Ball-screw selection resolves to four binding decisions: dynamic load rating (Ca) versus the calculated equivalent axial load, accuracy grade per ISO 3408 (commonly C0, C3, C5, C7, C10), lead versus linear travel rate, and the DN limit (nominal diameter × rpm) the chosen end-fixity can sustain without overheating [S1][S4].
For a single-axis pick-and-place or general automation, an ungrounded rolled ballscrew (C7–C10) is the default; for CNC machine-tool axes, optical or inspection equipment, and semiconductor handlers, a precision-ground screw (C3–C5) is the working envelope, with C0 reserved for ultra-precision machines and metrology [S1].
Definition and Operating Envelope
A ball screw converts rotary to linear motion through recirculating ball bearings running between a screw shaft and nut; mechanical efficiency sits above 90%, which is why the device is not self-locking and a brake or counter-load is mandatory on vertical axes [S4]. Standard catalog diameters span 6 mm up to 160 mm, with leads from 1 mm to 50 mm and standard lengths typically cut between 1 m and 6 m before end-machining [S1].
Two mechanical limits govern the upper operating envelope: critical speed (N_c) set by shaft diameter, length between bearings, and end-fixity, and the DN product, which is normally capped at roughly 100,000 for standard lubrication regimes — exceeding it shortens ball-track life and risks brinelling of the raceway [S1].
Selection Criteria and Calculation Path
The canonical sequence is: (1) determine axial load and mean load (F_m) from the duty profile; (2) compute required dynamic load rating C_a from the target L_10 life in millions of revolutions using C_a = F_m · (L_10)^(1/3); (3) select a screw whose catalog C_a exceeds that value; (4) check critical speed and DN against the application rpm; (5) pick an accuracy class consistent with the machine's positioning repeatability budget [S1].
Equivalent axial load must include any preload preload, the cutting or process force on the driven axis, and weight for vertical axes; a common engineering shortcut is to apply 1.0–1.2 × maximum process force as F_max when the duty cycle is poorly characterised, then derate to F_m using the time-weighted average. Nut preload as a percentage of dynamic load — typically 5–10% for machine tools and 2–5% for transfer axes — directly shortens rated life, so it must be carried into the calculation rather than added as a free margin [S1].
Accuracy Classes and Tolerance Bands
ISO 3408 defines the grades most engineers quote on drawings: C0 is the tightest, used on coordinate-measuring and surface-grinder spindles; C3 and C5 cover the bulk of CNC machining centres, lathes, and EDM; C7 and C10 suit injection moulding, packaging, and general material handling [S1].
The class controls three measurable quantities on a 300 mm reference length: specified travel deviation (the E_300 figure), backlash at zero load, and the variation in dynamic preload. As a rule of thumb, an axis that needs ±5 µm repeatability should not be paired with a screw worse than C3, and a C5 screw is normally the cost optimum for a 3-axis vertical machining centre with 0.01 mm positioning resolution [S1].
Rolled versus Ground: Decision Matrix
Rolled (cold-rolled) screws are produced by swaging a thread into medium-carbon or alloy steel, with the nut using a clearance-fit ball track; the cost is roughly one-third to one-half of a comparable ground screw, lead accuracy sits in the ±0.05 mm/300 mm range, and backlash runs 0.05–0.15 mm depending on preload method [S1].
Ground screws are machined and then ground against a master, with the nut track matched to the screw; they deliver ±0.005–0.012 mm/300 mm lead accuracy and zero-backlash preload options (oversize-ball or single-point preloading), at 2–4× the price and 4–6× the lead time on non-standard dimensions [S1]. The decision pivots on three tests: is positioning repeatability below 10 µm? — pick ground. Is the duty cycle under 50% with sub-1 m travel? — rolled is fine. Is the axis subject to reversal shock loads? — ground with oversize-ball preload.
Lead, Preload and End-Fixity Choices
Lead is set by linear speed and motor rpm: lead = v × 60 / n. A 1,000 mm/s axis on a 3,000 rpm servo gives a 20 mm lead; pushing the same speed with a 5 mm lead forces the motor past 12,000 rpm, which most industrial servos cannot sustain without gearbox derating. The interaction between lead and accuracy is direct — finer leads mean more thread engagements per unit travel, which reduces elastic compliance but raises the torque required to drive the nut [S1].
End fixity controls critical speed: fixed-free is the worst (β = 0.36), fixed-supported sits at β = 1.0, and fixed-fixed reaches β = 2.0 in textbook values, doubling allowable rpm for a given length. A fixed-fixed arrangement with a 32 mm-diameter, 1,500 mm-long screw runs comfortably to ~3,000 rpm in machine-tool service, whereas the same screw in fixed-free stalls near 1,000 rpm [S1].
Application Fit and Common Failure Modes
The dominant failure modes in field service are fatigue spalling of the ball track (linked to under-rated C_a or excessive preload), brinelling from static overload or shock (lifted loads, E-stop events), and premature wear from contamination when the wipers are omitted. Lubrication interval is the single largest controllable variable: grease-lubricated screws in a clean environment run 1–2 years between relubrication cycles, while oil-mist or oil-bath extends that to the bearing's rated L_10 life [S1].
For engineers sourcing from Chinese production hubs, large-volume ball-screw and ball-bearing catalogues from Jiangsu and Zhejiang typically stock SFU, SFK, and SFE nut families in C7/C10, with custom ground units available on 4–6-week lead times. Engineers sizing linear-motion stacks should treat the screw, the ball spline where one is used, and the end-bearing set as a coupled thermal and stiffness problem rather than three independent purchases.
Standards, Sourcing Levers and Cross-Reference
The governing specification is ISO 3408-1 through -3 for tolerance grades, geometry, and acceptance inspection, with JIS B 1192 historically serving as a near-equivalent for Japanese-supplied screws. Material expectations default to bearing steel (e.g. 50CrMo4 or SUJ2 equivalent) case-hardened to HRC 58–62, with stainless 440C or 1.4034 available for medical and food-grade axes [S4].
Practical sourcing levers in 2026: standardise on a small set of nut formats (SFU1605, SFU2005, SFU2505) to share stock across a machine family; specify preload as a percentage of Ca rather than as a torque value; and ask the vendor for an actual E_300 travel-test report, not just a class declaration. Buyers evaluating linear-motion stacks alongside other industrial categories can use the concrete pump truck sizing guide as a reference for the same kind of multi-parameter spec band logic, since both purchases trade off reach, payload, and duty cycle. For applications where the ball screw drives a ball valve or electric ball valve actuator, lifetime calculations must include the breakaway torque contribution from the seated valve, not just the process-line load [S1].
Trackable signals for the next design review: vendor-published test certificates aligned to ISO 3408-3, and any updates to standard factory-stock lead-times for C3/C5 ground screws — both are the leading indicators of whether the current sourcing landscape is tightening or loosening [S1][S4].