The 3D printing field: how the processes actually work

• 1
  What defines the 3D printing field
• 2
  How the main processes compare in practice
• 3
  Where printed tolerances stop holding
• 4
  Where CNC machining replaces printing
• 5
  Printed processes versus CNC machining

Every process in the 3D printing field builds a part by adding material, but the joining method changes everything downstream. FDM extrudes a thermoplastic filament through a heated nozzle and fuses it to the layer below. SLA and DLP cure liquid resin with a laser or projector. SLS and SLM sinter or melt powder with a laser, and the powder bed itself supports the part.

That last difference matters for design freedom. FDM and SLA need printed supports that you cut or wash away, so overhangs cost time and leave witness marks. Powder-bed processes float the part in unsintered powder, so internal channels and nested geometry come out supported by default. If your part has a closed internal cavity, choose powder bed.

The second defining trait is anisotropy. FDM parts are weaker across layer boundaries than along them, often by a wide margin. SLA and SLS behave closer to isotropic, but still show a build-direction effect. Any load-bearing printed part should be oriented so the main tensile load runs in the build plane, not across layers.

The third trait is resolution versus build volume. A resin printer holding ±0.1 mm on a small part will not hold it on a 300 mm tall build. Laser spot size, layer thickness and thermal shrinkage all scale with the envelope. Ask for tolerance on your actual part size, not on a spec sheet.

FDM is the cheapest route and the easiest to run in-house. A 0.4 mm nozzle gives roughly 0.4 mm wall thickness and 0.1–0.3 mm layer heights. It suits brackets, jigs, enclosures and fit checks where surface finish does not matter. Typical as-printed surfaces sit around Ra 12–20 μm, so you will not get a sealing face without post-machining.

SLA and DLP deliver the finest detail: layer heights from 0.025 mm and features down to roughly 0.1 mm. The trade is material. Standard resins creep under sustained load and degrade in UV light, so they belong in visual models, mold masters and flow-test parts. Engineering resins improve toughness and temperature resistance but cost several times more per liter.

SLS with PA12 is the workhorse for functional plastic parts. It holds ±0.3 mm on small features, needs no support removal, and takes threads and snap fits reasonably well. Glass-filled PA12 raises stiffness and heat deflection. Porous surfaces trap moisture and release it in a vacuum or cleanroom, so plan a sealing coat for those environments.

SLM and DMLS print metal in 20–60 μm layers with a laser. AlSi10Mg, 316L, Ti-6Al-4V and Inconel 718 are the common alloys. Parts come out with a rough, partially attached surface and residual stress, so they normally go through stress relief, support removal, and CNC finishing on every mating face. The printed blank is a near-net shape, not a finished part.

Printed accuracy splits into machine accuracy and part behavior. The machine positions the tool within a known band. The part then shrinks as it cools or cures, and thin walls warp more than thick ones. On a 100 mm dimension, expect ±0.3 mm from SLS and ±0.5 mm or worse from FDM. Resin holds ±0.1 mm on features under 50 mm.

Metal printing adds another layer of error. Melt pools leave internal porosity, and thermal gradients pull thin sections out of flat. A 3 mm thick plate printed flat may bow 0.2–0.5 mm over 100 mm. You can reduce this with build orientation, support design and stress relief, but you cannot remove it entirely.

Surface finish is the other hard limit. As-built SLM surfaces run Ra 8–15 μm and need media blasting or machining to reach a functional finish. That is far from the Ra 0.8–1.6 μm we hold on machined aluminum, and further still from the Ra 0.2–0.8 μm possible on a finishing pass.

Threads deserve a mention. Printed threads below M6 are unreliable in plastic and nearly unusable in metal. Design a pilot hole and cut the thread with a tap, or use a heat-set insert in plastic. Feature size also matters: holes below 1 mm in FDM and 0.5 mm in resin tend to close up or clog with debris.

Printing wins when geometry is complex, quantity is low, and loads are moderate. Machining wins when the part carries load, seals against fluid, mates to a tight datum, or must survive heat and fatigue. A printed bracket can be excellent for a fixture. The same bracket on a chassis mount is a different question.

Machined metal is fully dense and isotropic. A 6061-T6 or 17-4PH part holds ±0.005 mm and keeps it after thousands of load cycles. Printing a metal part and then machining it can be reasonable for internal cooling channels, but for a solid block with a few bores, cutting from bar stock is faster and cheaper.

We run 127 high-precision CNC machines across three wholly-owned plants, including 16 simultaneous 5-axis machining centers and 16 mill-turn centers. Maximum processing size reaches 4,000 mm, with work envelopes from 500 × 500 × 450 mm up to 4,000 × 400 × 150 mm. That range covers most printed parts you would otherwise finish by hand.

The practical workflow is hybrid. Print the complex geometry as a near-net blank, then machine the datums, bores, faces and threads. This keeps the internal channels that only additive can make, and puts the tolerance where it belongs. For a solid part, skip printing entirely. From one prototype to 10,000+ part runs, with no minimum order quantity, machining stays predictable.