"My axis has play in it" is one of those messages that can mean five different things, and we have learned to ask before assuming. But when the play shows up as a small lost distance every time the direction reverses - the part measures fine one way and a couple of hundredths off the other - nine times out of ten it is backlash in the ball screw nut. And backlash in a ball screw is a selection problem, not a wear problem.
This is the part of ball screw selection that catalogs skip over, because it is hard to put in a table: preload. Whether your axis has backlash, and how much, was decided the day the nut was chosen - not by the machine, and not by the controller.
Where the play comes from
A ball screw nut has balls rolling between the screw and the nut. For the balls to roll freely, there has to be a tiny bit of space - that space is backlash. It is not a defect. A nut with visible free play is a nut that was ordered without preload, and for positioning axes that was probably a mistake, because the controller cannot compensate for mechanical play it does not know about.
Preload removes the space by squeezing the balls between slightly oversized balls, or by using a double nut that pushes against itself. The result is zero measurable play at the cost of a little efficiency and a little heat. A light preload costs you a couple of percent of efficiency; a heavy preload can cost you ten.
Preloaded or not?
No preload
Free-running, maximum efficiency, near-zero heat. Right for transport axes, feeders, anything where a little positional slop does not matter and the load is mostly one-directional.
Preloaded
Zero play, higher stiffness - the axis holds position when the direction flips. The standard for milling, grinding, engraving and any cutting where tool pressure reverses the load. Costs efficiency and runs warmer.
The numbers matter more than the label. A preloaded ball nut will typically hold positioning repeatability in the low microns on a properly mounted axis; the same screw with a no-preload nut will show tenths-of-a-millimeter play under reversing load. That gap is why two machines with the same screw model can feel completely different to run.
The mistakes we see are consistent. Buying the cheapest nut and wondering why the axis "drifts" - it does not drift, it has backlash, and no amount of tuning fixes mechanical play. Over-preloading everything because zero play sounds safest - a heavy preload on a long screw heats the screw, and a hot screw grows, which is its own positioning error. Ignoring lubrication on preloaded nuts - a preloaded nut has more contact, and dry running wears it out in months instead of years.
If you are ordering a linear guide and ball screw set for a new axis, tell us what the axis does - cutting or positioning, how often it reverses, and how long the screw is. Through the inquiry form, that is enough for us to pick the preload class, and usually to talk you out of over-specifying it. We have been through this trade-off a few thousand times.
