Precision Worm Gears for Robots

A worm gear is rarely the first reducer people name for a robot joint. Harmonic and planetary reducers get most of the attention. Still, worm gearing keeps turning up in robots where a compact right-angle drive, high single-stage reduction and quiet running matter more than peak efficiency: pan-tilt heads, turrets, rotary stages, mobile-robot drive units and lifting axes.

Where worm gears fit in robotics

A worm and wheel convert rotation between non-intersecting, perpendicular shafts at a single-stage ratio commonly between 5:1 and 100:1. The contact is sliding rather than rolling. That gives smooth, quiet meshing and, depending on lead angle, the option of resisting back-driving. ESSOR's own engineering guide lists robotic joints among the places worm drives remain in use alongside planetary and harmonic reducers (Worm Drive Systems: The Complete Engineering Guide).

ESSOR lists these robotics-related uses on its site:

• Tailor-made worm drives for robotics, area-scan turret cameras and mobile robots for logistics.
• Rotary stages with a hollow table, a laterally mounted servo motor and full 360° rotation. Robotic-arm joints are one of the listed applications.
• A patent list on the Measurement page that includes titles on AGV worm gear reducers, a rotating mechanism for a vehicle-moving robot and a dual-lead steel worm gear reducer for drone drives.

A practical filter. In my experience reviewing drive selections, a worm gear is a good candidate when the axis is positioned and then held, when the envelope is tight in one direction, and when noise matters. It is a poor candidate when the joint runs near continuous duty, when battery runtime is tight, or when you need a back-drivable joint for force control. This is general engineering judgment rather than an ESSOR claim, so check it against your own duty cycle.

The four specifications that decide performance

1. Accuracy grade

ESSOR states that it manufactures to DIN 1, with serial production typically reaching DIN 3–4 and DIN 1 available as the top specification (Worm Wheel & Worm Gear Guide). Read that carefully: the nominal class and the production-typical class are different numbers, so state which one your drawing requires. Tighter grades mean lower cumulative pitch error and lower single-flank composite deviation, which is what gives a rotary axis repeatable positioning.

2. Backlash, and what happens as it wears

The wheel is usually softer bronze, so it wears slowly, and the free play at the output grows with it. ESSOR's backlash article gives a simple model: backlash at the pitch line equals an initial clearance plus wear, minus the lead difference times the axial shift. Its example uses an 80 mm pitch-diameter wheel (module 2, 40 teeth). I converted its linear figures to angle using j ÷ pitch radius:

Source of the geometry: assumptions in the ESSOR backlash article. The angle values are my own calculation. A smaller wheel turns the same linear gap into a larger angle.

For a robot, this matters most when an axis reverses often. Each reversal costs the free play as positioning error, and a camera or end-effector on the output shows it directly.

3. Lead angle: the efficiency and self-locking trade-off

Efficiency follows η = tan γ ÷ tan(γ + φ), where γ is the lead angle and φ = arctan(μ). Using the same assumptions as ESSOR's ratio simulator (2 mm module, 30 mm worm pitch diameter, 40 wheel teeth, μ ≈ 0.05), I worked out how the start count changes the result:

Calculated values for the gear mesh only. Bearing, seal and churning losses reduce real-world efficiency further.

The self-locking surprise. At μ ≈ 0.05 the self-locking limit is arctan(0.05), about 2.9°. Even the single-start, 3.8° pair above exceeds that, so it would back-drive. Rules of thumb such as "under 5° locks" depend heavily on lubrication and friction. ESSOR's own backlash article says not to rely on self-locking as a safety brake, because vibration can release a marginal lock. For a robot that must hold a load when power is lost, specify a brake and treat any self-locking as a bonus.

4. Heat

Sliding contact generates more heat than planetary or helical gearing. ESSOR notes that a sump temperature above about 90 °C degrades the oil and speeds up wear, and that after any backlash adjustment you should confirm the running temperature rise at your actual duty cycle. Compare the worm drive's 40–90% efficiency range with 90–98% per planetary stage and 95–99% for helical gearing (ESSOR engineering guide). For a battery-powered mobile robot, that gap is an energy cost, not just a thermal one.

Keeping backlash low as the gear wears

A standard cylindrical worm has constant thread thickness, so backlash is fixed at assembly by the housing center distance. Two designs let you take it back up later without opening the housing to replace the wheel.

Based on ESSOR's backlash article and product pages.

The dual-lead idea is that the two flanks of the thread have slightly different leads, so the thread gets gradually thicker along the axis. Shifting the worm toward the thick end closes the gap. The ratio does not change, because it is still wheel teeth divided by worm starts. Two limits are worth designing around:

• The adjustment range is finite. Once the worm reaches the thickest part of the thread, nothing is left to take up.
• Aim for minimum working backlash, not zero. Zero backlash creates preload, frictional heat and binding, especially on low-lead-angle sets.
• The mounting must allow it. The worm shaft has to move axially in a controlled way and then lock securely. Bearing play shows up as backlash in the same way as flank wear.

Material, profile and lubrication choices

• Worm: hardened and ground alloy steel (the ESSOR guide gives 20CrMnTi and 42CrMo as examples).
• Wheel: tin bronze for standard duty, aluminum or nickel-aluminum bronze for heavy duty, and engineering plastics such as POM only for light, instrument-grade loads.
• Profile: ZA, ZI, ZN, ZK and ZC each trade load capacity, efficiency and cost. The guide notes that most premature failures trace to poor or contaminated lubrication rather than geometry.
• Lubricant: use grease or oil formulated for bronze-on-steel sliding contact. ESSOR's dual-lead worm drives list a high-performance synthetic lubricant among their features.

The standards behind these terms are DIN 3975 (terms, parameters and flank forms for cylindrical worm gear pairs), DIN 3996 and ISO/TS 14521 (load-capacity calculation) and ANSI/AGMA 6034 (fine-pitch worm gearing), as summarized in ESSOR's engineering guide. Obtain the current editions from the issuing standards bodies before basing a design on them.

Can the supplier prove the precision?

A precision claim is only as good as the inspection behind it. ESSOR's Measurement page lists a Klingelnberg P65 gear measuring center, a Leitz coordinate measuring machine, a metallurgical microscope, and Brinell and Rockwell hardness testers. It also shows ISO 9001, IATF 16949, ISO 45001 and ISO 14001 certificates. The About page describes a company founded in 2000, with design validation in KISSsoft and SolidWorks, more than 100 patents and more than 300 global customers. These are the company's own statements, so request the relevant certificate copies and a sample inspection report for your part.

A short supplier checklist

• Which accuracy class will be certified for my part, and on which machine is it measured?
• Can the same tolerance hold from prototype to mass production?
• Is backlash adjustment available (dual-lead or split)?
• Does the lead angle meet both my efficiency target and my holding-torque margin, with a brake where needed?

Custom and OEM options

Robot housings seldom match catalog dimensions. ESSOR's Custom Solutions page describes design and prototyping from physical parts or drawings, small-batch trial production and performance verification, plus one-stop OEM manufacturing covering design, machining, assembly and inspection. For gear sets alone, see custom worms and wheels, which covers ZK, ZI, ZA and other profiles.

To get an accurate quote, prepare the following:

• A drawing or sample, or at least the center distance, module and envelope limits.
• Required torque, ratio, input speed and duty cycle.
• Target backlash and accuracy class, and whether the axis reverses often.
• Whether the joint must hold position when unpowered.

You can also browse dimensional data in the Worms & Wheels catalog and the Worm Drive catalog. ESSOR also exhibited at the 2nd Hangzhou International Humanoid Robot and Robot Technology Exhibition on 14–16 May 2026 (booth B224–225), according to its company news. The news post gives logistics only, not product details.

Frequently asked questions

Are worm gears accurate enough for robot joints?

Yes, for the right axes. Wheel accuracy up to DIN 1 is stated for ESSOR's dual-lead and split products, and adjustable backlash helps keep that accuracy through wear. For joints that need minimal friction and back-drivability, other reducer types may fit better.

Does a dual-lead worm change the ratio?

No. The ratio stays wheel teeth divided by worm starts. The lead difference only varies the thread thickness for backlash adjustment.

Can I swap a dual-lead worm into my existing housing?

Only if the housing lets the worm shaft move axially and lock, and the wheel is cut to match. A standard wheel will not mesh correctly with a dual-lead thread, so the pair must be made as a set.

Will a worm gear hold my robot arm when power is off?

Do not count on it. Self-locking depends on lead angle, lubrication and vibration, and a marginal lock can release. Use a brake for safety-critical holding.