Processing Accuracy Does Not Meet Standards? The Forgotten Machine Tool Preheating

• 1
  Why Processing Accuracy Does Not Meet Standards on the First Part
• 2
  What Actually Grows: Spindle, Ballscrew, and Column
• 3
  When Preheating Is Not the Fix
• 4
  How to Build a Warm-Up Routine That Fits the Job
• 5
  What We See on 127 Machines in Three Plants
• 6
  Thermal Drift vs Other First-Part Errors

A cold machine holds a different geometry than a warm one. On a Monday morning start, the spindle housing, front bearings, ballscrew and the bed casting are all at room temperature. Twenty minutes of cutting adds heat to each of them at different rates. The spindle nose grows forward, the ballscrew stretches along its length, and the column leans slightly. The part you cut at minute two is not the part you cut at minute forty.

This is why the failure looks intermittent. The operator checks the first part, sees a 0.03 mm error on a bore pitch, adjusts the offset, and the next three parts run clean. By lunch the offset is wrong again. Nothing in the program changed. The thermal state did.

The old page on this topic called preheating the forgotten step. That label still fits. Shops invest in spindle taper cleanliness and tool holder runout, then run a ±0.005 mm job on a machine that has not moved its axes under load since the previous evening.

The error does not appear on every part. It shows up on bores, on long-axis features, and on anything measured within the first ten minutes of the shift. That pattern is the fingerprint of a thermal problem rather than a mechanical one.

Steel expands about 11 to 12 μm per meter per degree Celsius. A 400 mm ballscrew on a mill-turn center, warmed 5 °C above ambient, grows roughly 22 μm along its length. That is more than the entire ±0.005 mm tolerance band on a tight bore. The screw is not damaged. It simply got longer.

The spindle grows in two directions. Axial growth pushes the tool tip away from the work, so a facing cut lands shallow. Radial growth at the front bearing changes the effective centerline, so a bored hole drifts off position. Front bearings usually run hotter than the rear, which tilts the spindle axis slightly. That tilt is small in angle and large in effect over a 300 mm part.

The column and bed react slower because they carry more mass. A cast iron column may still be moving thermally two hours after the spindle has stabilized. This is the reason a short warm-up can look successful and then fail at hour three.

Coolant matters too. Flood coolant pulls heat out of the cut zone but also chills the bed in the area under the nozzle. Uneven cooling bends the machine in a slow, unpredictable way. On high-accuracy work, coolant temperature control is part of the thermal budget, not a separate topic.

Thermal drift is not the only cause of a first-part miss. Before blaming the machine, rule out the tool. A holder with 0.02 mm runout will cut an oversized slot no matter how long you warm the spindle. Measure runout at the tool tip, not at the holder taper.

Fixture clamping is the second suspect. A part clamped on three points and shimmed on the fourth will spring when the vise releases. The error appears after unclamping, which makes it look like thermal growth in reverse.

Material stress relief is the third. A 7075 block hogged out at 8,000 rpm will move as internal stresses release. No warm-up routine fixes that. The answer is a roughing pass, a stress-relief pause, then a finishing pass.

If the error repeats at the same magnitude on every part, every shift, hot or cold, preheating will not help. That is a geometry or setup problem. Thermal error is recognized by drift, not by a fixed offset.

A warm-up is not a fixed thirty-minute ritual. It is a short program that loads the same axes and spindle speeds the real job will use, run until the machine's thermal state stops changing. For a 5-axis job with heavy trunnion motion, that may mean 40 minutes. For a 3-axis aluminum bracket at moderate speed, 12 minutes is often enough.

The routine should include spindle rotation at the job's target rpm, rapid moves across the full X and Y travel, and Z moves over the working height. Rotary axes need their own motion. A trunnion that sits still during warm-up is still cold when the first part starts.

Measure something after warm-up, not before. Touch off a known master or probe a reference feature, then set the work offset. If you set the offset on a cold machine and start cutting, the offset is stale before the second tool finishes.

For runs that last all day, a mid-shift recheck helps. Probe the reference feature again after four hours. If it moved more than half the tolerance band, the thermal state is still changing and the offsets need refreshing. This is standard on automotive and medical work at GreatLight, where a 99.99% qualification rate depends on it.

Across our three plants in Dongguan and Singapore, the machines that hold ±0.005 mm most consistently are not the newest ones. They are the ones with a written warm-up program stored next to the job program, and an operator who runs it before the first part.

The pattern is consistent. Machines left overnight in an air-conditioned shop at 22 °C need less warm-up time than machines in a bay that swings between 18 °C and 30 °C. Thermal stability of the room is worth as much as thermal stability of the spindle.

Direct sunlight is a real problem. A machine near a west-facing window can shift through the afternoon as the sun moves. We have moved machines for this reason alone. It looks like a small thing until a bore pitch drifts 0.02 mm at 3 p.m. every sunny day.

The practical takeaway: treat warm-up as part of the process sheet, not as an operator preference. If it is not written down, it will be skipped on a busy morning, and the first part will be the one that shows it.