Introduction to CNC Machining

• 1
  What a CNC machine actually does
• 2
  How a CAD model becomes a machined part
• 3
  Milling versus turning: the dividing line
• 4
  Holding ±0.005 mm and knowing when not to chase it
• 5
  Where CNC machining stops being the right process
• 6
  Axis count and what it buys you

A CNC machine does not read your drawing. It reads coordinates. The CAM step takes the CAD solid and decides, in order, which tool touches which surface at which feed and speed. Those decisions come out as G-code: a list of positions, spindle speeds and feed rates the controller executes without an operator turning handwheels.

That distinction matters when you review a quote or debug a first article. The part geometry you send is fixed. Everything downstream is a machining strategy: tool reach, workholding, pass depth, coolant. Two shops can quote the same STEP file and produce different surfaces, because one chose a long tool that deflects and the other paid for a stiffer setup.

So the useful question is never "is this part machinable?" Almost anything is, at a price. The real question is which setup, which tool, and how many operations it takes. Those three numbers drive cost, lead time and how repeatable the result will be across a run.

Keep the controller out of your mental model. Think in terms of a spinning cutter pushed along a programmed path, with a fixture holding the blank rigid enough that the path lands where the CAM software expected it to.

The chain is short but each link can break a part. You send a 3D solid, usually STEP or IGES. We check that the model is watertight and that the tolerances on the drawing match what the geometry can actually deliver. A 0.01 mm callout on a deep pocket wall is a different job than the same callout on a flat face.

CAM programming then picks tooling. A roughing pass removes most of the stock with a larger cutter at aggressive depth; a finishing pass follows with a smaller tool at light radial engagement to hit the surface finish. For aluminum 6061 we typically run finishing at Ra 1.6–3.2 μm as-machined, and reach Ra 0.8–1.6 μm with adjusted stepover or a dedicated finishing strategy.

Setup comes next, and this is where most surprises live. The blank has to be held without crushing thin walls or leaving marks on finished faces. A part that needs four sides machined may need two vises and an index, or a single 5-axis setup that rotates the part instead.

Finally the machine runs, and inspection closes the loop. We check raw material before cutting, monitor critical dimensions in process, and inspect 100% before shipment with reports on request. If a dimension drifts mid-run, the fix is a tool offset change, not a re-quote.

Milling spins the tool and moves it around a stationary blank. Turning spins the blank and pushes a single-point tool into it. The dividing line is part symmetry: if the dominant feature is a surface of revolution, turning is usually faster and cheaper per piece. If the part is mostly prismatic, milling wins.

Many real parts are both. A housing with a turned bore and milled mounting pads is a mill-turn job. Running it as two separate operations means two fixtures, two alignments, and a stacked tolerance between the bore centerline and the pad faces. A mill-turn center holds that relationship in one chucking.

For long slender shafts, turning with a tailstock or steady rest controls deflection better than milling a round profile. For thin plates, milling on a vacuum plate or soft jaws beats trying to turn a disc that will chatter.

Material changes the answer too. Titanium TC4 (Ti-6Al-4V) and Inconel generate heat at the cutting edge and work-harden if the tool rubs, so we keep radial engagement light and never let the cutter dwell. Aluminum 7075 cuts fast but moves after machining if stock removal is uneven.

±0.005 mm (±0.0002 in) is achievable on our equipment, but it is a process capability, not a default. It depends on feature size, material, and how many datums the inspector has to stack. A 5 mm bore in aluminum is a different proposition from a 300 mm face on a steel weldment.

Temperature is the quiet variable. A shop floor that swings a few degrees will move a 400 mm steel part more than the tolerance you asked for. We let parts stabilize before final inspection on tight work, and we flag any callout that is tighter than the geometry supports.

Surface finish and tolerance trade against each other. Pushing Ra 0.2–0.8 μm means light passes and more time. If the mating surface is a gasket face, Ra 1.6–3.2 μm is often enough and costs less. Tell us the function, not just the number.

The honest answer on any tight callout: send the drawing and let us run a DFM review. We return quotation and free DFM analysis within 12 hours, and we will say plainly if a dimension should be relaxed.

CNC machining removes material, so it is wasteful by definition. On a part where 90% of the blank becomes chips, casting or forging the near-net shape first and machining only the critical faces usually wins on cost at volume. We offer die casting and vacuum casting for exactly that reason.

Very thin walls are a second boundary. Below roughly 0.5 mm in aluminum, chatter and distortion become the dominant problem, and holding flatness gets expensive. Sheet metal fabrication handles thin gauges better because it forms rather than cuts.

Hollow internal channels are a third. A curved cooling passage inside a block cannot be cut by a rotating tool. That is additive territory, which is why we run custom 3D printing alongside the CNC floor.

For one to a few hundred parts with tight tolerances, CNC is hard to beat. From one prototype to 10,000+ part runs, no minimum order quantity applies, so the same process can carry you from first article to production without a tooling investment.