Makino CNC Precision Engineering: How the Machine Holds Microns

• 1
  Makino CNC precision engineering starts with thermal stability
• 2
  Axis accuracy and the rotary table decide the real error budget
• 3
  Spindle dynamics and tool holding set the surface you actually get
• 4
  When Makino CNC precision engineering is worth the machine hour
• 5
  Verification: how you prove the tolerance was held
• 6
  Which machine class fits the part

A machine does not lose accuracy because a ball screw suddenly changes pitch. It loses accuracy because the structure grows. Cast iron and steel expand roughly 10–12 μm per meter per °C. On a 500 mm part, a 4 °C drift between the morning and afternoon shifts moves the tool about 2 μm relative to the workpiece. That is already a third of a ±0.005 mm band, and it happens before any cutting force is applied.

Makino machines address this with symmetric castings, cooled ballscrews, and spindle jackets that carry temperature-controlled oil. The design goal is not zero growth. It is growth that stays even across the structure, so the tool center point moves less relative to the table. A machine that grows 15 μm uniformly is easier to compensate than one that twists 5 μm.

What this means on the shop floor: a cold machine cuts differently from a warm one. The first 30 to 45 minutes of a shift are the least stable. For tight bores and flatness, we run a warm-up cycle and let the spindle reach steady state before touching the first feature. On a ±0.005 mm job, that step is not optional.

Ambient conditions matter too. A shop that swings 8 °C between day and night will fight the machine all year. We keep our grinding and five-axis cells in a controlled area, and we log temperature when a job is flagged as critical. If you are qualifying a new supplier, ask how they handle warm-up. A vague answer usually means they do not.

• 1
  Warm-up cycleRun 30–45 minutes before the first critical cut.
• 2
  Cooled ballscrewsKeeps screw growth close to the casting growth.
• 3
  Symmetric structureEven growth is easier to compensate than twist.

A linear axis is easy to describe and hard to hold. Positioning accuracy, repeatability, and backlash all stack on top of each other. A machine quoted at ±0.002 mm positioning still has to fight reversal error and servo lag when the tool changes direction. On a part with 40 direction changes, those small errors accumulate in the surface, not in the dimension.

Rotary axes are the harder problem. Every five-axis move is a combination of three linear and two rotary motions, and each has its own error. The pivot distance, the rotary encoder resolution, and the thermal state of the trunnion all feed into the final position. A Ø400 mm rotary table with a small angular error produces a larger linear error at the part edge. At 200 mm from center, 0.001° is about 3.5 μm.

This is why a five-axis machine is not automatically more accurate than a three-axis one. It removes setup error by cutting five faces in one clamping, but it adds kinematic error. The trade only pays off when the part has features on multiple faces and the tolerance is tight enough that re-fixturing would cost more accuracy than the rotary axes do.

For parts that fit in a single orientation, a three-axis machine with a rigid setup can hold ±0.005 mm with less risk. We run both. The decision is made per part, not per customer preference.

• 1
  Reversal errorShows up as surface marks, not dimensional drift.
• 2
  Rotary errorGrows with distance from the table center.
• 3
  Three-axis caseBetter when all features are reachable in one setup.

The spindle is where the machine meets the cut. Its stiffness, runout, and thermal growth determine whether you get Ra 0.2–0.8 μm or a chatter pattern. A spindle with 2 μm runout will cut a hole that is 2 μm off before any other error is counted. On a bearing bore, that is half the tolerance.

Tool holding matters as much as the spindle. A shrink-fit holder puts the tool closer to the taper and adds stiffness. A collet chuck is faster to change but adds runout. On deep pockets, the tool length becomes the weak point. A 4× diameter overhang can triple the deflection compared with a 2× overhang. We shorten tools and use stub drills whenever the geometry allows.

Cutting parameters follow from this. In 6061 aluminium, a 12 mm carbide end mill at 8,000 rpm and 0.08 mm per tooth removes material quickly with good finish. In 17-4PH stainless, the same tool at the same speed will burn the edge. We drop to 2,500–3,500 rpm and reduce feed per tooth. The machine can hold the tolerance either way. The tool decides whether the surface passes.

Coolant strategy is part of the same picture. Through-spindle coolant clears chips from deep pockets and keeps the cutting zone at a stable temperature. On titanium and Inconel, high-pressure coolant reduces tool wear and stops recutting. On a ±0.005 mm job, chip evacuation is an accuracy issue, not a housekeeping issue.

• 1
  Spindle runout2 μm runout eats half a tight bore tolerance.
• 2
  OverhangKeep it under 4× diameter on finishing passes.
• 3
  Through coolantClears chips and stabilizes the cut zone.

A high-end machine costs more per hour than a standard VMC. That cost has to buy something. It buys accuracy that survives a full production run, not just the first article. It buys surface finish that reduces hand polishing. It buys fewer setups on complex parts, which removes a whole class of fixture error.

The parts that justify it usually share three traits: tight tolerance, hard geometry, and a material that is difficult to cut. A titanium implant with bone-contact surfaces, an aerospace bracket with pockets on five faces, a mold insert with deep ribs and a fine finish. Those jobs punish a weak setup and reward a stable machine.

The parts that do not justify it are just as important to name. A flat plate with holes in one face does not need five axes. A bracket with ±0.05 mm tolerance does not need a thermal-controlled cell. A prototype with loose tolerances is better served by a fast three-axis setup, because the cost goes into delivery speed, not into microns.

We make that call during DFM review, before quoting. If a part can be held on a three-axis machine with a simple fixture, we say so. The right process is the one that meets the drawing with the least risk. Sometimes that is a five-axis Makino class machine. Sometimes it is not.

• 1
  Worth itTight tolerance plus multi-face geometry plus hard material.
• 2
  Not worth itSingle-face features and loose tolerance bands.
• 3
  DFM firstProcess choice is made before the quote, not after.

A machine that can cut to ±0.005 mm still needs a way to prove it did. Calibration is the starting point. Ballbar tests, laser interferometry, and rotary axis checks tell you the machine is in spec before the job starts. Without that, you are measuring the part and guessing at the cause.

CMM inspection is the usual answer for finished parts. But a CMM has its own uncertainty, and the measurement strategy changes the result. A hole checked with a two-point diameter is not the same as one checked with a full circular scan. On tight bores, we specify the strategy on the inspection plan so the number means something.

In-process probing closes the loop on long runs. Touch probes check a datum or a critical feature between operations and feed offset corrections back to the control. This catches thermal drift and tool wear before they become scrap. On a 500-piece run, that is the difference between a stable process and a sorting operation.

We inspect 100% of parts before shipment, with raw material checks, in-process monitoring, and final inspection. Reports go out on request. If a drawing calls for a first article inspection report, we build the plan around it rather than bolting it on at the end.

• 1
  Machine calibrationBallbar and laser checks before the job starts.
• 2
  Inspection planDefine the measurement strategy, not just the tolerance.
• 3
  In-process probingCorrects offset before the error becomes scrap.