Warm Warming Up Forgotten: Why a Cold CNC Machine Loses Precision

• 1
  Warm warming up forgotten: what actually moves when a machine is cold
• 2
  When the effect is large enough to matter
• 3
  The warm-up routine engineers actually run
• 4
  Three mistakes that keep the problem alive
• 5
  When warm-up is essential, useful, or unnecessary

Cast iron and steel grow when they get warmer. The growth is small per degree but it is not small per meter. A ballscrew on a 750 mm X axis adds roughly 8.5 μm for every 1 °C rise across that length. From a 20 °C Monday morning shop to a 26 °C afternoon, that is about 50 μm of length change waiting to happen before the tool ever touches metal.

The spindle is the second source. It runs at 8,000–15,000 rpm, and the front bearing pair can be 4–8 °C hotter than the column casting within the first hour. That heat travels down the spindle nose and into the Z slide. The tool center point moves. On a vertical mill the shift is mostly along Z, which reads directly as a depth error on a pocket floor.

The third source is the structure itself. Bed, column and table warm at different rates because they carry different loads. A machine that has been idle for three days is not in thermal equilibrium with the shop floor; it is in equilibrium with whatever the night temperature was. Warm warming up forgotten simply means the first good part of the day is cut while the geometry is still drifting.

Put together, these three effects explain why a shop can hold ±0.005 mm on Tuesday afternoon and miss it at 9 a.m. on Monday. Nothing broke. The machine was not ready.

• 1
  Ballscrew growthAbout 8.5 μm per 1 °C over a 750 mm axis.
• 2
  Spindle bearing heatFront bearings 4–8 °C above the column casting.
• 3
  Structural lagBed, column and table reach equilibrium at different times.

Thermal drift is a length problem, so it scales with the dimension you are cutting. A 10 mm slot on a tolerance of ±0.05 mm will usually survive a cold start. A 600 mm bore pattern on ±0.02 mm will not. If the tightest feature on the print is under 50 mm, the geometry error from a cold spindle is often lost in the noise of tool wear and fixturing.

The second variable is the tolerance band itself. At ±0.005 mm, which is what our own machines are held to, a 20 μm thermal shift is four times the whole tolerance. At ±0.1 mm the same shift is background. The question is never whether the machine drifts; it is whether the drift is a meaningful fraction of the band.

The third variable is the wait. A machine stopped for one lunch break is still close to equilibrium. A machine stopped for a long weekend, a plant shutdown, or a move between buildings is not. The longer the idle time, the further the starting point sits from operating temperature, and the more of the first shift is spent chasing a moving target.

Humidity matters less than temperature but it is not nothing. A shop that swings from 30% to 70% relative humidity changes the way coolant evaporates from the part, which changes the part temperature, which changes the finished size on a thin wall. Thermal control is not only about the machine.

The standard practice is a spindle warm-up cycle before the first production part. Run the spindle in steps, typically 25% of maximum speed for 5 minutes, 50% for 5 minutes, 75% for 5 minutes, then 100% for 5 to 10 minutes. The steps matter more than the total; a single jump to full speed heats the bearing race unevenly and can shock the preload.

Axis warm-up comes next. Command a repeat program that moves all three linear axes plus any rotary axes through most of their travel. A diagonal move across the work envelope is the usual pattern because it exercises X, Y and Z together. Ten to fifteen passes at moderate feed is enough to distribute lubricant and bring the screw and guide surfaces up to a steady state.

The rule of thumb many shops use: if the machine has been idle more than a few days, allow 30 minutes or more before the first tight-tolerance cut. If it has been idle overnight, 10 to 15 minutes is often sufficient. If it runs continuously, a short cycle at the start of each shift keeps it stable.

Do not skip the measurement step. Cut a test feature, measure it, and compare against the last known good value. If the number has moved, the machine is still warming. If it repeats, you are ready.

• 1
  Spindle steps25% / 50% / 75% / 100%, 5 minutes each.
• 2
  Axis sweep10–15 diagonal passes across the work envelope.
• 3
  Idle ruleDays idle: 30 minutes or more before tight work.
• 4
  VerifyCut a test feature and compare to the last good value.

The first mistake is treating warm-up as a spindle-only task. Shops that spin the spindle for ten minutes and then cut a long part often see the error appear halfway through the run, when the screw and the structure finally catch up. The spindle is the fastest heat source, not the only one.

The second mistake is measuring too early. A first-article inspection taken five minutes after the warm-up cycle will pass or fail on a machine that is still moving. Inspection should follow the warm-up, not overlap it. On a ±0.005 mm job, a 15-minute gap between the last warm-up pass and the first measured cut is not wasted time.

The third mistake is ignoring the shop. A machine in a bay next to an open loading door sees a different thermal environment than one in the middle of a climate-controlled floor. If the ambient temperature swings more than 2–3 °C over a shift, no warm-up routine will hold the tightest tolerances for a full day.

The fix in each case is the same: understand which heat source dominates your process, and design the routine around it. There is no universal warm-up time. There is a thermal budget for each machine and each tolerance band.