Two inspection machines, same rails, same carriages, different preload. One machine held 1 μm repeatability on a measuring axis for years. The other drifted and the carriage ran stiff from day one. The difference: the good machine used light preload (class Z0), the stiff one used heavy preload (class Z3). The spec sheet had said heavier preload gives higher stiffness, so the builder picked the stiffest. The stiffness came, and so did the problems.
What preload does inside the carriage
A linear guide carriage has rows of balls between the rail and the carriage raceways. Preload is built in by using balls that are slightly oversized for the raceway gap — the manufacturer selects the ball diameter to create an internal interference. The balls are squeezed between rail and carriage. The result: the carriage has zero clearance and higher stiffness, because the balls are already in contact under load. Without preload, there’s a few microns of clearance, and the carriage moves a tiny amount before the balls pick up load.
The preload classes run from Z0 (no preload) through Z1 (light), Z2 (medium), and Z3 (heavy). The stiffness increases from Z0 to Z3, but the friction does too, and the running-in behavior changes. The carriage doesn’t just get stiffer — it gets harder to move, and the drag force varies across the stroke.
What heavy preload actually costs
On the inspection machine, the heavy preload showed up as two problems. First, the friction was high enough that the servo couldn’t hold position smoothly — the axis moved in small jumps as the balls climbed over the tiny variations in raceway geometry. The repeatability was worse than the machine with light preload. Second, the carriage ran warm — the extra friction heated the rail, and thermal growth moved the measuring reference. The axis drifted through the day as the machine warmed up.
The Z0 machine, by contrast, had a few microns of clearance and a slightly softer stiffness curve. For an inspection axis that carries a light probe and moves slowly, that clearance doesn’t show up as error — the positioning is done by the servo with position feedback, and the few microns of mechanical play are absorbed in the control loop’s settling. The stiffness that matters for a measuring axis is the stiffness at very low load, where the Z3 carriage’s preload was actually a liability.
The stiffness curves tell the story
Every guide manufacturer publishes a load-deflection curve for each preload class. The curves all flatten as load increases — at high load, all preloads behave similarly because the preload is a small fraction of the applied load. The difference is at low load. A Z3 carriage has a high initial stiffness — a 500 N load deflects it maybe 2 μm. A Z0 carriage deflects 8 μm under the same load. But the Z0 deflection is a smooth elastic curve. The Z3 carriage’s advantage only matters when the axis carries a significant, predictable load and needs that load held rigidly.
The rule that has held up in machine building: heavy preload is for axes that carry a real load — a milling spindle head, a grinding carriage — where cutting forces would push a light-preload carriage around. Light preload (Z1) or none (Z0) is for axes that carry the tool or probe itself, move fast, and need low friction and smooth motion. The cutting machine gets Z3 and pays the friction cost. The inspection machine gets Z0 and never pays it.
The moment Z3 becomes a wear problem
Heavy preload also changes wear behavior. A Z3 carriage has high contact stress at the ball-raceway contacts. Under light load, that stress is a constant — the balls are squeezed whether the axis is moving or not. On a machine that idles most of the time, the Z3 carriage wears faster than a Z2, simply because the contact stress is always there. The wear shows up as an increasing running resistance and, eventually, a loss of the very preload it was bought for. The Z3 advantage erodes in a year of light-duty running.
The manufacturer’s catalog shows a load rating for each preload. That rating assumes the preload is matched to the application. Running a Z3 carriage at 10% of its rated load, on an axis that mostly sits idle, is the worst case — maximum preload stress, minimum benefit. The correct pick for a lightly loaded axis is Z1 or Z0 even if the rails are the same size.
The checks before you switch preload
If an axis runs stiff, runs warm, or drifts with temperature, check three things before changing anything. Measure the running resistance with a spring scale at several positions along the stroke — a healthy Z3 carriage drags more than a Z1, but it should be uniform; a drag that climbs along the stroke is a geometry or contamination problem, not a preload problem. Check the rail temperature after an hour of running — more than 10°C above ambient with a light load points to excess preload or misalignment. Check the alignment — a rail mounted with the mounting screws over-torqued can bind the carriage exactly like heavy preload. Misaligned rails are the most common false preload problem.
Heavy preload buys stiffness at low load and charges friction, heat, and wear for it. Cutting axes with real cutting forces need it. Measuring axes and light carriages don’t. The stiff inspection machine was over-preloaded — the servo fought the friction and the heat moved the reference. Pick the preload class by the load the axis actually carries, not by the stiffest available.
