CNC Machine Installation Services: How to Choose and Verify
Machine installation is a measured process, not a delivery. These checks cover foundation, leveling, geometry, power, spindle run-in and acceptance cutting, so an engineer can judge a supplier before signing the report.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
Key takeaways
Match the installation scope to your machine
Scope grows with axis count, table size and spindle speed. Use this to size the work before asking for a price.
| Machine type | Typical scope | Time on site | What to verify |
|---|---|---|---|
| 3-axis VMC, 500 × 500 × 450 mm | Level, align, power, run-in | 2–3 days | Squareness, spindle runout |
| 4-axis mill with Ø400 mm rotary table | Add rotary axis alignment | 3–4 days | Rotary centerline to X/Y |
| 5-axis simultaneous center | Add kinematic calibration | 5–8 days | RTCP error, pivot distance |
| Large gantry, 4,000 mm travel | Add foundation survey, rail splice | 7–12 days | Straightness over full travel |
| Mill-turn center | Add sub-spindle coaxiality | 4–6 days | Main/sub spindle alignment |
What professional CNC machine installation actually includes
Installation starts before the truck arrives. The anchor plan, concrete cure time and floor flatness decide how much leveling work is left on site. On a 4,000 mm gantry, a floor that drifts 0.05 mm/m across the footprint will fight you for days. Survey first, pour second.
On site the sequence is fixed: rough level, wait for the casting to settle against a stable base, then fine level. We hold fine level to 0.02 mm/m on the bed. Skipping the settle step is the most common shortcut, and it shows up two weeks later as a twisted bed and a part that tapers.
Geometry comes next. Squareness between X and Y, parallelism of the Z axis to the column, and for anything with a rotary table, the centerline of that table against the linear axes. On a Ø400 mm rotary table, a 0.02 mm centerline offset becomes visible on any part machined at the table edge.
Only after geometry is signed off does the machine get power, air and coolant. Then spindle run-in and a test cut. If the acceptance part is not measured and logged, the installation is not finished, no matter what the checklist says.
- 1Foundation and anchorsFlatness survey, cured concrete, correct anchor torque.
- 2LevelingRough level, settle, fine level to 0.02 mm/m.
- 3GeometrySquareness, parallelism, rotary centerline, kinematic calibration.
- 4Utilities and run-inPower quality, air supply, spindle break-in, acceptance cut.
How to judge a CNC machine installation supplier
Ask what they measure, not what they promise. A supplier who quotes leveling to 0.02 mm/m and hands over the raw readings is telling you something. One who only says "leveled and aligned" is not.
Tooling matters as much as method. Granite squares, electronic levels, laser interferometers and ballbar kits are what turn a claim into a number. If a team arrives with a spirit level and a dial indicator for a 5-axis center, the kinematic calibration step will be skipped.
Scheduling honesty is a fair test. A 5-axis center needs 5 to 8 days on site once power and air are live. A quote that says two days for the same machine is either compressing the alignment or planning to leave parts of it out.
Finally, ask who signs the acceptance report and what happens if the test part fails. A written re-work clause costs nothing to include and tells you whether the supplier expects to pass on the first cut.
- 1Numbers, not adjectivesLeveling records, runout figures, backlash values, ballbar plots.
- 2Right instrumentsLaser interferometer and ballbar for multi-axis work.
- 3Realistic schedule5–8 days for simultaneous 5-axis, not two.
- 4Written re-work termsWhat happens if the acceptance part is out of tolerance.
Common traps in DIY or unsupervised installation
The first trap is power. Voltage that sags under spindle load causes alarms that get blamed on the control. Measure the supply under load, not at idle, and check the ground resistance before the first run.
The second is thermal. A machine leveled in a cold shop and run in a warm one moves. Castings and concrete both respond to a 10 °C swing. Log bed temperature during leveling and again after 8 hours of running.
The third is the rotary axis. It is easy to bolt down and hard to align. An unaligned table passes a static check and fails on a part cut 150 mm from center, where the error is multiplied by the radius.
The fourth is documentation. Without a baseline, every future accuracy complaint becomes an argument. Record the numbers on day one and the next service visit has something to compare against.
- 1Idle voltage looks fineLoad the spindle and re-measure.
- 2Cold-shop levelingRe-check after the machine reaches running temperature.
- 3Rotary axis bolted, not alignedCheck centerline against X and Y before cutting.
- 4No baseline recordLog geometry numbers at handover.
What the acceptance test should prove
A good acceptance part is not a simple block. It should touch every axis, include a circular interpolation move, and have features at different distances from the rotary center so radius error shows up.
Measure the part on a calibrated CMM, not on the machine that made it. That sounds obvious and gets skipped when a schedule is tight. A machine checking its own work proves nothing about absolute accuracy.
Compare the results against the tolerance you need to hold in production, not the machine's published spec. If you need ±0.005 mm on a 200 mm part, the acceptance part should be sized and toleranced to show that.
Keep the part. It becomes a reference for future service visits and a physical record of what the machine could do on installation day.
- 1Exercise every axisInclude circular interpolation and off-center features.
- 2Measure off-machineUse a calibrated CMM, not the machine under test.
- 3Tolerance the test part to your jobNot to the builder's brochure.
- 4Retain the partFuture service visits get a real baseline.
When professional installation is worth the cost
Single 3-axis machines with a stable floor are the easiest case to handle in-house. If you have a laser level, a granite square and someone who has done it before, the risk is manageable.
Anything with four or more axes, a rotary table over Ø400 mm, or a travel beyond 1,000 mm is a different job. The alignment chain gets longer and errors compound along it. That is where a specialist pays for itself.
If the machine will run medical or automotive production under ISO 13485 or IATF 16949, the installation records become part of your quality system. Auditors ask for leveling data and acceptance results. A supplier who documents the work saves you that effort.
If the machine is the constraint in a production line, downtime is the real cost. Bringing in a team that has the instruments and has done the sequence before shortens the ramp from power-on to first good part.
- 1In-house is fineSingle 3-axis, stable floor, experienced staff.
- 2Bring in a specialist4+ axes, large rotary tables, long travel.
- 3Regulated industriesInstallation records feed the quality system.
- 4Line constraintFaster ramp to first good part.
Step by step: a verifiable installation
Use this order. Reversing leveling and alignment wastes days.
- 11. Survey the floor and anchorsCheck flatness across the full footprint and confirm concrete cure. Target under 0.05 mm/m variation before the machine lands.
- 22. Rough level the bedBring the bed within 0.1 mm/m using the leveling pads. Do not chase final numbers yet.
- 33. Let it settle 24 hoursLeave the machine under its own weight. Re-check and adjust pads if the reading drifted.
- 44. Fine level to 0.02 mm/mWork pad by pad, checking in both axes. Lock the pads and re-read.
- 55. Align geometrySet X–Y squareness, Z parallelism, then rotary centerline. Log every reading with the instrument used.
- 66. Connect power, air and coolantMeasure supply voltage under load and ground resistance. Confirm air pressure and dew point at the machine inlet.
- 77. Run in the spindleFollow the builder's speed ladder. Watch for temperature rise and vibration at each step.
- 88. Cut and measure the acceptance partMachine a test part that exercises all axes. Measure it and compare against the agreed tolerance. Sign off only if it passes.
Frequently asked questions
How long does CNC machine installation take?
A 3-axis vertical machine with a 500 × 500 × 450 mm envelope typically needs 2 to 3 days on site once power and air are available. A 4-axis mill with a Ø400 mm rotary table runs 3 to 4 days.
A simultaneous 5-axis center needs 5 to 8 days because kinematic calibration is added to the geometry work. A large gantry with 4,000 mm travel can take 7 to 12 days, mostly for foundation survey and rail alignment.
Does the foundation really matter for a small machine?
Yes, but the tolerance is looser. For a compact machine, floor variation under 0.05 mm/m across the footprint is usually enough, provided the concrete is fully cured.
The failure mode is not the machine sinking. It is the bed twisting as one corner settles differently from the others. That shows up as taper or as a part that is good at one end of the table and out at the other.
What instruments should the installation team bring?
At minimum: an electronic level good to 0.02 mm/m, granite squares, dial indicators with magnetic bases, and a laser interferometer for linear and angular positioning.
For any machine with a rotary or tilting axis, a ballbar kit is the practical way to check circularity and kinematic error. If the team cannot name these instruments, ask how they plan to verify the geometry.
Can we install the machine ourselves and call a specialist for alignment only?
It is common and it can work. The split usually lands after fine leveling: your team handles the rigging, pads, power and air, and a specialist does geometry and acceptance.
The risk is in the handover. If leveling was done in a cold shop or without letting the machine settle, the alignment team inherits a moving target. Fix the level first, then book the specialist.
What should be in the handover package?
Leveling records with the instrument and resolution used, squareness and parallelism readings, rotary centerline offset, spindle runout, backlash values, and the acceptance part measurement report.
Also include the machine's baseline geometry numbers and photos of the anchor and pad installation. These are the documents you will reach for when accuracy drifts two years later.
How does installation affect the tolerance a shop can hold?
Geometry error does not stay local. A machine out of square by 0.03 mm/m accumulates 0.03 mm of error over a 1,000 mm move. On a long part, that can consume most of a ±0.005 mm budget before the tool touches the material.
Good installation also holds thermal behavior steadier, because the bed is not fighting a twisted base. That makes the first-part and last-part measurements closer together over a long run.
Plan the installation before the machine ships
Send us the machine model, envelope and floor survey. We will come back with a scope, a realistic schedule and the acceptance criteria in writing.
12-hour quoteFree DFM analysis100% inspection