Reading An ISO 230-2 Report: What The Numbers Tell You

Every machine we ship leaves with a serial-numbered report, and almost every buyer glances at the first number and files it. That is a waste, because the first number is rarely the one that governs whether the machine will hold your tolerance in production.

What the standard actually measures

ISO 230-2 is the test code for determining the accuracy and repeatability of positioning of numerically controlled axes. It is worth being precise about that wording: it measures how well an axis gets to a commanded position, one axis at a time, with nothing cutting. It is a geometric and positioning test, not a cutting test.

The method is straightforward. A laser interferometer is set along the axis, the axis is commanded to a series of target positions spread across the full travel, and the difference between the commanded position and the measured position is recorded at each one. Each target is approached repeatedly, and from both directions, because a machine frequently behaves differently arriving from the left than it does arriving from the right.

Positioning accuracy against repeatability

Two figures come out of that data and they answer different questions.

Positioning accuracy is how close the axis gets to the commanded position in absolute terms. It bundles together every systematic error in the axis: ballscrew pitch error, thermal growth during the test, scale errors, and any deviation the control has not compensated.

Repeatability is how tightly the axis clusters when it returns to the same commanded position over and over. It is a measure of scatter rather than offset.

The distinction matters commercially, and this is the part most buyers miss. A systematic positioning error can be compensated. If the axis is consistently 8 µm short at the same place, the control can be given a compensation table and the error largely disappears. Scatter cannot be compensated, because by definition it is not the same twice. A machine with mediocre positioning accuracy and excellent repeatability can be corrected into a good machine. A machine with the reverse cannot.

If you only have time to read one line of the report, read the repeatability figure for the axis that carries your critical dimension.

Reading the report line by line

A report that is worth anything gives a separate block of figures for every axis, not a single headline number for the machine. Within each block, look for:

• Both directions. Unidirectional and bidirectional figures should both appear. A bidirectional figure that is much worse than the unidirectional one points at backlash or a reversal problem, which shows up in your part as a step wherever the tool changes direction.
• The reversal value. Sometimes listed separately as backlash. On a machine that will do a lot of contouring, this figure matters more than the headline accuracy.
• The travel that was tested. Measuring a 200 mm window in the middle of a 1000 mm axis produces a flattering number that tells you nothing about the ends of the travel, where ballscrew wear and sag show up first.
• The number of target positions and runs. More targets and more repeat runs give a result you can rely on. A handful of points measured once is an indication, not a measurement.
• Ambient temperature. A steel ballscrew grows roughly 11 µm per metre per degree. A report with no stated temperature is an incomplete report.

What the report will not tell you

An ISO 230-2 report is a narrow document, and treating it as a complete statement of machine quality is the most common mistake we see. It says nothing about:

• Contouring accuracy. Two axes moving together, reversing at a quadrant, is a different problem. That is a circular test under ISO 230-4, usually with a ballbar.
• Thermal behaviour in production. The test is short and the machine is warm but not working. Spindle heat over an eight-hour shift is covered by ISO 230-3.
• Rigidity under cut. Nothing is being removed during the test. A machine can position beautifully and still chatter the moment it takes a real depth of cut, which is where the casting, the guideway type and the spindle earn their money.
• Surface finish. The only honest answer to a finish question is a cut sample in the material you actually run.

This is why every machine here also cuts a test piece that is measured rather than admired, and why we would rather cut your part than a standard test block if you send one. A generic test piece proves the machine against a generic tolerance. Your part proves it against yours.

A practical way to use it

When you are comparing two quotations, put the two reports side by side and check that they were produced the same way: same standard, same number of target positions, comparable travel, stated temperature. A better-looking number measured over a shorter travel at a friendlier temperature is not a better machine. It is a better-arranged test.

Then ask for the report of a machine already built rather than a specification sheet. A specification is a promise about a machine that does not exist yet. A serial-numbered report is a record of one that does.