CNC machining corona: how the process actually works

• 1
  What CNC machining corona actually means on the floor
• 2
  Tolerance, finish, and the limits that matter
• 3
  Material behavior that changes the cut
• 4
  Why finishing decides the real tolerance
• 5
  From one prototype to a 10,000-part run
• 6
  Which machine suits which part
• 7
  Judging whether a part suits CNC machining

CNC machining corona is a subtractive process. A cutting tool is held in a spindle, the workpiece is clamped to a table or chuck, and a control reads a toolpath and moves the axes. Nothing is formed or poured. Material is removed until the remaining shape matches the CAD model.

The cutting edge does the real work. Each tooth of an end mill or insert peels a chip. If the chip is too thin, the edge rubs and work-hardens the surface. If it is too thick, the tool deflects and the wall goes thin. Feeds and speeds are chosen to keep the chip in a stable range for the material. Aluminum 6061 runs fast, 17-4PH runs slow.

The machine only follows instructions. A poor toolpath, a weak fixture, or a wrong cutter will show up in the part no matter how tight the tolerance block reads. Most dimensional problems on a first article trace back to setup, not to the control.

That is why a drawing alone is not enough to quote a job. We look at wall thickness, depth-to-diameter ratio on bores, and where the datum sits. Those three items decide whether the part is simple or painful.

• 1
  Subtractive, not additiveMaterial leaves the block; geometry comes from the remaining solid.
• 2
  Chip load drives finishToo light or too heavy both hurt the surface.
• 3
  Setup dominates accuracyFixture rigidity matters as much as spindle accuracy.

Quoted tolerance is ±0.005 mm on critical features, and ±0.0002 in in imperial drawings. That is a real number, not a marketing line, but it applies to specific features. A 4,000 mm long part will not hold ±0.005 mm end to end because thermal drift and machine geometry take over.

Surface finish follows the same logic. Ra 0.2–0.8 μm needs fine stepover, sharp tooling, and often a finishing pass at low feed. Ra 0.8–1.6 μm is a normal machined finish for sealing faces. Ra 1.6–3.2 μm is as-machined and fine for brackets and covers. Asking for Ra 0.2 μm on a non-functional face adds cost without adding value.

Deep bores are the classic trap. A bore with a 6:1 depth-to-diameter ratio needs a long tool, and long tools chatter. We can usually hit the tolerance with a boring head and a peck cycle, but cycle time doubles. Below 4:1 the job is straightforward.

Thin walls behave the same way. Under 0.8 mm on aluminum, the wall deflects away from the cutter. Roughing leaves 0.3–0.5 mm, then a light finishing pass with a sharp tool keeps the wall straight. On stainless, the same wall is harder to hold.

• 1
  Tolerance is feature-specific±0.005 mm on critical features, not across a 4,000 mm part.
• 2
  Pick finish by functionRa 0.8–1.6 μm for seals; Ra 1.6–3.2 μm for covers.
• 3
  Watch the 4:1 ratioDeeper bores mean longer tools and more chatter risk.

Aluminum is the easy case, and that is why 6061 and 7075 show up in most first articles. 6061-T6 machines clean and holds a good finish. 7075 is stronger but gummier, so we slow the spindle and keep the chip heavy. 2024 sits between them and is common on aerospace brackets.

Stainless punishes light cuts. 303 is the free-machining grade and behaves well. 304 and 316 work-harden the moment the tool rubs, so we never let the cutter dwell. 17-4PH in the H1150 condition is tough on inserts and needs a rigid setup. If a design can use 303 instead of 316, cycle time drops noticeably.

Titanium TC4 (Ti-6Al-4V) and Inconel are the hard end. Heat stays in the cut, not in the chip, so tool life is short and feeds stay conservative. These parts get quoted with extra stock for finishing passes. On Inconel, a 0.5 mm finishing allowance is normal.

Plastics and composites cut fast but move. POM and PEEK hold tolerance well with sharp tooling and air blast. Carbon fiber eats carbide, so we plan for tool changes. ABS and PC are simple as long as clamping pressure is light enough not to mark the surface.

• 1
  Aluminum: fast and stable6061, 7075, 2024 cover most brackets and housings.
• 2
  Stainless: never rub303 machines clean; 304 and 316 work-harden on light cuts.
• 3
  Titanium and Inconel: slowHeat stays in the part, so tool life and feeds drop.
• 4
  Plastics: light clampingSharp tooling and air blast keep POM and PEEK stable.

A machined part is rarely the final part. Anodizing adds 2–10 μm per surface, and that shifts a tight bore. Hardcoat anodizing builds more. If a bore must stay within ±0.005 mm after coating, we machine it undersize on purpose and note the allowance on the drawing.

Plating behaves the same way. Electroless nickel and zinc both add thickness, and both add it unevenly on edges. Threads are usually masked so the pitch diameter does not close up. If a design needs a plated thread, say so before the first article, not after.

Bead blasting and tumbling change the finish without changing the size much, but they do knock sharp edges down. Laser marking needs a minimum character height of 1.5 mm to stay readable after anodizing. Smaller text fills in.

The point is simple. Lock the finish before you lock the tolerance. A drawing that calls for ±0.005 mm and a hardcoat on the same bore is a drawing that will come back for rework.

• 1
  Anodizing builds 2–10 μmMachine undersize if the bore must stay tight after coating.
• 2
  Mask plated threadsOtherwise the pitch diameter closes up.
• 3
  Laser text 1.5 mm minimumSmaller characters fill in after anodizing.

CNC machining does not need a mold, so the first part and the ten-thousandth part come off the same process. There is no minimum order quantity. A single prototype is a normal job, and so is a 10,000+ part run.

The economics shift as volume rises. At one to fifty parts, the setup dominates the price. At a few hundred, cycle time dominates. Past a few thousand, a dedicated fixture and a pallet system cut the cost per part, and someone should compare machining against die casting or vacuum casting before committing.

Cycle time is the number to watch. A part with three setups and two hours of run time is expensive at any volume. The same part redesigned with one setup and a 20-minute cycle is cheap. That is the conversation worth having with the engineer, not the discount conversation.

Lead time follows the same pattern. We return a quote and a free DFM analysis within 12 hours, production can start within 24 hours, and parts ship in 3–5 days. Those numbers hold because the tooling is standard and the planning happens before the spindle turns.

• 1
  No MOQOne prototype or a 10,000+ part run, same process.
• 2
  Setup dominates at low volumeCycle time takes over past a few hundred parts.
• 3
  Quote in 12 hoursFree DFM analysis comes with it.