A vertical axis on a drilling machine lost 0.02 mm of repeatability over a year. The axis was a 32 mm ball screw, 800 mm travel, driven by a 1 kW servo. The repeatability drifted slowly, then jumped. The customer checked the nut preload — fine. They checked the coupling — fine. They checked the servo — fine. The problem was the screw’s mounting arrangement. The screw was mounted fixed-fixed, with angular contact bearings at both ends, and the fixed-fixed arrangement was fighting the screw’s thermal growth.

The four mounting options and what they buy

A ball screw has two ends, and each end can be fixed (axially located, takes thrust both ways) or supported (radially located, free to slide axially). The combinations:

  • Fixed-fixed. Both ends fixed. Highest axial stiffness, highest critical speed. The screw can’t grow thermally — the growth becomes internal compression load.
  • Fixed-supported. One end fixed, one end supported. The supported end slides in a plain bearing. Stiffness slightly less, critical speed slightly less, but the screw can grow axially without loading up.
  • Fixed-free. One end fixed, the other free. Lowest stiffness, lowest critical speed. Only for short, low-speed, light screws.
  • Supported-supported. Neither end takes thrust. The nut must take all axial load. Rare, used on very long screws where thermal growth dominates.

On the drilling machine, the screw was 800 mm with fixed-fixed bearings. The motor ran at high duty, the screw got warm — maybe 15°C above the frame. A 32 mm screw, 800 mm long, grows about 0.015 mm per 100 mm per 100°C. At 15°C rise, that’s 0.18 mm of growth. Fixed-fixed won’t allow it. The growth turned into a compressive load on the screw. The screw bowed slightly, the effective length changed, and the axis lost repeatability.

Why fixed-fixed isn’t always best

The catalogs push fixed-fixed as the highest performance option — it is, for stiffness and critical speed. But it only works if the screw and the machine frame are at the same temperature. On a machine that runs warm or has a heat source near one end of the screw, fixed-fixed turns thermal growth into an internal load. The screw pushes against both bearing sets. The preload on the drive-end bearing changes. The axis drifts.

The drilling machine ran 16 hours a day with the servo motor mounted right next to the drive-end bearing. The motor radiated heat into that bearing housing. The far end of the screw stayed at shop temperature. The temperature gradient across the 800 mm was maybe 12°C. The differential growth was the entire problem.

The change that fixed it

The drive end stayed fixed. The far end was changed from a fixed bearing pair to a supported plain bearing (a needle or sleeve bearing in the housing). Now the screw grows freely toward the supported end. The thermal growth slides into the support instead of loading the screw. The repeatability came back to 0.005 mm and stayed there.

The stiffness penalty was small. The axial stiffness of a fixed-fixed 32 mm screw over 800 mm is about 200 N/μm. Fixed-supported drops to about 160 N/μm. For a drilling axis, the loss was invisible. The thermal stability it bought was the whole game.

When fixed-fixed is actually required

Fixed-fixed is worth its thermal trouble when the axis needs maximum stiffness and the screw is short relative to its diameter (length-to-diameter under about 30). A 32 mm screw at 800 mm is L/D of 25 — the boundary. For L/D under 20, fixed-fixed with careful cooling or temperature control is fine. For L/D over 30, or for any screw on a machine with an uneven heat source, fixed-supported is the safer call. The critical speed penalty is small — fixed-fixed critical speed is about 10% higher than fixed-supported for the same screw, and most axes run well below critical speed anyway.

The check to run before you change anything

If an axis drifts and you suspect thermal growth, do this: run the machine at full duty for an hour, then measure the axis position against a reference (a dial indicator on the table against a fixed stop). Mark it. Let the machine sit overnight, cold. Measure again. If the position differs by more than 0.01 mm between hot and cold, thermal growth is in play. Then check the screw surface temperature along its length — if there’s a gradient over 5°C, the mounting arrangement is fighting itself. Change the far end to supported and the problem goes away.

Fixed-fixed mounting is the stiffest on paper and the most thermally sensitive in practice. The drilling axis drifted because a 12°C gradient turned into internal screw load. Fixed-supported lets the screw breathe and keeps the repeatability. Use fixed-fixed for short screws on temperature-controlled machines. Use fixed-supported for long screws or machines that run warm.