Carbon CNC machining with complex geometric shapes

Carbon fiber reinforced polymer is not a homogeneous block. It is layers of fiber held in resin, and the fiber runs in one direction inside each ply. A cutter that moves across the fiber gets a different result than one that moves along it. Edge quality, delamination risk and tool wear all shift with direction. This is the single biggest difference between carbon CNC machining and cutting aluminum or steel.

In practice we plan the toolpath around the ply orientation the customer specifies. Climb milling with a sharp, coated cutter keeps the fiber in compression and reduces fraying at the exit edge. Where the geometry allows, we enter and exit the cut at a shallow angle instead of dropping straight into a wall. On thin ribs this is the difference between a clean edge and a delaminated one.

The second factor is resin. Resin softens with heat, and heat builds up fast when a small cutter removes a lot of material in one pass. We control this with moderate spindle speeds, conservative feed per tooth and air blast instead of flood coolant. Water-based coolant on a carbon part can wick into exposed edges and damage the resin over time.

Graphite behaves differently again. It is brittle and abrasive, cuts like a powder, and produces a very fine conductive dust. Graphite electrodes for EDM are usually machined dry with strong extraction, and the toolpath is written to avoid sharp internal corners where the material chips out.

• 1
  Fiber direction sets the cutPlan the toolpath around ply orientation, not around the model only.
• 2
  Sharp tools, light passesDull tooling delaminates carbon faster than it wears out.
• 3
  Air blast over flood coolantKeeps the resin cool without letting moisture into exposed edges.
• 4
  Graphite cuts dryFine conductive dust has to be extracted at the source.

A five-axis machine reaches features that a three-axis setup cannot, because the tool stays normal to the surface instead of approaching from one direction. That matters for carbon parts with curved pockets, angled ribs or undercut flanges. We run 16 simultaneous five-axis machining centers, with a 4,000 mm maximum processing size for long parts and a Ø400 mm rotary table for parts that need indexing around a bore.

Typical carbon geometries that go on the five-axis machines include monocoque panels with sculpted pockets, drone and UAV frame plates with tapered ribs, thin-walled housings with internal bosses, and brackets that blend three or four faces in one continuous surface. If a design can be reached from five directions without the tool shank hitting the part, it is usually machinable.

The geometry that gives us trouble is not the curved surface. It is the combination of thin walls and tight internal corners. A wall under about 1 mm deflects under cutting force, and a sharp internal corner concentrates stress and chips out. Where the drawing allows, we ask for a corner radius of at least one third of the wall thickness and a uniform wall section. Small design changes at that stage save a lot of scrap later.

Deep cavities with a high depth-to-diameter ratio are the other common limit. A long, thin cutter has to run at reduced feed and reduced depth of cut, and on abrasive carbon that means fast tool wear. Sometimes the right answer is to split the part into two pieces that are bonded or fastened, rather than to machine one deep pocket.

• 1
  Five-axis reachCurved pockets and multi-face blends machine in one setup.
• 2
  Thin walls need supportBelow roughly 1 mm, deflection shows up in the final inspection.
• 3
  Radius the internal cornersOne third of wall thickness is a workable rule of thumb.
• 4
  Deep pockets slow downLong small cutters wear quickly in abrasive carbon.

Carbon laminates are stiff along the fiber and weak through the thickness. A vise tightened the way you would tighten it on a steel block will crush the laminate or leave a witness mark that cannot be polished out. We fixture carbon on custom soft jaws, vacuum chucks or sacrificial backing plates, and we spread the clamping load over a wide area rather than a point.

Vacuum workholding suits flat panels and thin plates. The whole back face is supported, so the part cannot bow while the cutter pushes down on it. For a panel with pockets on both sides, we machine the first side on vacuum, flip it onto a machined backing plate that matches the pocket pattern, and cut the second side with the first side fully supported.

For contoured parts we often machine a sacrificial fixture from aluminum or tooling board that matches the part's outer surface. The carbon blank sits in that nest for the finishing passes. It costs one extra setup, but it holds wall thickness far more consistently than clamps, and it protects the finished surface from the jaw.

One more point on sequencing. Rough and finish in separate operations. Roughing removes the bulk with a stronger cutter and leaves a small allowance, then a finishing pass takes that allowance with a sharp tool at high spindle speed. Trying to do both in one pass is where most delamination and chatter marks come from.

• 1
  Spread the loadSoft jaws, vacuum or a matched nest instead of point clamps.
• 2
  Support the back faceUnsupported thin panels chatter and bow during cutting.
• 3
  Rough then finishTwo operations keep edge quality and wall thickness stable.
• 4
  Protect the surfaceA nested fixture keeps finished faces away from metal jaws.

Inspection on a complex composite part is mostly about features that are hard to reach with a caliper. We use CMM probing for hole position, wall thickness and profile on curved surfaces, and we can supply inspection reports on request. Every part gets a raw material check, in-process monitoring and a final inspection before it leaves the shop.

Tolerance is a conversation, not a default. We hold ±0.005 mm on machined metal features, and that number is realistic on machined carbon features too when the part is rigidly supported and the feature is measured at a stable temperature. On a thin, unsupported composite skin, the part itself moves more than the machine does, and the honest answer is that the tolerance should be opened or the geometry changed.

Edge finish matters more on carbon than on metal. A sawn or roughly milled edge exposes fiber and is the usual starting point for delamination. We deburr and sand machined edges, and where the drawing calls for it we apply a sealing coat. Bead blasting gives a uniform matte surface on flat areas but will round a sharp edge, so it is specified only where the edge is not functional.

Carbon fiber appears in our plastics list alongside ABS, POM, PEEK and PA. If your part is a composite with a metal insert, we machine the metal features and the composite in the same program where the geometry allows, which keeps the insert position tied to the composite surface instead of to a second setup.

• 1
  CMM on curved featuresProfile, wall thickness and hole position, report on request.
• 2
  Tolerance follows rigidity±0.005 mm holds when the part is supported and stable.
• 3
  Edge sealing stops delaminationExposed fiber at a cut edge is where failure starts.
• 4
  Metal inserts in one programKeeps insert position tied to the machined composite face.

Carbon earns its cost when stiffness per kilogram drives the design. If the part is a stiff bracket that sees modest load, aluminum 6061 or 7075 will be cheaper, faster and easier to rework. Carbon also loses on parts with many small threaded holes, because threads in a laminate pull out under load and need metal inserts, which adds operations.

Cost is geometry-driven, not material-driven, on most carbon jobs. A simple flat plate in carbon is not much harder to make than the same plate in aluminum. A deep pocket with a 0.8 mm wall is a different project, and the price reflects the fixturing, the slower feeds and the scrap risk. That is why we quote from the model rather than from a material list.

If your part is still in the prototype stage and the geometry is not frozen, it is worth running the first version in aluminum or a cheaper composite to prove the shape, then switching to carbon once the design settles. We support both routes with no minimum order quantity, from one prototype to 10,000+ part runs.

Where carbon does win clearly: UAV and drone structures, robotic arms where mass at the end of the arm matters, motorsport and automotive body panels, medical imaging components that must be radiolucent, and any bracket where vibration and weight combine. Our aerospace and robotics teams handle most of that work.

• 1
  Choose carbon for stiffness per kilogramNot for parts that are simply flat and lightly loaded.
• 2
  Threads need insertsDirect threads in laminate pull out under load.
• 3
  Prototype in aluminum firstProve the geometry before paying for composite stock.
• 4
  Good fitsUAV frames, robot arms, radiolucent medical parts.