What a CNC Processing Expert Actually Controls

• 1
  What the title means on a shop floor
• 2
  Tolerance stack-up decides the process, not the reverse
• 3
  Where the machine stops and the setup begins
• 4
  Runout, tool life, and the finish you can actually promise
• 5
  Inspection closes the loop, and it has to be planned
• 6
  Which process route fits which part

A CNC processing expert sits between the CAD model and the machine. They read the drawing, decide the order of operations, pick the stock, and set the datums that every later cut will follow. On a simple bracket that takes an hour. On a 400 mm aluminum housing with three bores that must stay coaxial, it takes most of a day before any metal moves.

The distinction matters to buyers. Two shops can own the same machine model and quote the same part at very different risk levels. The difference is usually not spindle speed. It is whether someone checked tool runout, planned a re-chuck strategy, and knew which feature would be measured first.

In our plants in Dongguan and Singapore, 127 high-precision CNC machines run across three sites. Machine count sets capacity. Process decisions set whether the parts that come off those machines are usable.

A useful test: ask for the setup sheet, not the machine list. The setup sheet shows datum choice, workholding, tool sequence, and in-process checks. That document is where expertise becomes visible.

• 1
  Drawing and datum reviewWhich face locates the part, and which faces are free.
• 2
  Operation sequencingWhat gets cut before the part loses its reference surfaces.
• 3
  Tool and fixture choiceReach, rigidity, runout, and how the part is held.
• 4
  Measurement planWhich features are checked, with what instrument, at which stage.

A single ±0.05 mm callout is easy on almost any three-axis mill. The trouble starts when several tolerances chain together. If a bore position, a pocket depth, and a face flatness all reference the same datum, the errors add. A process plan that ignores the chain will hold each feature individually and still miss the assembly.

The usual fix is to cut the chain. Move the critical feature so it is machined in the same setup as its datum, or add an intermediate datum that the drawing allows. This is a conversation to have at quote time. After the first article is cut, changing datums means new fixtures and lost days.

Material behavior belongs in the same calculation. Aluminum 6061 and 7075 move differently after roughing. Thin walls in 316L stainless will deflect under tool pressure even when the toolpath is correct. A 0.5 mm wall on a 60 mm deep pocket is a different job from a 3 mm wall on the same pocket.

The rule we use: if the tolerance band is tighter than the expected thermal and clamping movement, the part needs stress relief or a semi-finish pass before final dimensions. That is a process decision, not a machine setting.

• 1
  Chain the tolerancesAdd up the values that share a datum before promising capability.
• 2
  Separate rough and finishLeave 0.3–0.5 mm on walls, then finish after the part relaxes.
• 3
  Check wall-to-depth ratioDeep thin walls need reduced radial engagement, not more speed.

Five-axis capability expands what is reachable, not what is accurate. A simultaneous five-axis center can cut an impeller in one setup, which removes the positional error of four re-chucks. It cannot fix a part that is held in a vise by 4 mm of stock on one end.

Rigidity is the real ceiling. Long tools chatter. Thin fixtures ring. A part that hangs 200 mm out of the chuck will not hold ±0.005 mm no matter how the toolpath is smoothed. The expert response is to add a tailstock, a steady rest, or a sacrificial support that gets cut away in the last operation.

Size changes the calculus too. Our largest travel is 4,000 × 400 × 150 mm, which suits long extrusions and rails. Medium travels of 750 × 1,150 × 550 mm and 600 × 600 × 600 mm cover most housings and manifolds. Small travels of 500 × 500 × 450 mm and 500 × 310 × 200 mm do the tight work where thermal drift over a long axis would hurt.

A Ø400 mm rotary table adds a fourth axis for parts with features on multiple sides. It also adds a setup constraint: the part must be balanced and clamped without distorting the bore you are about to cut.

• 1
  Reach is not rigidityA five-axis move can reach the feature and still chatter on it.
• 2
  Support the overhangTailstock, steady rest, or sacrificial material beats a slower feed.
• 3
  Match travel to partLong parts on a small machine lose accuracy to thermal growth.

Tool runout is the least discussed and most common cause of a missed surface finish. A holder with 0.02 mm of runout cuts with one flute doing most of the work. The result is a rough patch on one side of a bore, a shorter tool life, and a finish that drifts across the part.

Measure runout at the tool tip, not at the holder. Then match the finishing operation to the finish callout. As-machined surfaces typically land at Ra 1.6–3.2 μm. A good high-speed finishing pass reaches Ra 0.8–1.6 μm. Getting to Ra 0.2–0.8 μm usually means a finer stepover, a dedicated finishing tool, or a secondary process such as polishing or lapping.

Cutting data follows from material, not from habit. Aluminum 6061 and 6082 run fast with high rake angles. Titanium Ti-6Al-4V and Inconel need lower surface speed and more coolant, and they work-harden if the tool rubs instead of cuts. Copper alloys like C110 and beryllium copper are gummy and want sharp, polished flutes.

Tool life is a schedule item. If a finishing tool dulls after 30 parts, the 31st part is out of tolerance before anyone notices. Logging tool changes by count and by measured wear keeps the drift inside the band.

• 1
  Check runout at the tipAim under 0.01 mm for finishing operations.
• 2
  Separate rough and finish toolsNever finish with the tool that took the roughing load.
• 3
  Log tool changesReplace by part count or measured wear, not by sound.

Inspection is not a final gate. It is part of the process plan. The first question is what to measure, and the second is when. A bore that will be re-chucked should be checked after the re-chuck, not before. A flatness callout on a face that gets clamped again at the end should be checked in the free state.

Metrology must match the tolerance. A ±0.05 mm feature can be checked with calipers. A ±0.005 mm feature needs a coordinate measuring machine, a micrometer with the right anvil, or a bore gauge that has been set with a ring standard. Using the wrong instrument produces numbers that look fine and mean nothing.

We inspect 100% of parts before shipment, with raw material checks, in-process monitoring, and a final inspection. Reports are available on request. That sequence is what keeps a qualification rate at 99.99% instead of relying on a sample at the end.

Temperature belongs in the plan. A part measured at 28 °C and used at 20 °C has already moved. For tight work on aluminum and stainless, let the part stabilize before the final measurement. It is a fifteen-minute wait that saves a rejected lot.

• 1
  Measure in the free stateClamping hides distortion that shows up after unload.
• 2
  Match instrument to toleranceCalipers cannot resolve ±0.005 mm.
• 3
  Control temperatureLet parts stabilize before final measurement on tight work.