Ball Screw

A ball screw is a precision linear-motion element that converts rotary motion into linear motion through recirculating steel balls that roll between matched helical raceways on a screw shaft and a ball nut. By substituting rolling contact for the sliding contact of an ordinary lead screw, it reaches mechanical efficiency of 90 percent or more, drastically reducing the drive torque, heat, and wear of a positioning axis.

Ball screws are the dominant feed-drive element in CNC machine tools, semiconductor stages, electric injection presses, aircraft actuators, and industrial robots. Selecting one correctly means matching accuracy grade, preload, load rating, and speed limit to the duty cycle, with every parameter traceable to the manufacturer datasheet and the ISO 3408 and DIN 69051 standards that govern the category.

Two ball screws showing the threaded screw shafts and ball nuts, with detail insets of the recirculating ball assembly and recirculation channel

Photo: Graibeard, CC BY-SA 3.0, via Wikimedia Commons

This guide is written for procurement engineers and design engineers specifying linear-motion drives. Six chapters move from working principle and history through recirculation types, manufacturing grades, accuracy and load standards, spec-sheet decoding, and a selection decision sequence, with two comparison tables and seven selection FAQs. All parameters reference the ISO 3408 series (Parts 1 to 5), DIN 69051, and JIS B 1192 public standards and published manufacturer handbooks.

Chapter 1 / 06

What is a Ball Screw

A ball screw is a mechanical linear actuator that translates rotary motion into linear motion with very low friction. Its three working parts are the screw shaft, the ball nut, and a closed circuit of hardened steel balls. The shaft and nut both carry a precision helical groove, and the balls roll in the gap between the two grooves, forming a load-carrying ball train. When the shaft turns, the balls roll along the raceway, drive the nut axially, and are continuously picked up at one end of the loaded zone and returned to the other end through a recirculation path, so the same balls cycle endlessly. This rolling-contact principle is the entire reason a ball screw exists: it replaces the sliding contact of an Acme lead screw with rolling contact, the same substitution that distinguishes a ball bearing from a plain bush.

The consequence is efficiency. A ball screw converts roughly 90 percent or more of input torque into linear thrust, while a comparable Acme lead screw averages near 40 to 50 percent and can drop below 20 percent at fine leads. Higher efficiency means a smaller, cooler motor delivers the same thrust, and it also means the screw back-drives readily: applied axial force will spin the shaft unless a brake or counterbalance holds it, which is why vertical ball screw axes almost always pair the screw with a braked servo motor to hold position. The lead screw trades that efficiency away in exchange for low cost, quiet operation, and the ability to be self-locking, so the two are complementary rather than competing.

Historically the rolling screw concept dates to patents in the late nineteenth century, but practical, mass-produced ball screws emerged in the 1940s, accelerated by automotive steering gear and by the demand for accurate feed drives in numerically controlled machine tools through the 1950s and 1960s. The arrival of grinding processes for hardened thread forms, and later the standardisation of accuracy classes under DIN 69051 and ISO 3408, turned the ball screw into the default precision feed element it is today. Modern variants extend from sub-3 mm miniature screws for medical and optical stages to shafts over 200 mm in diameter for heavy presses and steelworks.

Functionally, a ball screw is judged on four families of specification: accuracy (how faithfully commanded travel becomes actual travel), preload and rigidity (how much backlash and elastic deflection remain under load), load rating and life (how long the rolling surfaces survive fatigue), and speed limits (critical speed and DN value). No single screw optimises all four at once, so engineering selection is the disciplined act of mapping a duty cycle, a positioning tolerance, a thrust profile, and a speed envelope onto a specific diameter, lead, accuracy grade, nut style, and preload class.

It is worth being precise about terminology. The screw shaft is the rotating or sometimes stationary threaded rod; the ball nut is the carriage that moves; the lead is the linear distance the nut travels per shaft revolution, not to be confused with pitch, which equals lead only on a single-start thread. The ball-circle or pitch-circle diameter is the diameter on which the ball centres lie and is the figure used in DN-value speed calculations. Getting these definitions straight is the foundation for reading any ball screw datasheet without error.

Chapter 2 / 06

Recirculation Types and Nut Styles

The defining engineering feature of a ball nut is how it returns balls from the end of the loaded zone back to the start, because that recirculation path sets the nut's size, noise, maximum lead, and speed limit. Manufacturers such as THK and NSK group nuts into three families by circulation method: external return-tube, internal deflector, and end-cap. The table below compares them on the parameters that drive selection.

Recirculation typePathBest forDN ceiling (Dm-N)Trade-off
Return tube (external)External tube clamped to nut bodyLarge leads, large ball diameter, low cost~70,000Bulky outer profile, lower speed
Deflector (internal)Finger/button turns balls over one threadFine leads, compact precision stages100,000 to 160,000Limited to one ball turn per circuit
End cap (internal)Caps at both ends with through-boreHigh lead, high-speed transfer axesup to ~180,000More complex, costlier nut

External return-tube nuts route the balls out of the raceway, through a metal tube that protrudes from the side or top of the nut body, and back into the start of the loaded zone. The design is mechanically simple, tolerant of large ball diameters, and the cheapest to make, which keeps it common on heavy-lead transport screws. Its drawbacks are a larger outer envelope (the tube must be guarded against impact) and a comparatively low DN ceiling, so it is rarely the choice for high-speed axes.

Internal deflector nuts, sometimes called finger or button deflectors, use a small insert that lifts each ball, deflects it over the crest of a single thread, and drops it back into the adjacent groove. Because recirculation happens entirely inside the nut body across just one thread turn, deflector nuts are the most compact and the quietest, and they excel at fine leads. They are the workhorse of precision positioning screws, including the THK deflector type. The limitation is that each circuit spans only one ball turn, so load capacity per circuit is modest and several circuits are stacked to build up rating.

End-cap nuts collect the balls at one end of the loaded zone into a cap, carry them axially through a bore drilled the length of the nut, and reinject them through the cap at the opposite end. This straight, smooth path handles the high ball velocities of large leads and high speeds without the balls jamming, so end-cap nuts dominate high-speed transfer, palletising, and electric-press axes, and they reach the highest DN values. They cost more to manufacture and are physically longer than a deflector nut of equal capacity.

Beyond the circulation method, nuts are also classified by external form: flanged cylindrical nuts (the most common, bolting to a carriage through a machined flange) and plain cylindrical nuts clamped in a housing. Single-nut versus double-nut construction is a separate axis covered under preload in Chapter 5. The practical rule for Chapter 2 is to pick the circulation type first from lead and speed, then choose flange geometry to match the mounting interface.

Chapter 3 / 06

Manufacturing: Ground vs Rolled

How the thread form is produced sets the achievable accuracy, surface finish, noise, and price. Two processes dominate: thread grinding and thread rolling. Grinding cuts the helical groove into a hardened blank with a profiled abrasive wheel, removing heat-treatment distortion and leaving a fine, true surface; rolling cold-forms the thread by pressing a soft blank between hardened dies, displacing material rather than cutting it. The two routes span overlapping but distinct accuracy ranges, and they are the single biggest lever on unit cost.

AttributeGround ball screwRolled (precision) ball screw
Accuracy gradesC0 to C5C5 to C10
Typical travel deviation6 to 23 um / 300 mm23 to 210 um / 300 mm
Surface finishFine ground, low frictionCoarser, work-hardened skin
Relative costHighLow
Lead timeLongerShorter
Typical useMachine tools, semiconductor, metrologyGeneral automation, transport, packaging

Ground ball screws are hardened, then finish-ground, which corrects the dimensional drift that heat treatment introduces and produces the truest raceway form. This is why ground screws reach the tightest classes, C0 to C5, and deliver the best repeatability and the smoothest, quietest running. The cost is real: grinding is slow, demands expensive machines and skilled operators, and adds lead time. Ground screws are specified where positioning error must stay below roughly 10 micrometres, in CNC machining centres, grinding machines, wafer steppers, and coordinate measuring stages.

Rolled ball screws are cold-formed, a fast, material-efficient process that suits volume production and large diameters. Precision rolling has improved to the point that a rolled screw can meet C5 (about 23 micrometres travel deviation per 300 mm), which covers the great majority of factory automation. Most rolled product, however, falls in the C7 to C10 transport classes, which is entirely adequate for clamping, feeding, palletising, and material handling where repeatability matters more than absolute accuracy. The rolled thread also has a cold-worked surface skin that can be marginally beneficial for fatigue.

A persistent myth deserves correction: no ISO or DIN clause mandates a manufacturing method for a given accuracy class. The standards specify the permissible deviations, not how to achieve them. A precision-rolled C5 screw and a ground C5 screw are, by definition, both C5. The honest selection logic is therefore to fix the accuracy class and surface-finish requirement from the application, then let cost and lead time decide rolled versus ground, rather than reflexively over-specifying a ground screw when a rolled C5 would serve and cost a fraction as much.

Regardless of process, the finished shaft is heat-treated to a hardened raceway, commonly 58 to 62 HRC on the rolling surfaces, with the ends machined and sometimes induction-hardened to carry the support bearings (commonly a duplex angular contact bearing set that takes the axial thrust), locknuts, and the shaft coupling that joins the screw to the motor. Surface treatments such as black oxide or thin chrome are offered for corrosion resistance, and stainless grades exist for food, medical, and washdown duty at some cost in load rating.

Chapter 4 / 06

Accuracy and Load Standards

Ball screws are governed by a coherent body of international standards, and reading them is the difference between a defensible specification and a guess. The ISO 3408 series is the master reference: Part 1 fixes vocabulary and designation, Part 2 lists preferred nominal diameters, leads, and nut and mounting dimensions for the metric series, Part 3 sets acceptance conditions and the permissible deviations measured at acceptance, Part 4 covers static axial rigidity, and Part 5 defines static and dynamic axial load ratings and the operational-life calculation. DIN 69051 is the closely aligned German standard frequently cited as DIN/ISO, and JIS B 1192 is the Japanese equivalent whose C-grade notation (C0, C1, and so on) is the notation most datasheets actually print.

Accuracy grade is the headline figure. Grades run C0, C1, C2, C3, C5, C7, and C10, finest to coarsest. The standard splits them in two: the high-precision positioning grades C0 to C5 are controlled by travel deviation within any 300 mm length, while the transport grades C7 and C10 are controlled by travel deviation over the full thread length. The table below lists the controlling 300 mm figures and typical applications so the grade can be read straight onto a duty.

GradeTravel dev. / 300 mmRun-out per rev (typ.)Typical application
C0~6 um~4 umMetrology, ultra-precision stages
C1~8 um~5 umJig grinders, optics
C3~12 um~8 umCNC machining centres, semiconductor
C5~23 um~18 umGeneral CNC, precision automation
C750 um (per 300 mm)n/a (transport)General automation, feed axes
C10210 um (per 300 mm)n/a (transport)Transfer, clamping, material handling

Load capacity is described by two ratings. The basic dynamic axial load rating, Ca, is the axial load under which a population of identical screws reaches a nominal life of one million revolutions with 90 percent survival; it drives the fatigue-life calculation. The basic static axial load rating, C0a, is the axial load that produces a defined permanent deformation (a total of about 0.0001 times the ball diameter) at the most heavily loaded ball-raceway contact; it guards against brinelling under standstill shock or clamping loads. Both ratings rise steeply with shaft diameter and ball size, which is why up-sizing the screw is the standard cure for short life.

The operational-life method in ISO 3408-5 mirrors rolling-bearing practice. Nominal L10 life in revolutions is L equals the cube of (Ca divided by the product of load factor fw and applied axial load Fa), multiplied by one million revolutions. The cubic exponent is the same one used for ball bearings, so a modest increase in Ca pays off dramatically: doubling the dynamic rating multiplies predicted life by eight. The load factor fw inflates the working load to account for vibration and shock and typically runs 1.0 to 1.2 for smooth motion, 1.2 to 1.5 for normal operation, and 1.5 to 2.0 under heavy shock.

Two further standardised concerns belong here. Static axial rigidity (ISO 3408-4) quantifies how far the nut deflects elastically per unit axial load, dominated by the ball-contact stiffness and the support-bearing and shaft compliance, and it directly limits positioning accuracy under cutting force. Acceptance testing (ISO 3408-3) defines exactly how lead deviation, run-out, and preload drag are measured and recorded on the inspection certificate that should accompany any precision screw, so insist on that certificate for grades C5 and finer.

Chapter 5 / 06

Key Specification Parameters

A ball screw datasheet lists many figures, but only a handful drive the buying decision: nominal diameter and lead, accuracy grade, preload class, dynamic and static load ratings, critical speed, DN value, and rigidity. Each is decoded below so a procurement engineer can read a datasheet without ambiguity.

Nominal diameter and lead are the geometric core. Nominal (shaft) diameter, paired with lead, is written as, for example, 32 x 10, meaning a 32 mm shaft advancing 10 mm per revolution. Lead sets the trade between speed and resolution: a large lead gives high linear speed and lets the load back-drive easily, while a small lead gives fine positioning resolution and higher mechanical advantage. Multi-start threads achieve large leads on a small diameter. Preferred diameter and lead combinations are tabulated in ISO 3408-2, and staying on that preferred list shortens lead time and widens supplier choice.

Preload class describes the internal axial force built into the nut to eliminate backlash and stiffen the assembly. The three production methods are oversized balls in a single nut (light preload, generally up to about 5 percent of Ca), a lead-offset single nut where the raceway lead steps part way along the nut, and a double-nut design spaced by a precision shim. Preload is usually quoted as a percentage of dynamic capacity Ca, around 2 to 5 percent for light service and 5 to 8 percent for high-rigidity machine-tool axes. The penalty is drag torque and heat: every increment of preload raises no-load running torque and accelerates wear, so preload is specified to the minimum that meets the backlash and rigidity target, not maximised.

Dynamic and static load ratings (Ca and C0a) were defined in Chapter 4 and feed directly into life and shock checks. On the datasheet they appear in newtons; verify both, because a screw can satisfy the fatigue-life check on Ca yet still brinell under a clamping shock that exceeds C0a. Backlash is the axial lost motion in a non-preloaded nut, typically 0.05 to 0.1 mm for standard rolled assemblies and near zero for preloaded precision nuts; it is the figure that decides whether an open-loop or bidirectional positioning task is feasible.

Critical speed is the rotational speed at which the unsupported shaft reaches its first bending resonance. It falls with the square of the unsupported length and rises with root diameter, and it is multiplied by an end-fixity factor: fixed-free supports tolerate the least speed, simple-simple more, and fixed-fixed the most (roughly six times the fixed-free value). Engineers run below 80 percent of the calculated critical speed to keep a safety margin against whirl.

DN value (or Dm-N) is the product of ball-circle diameter in millimetres and rotational speed in rpm, and it limits the recirculation circuit rather than the shaft. Exceeding the DN ceiling makes balls collide and jam at the return entry, spiking wear and noise. Typical ceilings are about 70,000 for return-tube nuts, 100,000 to 160,000 for deflector nuts, and up to roughly 180,000 for high-speed end-cap designs. The governing speed limit is always the lower of critical speed and the DN ceiling.

Buckling load matters for long, slender screws loaded in compression: the permissible axial compressive load scales with root diameter to the fourth power, inversely with the square of unsupported length, and with an end-fixity factor (fixed-free smallest, fixed-fixed largest). A screw can pass its fatigue and speed checks yet still buckle elastically under thrust if the column is too slender, so this check is mandatory on vertical-press and long-stroke axes. Finally, lead accuracy and run-out values on the inspection certificate close the loop back to the accuracy grade decoded in Chapter 4.

Chapter 6 / 06

Selection Decision Factors

To convert the preceding chapters into a specific part number, follow the ordered sequence below. Most selection errors are not single wrong numbers but decisions taken in the wrong order, for example fixing the lead before confirming the speed limit. This list doubles as a fixed RFQ template.

  1. Stroke, mounting, and diameter: Set required stroke and end-fixity (fixed-fixed, fixed-supported, or fixed-free), then choose a nominal diameter that survives the buckling and critical-speed checks for that unsupported length. Long, fast, or vertical axes push diameter up first. In a typical feed axis the screw carries thrust while a parallel linear guide takes the moment and side loads, so the two are sized together.
  2. Lead: Derive lead from the required linear speed and motor speed (linear speed equals lead times rpm), then confirm the resolution and back-drive consequences. Larger lead means faster but easier back-drive; smaller lead means finer resolution and more self-help against gravity.
  3. Accuracy grade: Map the positioning tolerance onto C0 to C10 using Chapter 4. Do not over-specify: a rolled C5 covers most automation, while ground C3 or finer is reserved for sub-10-micrometre work. Each finer grade adds cost and lead time.
  4. Preload and rigidity: Choose zero, light, or medium preload from the backlash and stiffness target. Bidirectional positioning needs preload; unidirectional or low-duty transport often does not. Remember preload raises drag torque and heat.
  5. Load rating and life: Compute L10 with L equals the cube of (Ca over fw times Fa) times one million revolutions, convert to hours with Lh equals L over (60 times n), and confirm C0a exceeds the worst-case static and clamping shock. If life is short, enlarge diameter before enlarging the motor, because the cubic law rewards it.
  6. Speed limits: Verify both critical speed (run below 80 percent) and the DN ceiling for the chosen nut style. The lower limit governs, and it may force a change of recirculation type or a support change.
  7. Nut style and recirculation: Pick return-tube, deflector, or end-cap per Chapter 2 from lead and speed, then select flange geometry and ingress protection (wipers, scrapers, bellows) for the contamination environment.
  8. Lubrication and environment: Specify grease or oil, the relubrication interval, and any stainless or coated execution for corrosion, food, or cleanroom duty. Lubrication failure, not fatigue, is the most common field cause of premature ball screw death.

One dimension that buyers routinely overlook is serviceability and documentation: availability of an ISO 3408-3 inspection certificate for precision grades, published Ca, C0a, rigidity and DN data, support-bearing and end-machining drawings, and a repair or reman path for large shafts. THK, NSK, Bosch Rexroth, Steinmeyer, Kuroda, KSS, Hiwin, PMI, and TBI Motion all publish full technical handbooks and operate regional support, which makes them defensible choices on long-running production lines. Match the supplier tier to the accuracy class and the documentation you can defend at audit, not to brand prestige.

FAQ

What is the difference between a ball screw and a lead screw?

A ball screw transmits load through recirculating steel balls that roll between the shaft and nut raceways, so it relies on rolling friction and reaches mechanical efficiency of 90 percent or higher. A lead screw (often an Acme or trapezoidal thread) lets the nut slide directly against the screw thread, relying on sliding friction, so efficiency typically averages around 40 to 50 percent and can fall below 20 percent. The practical consequences are large: a ball screw needs a smaller motor for the same thrust, generates less heat, and lasts far longer under continuous duty, but it back-drives easily and usually needs a brake to hold a vertical load. A lead screw is cheaper, quieter, self-lubricating in plastic-nut form, and can be self-locking, which suits low-duty or hold-position axes.

What do the ball screw accuracy grades C0 to C10 mean?

Accuracy grade defines how closely the actual travel of the nut matches the commanded travel. Under ISO 3408-3 and DIN 69051, grades run C0, C1, C2, C3, C5, C7, and C10, where C0 is the most accurate and C10 the loosest. For positioning grades C0 to C5 the controlling figure is travel deviation over any 300 mm: C0 permits about 6 micrometres, C3 about 12 micrometres, and C5 about 23 micrometres per 300 mm. The transport grades C7 and C10 are specified instead as travel deviation over the full thread length: C7 allows 50 micrometres per 300 mm and C10 allows 210 micrometres per 300 mm. Choose C3 or C5 for CNC and semiconductor positioning, C7 for general automation, and C10 for transfer and clamping axes where repeatability matters more than absolute position.

Is a ground ball screw always more accurate than a rolled one?

Ground screws generally reach higher accuracy because the thread is finish-ground after hardening, removing heat-treat distortion, and they cover grades C0 to C5. Rolled screws are cold-formed between dies, which is faster and far cheaper, and they typically cover C5 to C10. However, no standard dictates a manufacturing method for a given class: a precision-rolled screw can meet C5 (about 23 micrometres per 300 mm), which is adequate for most automation. For sub-10-micrometre positioning, custody-grade machine tools, and metrology stages you still want ground C3 or better. The honest rule is to specify by the accuracy class and surface finish you need, then let price decide rolled versus ground, rather than assuming ground is mandatory.

How does ball screw preload work and how much should I specify?

Preload is an internal axial force that removes backlash and stiffens the nut, improving repeatability and positioning rigidity. Three methods dominate: oversized balls in a single nut (used for light preload, typically up to about 5 percent of dynamic load rating), a lead-offset single nut where the raceway lead steps part way down the body, and a double-nut design that is spaced apart by a shim or spacer. Preload is commonly quoted as a percentage of dynamic load capacity Ca: 2 to 5 percent for light service, 5 to 8 percent for high-rigidity machine tools. More preload buys stiffness and zero backlash but raises drag torque, heat, and wear, so over-preloading shortens life. As a guideline, keep external axial load below roughly 2.8 times the preload so the unloaded ball row never fully separates.

How do I calculate ball screw service life?

Ball screws use the same rolling-fatigue model as ball bearings, with a cubic exponent. The nominal L10 life in revolutions is L = (Ca / (fw x Fa)) cubed x 1,000,000, where Ca is the basic dynamic axial load rating, Fa is the applied axial load, and fw is a load factor (about 1.0 to 1.2 for smooth running, 1.2 to 1.5 with normal operation, and 1.5 to 2.0 under shock and vibration). L10 means 90 percent of a population survives without surface fatigue. Convert to hours with Lh = L / (60 x n), where n is the mean rotational speed in rpm. Because of the cubic relationship, doubling the load rating raises life eightfold, so when life is short the cheapest fix is usually a larger shaft diameter rather than a stronger motor.

What limits ball screw speed: critical speed or DN value?

Two independent limits apply. Critical speed is the rotational speed at which the shaft hits its first bending resonance; it falls with the square of unsupported length and rises with root diameter, and it depends strongly on end fixity (fixed-fixed mounting tolerates roughly six times the speed of fixed-free). Engineers normally run below 80 percent of the calculated critical speed. The DN value (sometimes Dm-N), the product of nominal or ball-circle diameter in millimetres and speed in rpm, limits the balls in the recirculation circuit; once exceeded, balls jam at the return entry and wear accelerates. Typical DN ceilings are 70,000 for conventional return-tube nuts, 100,000 to 160,000 for internal-deflector nuts, and up to 180,000 for high-speed end-cap designs. The lower of the two limits governs.

Which ball nut recirculation type should I choose?

There are three families. External return-tube nuts route balls through a tube clamped outside the nut body: simple, low cost, suited to large leads and large ball sizes, but bulky and DN-limited. Internal-deflector (finger or button) nuts turn the balls back over a single thread inside the nut: compact, quiet, and ideal for fine leads, used in most precision screws such as the THK deflector type. End-cap nuts collect balls at one end, carry them through a bore in the nut, and reinject at the other end: they handle the highest leads and DN values, so they dominate high-speed transfer axes. Choose deflector for precision and fine lead, end-cap for high lead and high speed, and return-tube for cost-sensitive heavy-lead duty.

Ask SpecForge AI