A linear axis that wandered 0.05 mm at the same position. The ball rail was new. The balls were fine. The issue: the rail was installed with clearance, not preload. At low speeds, the carriage wobbled. The customer thought the balls were worn. They weren’t — the rail was selected with C0 (clearance) preload class for a precision axis. This is about ball rail preload selection.

The three preload classes

Linear ball rails come in three preload classes:

  • C0 (Clearance): balls slightly smaller than raceway gap. Free movement, no preload. Low friction. Use for low-accuracy, high-speed applications where friction matters more than stiffness.
  • C1 (Light preload): balls slightly larger than raceway gap. Light interference. Moderate stiffness. Use for general automation.
  • C3 (Heavy preload): balls significantly larger. Strong interference. High stiffness, high friction. Use for precision machine tools and high-acceleration axes.

What preload does

Preload eliminates clearance between the balls and raceways. Without preload (C0), there’s a small gap — the carriage moves slightly (5-10 μm) before the balls engage. Under load, the carriage deflects. With C1 preload, the gap is eliminated. The carriage deflects less. With C3, the stiffness is maximum — but the friction is 2-3x higher, and the life is shorter (because the balls are overloaded).

The stiffness difference is dramatic. For a standard 25 mm rail: C0 stiffness is about 30 N/μm. C1 is 60 N/μm. C3 is 120 N/μm. At 1000 N load: C0 deflects 33 μm. C1 deflects 17 μm. C3 deflects 8 μm. For a precision axis needing ±5 μm accuracy, C3 is required. For a pick-and-place axis needing ±50 μm, C1 is fine.

What I changed

1. Changed C0 to C1. The wandering axis was a CNC grinding spindle feed. It needed C1 at minimum. I replaced the C0 carriage with C1. The deflection dropped from 33 to 17 μm. The axis repeated to ±10 μm. That met the spec.

2. Matched preload to load. C3 preload increases stiffness but reduces life. The ball life formula includes preload as a factor: L10 = (C/P)^3 × 50 km. With C3 preload, the effective load increases by about 20% (preload acts as a constant load). Life drops by about 1/(1.2)^3 = 58%. I don’t spec C3 unless the stiffness is absolutely required.

3. Checked mounting surfaces. Preload only works if the rail is mounted flat. A twisted mounting surface adds an external load that mimics preload — but unevenly. The carriage on a twisted rail runs hard on one side. The balls wear prematurely. I always check the mounting surface flatness with a straightedge before installing preloaded rails.

The friction trade-off

C3 preload has about 0.015 friction coefficient vs 0.002 for C0. For a 25 mm rail with 5000 N normal force, the C0 friction is 10 N. C3 friction is 75 N. On a horizontal axis, that’s fine. On a vertical axis, the motor must hold the load plus overcome the higher friction — I downsize the preload to C1 to avoid a bigger motor.

The class I choose: C1 for most automation, C3 for precision machine tools, C0 only for low-friction high-speed where stiffness doesn’t matter. The wandering axis wasn’t worn — it was C0 preload on a precision application. Swap to C1 and the deflection drops in half.