
A “±0.005 mm” figure on a specification sheet is not a test result. It is a number with no defined method behind it until you know three things: which axis, which test standard, and whether it was measured on your serial number or copied from a catalogue.
A defensible CNC lathe acceptance test is a package of six checks, not one number:
| # | What is tested | Standard |
|---|---|---|
| 1 | Geometry, no load | ISO 230-1, applied through ISO 13041-1 (horizontal spindle) or -2 (vertical spindle) |
| 2 | Positioning accuracy and repeatability | ISO 230-2, applied through ISO 13041-4 |
| 3 | Circular interpolation | ISO 230-4 |
| 4 | Thermal distortion | ISO 13041-8, referencing ISO 230-3 |
| 5 | Cutting accuracy | ISO 13041-6 |
| 6 | Functional and safety verification | Machine-specific; CE file where applicable |
Geometry alone passes machines that drift when warm. Positioning alone passes machines with backlash in the contour. You need all six to know what you bought.
Why a Single Accuracy Figure Proves Almost Nothing
Three failure modes hide behind one decimal number.
1. The axis is undefined. Positioning accuracy is measured per axis, per direction. A figure quoted as “positioning accuracy ±0.005 mm” without saying whether it is X or Z, unidirectional or bidirectional, is a marketing statement. Bidirectional values are always worse than unidirectional, because they include backlash.
2. The method is undefined. ISO 230-2 defines how the test is run: the number of target points, the number of approaches, the direction sequence, and how the results are reduced. Change the method and the number changes without the machine changing at all.
3. The unit is undefined. A catalogue figure applies to a model. An acceptance figure applies to a serial number. Only the second one is evidence.
There is a fourth, subtler problem: accuracy and repeatability are different things. A machine can hit the target on average and still scatter around it. Repeatability is what production actually depends on, because a scatter of ±0.01 mm cannot be compensated away by an operator, while a consistent offset usually can.
The Six Tests, and What Each One Catches
1. Geometry under no load — ISO 230-1 / ISO 13041-1 and -2
This is the foundation: straightness of bed or column, squareness between axes, parallelism of spindle axis to guideways, runout of the spindle nose and centre taper.
The test code is ISO 230-1, geometric accuracy of machines operating under no-load or quasi-static conditions. For turning machines it is applied through ISO 13041-1:2020 for horizontal workholding spindles or ISO 13041-2:2020 for vertical workholding spindles, each specifying the geometric tests and the applicable tolerances (ISO 13041-1, ISO 13041-2).
What it catches: a machine that was not scraped and aligned correctly, or one whose column moved in transit.
What it misses: everything that happens once the spindle is turning and the machine is warm.
2. Positioning accuracy and repeatability — ISO 230-2 / ISO 13041-4
ISO 230-2 specifies the procedures used to determine the accuracy and repeatability of positioning of numerically controlled axes (ISO 230-2 sample). For turning machines, the positioning tests are set out in ISO 13041-4, covering linear and rotary axes.
Ask for the raw data, not the summary. A usable report shows, per axis and per direction: target positions, measured values, the mean deviation, the scatter, and the reversal value that quantifies backlash.
What it catches: leadscrew or linear-scale error, backlash, and non-repeatable servo behaviour.
What it misses: how the two axes behave together — which is the next test.
3. Circular interpolation — ISO 230-4
A circular test commands the machine to interpolate a circle and measures the deviation from true circularity. ISO 230-4 defines circular tests for numerically controlled machine tools.
This is the test most often skipped by buyers and most revealing to a machinist. A machine can pass geometry and pass single-axis positioning, and still cut an oval because the two axes do not accelerate and reverse in sympathy. Circularity error exposes servo mismatch, backlash in the reversing axis, and scale or pitch errors that cancel when each axis is tested alone.
What it catches: contour error on tapered and radiused work — exactly the features that decide whether a part fits.
What it misses: thermal behaviour.
4. Thermal distortion — ISO 13041-8 / ISO 230-3
A lathe is cold at 08:00 and warm by 11:00. The headstock grows, the ballscrew grows, and the tool-to-work relationship moves. ISO 230-3 defines the thermal distortion tests, including distortion caused by rotating spindles; ISO 13041-8 applies thermal distortion testing to turning machines (ISO 13041-8).
What it catches: the taper that appears on the third hour of a shift and vanishes after lunch. Buyers who only test a cold machine discover this in production, not in acceptance.
What it misses: nothing important, if you run it — but it takes time, which is why it is usually the first test cut from a schedule.
5. Cutting accuracy — ISO 13041-6
Geometry and positioning test the machine. A cutting test tests the system: machine, tooling, fixture, program and control together. ISO 13041-6 sets minimum requirements for assessing cutting accuracy on turning machines (ISO 13041-6).
The practical version: a test part with a facing cut, a diameter, a taper, a radius, a thread and a groove, measured and recorded. It is the only test that produces a number a production manager recognises.
What it catches: everything the previous four tests can miss in combination — thermal growth under real cutting load, tool deflection, control tuning.
6. Functional and safety verification
Not a single ISO number: interlocks, guards, emergency stop, coolant and lubrication systems, hydraulic chuck pressure, door interlocks, alarm behaviour, and — where the machine ships to the EU — the CE technical file built against the regime in force at completion. Regulation (EU) 2023/1230 becomes mandatory on 20 January 2027, replacing Directive 2006/42/EC (EUR-Lex).
For vertical machines, the accuracy standard is different again: ISO 3655 covers acceptance conditions for vertical turning and boring lathes with one or two columns and a single fixed or movable table (ISO). If you are buying a VTC, that is the standard to name in the purchase order, not ISO 13041-1.
Reading a Published Accuracy Figure Correctly
Take a real example. The VTC 900 lists positioning accuracy ±0.008 mm alongside a 900 mm maximum turning diameter and a 630 mm hydraulic chuck (VTC 900 specification).
That figure is a useful starting point, and it is still not acceptance evidence. Three questions convert it into one:
- Which axis, and which direction? X and Z behave differently; bidirectional includes backlash.
- Measured to which standard? ISO 230-2 via ISO 13041-4 is the defensible answer for a turning machine.
- On which serial number, at what ambient temperature, after what warm-up?
A supplier who can answer all three is selling an engineered product. A supplier who answers only the first is selling a brochure.
What to Physically Do at Run-Off
A witnessed test is not a tour. Bring these five things and the day produces evidence rather than impressions.
| Do this | Why it matters |
|---|---|
| Bring your own test-part drawing | Proves the machine can make your part, not a convenient demo part |
| Witness the test at operating temperature | Cold tests hide thermal drift; run a warm-up cycle first |
| Ask for raw data files, not a pass/fail sheet | You cannot audit a conclusion, only measurements |
| Photograph the setup and the readings | Disputes later are about what was measured, not whether it was |
| Run one repeated cycle and measure the drift | A two-hour repeat tells you what the machine does on a shift, not a demo |
Two more items belong in the purchase order rather than the visit:
- The acceptance criteria, by standard number and tolerance, agreed before the deposit — not “meets international standards”
- The remedy if a test fails: rework, replacement, or price adjustment, with a deadline
Machines rarely fail acceptance outright. They fail on the second visit, when the failed item was never written down the first time.
Where Wojie Fits
Wojie CNC builds slant-bed and flat-bed lathes, vertical lathes and vertical machining centres across six production bases, holds ISO 9001 and CE certification, and offers OEM/ODM configuration. Its castings are resin-sand with heat ageing and vibration treatment before machining — a foundry-side measure that matters for the geometric tests above, because a column that moves after assembly cannot be corrected by control tuning.
For buyers running the process described here, the practical request is simple: name the standard, ask for the recorded values on your serial number, and agree the remedy before the deposit.
FAQ
1. What is a CNC lathe acceptance test? A set of measured checks — geometry, positioning, circular interpolation, thermal distortion, cutting accuracy and functional safety — performed on the machine you are buying, against named ISO standards, with recorded values you can audit.
2. Which ISO standard applies to CNC lathe accuracy? For horizontal-spindle turning machines, ISO 13041-1 (geometry) and ISO 13041-4 (positioning); the measurement method comes from ISO 230-1 and ISO 230-2. Thermal tests are ISO 13041-8 and cutting accuracy is ISO 13041-6. Vertical turning lathes use ISO 3655.
3. What is the difference between positioning accuracy and repeatability? Accuracy is how close the average position lands to the commanded target. Repeatability is how tightly repeated approaches cluster. Repeatability usually matters more in production, because a consistent offset can be compensated while scatter cannot.
4. Why is a circular test necessary if each axis passes? Because axes fail together in ways they do not fail alone. Circular interpolation exposes servo mismatch and reversal backlash that single-axis positioning tests cancel out — and those errors appear as ovality on tapered and radiused work.
5. How long should an acceptance test take? Long enough to include a warm-up and a thermal drift check. A test completed on a cold machine in ninety minutes proves geometry, not production capability.
6. Should I accept a certificate of accuracy? Only if it names the standard, the axis, the test conditions and your serial number. A generic model certificate is a marketing document, not evidence.
7. Can I reject a machine that fails one test? Only if the criteria and the remedy were agreed in writing beforehand. Put the standard numbers, tolerances and consequences in the purchase order, not in an email after the visit.
Summary
- One number is not a test. Accuracy needs an axis, a method and a serial number.
- Six checks make a package: geometry, positioning, circularity, thermal, cutting, functional.
- Name the standard in the PO — ISO 13041-1/-4/-6/-8 for horizontal lathes, ISO 3655 for vertical.
- Test warm, not cold, and ask for raw data.
- Agree the remedy before the deposit, or a failed test becomes a negotiation.
The purpose of acceptance is not to catch a bad machine. It is to make the machine’s condition a documented fact on the day it changes hands — so that every conversation afterwards starts from data.

