and good rigidity, at the cost of backlash and the need to keep the pinion precisely engaged along the whole length. It's common on large gantries and robot arm extensions where a belt would be too springy.
Integrated linear actuators package a motor, gearbox, screw and housing into one sealed unit with a mounting eye at each end. They're the pragmatic choice when you need a specific stroke and force and don't want to design a mechanism at all — deployment legs, tilting platforms, hatch openers. You pay a premium and give up flexibility, but the engineering is done and they're usually self-locking.
| Lead screw | Ball screw | Belt | Rack & pinion | |
|---|---|---|---|---|
| Efficiency | 25–40% | ~90% | 95%+ | ~90% |
| Force from a given motor | High | Very high | Low | Moderate |
| Speed | Low | Moderate | High | High |
| Holds without power | Usually yes | No | No | No |
| Rigidity | Good | Excellent | Springy | Good |
| Practical travel | Short (whip limits) | Short to medium | Very long | Very long |
| Cost | Low | High | Very low | Moderate |
Take a robot arm that must lift a 2 kg payload 300 mm vertically in about three seconds. Force needed: roughly 20 N plus the mechanism's own weight, call it 30 N. Speed needed: 100 mm/s. A belt would manage the speed trivially but needs the motor to hold the load continuously and would drop it on power loss. A T8×8 lead screw at 100 mm/s means 750 RPM — beyond comfortable stepper territory (see the Stepper Motor Calculator) and it doesn't self-lock anyway. A T8×2 screw at 100 mm/s needs 3000 RPM, which is out of the question. The honest resolutions are: accept slower motion with a fine screw, use a belt plus a brake, or pick an integrated actuator rated for the job. Running the numbers early is what surfaces that kind of conflict before parts are ordered.
M8 threaded rod works for slow, light, non-critical axes and it's cheap — but its V-thread form is less efficient than a trapezoidal screw, it's rarely straight, and the nuts have significant backlash. Fine for a first prototype, worth replacing for anything you care about.
A long, thin screw spun fast bows outward and whirls violently, limiting usable RPM. It scales badly with length — doubling the length cuts the critical speed by roughly four. For screws over about 500 mm, either keep the speed low or support the far end.
You need rails. A screw should only ever carry axial load; using it to resist side loads bends it and destroys the nut. Guides take the load, the screw provides motion — the topic of How to Build a Linear Axis.
Force and speed get all the attention, but how often the axis moves quietly decides several designs. A lead screw at 30% efficiency turns most of the motor's work into heat in the nut, and a POM nut running continuously will soften and wear far faster than the same nut making occasional moves. If your axis runs constantly — a conveyor, a scanning stage, a pump — the efficiency column in the table above stops being an abstraction and becomes a maintenance schedule.
The same applies to holding. A self-locking screw holds a load for free, drawing no current at all. A belt or ball screw holding the same load needs the motor energised continuously, which on a stepper means full current into a stationary motor producing nothing but heat — and on a DC motor with a position loop means constant small corrections. Over an eight-hour run that difference is substantial in both battery drain and component temperature, and it's worth checking against the Power Budget Calculator before assuming a belt axis is the efficient choice. For intermittent motion the calculus flips entirely: a belt that moves briefly and rests unloaded is far more efficient overall than a screw grinding away at 30%.
Screws for force and holding, belts for speed and length, ball screws when precision justifies the cost. Run your numbers through the Lead Screw Calculator to see the torque and self-locking verdict before committing to a mechanism.