Roller-bearing gear teeth replace solid involute teeth with rolling cylindrical pins or needle bearings that engage a matching cycloidal rack or wheel. The Nexen Roller Pinion System on heavy CNC gantries is one well-known implementation. Each tooth rolls into mesh instead of sliding, which kills the friction, heat, and backlash that conventional spur teeth generate under reversing loads. The result is sub-arcminute positioning accuracy, 99% mechanical efficiency, and service life past 100 million cycles in indexing duty.
Roller-bearing Gear Teeth Interactive Calculator
Vary input power and the two gear efficiencies to compare delivered power, heat loss, and savings for sliding versus rolling tooth contact.
Equation Used
This calculator compares the article's conventional sliding gear efficiency with roller-bearing rolling contact efficiency. For the same input power, the useful output is eta times input power, and the remaining power is treated as frictional heat loss.
- Both mechanisms receive the same input shaft power.
- Efficiency is entered as a steady-state percent.
- Heat loss is the mechanical power not delivered to the output.
- Default efficiencies use the article comparison: sliding contact about 95%, rolling contact 99%.
The Roller-bearing Gear Teeth in Action
A conventional spur tooth slides as it meshes — the contact point sweeps along the involute flank, and that sliding generates friction, heat, and wear. A roller-bearing tooth eliminates the sliding entirely. Each "tooth" is a hardened pin running on a needle or ball bearing, and the mating rack carries a cycloidal pocket profile shaped so the pin enters, rolls through engagement, and exits with pure rolling contact. No sliding means no stick-slip, no scuffing, and no measurable backlash when you preload two pinions against the rack.
The geometry has to be exact. The rack pocket is cut on a CNC gear shaper to a cycloidal profile generated from the pin diameter and the pitch radius — if the pocket is 50 µm too tight the pin binds and the bearing skids, if it's 50 µm too loose the pin rattles and you lose positioning accuracy. Pin diameter must match within ±5 µm across a full pinion. The Nexen RPS rack, for example, holds a tooth-to-tooth pitch tolerance of ±0.013 mm over a 1 m segment, and segments butt together with ground reference faces so a 6 m gantry rail meshes as one continuous track.
Failure in the field almost always comes from contamination or preload loss. Grit driven into the cycloidal pocket scores the pin and seizes the needle bearing — you'll hear a chirp at one rotational frequency, then the pinion starts shedding rollers. Preload backed off by even half a turn introduces backlash that shows up as a position error during reversal under load, not during constant-direction motion, which is why operators often misdiagnose it as a servo tuning issue.
Key Components
- Roller Pin (Tooth): A hardened steel pin, typically 52100 bearing steel through-hardened to 60-62 HRC, mounted on a caged needle bearing. Diameters run from 6 mm on small indexing pinions up to 25 mm on heavy gantry drives. Pin runout must stay under 5 µm or the rack pocket sees a varying contact angle and accuracy degrades.
- Needle Bearing: Sits between the pin and the pinion hub, letting the pin rotate freely as it rolls through the cycloidal pocket. A typical NA4900 series needle bearing carries 8-12 kN dynamic load. Loss of grease here is the number-one failure path — once the needle skids the pin grinds a flat and the whole pinion is scrap.
- Cycloidal Rack or Wheel: Carries the pockets that the pins roll into. Profile is mathematically derived from pin diameter D and pitch p so that contact is purely rolling at every angle of engagement. Surface ground to Ra 0.4 µm or better and induction-hardened to 55-58 HRC on the contact flanks.
- Pinion Hub: Holds 8 to 24 pin-and-bearing assemblies in a circular array. Hub bore is precision-ground to mate with the reducer output shaft within H6 fit. Two pinions on the same shaft, preloaded against each other, give zero backlash without spring-load schemes that introduce compliance.
- Preload Adjustment: On dual-pinion designs, an eccentric mounting block lets you crank one pinion into the rack until pin-to-pocket clearance hits zero, then back off 5-10 µm of working clearance. Set this with a torque wrench to a documented value — typically 8-12 Nm on the eccentric — and recheck after the first 100 hours of running.
Industries That Rely on the Roller-bearing Gear Teeth
Roller-bearing gear teeth show up wherever sliding-tooth gears can't deliver the accuracy, life, or efficiency required. The mechanism is overkill for a conveyor drive but indispensable on a 12 m laser cutter where you need ±15 µm repeatability over the full travel and you need it after 5 years of three-shift operation. Cost is the gating factor — a Nexen RPS rack runs roughly 8-10× the price of equivalent-module ground steel rack — so you see them on machines where downtime or rework cost dwarfs the gear bill.
- Machine Tools: X-axis drive on a Cincinnati CL-940 fibre laser cutter with a 12 m × 2.5 m bed, where Nexen RPS pinions hold ±20 µm positioning across the full travel
- Robotics & Automation: Seventh-axis linear track for a KUKA KR 240 R3330 welding robot, where a roller pinion drive carries the 1,200 kg robot at 2 m/s without backlash on reversal
- Pharmaceutical Packaging: Rotary indexing table on an IMA Adapta capsule filler running 200,000 capsules per hour, where roller-bearing teeth give the dwell-and-index motion zero overshoot at the stop position
- Aerospace Manufacturing: Gantry drive on an Electroimpact wing-skin riveter at the Boeing Renton 737 line, where the rack runs 30 m and tooth-to-tooth pitch error must stay under 25 µm over the full length
- Solar Tracking: Slew drive on a Nextracker NX Horizon single-axis tracker, where a roller-pinion stage replaces a worm gear to cut backlash-induced wind chatter in the panel array
- Heavy Material Handling: Turret rotation on a Liebherr LTM 1100-4.2 mobile crane, where rolling contact teeth handle reversing loads from luffing without the pitting that plagued earlier involute-tooth slew rings
The Formula Behind the Roller-bearing Gear Teeth
The pitch-circle relationship sets the pin spacing on the pinion against the pocket spacing on the rack. Get this wrong and the pins climb out of the pockets at one end of the engagement arc, which is exactly the failure mode you see on knock-off racks where the manufacturer guessed the cycloidal profile from photographs. At the low end of the typical pin-count range — 8 pins — you get a chunky pinion with high tangential force per tooth but rougher motion; at the high end — 24 pins — motion smooths out but each pin carries less load and the pinion gets physically bigger. Most CNC gantry designs sit at 12-16 pins, which is the sweet spot for tangential force capacity versus motion smoothness.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| p | Tooth pitch — centre-to-centre distance between adjacent pins on the pinion, measured along the pitch circle | mm | in |
| Dp | Pitch circle diameter of the pinion | mm | in |
| Z | Number of roller pins on the pinion | dimensionless | dimensionless |
| Ft | Tangential force per pin = T / (Zeng × Dp / 2), where Zeng is pins simultaneously engaged (usually 2-3) | N | lbf |
Worked Example: Roller-bearing Gear Teeth in a 9 m vertical CNC stone-cutting gantry
Sizing the Y-axis roller pinion drive on a Breton Contourbreton NC500 5-axis stone-cutting gantry at a granite countertop fabricator in Verona. The gantry beam weighs 2,400 kg, travels 9 m of Nexen-style rack, and must hold ±25 µm repeatability while cutting 30 mm granite at 4 m/min feed. Servo is a Yaskawa SGM7G 1.8 kW with a 10:1 planetary reducer ahead of the pinion. We need to set the pin count and verify the per-pin tangential load.
Given
- Dp = 120 mm
- Z (nominal) = 16 pins
- T (peak servo torque after reducer) = 180 Nm
- Zeng = 2 pins simultaneously in mesh
- Pin diameter = 16 mm
Solution
Step 1 — at the nominal 16-pin design, compute the tooth pitch on the pinion:
This pitch must match the rack pocket spacing within ±0.013 mm or the engagement skews. The rack we're specifying is cut at exactly 23.56 mm pitch, so we're locked in.
Step 2 — at nominal Z = 16 with 2 pins engaged, compute the per-pin tangential force at peak servo torque:
The NA4900 needle bearing inside each pin assembly is rated for 8 kN dynamic load, so we're at roughly 19% of dynamic rating. That gives an L10 life well past 100 million cycles — comfortable.
Step 3 — at the low end of the typical pin-count range, Z = 8 pins:
An 8-pin pinion gives a chunky 47 mm pitch and you can feel the mesh frequency through the gantry — at 4 m/min feed the tooth-passing frequency drops to 1.4 Hz, which is right in the resonant band of a heavy stone gantry. You'd see surface chatter on the cut.
Step 4 — at the high end of the typical range, Z = 24 pins:
24 pins lets 3 engage simultaneously, dropping per-pin load to 1,000 N (12% of bearing rating) and pushing tooth-passing frequency to 4.2 Hz — out of the resonance band. Downside: pinion diameter holds at 120 mm but pin diameter has to shrink to ~10 mm to fit, which lowers individual pin shear strength. For granite cutting at this scale, 16 pins is the right answer.
Result
16 pins on a 120 mm pitch circle gives 23. 56 mm pitch and 1,500 N per pin at peak torque — about 19% of the needle-bearing dynamic rating, which translates to an L10 life past 100 million cycles or roughly 8-10 years of three-shift production at this fabricator. At 8 pins you'd save money on the pinion but the 1.4 Hz tooth-passing frequency lands inside the gantry's resonance band and you'd see chatter marks on every cut; at 24 pins you'd get smoother motion but small-diameter pins limit overload margin during a tool crash. If your built drive measures more than 50 µm backlash on a reversing-load test, check three things in order: (1) eccentric preload block backed off — most common, fixed by retorquing to 10 Nm; (2) rack segment butt joint with a step error over 13 µm — shows up as a periodic position error every rack length; (3) needle bearing grease loss on one or two pin assemblies, which you'll hear as a chirp at pin-pass frequency before the pin starts shedding rollers.
Choosing the Roller-bearing Gear Teeth: Pros and Cons
Roller-bearing gear teeth aren't the right answer for every drive. They cost a lot, they need precise installation, and they reward applications where backlash and efficiency really matter. Below is how they stack up against ground involute rack-and-pinion and against a precision cycloidal reducer driving a conventional pinion.
| Property | Roller-bearing gear teeth | Ground involute rack & pinion | Cycloidal reducer + pinion |
|---|---|---|---|
| Backlash | 0 arcmin (preloaded dual pinion) | 3-6 arcmin typical, 1 arcmin with split pinion | <1 arcmin at reducer, plus pinion backlash downstream |
| Mechanical efficiency | 97-99% | 94-97% | 85-92% (cycloidal stage losses) |
| Maximum linear speed | 5 m/s sustained | 3 m/s before tooth-mesh noise | 2 m/s limited by reducer input speed |
| Positioning repeatability over 6 m | ±15-25 µm | ±50-80 µm | ±30-50 µm (depends on pinion quality) |
| Cost per metre of rack (relative) | 8-10× | 1× (baseline) | 2-3× plus reducer cost |
| Service life in indexing duty | 100M+ cycles | 20-40M cycles before pitting | 50-80M cycles, limited by needle wear |
| Tolerance to contamination | Poor — sealed covers required | Good — open running tolerated | Sealed unit, very good |
| Best application fit | Long-travel high-accuracy gantries | General machine axes, conveyor drives | Compact high-ratio rotary axes |
Frequently Asked Questions About Roller-bearing Gear Teeth
You're seeing thermal growth in the rack. A 6 m steel rack expands roughly 70 µm per °C, and a busy machine room can swing 5-8 °C across a shift. The roller pinion itself doesn't drift — the rack underneath it is moving, and your encoder is on the motor not the workpiece.
Fix is either a glass scale on the axis itself for closed-loop position feedback, or thermal compensation in the CNC using a rack-mounted temperature sensor. Don't blame the pinion preload — re-checking it just wastes time on a phantom problem.
Calculate your tooth-passing frequency at maximum traverse speed and check it against the natural frequency of the moving mass. A 12-pin pinion at 2 m/s on a 100 mm pitch circle gives roughly 6 Hz mesh frequency — fine for most 500-1500 kg robot loads which sit in the 15-25 Hz band. A 16-pin gives 8 Hz, also clear.
The real decision driver is overload capacity. If your robot ever crashes the tool against a fixture, the 12-pin sees 33% higher per-pin load on the engaged pin. Pick 16 if your duty cycle includes any chance of impact loading, pick 12 if cost dominates and the load profile is smooth.
The eccentric preload block is creeping. Most designs use a hardened steel eccentric clamped by a single shoulder bolt, and if the clamp torque sits below about 80% of the bolt's yield, micro-vibration walks the eccentric a few arcseconds at a time. After 200 hours you've lost 20-30 µm of preload.
Two fixes: (1) retorque the eccentric clamp bolt to manufacturer spec with a calibrated wrench — most installers undertorque by 30-40%, and (2) add a witness mark across the eccentric and housing so you can see angular drift at a glance during weekly inspection.
No. The needle bearing inside each pin assembly absolutely requires lubrication — it's a rolling-element bearing with cage, and dry running destroys it within hours. What you can do is switch to food-grade NSF H1 grease (Klüber Klübersynth UH1 or equivalent) and seal the pinion with a labyrinth cover.
For genuinely dry-cell applications, look at a different mechanism — typically a precision belt drive with a steel-corded belt, or a direct-drive linear motor. Roller pinions need that grease film between needle and pin or you'll be replacing pinions every few months.
Almost certainly not the pinion. Stiffness loss in a roller pinion drive almost always traces to the mounting interface. The pinion bolts to the reducer output via a flange, and if that flange has even 20 µm of compliance — from a missing pilot diameter, an undersized bolt circle, or a soft adapter plate — you'll measure axis stiffness well below the rated value.
Run this diagnostic: clamp the pinion teeth with a known load and dial-indicate the pinion face against the gantry beam. If the pinion-to-gantry deflection is more than 5 µm under 500 N tangential, the issue is upstream of the pinion. Stiffen the mounting and the number comes back.
An involute tooth slides — debris gets pushed out of the contact zone or embedded harmlessly in the flank. A roller pinion has a rolling needle bearing inside a sealed cap, and silica dust from stone cutting bypasses light seals, gets into the needle cage, and grinds the rollers in days. The pinion teeth themselves look fine, but the bearings are dead.
You need either bellows covers over the full rack length plus positive-pressure dry air inside the pinion housing, or accept a different drive — a hardened ground rack with a precision involute pinion and a worm-driven backlash compensator gives you 50-80 µm repeatability with much better dirt tolerance. For granite shops this is often the more honest answer.
References & Further Reading
- Wikipedia contributors. Rack and pinion. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
