How Accurate Are CNC Machine Tools? A 5-Check Shopfloor Method
Engineers ask a simple question: how accurate are CNC machine tools in my country, and how big is the gap with foreign-built machines? The honest answer is that the country label matters less than the repeatability test you run in your own shop. This guide gives you five checks you can run on any vertical or horizontal machining center, the values to expect, and the point where the gap stops mattering for your part.
In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
Key takeaways
How accurate are CNC machine tools: the three numbers that matter
When someone asks how accurate are CNC machine tools in a given country, they usually mean one number. In practice three numbers decide whether your part passes inspection. Positioning accuracy is how close the tool reaches a commanded point. Repeatability is how close it returns to the same point again and again. Thermal stability is how much both values move as the machine warms up over a shift.
Repeatability is the one to test first. A machine with ±0.010 mm positioning but ±0.003 mm repeatability will hold a boring diameter better than a machine with ±0.005 mm positioning and ±0.008 mm repeatability. The second machine looks better on the brochure and loses on the floor.
The ISO 230-2 test method defines how to measure these values with a laser interferometer or a ballbar. Ask any supplier for the raw test report, not a summary line. A report dated more than 18 months ago tells you little about the machine's condition today.
- 1Positioning accuracyDistance from commanded point to actual point along each axis.
- 2RepeatabilitySpread of results when returning to the same point 30 times.
- 3Thermal stabilityDrift measured from cold start to four hours of running.
What tolerance each class of machine can hold
A three-axis vertical mill with box ways and hand-scraped geometry typically holds ±0.02 mm on a 300 mm part. Add linear scales and a temperature-controlled room and that tightens to ±0.005 mm. A simultaneous five-axis machining center with a Ø400 mm rotary table adds rotary positioning error, so the effective tolerance on a trunnion part is usually ±0.010 mm to ±0.015 mm unless the table is calibrated.
High-speed spindles change the picture. At 20,000 rpm the spindle grows several micrometres from centrifugal force and heat. If your finish pass runs at 8,000 rpm but your roughing ran at 18,000 rpm, the tool tip position shifts between operations. Let the spindle idle at cutting speed for ten minutes before the finishing cut.
The practical floor for most job shops is ±0.005 mm with 100% inspection. Below that you are measuring in a metrology lab, not machining on the floor. If your drawing calls for ±0.002 mm on a 200 mm aluminium part, expect a grinding or lapping step after milling.
How big is the gap between regions, really
Ten years ago the gap was obvious. Machines from Japan, Germany and Switzerland held tighter geometry after five years of service. Today the hardware has converged. A modern machining center built in China, Taiwan, Korea or Europe uses similar linear guides, ballscrews and spindle bearings from the same handful of suppliers.
What still differs is the support layer. Metrology equipment, calibration frequency, coolant maintenance and operator training vary more between shops than between countries. A shop with a laser interferometer and a strict recalibration schedule will out-machine a better-equipped shop that skips maintenance.
So the honest answer to how big the gap is: on new machines, within a few micrometres. On ten-year-old machines, it depends on who maintained them. Ask for the calibration history, not the country of origin.
When a tighter machine is the wrong purchase
Buying accuracy you cannot measure is waste. If your incoming inspection uses a caliper with 0.02 mm resolution, a machine that holds ±0.003 mm gives you no verifiable benefit. The parts will be better, but you cannot prove it, and the customer cannot audit it.
The same logic applies to surface finish. A machine with a high-frequency spindle can produce Ra 0.2–0.8 μm on aluminium, but if the drawing calls for Ra 1.6 μm, you spent spindle hours for nothing. Match the process to the callout, then put the savings into inspection.
There is one exception. Regulated industries like medical devices and aerospace require documented process capability, not just good parts. There, the tighter machine plus a metrology report is the cheaper path, because it removes the rework loop and the audit findings that follow.
Step by step: five checks you can run this week
- 11. Warm up the spindle for 30 minutesRun the spindle at 60–70% of your finishing speed for 30 minutes with the axes cycling. Skip this and your first measurements include 10–30 μm of thermal growth. Log the ambient temperature before you start.
- 22. Measure bidirectional repeatabilityMount a dial indicator or test bar on the table and touch off a fixed point 30 times. Record the spread. Under 0.005 mm is a high-precision machine. Between 0.005 mm and 0.015 mm is normal production class.
- 33. Cut a test part with known geometryMachine a 100 mm × 100 mm square and a Ø50 mm bore in aluminium 6061. Measure with a calibrated micrometer and a bore gauge at three depths. Compare to the CAM nominal, not to the drawing tolerance.
- 44. Check backlash on each axisApproach a point from +X then from −X and compare the readings. More than 0.010 mm difference means the ballscrew or thrust bearing needs attention. Do this on all three linear axes.
- 55. Re-measure after four hours of cuttingRepeat step 2 at the end of a shift. If repeatability grew by more than 0.005 mm, the machine is drifting. Check coolant temperature, spindle cooling and the room's air conditioning.
- 66. Log everything in one fileKeep date, ambient temperature, spindle hours, and the three numbers. Trends over months tell you when to schedule recalibration better than any single reading.
Machine class versus achievable tolerance
Values assume a temperature-controlled room and a warmed-up spindle.
| Machine class | Typical repeatability | Best practical tolerance | Fit for |
|---|---|---|---|
| Three-axis box way mill | ±0.010 mm | ±0.020 mm | Fixtures, brackets, plates |
| Three-axis linear guide mill | ±0.005 mm | ±0.010 mm | Housings, manifolds, jigs |
| Four-axis with rotary table | ±0.008 mm | ±0.015 mm | Shafts, flanges, cross holes |
| Five-axis simultaneous | ±0.005 mm | ±0.010 mm | Impellers, medical, aerospace |
| Mill-turn center | ±0.005 mm | ±0.008 mm | Turned parts with milled features |
| Grinding or lapping step | ±0.002 mm | ±0.002 mm | Seal faces, bearing bores |
Questions engineers ask next
Is a ±0.005 mm machine always better than a ±0.010 mm machine?
No. Repeatability and thermal stability decide the outcome on a long run. A ±0.010 mm machine that holds its value for eight hours will outproduce a ±0.005 mm machine that drifts after two hours.
Check the ISO 230-2 report and the maintenance log before you compare catalog numbers.
How often should a machining center be recalibrated?
Most shops run a full laser interferometer check once a year and a ballbar check every three to six months. High-utilization five-axis machines with rotary tables benefit from a quarterly check.
After any crash, recalibrate before the next production run. Geometry does not heal itself.
Does the country of origin still matter for accuracy?
On new machines, the hardware gap is small. Linear guides, ballscrews and spindle bearings come from a narrow supplier base worldwide.
What differs is service response, spare parts availability and how strictly the builder calibrates before shipment. Ask for the pre-shipment test report.
Can I improve accuracy on an older machine without replacing it?
Yes, within limits. Reball the ballscrews, replace thrust bearings, re-scrape the ways and add linear scales. A retrofit can recover ±0.005 mm repeatability on a machine that had drifted to ±0.020 mm.
The retrofit costs a fraction of a new machine but takes the machine offline for two to three weeks.
What room temperature should a precision shop hold?
20 °C ±1 °C is the common target for tight work. Aluminium grows about 23 μm per metre per degree Celsius, so a 5 °C swing moves a 500 mm part by roughly 57 μm.
If you cannot control the room, control the part. Let it soak to room temperature before final measurement.
Send us your drawing and we will tell you what the machine can hold
Upload a STEP file and we return a quotation plus a free DFM analysis within 12 hours, with the tolerance class we would run and the inspection method behind it.
12-hour quote100% inspection±0.005 mm toleranceNo minimum order
More machining data and process notes
We publish setup notes, tooling trials and inspection data from the factory floor.
