CNC precision processing facts that decide your part

• 1
  CNC precision processing facts: what the cutter really removes
• 2
  Why setup count drives accuracy more than spindle speed
• 3
  Material behavior changes the tolerance you can hold
• 4
  Surface finish is a toolpath decision, not a polish problem
• 5
  What actually moves the price of a machined part
• 6
  How a precision job moves through the shop
• 7
  Process choice by part geometry

CNC precision processing is subtractive. A rotating cutter or a fixed turning tool removes material from a solid block until the remaining shape matches the CAD model. The geometry part is simple. The hard part is everything that moves between the tool tip and the workpiece: spindle bearings, ballscrews, thermal growth, fixture rigidity and the material's own springback.

That chain is why two shops can quote the same drawing and deliver different results. A machine rated to ±0.005 mm only holds that number when the setup is rigid, the tool is sharp, and the temperature in the room is stable. Move the same job to a fixture hanging off the table edge and the tolerance drifts before the first pass is finished.

So the useful question is not "how accurate is your CNC?" It is "which features on this part actually need the tight tolerance, and which ones can stay loose?" Engineers who answer that before quoting usually get lower prices and fewer rejected parts.

We machine prototypes and production runs at GreatLight in Dongguan on 127 high-precision CNC machines, including 16 simultaneous 5-axis centers. The numbers in this article come from that floor, not from a spec sheet.

Every time a part is unclamped and turned, a new datum is introduced. Each datum carries its own error. A part cut in one 5-axis setup keeps a single reference frame from start to finish. A part cut in four 3-axis setups stacks four alignment errors on top of each other.

That stacking is the quiet reason a simple bracket can miss a hole position by 0.05 mm while a complex impeller holds ±0.005 mm. The impeller never left the machine. The bracket was moved four times.

For parts with features on three or more faces, or with angular holes that meet at odd angles, 5-axis machining removes the repositioning risk entirely. For a flat plate with holes on one face, 3-axis is faster and cheaper. There is no prize for using more axes than the geometry needs.

A 3-axis job with a well-designed soft jaw and a single flip can still hold ±0.01 mm on most aluminium parts. That is often good enough, and it costs less.

• 1
  One setupBest accuracy, best for complex geometry and angular features.
• 2
  Two setupsCommon for plates; watch the flip datum carefully.
• 3
  Four or more setupsError stacks fast; consider redesigning the fixture or the part.

Aluminium 6061 and 7075 cut clean and hold tight numbers without much fuss. Stainless 316 and 17-4PH work-harden at the cut, so light passes and sharp tooling matter more than spindle speed. Titanium Ti-6Al-4V moves under heat and needs coolant pressure and slower feeds to stay in tolerance.

Thin walls are a separate problem. A 0.8 mm aluminium wall will deflect under clamping pressure and spring back after the clamp releases. The finished part looks right on the machine and measures wrong on the CMM. Machining it in two stages, with a stress-relief pause between roughing and finishing, is often the only way to hold the wall flat.

Plastics behave differently again. POM and PEEK expand with heat and need slower feeds and air blast rather than flood coolant. ABS and PC scratch easily, so the finish pass matters more than the roughing strategy.

The practical takeaway: pick the material for the function first, then tell your machinist which features are critical. We adjust the strategy around that, not the other way round.

Ra values come from the cutter marks left behind. A sharp tool, a small stepover and a stable setup produce Ra 0.8–1.6 μm directly off the machine. Getting to Ra 0.2–0.8 μm usually means a separate finishing pass with a smaller tool, sometimes followed by bead blasting or polishing.

Chasing a fine finish on a feature that does not need it is one of the most common ways to add cost without adding value. A sealing face needs Ra 0.8 μm or better. A bracket mounting surface at Ra 3.2 μm is fine. Mark the critical surfaces on the drawing and leave the rest as machined.

Anodizing and plating change the surface too. Hardcoat anodizing adds roughly 0.025 to 0.05 mm per side and can round sharp edges. If a bore is anodized after machining, the machinist needs to undersize it so the finished bore lands in tolerance.

Laser marking needs a minimum character height of 1.5 mm to stay legible after finishing. Smaller text fills in or fades.

Tolerance drives cost more than any other single factor. Going from ±0.05 mm to ±0.005 mm on a whole part can double the machine time because it forces slower feeds, more inspection and sometimes a second finishing pass. Tighten only the features that need it.

Setup count is the second lever. Each additional setup adds fixturing time, alignment time and risk. A part designed so that all critical features are reachable in one orientation is cheaper to make and more likely to pass inspection.

Material and finish are the third. Titanium and Inconel cost more to buy and more to cut. Hardcoat anodizing and electroless nickel add a plating step and a lead-time buffer. None of this is hidden, but it is easy to forget when the drawing only shows a tolerance block.

Quantity changes the picture too. There is no minimum order quantity here, so a single prototype and a 10,000-part run use different strategies. Prototypes run on 5-axis to avoid fixture cost. Production runs may move to a dedicated fixture and a 4-axis cycle to cut unit time.