You can measure machine repeatability in your own shop without a laser interferometer using a dial indicator, a test bar, and a repeated positioning cycle.
You can measure machine repeatability in your own shop without a laser interferometer using a dial indicator, a test bar, and a repeated positioning cycle. This shop test will not replace a full volumetric accuracy report, but it tells you whether a used European machine or a new Chinese machine holds position under your conditions before you sign the order.
The situation is common in a German Mittelstand shop. You are comparing a new Chinese vertical machining centre from a catalogue with a 12-year-old European machining centre from a dealer. Both quotes say repeatability is ±0.005 mm. You do not own a laser interferometer, and you do not want to spend €2,000–€5,000 on an external measurement service before you even buy. You can still make a useful measurement in 30–60 minutes. This article explains what to measure, what the numbers mean, and what to ask the supplier.
Accuracy is the difference between where the control says the axis is and where the tool actually is. Repeatability is the spread of positions when the axis returns to the same commanded point many times.
For batch production, repeatability often matters more than absolute accuracy. If a machine always stops 0.004 mm to the left, you can shift the work offset and compensate. If the machine stops in a different place every cycle, you cannot compensate. That spread is what a shop test can show.
Two repeatability terms matter in supplier data:
Ask the supplier which one they quote. A unidirectional figure of ±0.003 mm can hide a bidirectional figure of ±0.008 mm if backlash is poorly compensated.
You need three items:
Procedure:
A healthy machining centre in the 500–1000 mm travel class should show a range under 0.005 mm for a linear axis at one point. This is a typical industry range, not a guarantee. Check the official catalogue for the specific machine. If the range is over 0.010 mm, the machine cannot hold the repeatability figure quoted for that axis.
This test measures unidirectional repeatability at one point. It does not measure the whole travel, thermal drift, or squareness between axes.
Repeat the indicator test, but approach the point from the plus direction and then from the minus direction alternately.
For each direction, calculate the average approach position. The difference between the two averages is backlash. Backlash is the lost motion when the axis reverses, caused by clearance in the ball screw, nut, or guide. Modern controls compensate backlash electronically, but compensation cannot fix mechanical looseness.
Typical values:
These are typical industry ranges, not figures from any specific catalogue. Ask the dealer to measure backlash on the used machine in front of you. If the dealer refuses, treat the machine as unverified.
Run a short program that moves the machine through a typical tool path for 20–30 minutes. Use normal rapid traverses and feed rates. Then repeat the dial indicator test at the same target points.
Compare the cold reading to the warm reading. The difference is thermal drift. A machine expands as spindle, motors, and ball screws heat up. In a shop with temperature control of ±2 °C, a good machine should drift less than 0.010 mm on a 500 mm travel over 30 minutes. If your shop varies from 18 °C to 28 °C during the day, the drift will be larger.
Ask the supplier for the thermal stability specification, usually tested to ISO 230-3. If they cannot provide a thermal test report, assume the drift figure is unknown until you measure it yourself.
Use these questions when you compare a new Chinese machine with a used European machine. The same questions apply to both, but the evidence you need is different.
The framing matters. A 12-year-old European machine may have been built with high-quality castings and a rigid structure, but the ball screws, linear guides, and spindle bearings wear. A new Chinese machine may have fresh ballscrews and guides, but the assembly quality, thermal stability, and control tuning vary more between builders.
The table below shows what you can check in a shop test and what you should ask for from the supplier.
| Check | New Chinese machine | Used European machine | What it tells you |
|---|---|---|---|
| Unidirectional repeatability, 30 cycles, one point | Ask for FAT data; run SAT in your shop | Run 30-cycle dial indicator test on dealer floor | Current spread at one point under shop conditions |
| Bidirectional repeatability | Ask for ISO 230-2 bidir report | Run alternating-direction test yourself | Backlash and reversal error |
| Thermal drift over 30–60 minutes | Ask for ISO 230-3 thermal test report | Run warm-up test then repeat indicator test | Stability under your shop temperature and load |
| Full volumetric accuracy | Ask for laser interferometer report for all axes | Ask for last calibration certificate; if none, order a ballbar check | Overall positioning over full travel |
| Spindle runout and taper condition | Ask for factory test record | Measure runout with a test bar in the spindle | Tool holding quality and surface finish risk |
| Service and spare parts lead time in DACH/UK | Ask for local service response time and parts stock list | Ask for dealer support and availability of used spares | Downtime risk after a failure, and how realistic local support really is |
A dial indicator and 30 cycles give a useful signal, not a full accuracy picture. It is worth being clear about where the method breaks down, because a number on a gauge can feel more authoritative than it really is.
When a contract depends on tight tolerances across the whole working volume, the honest path is a laser interferometer or ballbar check to ISO 230-2. The shop test earns its place earlier: as a cheap way to reject a machine that is clearly bad before paying for the precise one.
Is a dial indicator reading the same thing as the ISO 230-2 repeatability figure? No. ISO 230-2 is a statistical result over several positions, several cycles, and both approach directions, with defined warm-up and temperature. The shop test is a single-point, mostly unidirectional sanity check. Agreement at one point is encouraging; it is not a certified figure.
How many cycles are enough to be useful? Thirty cycles at one position gives a stable picture of the spread at that point. Fewer than ten is hard to trust, because a single outlier can dominate the range. Five positions along each axis with five cycles each is a reasonable middle ground that still fits in an afternoon.
If both quotes say ±0.005 mm, why does a test matter? A brochure number is measured under controlled conditions in a specific direction, sometimes unidirectional. The bidirectional, warm, loaded, on-your-floor value can be two or three times larger. The test measures the machine as it actually arrives, which is the number production has to live with.
Does this replace the FAT or SAT? It does not. The factory acceptance test should still include full laser interferometer data for every linear axis, and the site acceptance test repeats the key checks after levelling on the buyer's foundation. The in-shop test is a pre-purchase screen, not a substitute for either acceptance gate.
One machine passes the test and one does not. Is the decision made? A failed or clearly worse test is a strong reason to walk away or renegotiate. A passed test only removes repeatability from the risk list; spindle condition, service history, control support, and total cost of ownership still decide the purchase. Treat the gauge as one vote, not the verdict.
Run the buyer self-check & book a machine selection call — contact the Gree CNC team.