Improve the performance of UK CNCs through regular maintenance

• 1
  Improve UK CNCs through regular maintenance: where accuracy actually goes
• 2
  Spindle and ballscrew checks that catch backlash early
• 3
  Thermal drift and geometry checks with a warm machine
• 4
  Daily and weekly routine that keeps the tolerance
• 5
  Symptom, likely cause and correct response

A CNC does not lose accuracy all at once. It drifts. The first sign is usually a small shift in a bore diameter that holds for a week and then moves again. On a machine cutting to ±0.005 mm, that drift is often 0.01 mm or more before anyone notices, because the part still fits and the drawing still passes at the bench.

Three mechanisms cause most of it. Ballscrew and guideway wear adds backlash that the control cannot fully compensate. Spindle bearing preload relaxes as grease breaks down, so the tool deflects more under cut. Thermal growth moves the whole structure as the spindle and ballscrews warm up. Wear is slow and predictable. Thermal drift is fast and repeats every morning.

That split matters for planning. Wear is caught by measurement and fixed by adjustment or replacement. Thermal drift is caught by warm-up routines and by keeping the machine running through the shift rather than stopping and starting. A maintenance schedule that only greases and wipes will miss the second category entirely.

There is also a fourth cause that is not mechanical at all: swarf and coolant mist settling on the scale and the linear rail. A 0.02 mm chip under a way cover is enough to shift a finishing pass. Cleaning is not cosmetic work on a machine held to ±0.005 mm. It is part of the accuracy budget.

Start with a spindle runout check using a dial test indicator on a clean taper. On a machine in good condition, runout stays inside 0.005 mm at the gauge line. If it reads 0.01 mm, the taper needs a clean and re-check before you blame the bearings. Embedded chips in the taper seat are more common than worn bearings.

Backlash on each axis is the next check. Command a 0.05 mm move and read the actual position from the scale or the indicator. Repeat in both directions five times. Backlash that is consistent and under 0.005 mm is normal on a ballscrew with preloaded nuts. Backlash that grows during a shift points to thermal movement, not to a worn nut.

Grease condition tells you more than grease quantity. Pull the spindle and push a small amount of grease onto a gloved finger. Dark, gritty grease with a metallic sheen means bearing wear is already happening. Clean, slightly amber grease means the bearings are fine and you can extend the interval. Change the grease on condition, not only on a calendar.

Check the ballscrew thrust bearings last. A thrust bearing losing preload shows up as a step in the surface finish right where the axis reverses. You will see it as a mark every time the tool changes direction, not as a dimensional error. That finish mark is the earliest warning you get, and it appears long before the part goes out of tolerance.

Geometry checks are only valid on a warm machine. A cold machine that reads square will read differently after three hours of cutting. Run the spindle at 8,000–10,000 rpm for 30 minutes before you measure squareness or parallelism. If your production starts cold every morning, that is where your first-hour scrap comes from.

Measure the machine, not the part, when you chase a dimensional problem. Cut a test piece, let it cool to room temperature, then measure. Aluminium 6061 grows about 23 μm per metre per degree Celsius. A part measured hot on the machine can read 0.02 mm larger than the same part measured an hour later on a granite table.

Coolant temperature is part of the thermal loop. A chiller running 2–3 °C above ambient is stable. A chiller that swings 8 °C drives the ballscrew and the workpiece together, and no amount of machine leveling will fix it. Check the chiller setpoint and the actual tank temperature on the same day you check the machine.

For long parts, check the machine travel rather than the part. A 4,000 mm axis will show thermal bow that a 600 mm test piece never reveals. If you cut long components in 6061 or 7075, run the same test bar at the start and end of a shift and compare the two readings. The difference is your real process capability.

Daily work is short. Wipe the taper, clear the way covers, check coolant concentration with a refractometer, and confirm the tool setter reads the same on a known tool. That is 15 minutes. It prevents most of the faults that turn into a half-day of rework later.

Weekly work adds measurement. Cut a test piece in the material you actually run, let it cool, and record the result in a log. A single number per week per machine is enough. When the number moves 0.01 mm over three weeks, you have a trend and you can plan the repair instead of reacting to a rejected batch.

Monthly work is where you check the air supply, the chiller, and the level of the machine. Water in the air line reaches the spindle and the tool changer, and it causes intermittent faults that are hard to trace. Drain the receiver and check the filter element. On a machine with a Ø400 mm rotary table, also check the table clamp force.

Keep the log in one place and keep it simple. Machine number, date, test piece result, coolant reading, and any adjustment made. A maintenance log that takes two minutes to fill in gets filled in. A twenty-field form does not, and then the history is gone when you need it most.