GreatLight CNC Machining Factory logo
CNC Machining
Rapid Prototyping
Materials
Industries
News
About GL

Get Instant Quote

Machining mechanics

Ball Screws at CNC Turn: How the Motion Loop Really Works

A plain explanation of what happens inside the axis drive of a CNC turning center, and what the screw geometry does to part quality. Written for machine builders, automation engineers, and buyers who need to judge whether a round part belongs on a lathe at all.

±0.005 mm toleranceRa 0.2–0.8 μm finish16 mill-turn centers3 plants
Ball screws at CNC turn: recirculating ball nut on a lathe axis drive
The mechanism

Why ball screws at CNC turn decide what the tool actually does

A ball screw turns rotary motion from the servo into linear motion of the carriage or turret. Steel balls recirculate between the screw groove and the nut groove, so contact is rolling rather than sliding. Friction drops to roughly a third of a sliding lead screw, and the screw turns with very little stick-slip. That matters because a lathe axis reverses direction thousands of times per shift. Any hesitance at reversal shows up on the part as a witness mark or a taper.

The groove profile is a Gothic arch or a circular arc, ground to a lead accuracy class. Lead error is the difference between commanded travel and real travel over one revolution. On a C3 screw that error stays inside a few micrometres across 300 mm. On a rolled C7 screw it can be three to five times larger. For a turning center holding ±0.005 mm on diameter, that difference is not academic. It sets how much the control has to compensate, and how much the compensation drifts as the screw warms.

Axial stiffness is the second half of the story. It comes from three places in series: the screw shaft, the nut, and the support bearings. The weakest link sets the limit. A 40 mm screw on a long span with a fixed-supported bearing pair behaves very differently from the same screw with a fixed-fixed pair. When we quote a turned part that needs tight axial location, we look at the machine's screw diameter and bearing arrangement, not just the control resolution.

  • 1
    Rolling, not slidingBalls cut friction and break the stick-slip that ruins reversal points.
  • 2
    Lead error is stackableGround C3 holds micrometre-level lead; rolled C7 does not.
  • 3
    Stiffness is a chainShaft, nut and bearings in series — the softest part wins.
Preload

Preload and backlash: the trade engineers keep re-making

A ball nut with no preload has axial play, usually 0.02 to 0.05 mm. Under a climb-milling or heavy turning cut, the nut shifts across that gap and the tool follows. Preload removes the gap by loading two ball circuits against each other. Double-nut designs use a shim or a ground spacer between two nuts. Single-nut designs use oversized balls or a shifted lead. Both create a light interference that keeps every ball in contact.

Preload is not free. It adds drag torque, generates heat, and shortens screw life. A common rule is to preload to about 5 to 8 percent of the screw's dynamic load rating. Go higher and you buy stiffness with heat. On a lathe running a long roughing cycle, that heat travels down the screw and changes its length. A 1 m steel screw grows about 12 μm per °C. Ten degrees of rise moves the tool 0.12 mm. The control can map that if the machine has scale feedback. Without scales, it cannot.

This is why two seemingly identical turning centers hold different tolerances on the same job. One has preloaded C3 screws with linear scales on X and Z. The other has rolled screws and rotary encoders only. The second machine is fine for a ±0.05 mm bracket. It is the wrong machine for a ±0.005 mm bearing seat, and no amount of careful programming fixes that.

  • 1
    Typical preload5–8 percent of dynamic load rating keeps drag and heat sane.
  • 2
    Thermal growthSteel grows roughly 12 μm per metre per °C of rise.
  • 3
    Scales change the gameLinear feedback compensates growth; rotary encoders cannot.
Boundaries

Critical speed, buckling and the limits of a turning axis

Every screw has a speed at which it starts to whip. Critical speed depends on diameter, unsupported length, and end fixity. A screw supported at both ends can run about twice as fast as one supported at one end. Long, thin screws hit this limit well below the servo's own maximum. When the axis runs near critical speed, the whipping shows up as surface chatter on the turned diameter and as noise from the nut.

Buckling is the compression-side twin. A screw pushing a heavy turret under high feed force acts as a column. If the compressive load passes the Euler limit, the screw bows. The symptom is a sudden loss of position on the return stroke, not a gradual drift. Machines with long Z travel and heavy saddles are the ones at risk.

There is also a practical floor on what the axis can resolve. A ball screw with 10 mm lead and a 10,000 count encoder gives about 1 μm per count on paper. Real reversal accuracy is far coarser, often 5 to 20 μm, because of nut compliance and thermal state. Anyone who quotes turning tolerances tighter than the machine's real reversal accuracy is quoting the encoder, not the part.

  • 1
    Whip limitRises with diameter and with fixed-fixed end support.
  • 2
    Buckling limitBites long Z axes under heavy feed force.
  • 3
    Resolution is not accuracyEncoder count says little about real reversal error.
The blank

When the screw itself is the part being turned

Not every ball screw question is about the machine axis. Some customers send us the screw as a component: journal diameters, bearing seats, thread reliefs, snap ring grooves. These parts are long, slender, and unforgiving. A 32 mm screw ground to a 20 mm journal over 900 mm of unsupported length deflects under its own cutting force. The tailstock is not optional. Neither is a steady rest.

The journal runout relative to the thread axis is the feature that matters most. If the bearing seat runs out 0.01 mm, the screw will not sit true in the nut, and the whole assembly loses its preload consistency. We hold ±0.005 mm on those journals and check runout between centers. The thread itself is usually ground, not turned, once the lead accuracy class goes beyond C7.

Materials matter less than people expect here. Most production screws are 4150, 4140, or a case-hardening grade. Stainless 17-4PH appears in cleanroom and medical handling equipment. Whatever the grade, the blank needs stress relief before finish grinding, or the journals move a few micrometres after the last cut.

  • 1
    Slender workTailstock plus steady rest, or the journal turns oval.
  • 2
    Runout rulesJournal runout to thread axis drives preload consistency.
  • 3
    Stress reliefDo it before finish grinding, not after.
Selection data

Matching the round part to the right machine and axis

Use this as a first filter before quoting. The middle column is the axis-drive side; the right column is the workpiece side.

Part or axis traitAxis drive choiceTurning setup choice
Tolerance ±0.005 mm on diameterGround C3 screw, preloaded nutMill-turn center, temperature-stable shop
Tolerance ±0.05 mm on diameterRolled C7 screw is usually enoughStandard 2-axis lathe
Length over 4× diameterNot an axis issueSteady rest plus tailstock, light radial cuts
Heavy interrupted cutFixed-fixed bearing pair, higher preloadRigid turret, reduced feed per rev
Long Z travel, heavy saddleCheck buckling load before orderingSplit the cycle between two setups
Cleanroom or medical handlingStandard screw, sealed nut17-4PH or 316L blank, passivated finish
High rapids on a long axisVerify critical speed with marginNot affected by turning setup
Journal runout under 0.01 mmNot an axis issueGrind between centers, check with an indicator

The trade you are actually making

If the part needs ±0.005 mm and a fine finish, pay for a preloaded ground-screw mill-turn center and accept the higher hourly rate. If it needs ±0.05 mm on a simple diameter, a rolled-screw 2-axis lathe is the cheaper and equally correct choice.

FAQs

Questions engineers ask before releasing the drawing

Can a turning center hold ±0.005 mm on a long shaft?

On a short, well-supported part, yes. On a slender shaft the limit is usually deflection, not the axis drive.

We hold ±0.005 mm on bearing journals with a steady rest and tailstock, but the diameter and length ratio decides whether it is realistic. Send the drawing and we will say so before quoting.

Does screw preload affect the surface finish on the part?

Indirectly. Preload raises axial stiffness, which reduces tool push-off during a heavy cut.

A machine with a loose nut can show taper or a step at the reversal point even when the tool and the program are correct.

How much does thermal growth move the tool over a shift?

A steel screw grows about 12 μm per metre for each degree Celsius of rise. A 1 m screw that warms 8 °C moves roughly 0.1 mm.

Machines with linear scale feedback on X and Z correct for this. Machines with rotary encoders only cannot, and the operator compensates by hand or by wear offset.

Is a rolled screw ever good enough for a turning machine?

Yes, for a lot of work. Rolled C7 screws are common on 2-axis lathes and on automation slides where the tolerance is ±0.05 mm or looser.

Above that, the lead error and the preload consistency stop being predictable enough to program around.

What material do you use for turned screw blanks?

4140, 4150 and case-hardening grades cover most production screws. We also turn 17-4PH, 304 and 316L for handling and medical equipment.

Whatever the grade, stress relief before finish grinding keeps the journals from moving after the last pass.

Can you turn a ball nut housing as well as the screw?

Yes. Nut housings are usually a bore-and-face job with a tight concentricity callout between the bore and the mounting flange.

We machine them on the same mill-turn centers we use for screw journals, so the two parts can be checked against each other before assembly.

Send the drawing, get a real answer on feasibility

Quotation and free DFM analysis within 12 hours, with no minimum order quantity from one prototype upward.

12-hour quote100% inspectionNo minimum orderNDA on request

Follow

More machining notes from the shop floor

We publish setup notes, tooling trials and inspection data from the factory floor.

FacebookTikTokYouTubeLinkedInInstagramThreadsPinterest

Trusted by engineers and manufacturers worldwide

Tesla Ford Motor Company BYD Auto Denso Magna International Boeing Airbus Medtronic KUKA FANUC