A comprehensive engineering guide to ball screw backlash: root causes (manufacturing clearance, wear, thermal expansion, elastic deformation), measurement methods (dial indicator, torque-to-movement), and compensation techniques (double-nut preload, oversized balls, CNC software compensation, closed-loop linear scale feedback) with a selection guide for zero-backlash applications.
What Is Ball Screw Backlash and Why It Matters
Backlash is the axial play between the ball screw shaft and the ball nut - the distance the shaft can rotate without producing any linear movement at the nut. In an ideal zero-backlash system, every fraction of a degree of screw rotation translates instantly into linear displacement. In practice, manufacturing tolerances, wear, and thermal effects all introduce a small dead band where input rotation produces no output motion.
For open-loop systems driven by stepper motors or servo motors with rotary encoder feedback only, backlash directly degrades positioning accuracy. If a ball screw has 0.05 mm of backlash and the axis reverses direction, the first 0.05 mm of commanded travel produces no actual movement - the motor turns, but the nut stands still until the clearance is taken up. In closed-loop systems with a linear scale, the controller compensates for backlash in real time, but the dead band still affects dynamic response, settling time, and contour accuracy during direction reversals.
Root Causes of Backlash
1. Manufacturing Clearance
Standard rolled ball screws (C7 accuracy grade) are produced by roll-forming the thread groove into the shaft. The rolling process cannot achieve the same dimensional precision as thread grinding, so the ball groove diameter, ball diameter, and nut bore all carry tolerance accumulation. The resulting clearance typically ranges from 0.05 to 0.15 mm for a standard single nut. Ground ball screws (C5 accuracy and better) have tighter tolerances - standard nut backlash of 0.003 to 0.008 mm - because the grinding process controls groove geometry within microns.
2. Wear During Service
Even a zero-backlash preloaded ball screw develops clearance over time as the ball-raceway contact surfaces wear. In a typical industrial duty cycle of 20 km of travel per day, the cumulative rolling contact fatigue gradually widens the raceway groove. After 12-18 months of continuous operation, a preloaded nut that started at zero backlash may develop 0.01 to 0.03 mm of lost preload - enough to affect bidirectional positioning on precision axes.
3. Thermal Expansion
The ball screw shaft heats up during operation from friction between the balls and raceway. A 1,000 mm steel screw heated by 5°C expands axially by approximately 0.06 mm. If the nut and shaft expand at different rates - for example, the nut body is aluminum while the shaft is steel - the differential expansion changes the internal ball clearance, creating transient backlash during warmup and thermal cycling.
4. Elastic Deformation Under Load
Under heavy axial load, the balls and raceways deform elastically at the Hertzian contact points. This deformation is not permanent, but it produces a compliant dead band that behaves like backlash during direction reversals. For a 25 mm diameter ball screw under 5,000 N axial load, elastic deformation contributes approximately 0.008 to 0.015 mm of apparent backlash - significant for sub-0.02 mm positioning applications.
Measuring Backlash: Practical Methods
Dial Indicator Method
The simplest backlash measurement requires a dial indicator (0.001 mm resolution) and a fixed reference point on the machine bed:
- Jog the axis in one direction to take up all clearance. Zero the dial indicator against the carriage.
- Reverse the jog direction by a small increment (e.g., 0.5 mm commanded move).
- Read the dial indicator: the difference between commanded move and actual move is the total backlash.
- Repeat 3-5 times and average the readings to account for measurement noise.
For a C7 rolled SFU2010 ball screw, expect 0.05 to 0.12 mm. For a C5 ground SFE2010 with preloaded nut, expect 0 to 0.005 mm.
Torque-to-Movement Method
For automated measurement on CNC machines, apply a known torque pulse to the servo motor and measure the resulting linear displacement. The torque-to-displacement ratio during the clearance phase is near zero (the shaft rotates freely), then jumps sharply once the balls engage. The angular rotation during the free phase, multiplied by the screw lead, gives the backlash value. This method integrates seamlessly into CNC controller diagnostics and can track backlash drift over time.
Backlash Compensation Methods
Method 1: Preloaded Double Nut (DFU Series)
The most effective hardware solution is a preloaded double nut. Two ball nuts are mounted on the same screw with a precision spacer between them. The spacer forces the balls in one nut to contact the right-hand side of the groove, while the balls in the other nut contact the left-hand side. This pre-loads both nuts against each other, eliminating clearance in both directions.
Dongfeng Bearing's DFU series double-nut ball screws achieve true zero backlash with preload levels adjustable from light (P0) through heavy (P2). The preload force is set at the factory using precision ground spacers matched to the specific screw and nut combination. Preload measurement data is provided per assembly, documenting the actual residual backlash (typically <0.003 mm) and the preload drag torque.torque.
The tradeoff of double-nut preload is higher friction and heat generation. A P2 (heavy) preload can increase drag torque by 40-60% compared to a non-preloaded single nut, which raises the motor torque requirement and may limit maximum traverse speed. For most precision applications, P1 (medium) preload provides the best balance of zero backlash and acceptable thermal behavior.
Method 2: Oversized Ball Preload
An alternative to the double-nut approach is selecting balls 1-2 microns larger than the standard ball diameter for the screw size. The oversized balls create interference fit in the single nut, pre-loading the ball-raceway contact in both directions simultaneously. This method is more compact (single nut envelope) and lighter than a double nut, but the preload level is fixed at manufacture and cannot be adjusted in the field.
Oversized ball preload is standard on Dongfeng's SFE series ground ball screws in P1 and P2 preload classes. It is preferred for applications where envelope size is constrained - such as semiconductor wafer stages and medical device assembly axes - where the double-nut length would exceed the available Z-axis stack height.ght.
Method 3: CNC Software Compensation
Most modern CNC controllers support backlash compensation: when the axis reverses direction, the controller injects a preset number of additional steps before commanding the actual move. The compensation value is measured during machine commissioning and stored in a controller parameter.
Software compensation is effective for static positioning but has limitations:
- It does not help during continuous contouring: In circular interpolation, the axis reverses direction at the quadrant points. The compensation step creates a brief dwell that produces a visible mark on the workpiece surface at the 0°, 90°, 180°, and 270° positions of a circle.
- It cannot track wear: The compensation value is set at commissioning. As the ball screw wears and backlash increases, the stored value becomes stale. Periodic re-measurement and parameter updates are required - typically every 6 months on production machines.
- It does not address elastic compliance: Load-dependent deformation is not a fixed value; it varies with cutting force. A fixed compensation parameter cannot adapt to the varying loads encountered during machining.
For these reasons, software compensation is a complement to - not a replacement for - mechanical preload. The best practice is to minimize mechanical backlash with a preloaded nut, then use software compensation for the residual 0.003-0.005 mm that even preloaded systems exhibit.
Method 4: Closed-Loop Linear Scale Feedback
The ultimate backlash solution is a closed-loop system with a linear encoder (glass scale or magnetic scale) mounted on the moving axis. The controller reads the actual carriage position directly from the scale, bypassing the ball screw entirely as the position reference. Backlash, lead error, and thermal expansion are all compensated in real time because the controller compares the commanded position to the measured position, not to the calculated position from motor rotation.
This approach is standard on jig borers, coordinate measuring machines, and semiconductor lithography stages where sub-micron positioning is required. The cost is significant: a 1,000 mm linear scale with 0.1 micron resolution adds $1,500-3,000 per axis, plus the controller must support full closed-loop operation. For general-purpose CNC machining at +/-0.01 mm tolerances, a preloaded ball screw with rotary encoder feedback is more cost-effective.
When to Replace vs Compensate
A ball screw that has developed 0.05 mm or more of backlash through wear should be evaluated for replacement rather than compensation. Here is the decision framework:
- Backlash < 0.01 mm:mm: Normal for a preloaded nut after 12 months of service. Use software compensation for the residual value. No action needed.
- Backlash 0.01 - 0.03 mm: The preload is degrading. If the screw is a double-nut (DFU) design, the spacer can be re-shimmed to restore preload. If single-nut oversized-ball, plan replacement within 3-6 months. Increase software compensation and monitor monthly.
- Backlash > 0.03 mm on a preloaded nut, or > 0.10 mm on a standard nut:d nut: The raceway is worn beyond field compensation. Replace the ball screw and nut assembly. Inspect the screw shaft raceway under magnification - if spalling or pitting is visible, the shaft must also be replaced; re-nutting a worn shaft will fail within weeks.
Selecting the Right Ball Screw for Zero-Backlash Applications
When specifying a ball screw for an application that requires zero or near-zero backlash, match the screw type to the application demands:
| Application | Recommended Type | Expected Backlash |
|---|---|---|
| CNC wood router (+/-0.1 mm tolerance) | SFU rolled, standard single nut | 0.05 - 0.12 mm (compensate in software) |
| CNC aluminum mill (+/-0.02 mm tolerance) | DFU rolled, P1 preload double nut | 0 - 0.005 mm |
| Surface grinder (+/-0.005 mm tolerance) | SFE ground, P1 preload, oversized ball | 0 - 0.003 mm |
| EDM die sinker (+/-0.005 mm, bidirectional) | DFU ground, P2 preload double nut | 0 mm (true zero) |
| Semiconductor wafer stage (sub-micron) | SFE ground, P2 preload + linear scale | 0 mm (closed-loop compensated) |
Xiamen Dongfeng Bearing Mechanical & Electrical Co., Ltd. manufactures both rolled (SFU) and ground (SFE) ball screw series with single-nut and double-nut (DFU) configurations. Preload classes from P0 through P2 are available with per-assembly preload measurement documentation, allowing machine builders to specify the exact backlash performance their application demands - from general-purpose CNC routers to jig borers and semiconductor equipment.ent.