Two worm-wheels with one tooth difference is a compound worm drive where a single worm shaft engages two coaxial worm-wheels whose tooth counts differ by exactly one — for example 30 and 31 teeth. Precision robotics and instrument drives use this mechanism where huge reduction must fit in one stage. The differential between the two wheels produces the output rotation, so each input revolution advances the output by a tiny fraction of a tooth. The result is reduction ratios of 500:1 to 3000:1 in a package shorter than a typical 2-stage spur gearbox.
Two Worm-wheels with One Tooth Difference Interactive Calculator
Vary the fixed and output worm-wheel tooth counts to see the differential reduction ratio, output step size, and worm revolutions per output tooth.
Equation Used
This calculator uses the worked example differential worm-drive relationship. For a single-start worm, the effective reduction is the product of the grounded and output wheel tooth counts divided by their tooth-count difference. With 30 and 31 teeth, the ratio is 30 x 31 / 1 = 930:1.
- Single-start worm, so one worm revolution advances one tooth.
- Fixed wheel is grounded and the second wheel is the output.
- Both worm-wheels have matching lead and pitch; backlash and efficiency losses are not included.
- Tooth counts must be different for a valid differential ratio.
How the Two Worm-wheels with One Tooth Difference Actually Works
The trick here is the differential. A single worm engages two worm-wheels mounted on the same axis — one wheel grounded to the housing, the other free to rotate as the output. If the worm has the same lead engaging both wheels but the wheels have different tooth counts, say 30 and 31, then one input revolution of the worm advances both wheels by one tooth. The grounded wheel cannot move, so the worm itself walks around it. That walking motion is what carries the free wheel forward — but only by the difference of 1 tooth out of 31, not by a full tooth. So you get a 31:1 ratio per single tooth difference, multiplied by the worm's own reduction, giving combined ratios that easily clear 1000:1 in a single stage.
Why build it this way? Because the alternative — stacking spur or worm stages — eats axial length, adds backlash at every mesh, and bleeds efficiency. A differential worm drive collapses two reductions into one volume. The compound worm gear reduction also keeps the self-locking behaviour of a normal worm drive, which means back-driving resistance stays high and the output holds position with the motor unpowered.
Get the tolerances wrong and the mechanism punishes you fast. The two worm-wheels must share an identical lead angle to within roughly 5 arc-minutes — if they don't, the worm cannot fully engage both wheels at once and you get tooth-tip contact, rapid wear, and a juddering output. Centre distance from the worm to each wheel must match within about 0.02 mm. The most common failure modes are: scuffing on the smaller wheel because the lead-angle mismatch concentrates load on a single tooth flank, and seizing under load when thermal expansion closes the already-tight backlash to zero.
Key Components
- Single Worm Shaft: One continuous worm spans both wheels with identical lead and identical pitch diameter across its working length. Lead error must stay below 0.005 mm over the engagement length or load splits unevenly between the two wheels.
- Fixed Worm-wheel (N teeth): The grounded wheel — typically 30 teeth in the classic example — is rigidly fastened to the housing. It acts as the reaction reference. Tooth-count accuracy and concentricity to the worm axis define output backlash; runout above 0.01 mm shows up directly as output wobble.
- Output Worm-wheel (N+1 teeth): The free wheel carries the output flange and has exactly one more tooth than the fixed wheel. The 1-tooth difference sets the per-revolution output advance. Both wheels share the same outer diameter despite the tooth-count difference, achieved by running a slightly finer pitch on the larger wheel.
- Coaxial Bearing Stack: Two precision angular-contact bearings or a crossed-roller bearing carry the output wheel coaxial with the fixed wheel. Coaxiality must hold to 0.01 mm TIR or the worm engagement walks between the two wheels under load.
- Housing: Carries the grounded wheel rigidly and locates the worm shaft at the correct centre distance. Bore-to-bore parallelism between the worm bore and the wheel axis is the single most important machining tolerance — 0.02 mm over 100 mm length is typical.
Where the Two Worm-wheels with One Tooth Difference Is Used
You see this mechanism wherever space and ratio matter more than peak efficiency. The differential worm drive shows up in instrument drives, antenna positioners, robotic joints, and any application where a harmonic drive alternative is wanted at lower cost. It is not a high-speed mechanism — efficiency hovers around 30-50% — but for indexing and positioning it is hard to beat on packaging.
- Aerospace tracking: Azimuth drive on the Kongsberg KSP-7000 satellite tracking pedestal, where a 1500:1 differential worm collapses the elevation reduction into a single compact stage.
- Robotics: Wrist roll joint on early SCARA arms such as the AdeptOne, where the differential worm gives high reduction with self-locking behaviour so the gripper holds payload with motor power off.
- Optical instruments: Slow-motion fine-adjustment drive on Carl Zeiss astronomical theodolites, where 2000:1 reduction lets a hand wheel produce arc-second pointing changes.
- Industrial indexing: Rotary table indexers on Hardinge HC-style turret lathes, used as a manual jog-fine adjuster for tool-station alignment.
- Defence: Turret traverse fine-aim mechanism on naval gun mounts where the operator's hand crank must produce sub-mil aim corrections through a high reduction with positive locking.
- Lab automation: Stage drive on Zeiss Axio Imager microscopes for the focus block, where the 1-tooth differential gives sub-micron Z-axis advance per knob revolution.
The Formula Behind the Two Worm-wheels with One Tooth Difference
The reduction ratio of a differential worm drive depends on the worm starts, the tooth counts of both wheels, and which wheel is grounded. At the low end of the typical range — say 20 and 21 teeth — you get around 420:1, fast enough to be useful but already at the edge of what a single-start worm can drive efficiently. At the nominal sweet spot, around 30 and 31 teeth with a single-start worm, the ratio sits near 930:1 with reasonable manufacturing tolerances. Push to 60 and 61 teeth and you can theoretically reach 3660:1, but tooth-flank load drops so low that any centre-distance error causes the worm to skip teeth under shock load. Most production designs live in the 800-1500:1 band.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| i | Overall reduction ratio (input revs per output rev) | dimensionless | dimensionless |
| N1 | Tooth count of the fixed (grounded) worm-wheel | teeth | teeth |
| N2 | Tooth count of the output worm-wheel | teeth | teeth |
| z | Number of starts on the worm | starts | starts |
| N2 − N1 | Tooth-count difference (always 1 for this mechanism) | teeth | teeth |
Worked Example: Two Worm-wheels with One Tooth Difference in an azimuth fine-aim drive on a radio telescope feed positioner
You are sizing the differential worm drive on the feed-leg positioner of a 12 m research radio dish — similar in scale to the smaller dishes at the Allen Telescope Array. The drive must convert a 1800 RPM stepper output into sub-arc-minute azimuth pointing. The selected design uses a single-start worm engaging a fixed 40-tooth wheel and an output 41-tooth wheel.
Given
- N1 = 40 teeth
- N2 = 41 teeth
- z = 1 start
- Input speed = 1800 RPM
Solution
Step 1 — compute the nominal reduction ratio with the 40/41 tooth pair:
Step 2 — convert input RPM to nominal output RPM:
Step 3 — at the low end of the typical instrument-drive range, drop to a 20/21 tooth pair to get a faster slew:
4.29 RPM is fast enough to slew the dish across half the sky in roughly 20 seconds — useful for coarse acquisition but too coarse for fine pointing because each stepper microstep advances the dish more than an arc-minute.
Step 4 — at the high end, push to a 60/61 tooth pair for finer resolution:
0.49 RPM gives roughly arc-second resolution per microstep, but the load per tooth drops so low that any 0.03 mm centre-distance drift from thermal expansion lets the worm skip a tooth under wind gust load. The 1640:1 nominal sits in the sweet spot — fast enough for slew, fine enough for pointing, and tooth load high enough to prevent skip.
Result
Nominal output speed is 1. 098 RPM at the dish. That feels almost stationary to the eye — the feed leg drifts across the sky at roughly 6.6° per minute, fine enough that an operator watching by hand sees no motion in real time. At the 420:1 low-end choice the dish slews 4× faster but pointing resolution collapses; at 3660:1 you get arc-second resolution but the drive becomes shock-sensitive. If you measure 1.5 RPM output instead of the predicted 1.1 RPM, the most likely causes are: (1) the worm has a 2-start thread instead of a 1-start — re-check the lead, (2) the fixed wheel has slipped on its housing pin and is rotating slowly with the output, halving the effective reduction, or (3) the output coupling has a hidden 1.5:1 ratio downstream that was not in the spec.
Choosing the Two Worm-wheels with One Tooth Difference: Pros and Cons
Differential worm drives compete directly with harmonic drives, cycloidal reducers, and stacked worm-and-spur reductions. The choice comes down to ratio per volume, efficiency, backlash, and cost. Here is how the differential worm drive stacks up against the two most common alternatives at the same target ratio of roughly 1000:1.
| Property | Two Worm-wheels with One Tooth Difference | Harmonic Drive | Stacked 2-Stage Worm |
|---|---|---|---|
| Reduction per stage | 500:1 to 3000:1 single stage | 30:1 to 320:1 single stage | 100:1 to 600:1 across 2 stages |
| Efficiency at rated load | 30-50% | 70-85% | 40-60% |
| Backlash (typical) | 3-10 arc-min | <1 arc-min (zero-backlash variants) | 8-20 arc-min |
| Self-locking / back-driving resistance | Yes, fully self-locking | Back-drivable | Yes, self-locking |
| Axial length for 1000:1 | ~80 mm | ~50 mm | ~180 mm |
| Relative cost (1=cheapest) | 2 — moderate, custom wheels | 4 — high, precision flexspline | 1 — off-the-shelf parts |
| Best application fit | Indexing, slow positioning, instrument drives | Robotics joints needing zero backlash and high efficiency | Heavy hoists where length is not critical |
| Lifespan at continuous duty | 2,000-5,000 hr (efficiency-limited heat) | 10,000+ hr | 5,000-15,000 hr |
Frequently Asked Questions About Two Worm-wheels with One Tooth Difference
Backlash compounds across both meshes. You have one worm-to-wheel mesh on the fixed side and another on the output side, and any clearance at either mesh shows up at the output. If each mesh has 0.02 mm circumferential clearance, the output sees roughly the sum, not the average — so 6-8 arc-minutes is normal even with carefully cut wheels.
If you need under 2 arc-minutes you typically need to preload the worm radially against both wheels using an eccentric bushing on the worm shaft, taking up backlash on both meshes simultaneously. A duplex worm with offset lead between the two wheel zones is the production solution.
It is not just about the ratio. Tooth load per mesh scales inversely with tooth count for a given output torque, so the 60/61 pair runs each tooth at half the load of a 30/31 pair. Sounds good — but the smaller tooth on a 60/61 wheel is roughly half the size, so bending strength drops faster than load drops. The net is that 60/61 only wins on smoothness, not capacity.
Pick the lower tooth count compatible with your ripple and resolution targets. For most instrument drives 30-45 teeth is the sweet spot.
Differential worm drives are heat-limited because efficiency sits at 30-50%. At continuous duty most of the input power becomes heat at the worm-wheel mesh, and bronze wheel temperatures above 90°C cause oil viscosity to drop and tooth flanks to gall. The speed loss you see is real — galling roughens the flanks and pushes the worm away from full tooth engagement.
Check oil sump temperature with an IR thermometer. If it climbs past 80°C in normal duty, you need either forced air cooling, a larger oil sump, or a duty-cycle limit. The Zeiss theodolite drives, for example, are explicitly rated for intermittent duty only.
You can, but you pay for it in ratio. The denominator in the reduction formula multiplies by the start count z, so a 2-start worm with 40/41 wheels gives 820:1 instead of 1640:1. Efficiency climbs from roughly 35% to 55% because the larger lead angle reduces sliding losses.
The honest tradeoff: if you have ratio to spare, multi-start is the right choice. If you are already at the minimum ratio that works for your resolution, stay single-start and accept the efficiency.
This is almost always a single bad tooth on one of the wheels — either a tooth-spacing error from indexing the gear-cutting machine wrong, or a nick from handling. Because the two wheels rotate at slightly different rates, a defect on the fixed wheel shows up at one specific worm position per output revolution, while a defect on the output wheel shows up at a fixed output angle.
Mark the output shaft and watch where the stutter occurs. If it tracks output angle, it is the output wheel. If it tracks worm rotation count, it is the fixed wheel. Replace the offending wheel — re-cutting one tooth almost never works because the surrounding teeth already carry the wear pattern.
For a robotic arm doing actual work, use a harmonic drive. The differential worm has three problems for robotics: efficiency near 35% means thermal limits at duty cycles above 30%, backlash of 5+ arc-minutes is too coarse for end-effector positioning, and the self-locking behaviour means you cannot back-drive for compliance or force control.
Where the differential worm wins in robotics is fixed-base indexing applications — rotary tool changers, payload turntables, fixture rotators — where the joint moves slowly, holds position with motor off, and never needs back-drive sensing.
References & Further Reading
- Wikipedia contributors. Worm drive. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
