A Screw Traversing Ball Bearing is a linear-motion mechanism that turns a rotating screw into precise carriage travel by running a recirculating ball nut along a hardened helical thread. Quality C5 ground ball screws hold positioning accuracy of 18 µm per 300 mm at travel rates up to 60 m/min. The mechanism eliminates the sliding friction of an Acme lead screw, replacing it with rolling contact through 30 to 60 captive balls. You will find it on every modern CNC mill, including the Haas VF-2 X-axis, where it drives the saddle with sub-thou repeatability.
Screw Traversing Ball Bearing Interactive Calculator
Vary screw lead, turns, target traverse speed, and efficiency to see carriage travel, required screw RPM, and motion in the ball-screw visualizer.
Equation Used
The screw lead converts rotation into linear travel: each revolution advances the constrained ball nut by one lead. For speed, the same relation is rearranged so required rpm equals 1000 times traverse speed in m/min divided by lead in mm/rev.
FIRGELLI Automations - Interactive Mechanism Calculators.
- Ball nut is keyed so it cannot rotate.
- Screw lead is constant over the stroke.
- Backlash, acceleration, screw whip, and load torque are not included.
- Efficiency is shown as a loss indicator and does not change geometric travel.
How the Screw Traversing Ball Bearing Actually Works
The principle is simple — a screw shaft turns, a nut wrapped around the screw cannot turn (it is keyed to a carriage), so the nut must travel axially. What makes the ball-bearing version different from a plain Acme lead screw is the load path. Instead of bronze threads sliding against steel threads, hardened steel balls roll between the screw groove and the nut groove, then recirculate through a return tube or insert. Friction drops from roughly 0.15 (sliding) to about 0.003 (rolling). That single change is why a 2 kW servo can drive a 500 kg machine tool saddle at 30 m/min — and why backdriving becomes a real concern on vertical axes, since efficiencies above 90% mean gravity will spin the screw if the brake fails.
The geometry has to be right or the bearing eats itself. The ball groove is a Gothic arch profile — two arcs blended at the centre — giving a 4-point contact with the ball. The contact angle sits between 40° and 45°. If the screw is reground sloppily and the contact angle drifts below 30°, axial stiffness collapses and you get the chattering, juddering feed that machinists call "screw whip." If the preload is too high, the balls indent the groove (Brinelling) within a few hundred hours and you lose the C5 grade you paid for. Preload typically runs 5-10% of the dynamic load rating — not more, not less.
Failure modes are predictable. Contamination is number one — a single 50 µm grinding chip will pit a ball within minutes of operation, then the pitted ball pits the groove, and the whole nut starts whining. Lubrication starvation is number two; the recirculating ball nut needs grease (NLGI 2) every 100 hours of duty or oil mist if the duty cycle is heavy. Number three is misalignment between the screw axis and the linear guideway — anything beyond 0.05 mm/300 mm parallelism will load the ball nut radially, and a ball nut hates radial load. It will fatigue the return tube fingers and start dropping balls inside six months.
Key Components
- Ground Screw Shaft: Hardened (58-62 HRC) and ground bearing-quality steel shaft with a Gothic-arch helical groove. Lead accuracy is graded — C7 holds 50 µm/300 mm, C5 holds 18 µm/300 mm, C3 holds 8 µm/300 mm. Shaft diameter on a typical mill X-axis runs 25-40 mm with a 5-10 mm lead.
- Recirculating Ball Nut: Houses 30 to 60 grade-10 chrome steel balls (typically 3.175 mm or 4.762 mm diameter) that roll in the matching groove. Includes a return tube, end-cap, or internal insert that lifts balls out of the load zone and feeds them back to the start. Two nuts can be paired with a spacer to apply axial preload.
- Return Path: External return tube or internal channel that carries balls back to the start of the load zone. This is the most fragile element — a dropped ball or a damaged tube finger destroys the nut. End-cap returns handle higher speed (DN values to 160,000) than tube returns.
- End Bearings: Angular contact bearings (60° contact angle, paired DB or DT) at the fixed end take the screw's axial load. The opposite end uses a deep-groove or floating support to allow thermal expansion — a 1 m screw running at 3000 RPM will grow 0.3 mm in 30 minutes.
- Lubrication Port: Single grease zerk on the ball nut housing. Standard fill is NLGI 2 lithium-complex grease, 1.5 cm³ every 100 operating hours. Oil-mist systems use ISO VG 68 at 0.03 cm³/cycle for high-DN applications.
Where the Screw Traversing Ball Bearing Is Used
You see this mechanism anywhere a screw needs to move a carriage with repeatable accuracy and decent speed. The reason it dominates over plain Acme lead screws in factory equipment comes down to three numbers: 90%+ efficiency, sub-10 µm positioning, and 20,000+ hour service life. Where you do not see it is in self-locking applications — vertical Z-axes without a brake, hand-cranked vises, anything where backdriving is dangerous — because rolling-element efficiency works both directions.
- CNC Machine Tools: Haas VF-2 vertical machining centre uses 32 mm × 10 mm lead C5 ball screws on all three axes, driving the saddle at 30 m/min rapid traverse.
- Injection Moulding: Engel e-motion all-electric press uses a large-diameter (80 mm) ball screw on the clamp axis, generating 1500 kN clamping force with sub-millisecond positioning.
- Semiconductor Lithography: ASML wafer-stage coarse positioning stages use precision-ground C0 ball screws as the long-stroke axis before handing off to a voice-coil fine stage.
- Precision Grinders: Studer S33 cylindrical grinder uses preloaded double-nut ball screws on the X-axis for 1 µm infeed resolution on the wheelhead.
- Press Brakes: Trumpf TruBend 5000 servo-electric press brake uses dual ball screws to drive the ram, replacing the hydraulic cylinders of older machines and cutting cycle time by 30%.
- Automated Assembly: ATS SuperTrak conveyor end-of-line stations use 16 mm × 5 mm ball screws on pick-and-place Z-actuators, cycling 1.5 million times per year.
The Formula Behind the Screw Traversing Ball Bearing
The fundamental equation links shaft RPM to carriage linear speed through the screw lead. What the practitioner cares about is where the operating sweet spot sits. At the low end of the range — say 200 RPM on a 10 mm lead screw — you get 2 m/min, fine for finishing cuts but slow for rapid moves. At the nominal mid-range, 1500-2000 RPM, you hit the productive sweet spot where servo torque is still healthy and DN values stay within bearing rating. Push beyond about 3000 RPM on a 32 mm screw and you cross the critical speed limit — the screw whips like a skipping rope and positioning accuracy collapses.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| v | Linear travel speed of the carriage | m/s | in/s |
| N | Rotational speed of the screw shaft | RPM | RPM |
| P | Lead of the screw (linear travel per revolution) | m/rev | in/rev |
| DN | Speed factor (ball circle diameter × RPM) — must stay below screw rating | mm·RPM | in·RPM |
Worked Example: Screw Traversing Ball Bearing in an aerospace 5-axis trunnion mill
An aerospace job shop in Wichita is commissioning a DMG Mori NMV5000 5-axis trunnion mill and needs to verify the X-axis ball screw selection. The screw is a 40 mm diameter, 16 mm lead, C5 grade ground ball screw with a single-nut preloaded ball nut. The servo motor is rated 3000 RPM continuous, and the operator wants to know rapid traverse speed at nominal operation, what happens at the low-feed end (200 RPM finishing cuts) and at the upper limit of the servo (3000 RPM rapid).
Given
- Dscrew = 40 mm
- P = 16 mm/rev
- Nnom = 1500 RPM
- Nlow = 200 RPM
- Nhigh = 3000 RPM
- DNlimit = 120000 mm·RPM
Solution
Step 1 — at nominal 1500 RPM, convert lead to metres and compute carriage speed:
That is the productive sweet spot. 24 m/min rapid is fast enough that air-cut moves between features feel snappy, but the servo is still well inside its torque envelope and the screw is nowhere near critical speed. The DN value here is 40 × 1500 = 60,000 mm·RPM, half the rated DN limit.
Step 2 — at the low end of the operating range, 200 RPM finishing feed:
This is slow enough that you can watch the saddle creep across the table. It is the right regime for an aluminium 7075 finishing pass at 0.05 mm chip load — the screw is operating at low DN, the balls dwell longer in each contact, and any preload-induced friction warmth has time to dissipate. Positioning accuracy is at its best here because thermal growth is negligible.
Step 3 — at the upper limit of the servo, 3000 RPM:
On paper, 48 m/min rapid. In practice, DN = 40 × 3000 = 120,000 mm·RPM — exactly at the rated limit for a tube-return ball nut. End-cap return designs handle this fine; tube-return nuts will start shedding balls inside a year. Also check critical speed for the screw length — for a 1 m screw between supports, fixed-simple mounting, critical speed is around 3200 RPM, so you are running at 94% of whip threshold. The screw will sing audibly.
Result
The nominal carriage speed is 0. 40 m/s (24 m/min) at 1500 RPM. That is the productive sweet spot — fast rapids without thermal or whip concerns. The range spans 3.2 m/min at the finishing end up to a theoretical 48 m/min at the servo limit, but real-world useful upper speed sits closer to 36 m/min (2250 RPM) for a tube-return nut on a 1 m screw — the last 20% of the speed envelope buys you noise, heat, and reduced screw life. If you measure actual rapid speed below the predicted value, check three things first: (1) servo current limit clipping near top speed, which shows up as the axis ramping fine but never reaching commanded velocity, (2) ball nut preload above 10% of dynamic rating, which adds drag torque the servo eats into top-speed headroom, and (3) end-bearing preload set wrong on the angular contact pair — a too-loose DB pair lets the screw shuttle axially and the encoder reads the missing motion as lower velocity.
Screw Traversing Ball Bearing vs Alternatives
Screw-driven linear motion comes in three main flavours, and the choice between them is rarely about the screw alone — it is about cost per micron, duty cycle, and whether you need the system to be self-locking. Here is how the Screw Traversing Ball Bearing stacks up against the two most common alternatives.
| Property | Screw Traversing Ball Bearing | Acme Lead Screw | Linear Motor |
|---|---|---|---|
| Efficiency | 90-95% | 30-50% | 85-95% |
| Positioning accuracy (per 300 mm) | 8-50 µm (C3 to C7) | 75-150 µm | 1-10 µm |
| Maximum linear speed | 60 m/min (end-cap return) | 15 m/min | 300 m/min |
| Maximum stroke length | 6 m (above this, whip dominates) | 3 m | Effectively unlimited |
| Self-locking (back-drive) | No — needs brake on vertical axes | Yes — friction holds load | No — needs brake or magnetic detent |
| Cost (1 m axis, ballpark) | $800-$2500 | $150-$400 | $4000-$12000 |
| Service life (rated load) | 20,000-50,000 hours | 2,000-8,000 hours | 30,000+ hours |
| Sensitivity to contamination | High — sealed nut required | Moderate | Low |
Frequently Asked Questions About Screw Traversing Ball Bearing
Thermal growth. A 1 m steel screw expands 11 µm per °C — at 30 minutes of duty cycle the screw can easily heat 15-20°C above ambient from internal friction, which adds 165-220 µm of length error end-to-end. The C5 grade specifies pitch accuracy at 20°C reference temperature, not at operating temperature.
Three fixes: pre-tension the screw on installation (typical 0.02-0.04% strain) so thermal growth pushes against tension rather than buckling, switch to a hollow-shaft oil-cooled screw on long axes, or use linear scale feedback rather than rotary encoder feedback so the control loop sees the actual position not the commanded one.
It comes down to DN limit and contamination. Tube returns lift balls fully clear of the nut body and run external — cheaper, easier to inspect, but the tube fingers cap the speed at around DN 70,000-90,000 and the external loop is exposed to chips. End-cap and internal-deflector returns keep the balls inside the nut envelope and handle DN 120,000-160,000.
Rule of thumb: if your servo × screw diameter exceeds 70,000 mm·RPM, or the application is wet (coolant, slurry, swarf), specify end-cap. For dry pick-and-place work under 30 m/min, tube-return is fine and saves 30-40% on ball nut cost.
The backlash you are reading is almost certainly not in the ball nut. Preloaded double-nuts hold zero backlash by construction — the spacer between the two nuts forces the balls into 4-point contact in opposite directions. If you see 30 µm of axial play, look elsewhere in the load path.
Check the angular-contact end bearings first — a DB pair with insufficient preload (or a worn-in pair after a few thousand hours) will show exactly this symptom. Second, check the screw-to-motor coupling — a worn jaw coupling or a slipping shrink-disc adds reversal play that looks like nut backlash. Third, check the saddle-to-nut bracket bolts; one loose M8 fastener will give you 20-50 µm of springy reversal that mimics backlash perfectly.
No. The self-locking threshold for a screw is when the lead angle drops below the friction angle — for an Acme screw at coefficient 0.15, that is around a 5° lead angle. A ball screw runs at coefficient 0.003, so the friction angle is about 0.17°. You would need a lead so fine the screw could not be manufactured.
Every vertical ball-screw axis needs either a fail-safe spring-set electromagnetic brake on the motor, a counterweight, or a pneumatic counterbalance cylinder. Skip this and the saddle drops to the bottom of the stroke the instant the servo de-energises — which is exactly what happens when you hit the e-stop.
One tick per revolution almost always means a single damaged ball or a single damaged spot in the screw groove that the balls roll past once each turn. The tick will get louder, then turn into a grinding feel through the handwheel within a few hundred hours.
Most common cause is a contamination event — a chip got past the wiper seal, embedded in one ball, and that ball has now spalled the groove on each pass. Less commonly it is a Brinell mark from over-preload at installation. Either way the nut is finished. Replace it before the damaged ball jams the return path and locks the axis mid-cut.
Almost never. C7 holds 50 µm per 300 mm, C3 holds 8 µm per 300 mm — the C3 screw costs 4-6× more. On a wood router or aluminium hobby mill, the dominant error sources are gantry flex, spindle runout, and tool deflection, all of which are 50-200 µm. Spending $1200 on a C3 screw to fix a 50 µm screw error when you have 150 µm of gantry flex is wasted money.
The crossover point is around grinding and aerospace finish work where everything else has been tightened down to single-micron territory. For everything else, C7 (or rolled C7 for budget builds) plus linear scales gives better real-world accuracy than C3 alone.
References & Further Reading
- Wikipedia contributors. Ball screw. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
