Measuring Machine Environment: What Actually Moves Your Numbers
A CMM does not measure a part. It measures a part plus the room it sits in. This page explains how temperature, vibration, humidity, power and air reach the probe, and where correction stops helping. Written for engineers who have to sign off on a dimensional report.
In this article
- 1
- 2
- 3
- 4
- 5
The measuring machine environment is a metrology input, not a background detail
Every CMM reading is a comparison between a probe position and a scale. Both sit in air. When that air changes, the steel and aluminium and granite around it change size, and the machine has no way to know the part moved unless someone told it.
The industry reference is 20 °C. ISO 1 fixes it, and every geometric tolerance on a drawing is written as if the part and the machine were both at that temperature. A shop at 24 °C is not measuring the same part that a shop at 20 °C measures. The difference is small on a 20 mm bore. It is not small on a 900 mm frame.
This is why a measuring machine environment is specified as a set of numbers, not a feeling. Temperature, its rate of change, its gradient across the room, floor vibration, humidity, supply voltage and compressed air quality all have limits. Stay inside them and the stated accuracy holds. Drift outside them and the error budget is gone before the probe touches anything.
We machine to ±0.005 mm and inspect 100% of parts before shipment. That number only means something when the inspection room is stable enough to support it.
Temperature and gradient: the largest error source in any measuring machine environment
Temperature is the biggest single factor. Aluminium expands about 23 μm per metre per °C. Steel is roughly 11 μm, and granite around 7 μm. A 500 mm aluminium part that warms by 3 °C from handling grows about 35 μm. That is seven times the tolerance we hold on a tight feature.
Room temperature alone is not the whole story. Three separate numbers matter: absolute offset from 20 °C, short-term fluctuation, and the vertical or horizontal gradient across the measuring volume. A room sitting steadily at 21 °C is usually easier to correct than a room averaging 20 °C that swings 1.5 °C every twenty minutes.
Gradients are worse than offsets because they bend the machine. A 1 °C difference between floor and ceiling tilts a column over its height. The CMM scale reads a straight line that is no longer straight. No software correction fixes a bent geometry.
Modern machines carry temperature compensation. Sensors track the scale and the part, and the controller applies the thermal expansion coefficient of the material. This works well for slow, uniform change. It cannot follow a cold part pulled from a cold shop into a warm room and measured immediately.
Vibration: the error you cannot see on the display
Vibration enters through the floor. Stamping presses, air compressors, surface grinders and forklifts all send energy into the slab. The CMM frame responds, the probe follows, and the reported point sits somewhere other than where the stylus actually contacted the surface.
Low-frequency vibration is the hardest to detect. A few micrometres of movement at 5–20 Hz will not look like noise on a form plot, but it will shift a diameter measurement. Operators often blame the probe when the floor is the problem.
Isolation helps, but it is not a licence to ignore the source. Pneumatic isolators work well above their natural frequency and poorly below it. A press running at 8 Hz next to an air-bearing CMM is a layout problem, not a mounting problem.
Practical checks are simple. Put a glass of water on the granite and watch the surface. Better, run a repeatability test on a reference sphere at different times of day. If the spread widens when the press department starts up, you have your answer.
Humidity, compressed air and power supply in the measuring machine environment
Humidity is a secondary factor until it is not. Below about 40% RH, bare steel gauge blocks, styli and guideways can oxidise. Above roughly 60% RH, condensation forms on cold surfaces when the room warms. Both change contact geometry at the micron level.
Compressed air matters more than most people expect. Air bearings need clean, dry, oil-free air at stable pressure. Water carryover causes stick-slip, and pressure droop changes the bearing gap. A filter that has not been changed in a year is a measuring error waiting to happen.
Power supply affects the controller, the computer and the scale electronics. Voltage sag, frequency drift and poor grounding all show up as unstable readings or random faults. Dedicated circuits, proper earthing and, where the mains are dirty, an uninterruptible supply are standard practice.
None of these factors is expensive to control. They are easy to forget because they fail slowly.
Typical limits for a measuring machine environment
Values below reflect common practice for a temperature-controlled inspection room. Always follow the machine builder's specification first.
| Factor | Typical limit | What it affects | Warning sign |
|---|---|---|---|
| Air temperature | 20 °C ± 1 °C | Scale and part expansion | Daily reading drift |
| Short-term change | < 0.5 °C per hour | Repeatability over a run | Spread grows during a shift |
| Gradient | < 1 °C per metre | Straightness and squareness | Form error on long parts |
| Floor vibration | < 2 μm/s RMS | Probe contact point | Worse results near presses |
| Relative humidity | 40–60% RH | Oxidation and condensation | Rust on gauge blocks |
| Air supply | Clean, dry, stable | Air bearing gap | Stick-slip on slow moves |
| Mains power | Stable, well earthed | Controller and scales | Random faults and resets |
When correction is enough, and when it is not
For slow, uniform temperature change on parts that have soaked for hours, software compensation is reliable and you can run at 20 °C ± 2 °C. For gradients, fast changes, or a cold part measured right after machining, no correction saves the reading — let the part soak, fix the airflow, or move the measurement.
Frequently asked questions
How long should a part soak before measurement?
The rule of thumb is 1 hour per 25 mm of section thickness for steel, and roughly double that for aluminium because its thermal conductivity is higher and it follows the room faster.
A 100 mm aluminium housing needs several hours in the inspection room before its core reaches room temperature. Touching it with bare hands adds local heat that takes minutes to fade.
Can a CMM be used on the shop floor?
Yes, if the machine is built for it and the environment is managed. Production CMMs typically use a wider temperature band and heavier isolation.
The trade-off is accuracy. A machine rated at 20 °C ± 1 °C will not hold that number on a floor that swings 8 °C across a shift.
Does temperature compensation fix a cold part?
No. Compensation assumes the part is in thermal equilibrium with the room and applies the material coefficient.
A part that is still cooling has a temperature gradient inside it. The sensor reads the surface, not the core, so the model is wrong by an unknown amount.
What vibration level is acceptable?
For a high-accuracy CMM, keep floor vibration below about 2 μm/s RMS in the 1–100 Hz band. Some builders specify tighter limits.
Measure it with a seismometer at the machine base, not at the wall. The slab and the frame path matter.
Is humidity really worth controlling?
It is the cheapest factor to control and the easiest to ignore. Keep it between 40% and 60% RH.
Below 40% you invite oxidation on steel gauges. Above 60% you invite condensation when the room warms after a cold night.
What records should we keep?
Log room temperature and humidity continuously, and record vibration checks at commissioning and after any layout change.
When a measurement is disputed, these logs are what let you separate a real part deviation from a room problem.
Send us your drawing and inspection requirement
We machine to ±0.005 mm and inspect every part before shipment, with reports on request. Tell us the tolerance and the feature, and we will tell you how we hold it.
12-hour quote100% inspectionNo minimum order quantity
More from the shop floor
We publish setup notes, tooling trials and inspection data from the factory floor.
