CNC and 3D Printing: How Each Process Builds a Part

• 1
  Key takeaways
• 2
  How material is added or removed
• 3
  What tolerance and surface finish each route can hold
• 4
  Where geometry decides the process, not cost
• 5
  How cost moves with quantity and material
• 6
  Using both processes on one part
• 7
  Choosing between CNC and 3D printing by part requirement

Subtractive machining starts with a billet, plate or casting and removes material with a rotating cutter or a single-point tool. The toolpath comes from CAM software driven by a solid model. Every removed chip leaves a surface whose quality depends on feed per tooth, cutting speed and tool rigidity. That is why a milled face carries visible stepover marks while a turned diameter can come off the lathe at Ra 1.6 µm and be polished further.

Additive processes do the opposite. A model is sliced into layers, and the machine deposits or cures material inside each layer boundary. FDM extrudes a thermoplastic filament, SLA cures resin with a laser or projector, SLS sinters polymer powder, and metal systems use a laser or electron beam over a powder bed. The part grows in a chamber, so supports are often needed under overhangs and inside holes.

That single difference sets the failure modes. Machining fails when a cutter cannot reach a feature, when the part deflects under clamping, or when a thin wall chatters. Printing fails at overhangs, at layer bonding, and at the shrinkage that pulls a warp into a long flat section. Knowing which failure you can tolerate usually decides the route faster than a price comparison.

• 1
  MachiningSolid stock in, chips out. Strength, grain direction and finish come from the parent material.
• 2
  PrintingLayer-on-layer build. Properties are directional and depend on build orientation.

On our 127 CNC machines, general machining tolerances sit at ±0.005 mm (±0.0002 in) on critical features, with as-machined surfaces at Ra 1.6–3.2 µm and finer finishing down to Ra 0.2–0.8 µm where the drawing calls for it. We reach those numbers on 16 simultaneous 5-axis centers, 12 four-axis mills, 27 three-axis machines and 16 mill-turn centers, with a Ø400 mm rotary table for round parts and up to 4,000 mm of travel on the largest frame.

Printed parts are a different story. Dimensional accuracy on a well-tuned FDM machine is typically a few tenths of a millimeter, and the layer lines are a permanent part of the surface unless you sand, vapor-smooth or tumble them. SLS and SLA do better, but the tolerance band is still wider than machining, and it drifts with build position, powder reuse and thermal history in the chamber.

The practical split: if a feature mates with a bearing, seals against a gasket, or sits in an assembly stack-up that adds up to less than 0.05 mm, machine it. If a feature only has to fit a hand, route a cable or hold a PCB, printing is fine. Mixed drawings are common, and we handle them by printing the shell and machining the interface.

• 1
  Check the stack-upA loose fit on one part can still break a tight assembly if three parts stack in the same direction.
• 2
  Name the critical facesMark datum and fit surfaces on the drawing so the shop knows where to spend time.

Internal cooling channels that spiral through a mold insert, lattice infill, and organic brackets with no flat reference are easy to print and hard to cut. A curved channel can be printed in one piece; machining it means drilling from multiple angles, plugging, or splitting the part and joining it later. If the channel has to hold pressure, the joint becomes the risk.

Machining owns the opposite cases. Sharp internal corners always need a cutter radius, so a square pocket gets a fillet of at least the tool radius unless you EDM it. Deep bores run into length-to-diameter limits, and a hole deeper than about 8× its diameter needs a longer, less rigid tool and peck drilling. Threads, keyways, splines and precision bores are routine on a mill-turn center and awkward on a printer.

Wall thickness cuts both ways. Printing handles thin walls down to roughly 0.8–1.0 mm for a 0.4 mm nozzle, but a tall thin wall can warp. Machining a 0.5 mm wall in aluminum is possible, yet clamping pressure will bow it, so we often leave support ribs and cut them off last.

• 1
  Print-friendlyCurved channels, lattices, hollow shells, low-load brackets, jigs and fixtures.
• 2
  Machine-friendlyBearing bores, sealing faces, threads, splines, flat datums, tight stack-ups.

Printing has almost no setup cost. Upload a file, orient it, build it. That makes one-off brackets, fit-check models and ergonomic test pieces cheap to iterate. The catch is cycle time: a large SLS build or a metal powder-bed job can run for many hours, and the cost per part barely drops as quantity rises because machine time scales with volume.

Machining carries setup cost in fixtures, programming and first-article checks, but the cycle time per part is short. Spread that setup across 50 or 500 parts and the unit price falls. Material also behaves differently: aluminum 6061-T6, 7075, 304 and 316L stainless, 17-4PH, Ti-6Al-4V, Inconel and engineering plastics such as POM and PEEK are all standard stock for us, and the machined part keeps the wrought properties of the parent material.

Printed metal parts come out of the build with residual stress and a rough surface, so they usually need stress relief, support removal, and machining on mating faces anyway. Once you add those steps, the cost advantage over a simple machined part often disappears.

• 1
  Low countPrinting usually wins below a few dozen parts when geometry is complex.
• 2
  Higher countMachining wins once setup amortizes and cycle time is short.

The strongest answer is often not either-or. Print a near-net blank with the complex internal geometry already formed, then clamp it and machine only the faces that need tolerance: bearing seats, sealing surfaces, bolt pads and datums. This keeps the hard-to-cut geometry and buys back the accuracy where it matters.

The same idea runs in reverse for tooling. A printed fixture or soft jaw can hold a part for first-article inspection, and once the design settles we cut the production fixture from aluminum or steel. Iteration stays cheap while the production route stays repeatable.

Confidentiality matters here because printed and machined files expose a complete design. Uploads to our portal are secure and confidential, and we sign an NDA on request before files move. Quotation and a free DFM review come back within 12 hours, and production can start within 24 hours of approval.

• 1
  Hybrid flowPrint the shell, machine the interfaces, inspect the assembly.
• 2
  Early DFMSend the model before tooling so we can flag features that will not cut or print cleanly.