CNC screws basic knowledge: what engineers actually need
Three different things get called a CNC screw: the fastener that clamps a part, the lead screw that moves an axis, and the ball screw that positions it. This page separates them, explains the mechanics behind each, and shows where the accuracy limits really come from. Written for design engineers and buyers who have to pick a thread, a nut, or a drive and defend the choice.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
Key takeaways
What an engineer means by a CNC screw
Ask five people what a CNC screw is and you get five answers. In a machine shop, the phrase usually means the threaded fastener that clamps a fixture plate, a vise jaw, or a workholding block. In a machine tool catalog, it means the drive screw that converts motor rotation into axis travel. Those are different components with different failure modes, and mixing them up leads to wrong specifications.
The fastener side is straightforward. A socket head cap screw, a flat head screw, or a shoulder screw holds parts together. What matters is thread form, class of fit, material, and preload. A shoulder screw also locates, so its shank diameter matters more than its thread.
The drive side is where the engineering gets interesting. A lead screw is a threaded shaft running inside a threaded nut. A ball screw replaces sliding contact with recirculating steel balls. Same job, very different friction, efficiency, and stiffness numbers.
This page treats the drive screw as the main subject because that is where design decisions have consequences. The fastener section stays short and practical, since most engineers already know how to call out a thread.
One more term worth clearing up: people say screw when the catalog says screw, and say screw when the drawing says shaft. Both appear in CNC work. Use the drawing, not the shop-floor shorthand.
How thread geometry turns rotation into precise travel
Every drive screw obeys the same relationship. One full turn moves the nut by one lead. Lead is not the same as pitch: a two-start screw with a 2 mm pitch has a 4 mm lead, so one turn moves 4 mm. That distinction matters when you size a motor or calculate resolution.
The angle of the thread flank decides how much of the motor torque becomes axial force. A shallow flank angle gives high thrust but high friction. A steeper angle is more efficient but produces less force per unit of torque. Ball screws sidestep this tradeoff by rolling instead of sliding, which is why they reach 90% efficiency where an acme lead screw sits near 30 to 50%.
Contact geometry drives stiffness. In a lead screw, the nut thread and shaft thread touch over a broad area, and that area grows under load. In a ball screw, each ball touches the raceway at two small elliptical points. Small contact areas mean high local stress but also very high stiffness, provided the raceway is hard enough.
Resolution is not accuracy. A 5 mm lead screw on a 1.8 degree stepper gives 0.025 mm per full step before microstepping, but the axis may still be 0.1 mm off after a meter of travel. Resolution describes the smallest commanded move; accuracy describes where the nut actually ends up.
Thermal growth belongs in the same conversation. A steel screw heated 5 °C over a 1,000 mm span grows roughly 0.06 mm. On a warm machine running all day, that shifts the zero point slowly and quietly.
Backlash, preload, and why the nut design decides repeatability
Backlash is the lost motion when you reverse direction. In a lead screw it comes from the clearance between the two thread flanks. In a ball screw it comes from the gap between the balls and the raceway. Either way, the axis moves a little before the load moves, and any bi-directional positioning job suffers.
Preload removes that gap by forcing the nut and screw into permanent contact. Ball screw makers do it in three common ways: oversized balls, a double nut with a spacer, or a shifted lead on one nut half. Oversized balls are cheap and compact, but the preload fades as the balls wear. A double nut holds preload longer and lets you adjust it with shims.
Preload is not free. A preloaded nut has higher drag torque, generates more heat, and wears faster than a zero-preload nut. That is the tradeoff. Light preload on a fast, lightly loaded axis; heavy preload on a slow, stiff, high-force axis.
For lead screws, an anti-backlash nut with a spring or a split nut does the same job at lower cost. It works well at low speed and light load. It does not survive high reversing loads for long, because the spring or the plastic insert takes the wear.
Measure backlash the same way every time. Zero an indicator against the nut, jog one direction, then reverse by a known amount and read the lag. Do it at three positions along the travel, because wear is rarely uniform.
Accuracy classes, pitch error, and what the number really covers
Screw accuracy classes describe allowable travel deviation over a defined length, not just a single-point tolerance. A rolled screw typically lands in a mid class, a ground screw in a tighter one. The class covers cumulative pitch error, plus a allowance for variation within one revolution.
Two errors hide inside that number. Cumulative pitch error accumulates along the screw and shows up as a position offset at the far end. Periodic error repeats once per turn and shows up as a ripple in surface finish or as a small oscillation in a positioning loop. They have different causes and different fixes.
Rolled screws are formed by pressing a die against a bar. It is fast and inexpensive, and the grain flow follows the thread, which helps fatigue life. Ground screws are cut and then ground after heat treatment, which removes distortion and produces tighter lead accuracy. Cost reflects the extra operations, not marketing.
For most positioning work, a rolled screw in a mid class plus a good feedback scale beats a top-class screw running open loop. The scale sees the real position, including thermal drift and pitch error. Closed loop does not fix a mechanically loose nut, though. Feedback cannot remove backlash.
Ask for the lead deviation chart when the axis is long. A single class number hides how the error is distributed, and where it sits matters if you only travel over part of the screw.
Materials, hardening, and the loads that end a screw's life
Most ball screws and lead screws start as medium-carbon or alloy steel, then get induction hardened or through hardened for the raceway. Surface hardness in the high 50s to low 60s on the Rockwell C scale is normal for a rolling element raceway. Below that, balls dent the surface and the preload goes away.
Stainless grades appear where corrosion matters, but they usually carry lower load ratings for the same size, because the alloy is softer and harder to harden uniformly. On a food or medical machine, that tradeoff is often worth taking. On a heavy press axis, it is usually not.
Aluminum and brass screws exist for light-duty adjustment and for non-magnetic or non-sparking environments. They wear quickly under repeated motion. Use them for setup adjustments, not for production cycling.
Lubrication is part of the load path. Grease stays put on slow vertical axes and carries the film through shock loads. Oil runs cooler on fast horizontal axes and flushes debris. Mixing the two, or letting a greased nut run dry, shortens life more than a one-class accuracy downgrade ever would.
Protect the screw from chips and coolant. Bellows, wipers, and a felt seal at the nut face add a little drag but keep abrasive debris out of the raceway. On any machine cutting metal nearby, that is the cheapest life extension you can buy.
When a screw fails, look at the failure pattern first. Flaking on one side of the raceway points at misalignment. Uniform wear points at lubrication or duty cycle. Cracking at the thread root points at a stress riser or an overload.
Which screw for which job
Load, speed, and repeatability decide the family before tolerance does.
| Type | Best for | Typical limit | Watch out for |
|---|---|---|---|
| Fastener screw | Clamping fixtures and joints | Preload set by torque spec | Thread stripping in soft aluminum |
| Acme lead screw | Slow, high-force, low-cost axes | Efficiency near 30–50% | Wear and backlash over time |
| Rolled ball screw | General positioning, cost-sensitive | Mid accuracy class | Lead error accumulates over length |
| Ground ball screw | Tight positioning, long life | Tight accuracy class | Higher cost per axis |
| Anti-backlash nut | Light-load reversing axes | Low speed only | Insert wear resets backlash |
| Double nut preload | Stiff bi-directional axes | Higher drag torque | Extra length and heat |
| Stainless screw | Corrosive or washdown areas | Lower load rating | Softer raceway wears faster |
The call we would make
For a slow, high-force axis on a budget, use an acme lead screw with an anti-backlash nut. For bi-directional positioning that has to repeat, use a ground or rolled ball screw with a double nut and a feedback scale, and stop trying to buy accuracy with a higher class alone.
Questions that come up after the basics
Is a ball screw always more accurate than a lead screw?
No. Accuracy comes from the lead accuracy class, the nut preload, and the mounting, not the rolling element alone. A tightly specified lead screw on a short, slow axis can hold position better than a loose ball screw on a long one.
What a ball screw reliably gives you is higher efficiency, lower heat, and better stiffness at speed. Those are real advantages, but they are not the same as accuracy.
How much preload should a positioning axis use?
Enough to remove backlash at the worst operating condition, and no more. Many machine axes run well with light preload, which keeps drag torque and heat low.
Heavy preload belongs on axes that see high reversing loads or that must resist vibration. It costs torque, heat, and life. If the axis never reverses under load, the extra preload buys little.
Does screw length change the accuracy class I need?
Indirectly, yes. The class number is defined over a length, so a long screw with the same class carries more total error than a short one. If you only travel over part of the screw, ask where the error sits.
For long spans, thermal growth also starts to dominate. Measure the machine warm before you blame the screw.
What causes a screw to lose preload in service?
Wear on the balls and raceway is the usual cause. Oversized-ball preload fades as the parts wear in. Debris in the raceway accelerates it sharply.
Poor lubrication and misalignment both accelerate wear. A screw that is overloaded or run dry will lose preload long before its rated life.
Should I grease or oil a ball screw?
Grease for slow vertical axes and shock loads. Oil for fast horizontal axes that need cooling and debris flushing. Follow one regime and keep it consistent.
Mixing grease and oil usually produces a paste that holds chips. That is worse than either choice on its own.
When is a lead screw the wrong choice?
When the axis must reverse often under load, or when speed matters. Sliding friction turns motor torque into heat, and efficiency drops fast as speed rises.
It is also wrong when backlash must stay near zero for years with no adjustment. A ball screw with proper preload holds that condition longer.
Send us the drawing and the duty cycle
We machine screw-based parts, housings, nuts, and fixtures to your print, with DFM feedback on thread callouts and preload fits before cutting starts.
12-hour quoteTolerances to ±0.005 mm100% inspection