When a CNC machine tool changes temperature, it changes shape.
When a CNC machine tool changes temperature, it changes shape. This phenomenon—called thermal growth—is one of the largest remaining sources of dimensional error in modern machining. Thermal growth is the dimensional change of machine components caused by temperature variation. Heat comes from spindle bearings, servo motors, ballscrews, friction in guideways, cutting chips, coolant, and the shop environment itself.
For overseas workshop owners and procurement managers evaluating Chinese CNC equipment, thermal performance is not a marketing detail. It determines whether a machine holds tolerance on the first part of the morning shift and the last part of an unattended night run. A machine with poor thermal control can easily drift 50 µm during warm-up—five times the total positional tolerance of many precision jobs.
This article explains where thermal growth comes from, how machine builders measure and compensate for it, and what a buyer can check before purchase.
Most machine tool structures are made from cast iron or steel. These materials expand roughly 10–12 micrometres per metre per degree Celsius. A micrometre (µm) is one thousandth of a millimetre. A vertical machining centre column measuring 1.2 m from base to spindle can grow about 60 µm if its average temperature rises 5 °C. That movement is enough to scrap parts with position tolerances of ±25 µm or less.
Spindle growth alone is often larger than structural growth. In a typical 40-taper vertical machining centre, the spindle nose can move 20–60 µm in the Z axis during the first hour of operation as bearing friction and motor heat build up. On a chilled spindle, that drift can be held below 15–25 µm after warm-up. On an unchilled spindle, it can remain unstable for much longer.
The business impact appears in several forms:
For buyers comparing machines across suppliers, thermal stability is therefore a cost-of-ownership issue, not only a precision issue.
The spindle is the motor-driven rotating shaft assembly that holds the cutting tool. In a BT40 taper—a 40 mm diameter tapered shank interface with two drive keys, common on 40-taper vertical machining centres—the taper locates the tool holder in the spindle nose. Heat from spindle bearings and motor windings flows into the spindle housing and shaft.
Spindle growth usually appears as Z-axis drift. The front bearings heat first, the housing expands, and the tool tip moves downward relative to the table. If the spindle is vertical, this directly changes depth of cut and hole depth. If the spindle is horizontal, the same effect shifts bore alignment.
Liquid-cooled spindles reduce this error by circulating temperature-controlled coolant around the motor and bearing area. Air-fan cooling is less stable because it depends on ambient air temperature. Liquid cooling does not eliminate spindle growth, but it reduces the temperature gradient and makes the remaining drift more predictable.
A ballscrew is a precision screw shaft with recirculating ball bearings between the nut and the screw. It converts rotary motion from a servo motor into linear axis motion. Friction inside the ball nut, especially in preloaded double nuts, generates heat. A steel ballscrew expands about 11 µm per metre per degree Celsius. A 1 m screw that rises 5 °C grows roughly 55 µm.
If the machine uses only a rotary encoder on the servo motor, the control does not directly see this ballscrew expansion. The table may be slightly out of position even though the motor rotation is correct. A linear scale—a glass or steel scale mounted along an axis that directly reads slide position—can correct the measured table position, but ballscrew growth still affects mechanical stiffness, backlash behaviour, and thermal gradient.
The bed, column, saddle, and table castings absorb heat from the spindle, guides, coolant, chips, and room air. Because their thermal mass is large, they respond slowly. A shop temperature change of 8 °C over a shift can cause a 1.5 m column to expand more than 100 µm in length. Uneven heating from direct sunlight, open doors, or nearby compressors can also tilt the column relative to the table.
Coolant is another heat source. Return coolant carries heat from chips and the workzone into the machine base. If the coolant tank is not temperature-controlled, coolant temperature can rise 5–10 °C during heavy milling, slowly distorting the machine frame.
| Heat Source | Typical Effect | Main Axis Affected | Common Control Methods |
|---|---|---|---|
| Spindle bearings and motor | Tool tip drift, bore depth change | Z | Liquid spindle chiller, warm-up cycle, software compensation |
| Ballscrew friction | Table position error | X, Y, Z | Core-cooled screws, linear scales, software compensation |
| Cast structure heating | Column growth, tilt | Z, Y, X | Symmetric design, ambient control, software compensation |
| Coolant and chip heat | Base distortion over time | X, Y, Z | Temperature-controlled coolant, chip flushing |
| Workshop ambient change | Gradual drift during shift | All axes | Ambient sensors, enclosure insulation, air conditioning |
Thermal compensation is a system that shifts axis positions to counteract predicted or measured thermal deformation. There are two broad families: passive mechanical design measures and software-based correction.
Machine builders often reduce thermal error before applying software compensation:
A typical warm-up cycle lasts 10–20 minutes and ramps the spindle through several speed levels while moving linear axes across their strokes. This brings the machine closer to thermal equilibrium before cutting, reducing drift during the first production hour.
Software-based compensation uses temperature sensors mounted on the spindle nose, spindle rear, ballscrew nuts, motor flanges, column, and ambient air. The control reads sensor values and applies small axis offsets every few seconds based on a thermal model.
The model may use the difference between spindle front and rear temperatures to estimate Z-axis growth. It may use screw temperature to estimate axis drift. The exact mapping and calibration method are machine-builder-specific, but the operating principle is the same: temperature inputs, axis correction outputs.
There are two control approaches:
Software compensation works best when the temperature gradient is repeatable. It cannot correct random or unknown heat sources. It also requires periodic recalibration if sensors are moved, replaced, or if the machine is crashed.
| Compensation Method | Typical Sensors | Typical Improvement | Main Limitation |
|---|---|---|---|
| Warm-up cycle only | None | Reduces cold-start drift by 30–60% | Wasteful of time and power; does not control long-run drift |
| Liquid spindle chiller | Chiller outlet/inlet temperature | Spindle Z drift can be held under 25 µm after warm-up | Adds chiller maintenance; does not fix column distortion |
| Core-cooled ballscrews | Fluid temperature | Screw growth can be cut to 30–50% of uncooled value | Not all designs accept core cooling; small residual growth remains |
| Multi-sensor software compensation | 4–8 machine temperature sensors | Can reduce thermal drift to 30–70% of uncompensated value | Needs accurate sensor placement and recalibration after changes |
| Linear scale feedback | Glass or steel scale | Directly corrects linear position from screw growth | Raises cost; will not correct spindle Z growth or structural tilt |
The main international standard for thermal testing of machine tools is ISO 230-3. This standard defines test procedures for evaluating thermal distortion under no-load operation, spindle rotation, and linear axis movement. A buyer can request an ISO 230-3 test report from the machine builder to compare thermal drift under identical test conditions.
Another relevant standard is ISO 230-2, which covers positioning accuracy and repeatability. Repeatability is the ability of the machine to return to the same commanded position over multiple attempts. Thermal drift can degrade repeatability if the test is done during warm-up.
Useful acceptance benchmarks for a 3-axis vertical machining centre include:
Requesting a thermal test report without knowing the test conditions is not enough. The report should state spindle speed, run time, ambient temperature, sensor locations, and which axis was measured.
Thermal compensation is powerful, but it has clear limits. A buyer evaluating claims should understand where the technology does not provide a solution.
A buyer can use the following checklist when reviewing quotations, factory audit documentation, and acceptance test proposals. The goal is not to demand a specific compensation method, but to verify that thermal performance has been designed, measured, and documented.
Thermal growth cannot be eliminated from a machine tool, but it can be managed. The difference between a low-cost machine and a production-ready machine often appears only after 45 minutes of operation, when the cold-start advantage has faded. By checking thermal design, test data, and compensation behaviour before purchase, a buyer can avoid paying for accuracy that disappears as soon as the spindle warms up.
Download the benchmark spec sheet & subscribe to the terminology hub — contact the Gree CNC team.