Basic Knowledge of CNC Metal Cutting

CNC metal cutting is a chip-forming process. A rotating cutting tool or a rotating workpiece meets the material, and a controlled amount of metal is sheared off as chips. The machine does not shape metal by force or heat; it removes material along a path the program defines. Everything else on this page follows from that one fact.

The path comes from coordinates. A CAM programmer takes the CAD model, chooses the tool, and outputs G-code: a list of positions, feed rates, spindle speeds and auxiliary commands. The controller reads that list and drives the axes. On a 3-axis mill the tool moves in X, Y and Z. Add rotary axes and the cutter can approach a face from an angle instead of only from above.

Accuracy in this process is not a single number. It is the sum of machine geometry, tool deflection, thermal growth, workholding stiffness and how the material behaves at the cutting edge. A machine rated to ±0.005 mm will only hold that if the setup supports it.

Chip evacuation matters more than most people expect. Aluminium stringers can wrap a tool and break it. Titanium and stainless steels work-harden if the tool rubs instead of cuts. Deep pockets need air blast, through-tool coolant or a peck strategy. When a job fails, it is often chips, not the program.

• 1
  MillingWorkpiece clamped, tool rotates and travels on multiple axes.
• 2
  TurningTool is fixed, workpiece rotates; best for round, symmetric parts.
• 3
  Mill-turnBoth in one setup, so a part keeps its position between features.
• 4
  Drilling and tappingAxial holes and threads, usually after the outer shape is stable.

The first decision is geometry, not price. If the part is a body of revolution with features around one axis, a lathe cuts it faster and holds concentricity without extra fixturing. Shafts, bushings, pistons, threaded inserts and connector shells fall into this group. If the part is prismatic, with pockets, faces and holes on several sides, it belongs on a mill.

Parts that mix the two are the interesting ones. A housing with a turned bore and a milled mounting face used to need two machines and two setups; each setup adds position error. A mill-turn center cuts both in one clamping, so the bore and the face stay related. That is why we run 16 mill-turn centers alongside 16 simultaneous 5-axis machining centers and 27 three-axis machines.

Do not ignore part size. Travel limits decide the machine class before any feature is considered. On our floor, the largest envelope is 4,000 × 400 × 150 mm, medium machines cover 750 × 1,150 × 550 mm and 600 × 600 × 600 mm, and compact machines cover 500 × 500 × 450 mm and 500 × 310 × 200 mm. A part slightly over a class boundary may need reorientation instead of a bigger machine.

Wall thickness is the other gate. Thin walls deflect under cutting force and vibrate. If a wall is under about 0.8 mm in aluminium or 1 mm in steel, expect to slow the cut and add support, or accept that the feature is a better fit for a different process.

• 1
  Choose turningRound parts, tight concentricity, threads and grooves on one axis.
• 2
  Choose millingPrismatic shapes, pockets, slots and faces on multiple sides.
• 3
  Choose 5-axisAngled faces, undercuts, or fewer setups on a complex part.
• 4
  Reconsider the designWalls below 0.8 mm, deep narrow slots, sharp internal corners.

G-code is not the hard part. A programmer selects a tool, sets a spindle speed and a feed per tooth, then defines the path. The real skill sits in the choices around it: which tool enters a corner, how much material is left for the finishing pass, and whether the part is reached in one setup or three.

Speeds and feeds follow the material. Aluminium runs fast with high rake tools and generous coolant. Stainless steels such as 304 and 316 need lower surface speed and a constant feed so the edge keeps cutting rather than rubbing. Titanium TC4 and Inconel are worse: heat stays at the edge, so tool life drops and the path has to avoid dwelling in a cut. 17-4PH sits between stainless and titanium in difficulty.

Workholding decides whether the program can be trusted. A part held in a weak vise will move during a heavy roughing pass, and no controller can correct that. For thin plates and rings we use custom soft jaws, vacuum plates or sacrificial tabs. For long parts, support the overhang. The setup is often the difference between a stable process and a scrapped batch.

Tool wear is normal. A finishing pass that held ±0.005 mm at the start of a run can drift as the edge dulls. In-process monitoring catches that: measure a critical feature, compare it to the nominal, and compensate before the next part. On a 10,000-part run this matters more than any single program trick.

• 1
  Rough then finishLeave 0.2–0.5 mm on walls for the finishing pass.
• 2
  Avoid full-width cutsRadial engagement of 30–50% of tool diameter lowers deflection.
• 3
  Control the cornerSharp internal corners need a small tool, which limits reach.

Tolerance is a system result. The machine contributes its own geometry error, the spindle adds runout, the tool adds deflection, and the material adds springback. We hold ±0.005 mm (±0.0002 in) on features that are designed to be reached, meaning a stable base, a sensible wall and a tool that can physically enter the feature. A deep slot with a 2 mm cutter will not hold that, because the tool bends.

Surface finish follows the same logic. A fine finish of Ra 0.2–0.8 μm is possible on a dedicated finishing pass with a sharp tool and a light cut. General machined surfaces land at Ra 0.8–1.6 μm, and as-machined surfaces at Ra 1.6–3.2 μm. If your drawing calls for a mirror finish across a large face, say so early: it changes the tool, the passes and the inspection plan.

Datum choice drives the result more than most engineers admit. If a bore is dimensioned from a face that is machined in a second setup, the two features carry the setup error between them. Ask which face is the primary datum and whether the critical dimensions share it. Parts with a single clear datum are easier to hold and cheaper to inspect.

Inspection closes the loop. We check raw material on arrival, monitor during the run, and inspect 100% before shipment, with reports on request. That sequence is what keeps a qualification rate of 99.99% meaningful instead of a slogan.

• 1
  Design to reachA feature must be physically reachable by a real tool.
• 2
  One datumKeep critical dimensions on the same reference face.
• 3
  Specify finish by functionOnly seal or bearing surfaces need the fine range.

Most cost and delay sits in the drawing, not the machine. A radius in the corner of a deep pocket forces a small tool and a long cycle. A thread that runs to the bottom of a blind hole cannot be cut. A toleranced dimension on a non-functional face adds inspection time for no benefit. These are the items a DFM review catches.

Build the part around the function. Tighten only the faces that mate, seal, slide or locate. Leave cosmetic and clearance surfaces at general tolerance. This is the single largest lever on both price and lead time, and it does not change how the part works.

Think about the setup count. Every additional orientation adds a fixture, a re-clamp and an error stack. A design that can be machined from two sides costs less than one that needs four, even if the two-sided version looks more complex on paper. Five-axis work exists mainly to reduce that count.

Send material grade and temper, not just a family name. 6061-T6 and 6061-O cut and finish very differently. 17-4PH in the annealed state behaves nothing like the H900 condition. The same applies to surface treatment: anodizing adds a few micrometres and can close a tight bore if the allowance was not planned.

• 1
  Corner radiiMatch the radius to the largest tool that can reach the pocket.
• 2
  Thread depthAllow runout space at the bottom of blind threads.
• 3
  Material stateGrade plus temper plus heat-treat condition.
• 4
  Finish allowancePlating and anodizing change the final dimension.