3D Printing or CNC Machining: How to Decide Per Part

Designers usually ask which process is better. The more useful question is which of the three constraints on your part is the tightest: geometry, tolerance or quantity. Once you know that, the process picks itself.

Geometry decides first

Milling removes material with a rotating tool, so every feature needs tool access. Internal cavities, closed lattice structures, conformal channels and undercuts either force multiple setups, a split design or they are simply not machinable. FDM does not care about access, but it cares about overhangs, which need support and leave a rougher surface. If your part is a housing with internal ribs and a routed channel, printing is usually the shorter path. If it is a flat plate with a bored pattern and a ground face, milling wins.

Tolerance is where expectations break

Milled plastic parts hold roughly plus or minus 0.05 mm without much effort. Well controlled FDM holds around plus or minus 0.2 mm on smaller dimensions, and the deviation grows with part length because of thermal shrinkage. That gap sounds decisive but rarely applies to the whole part. Most components have two or three features that need a fit and a lot of surfaces that need nothing at all. Printing the body and reaming or milling only the critical features is often cheaper than machining everything.

Cost drivers are not comparable

• Machining cost is driven by setups, tool changes and removed volume, so a bulky part with few features is cheap and a small part with many features is expensive
• FDM cost is driven by print time and build plate occupancy, so a bulky part is expensive and geometric complexity is nearly free
• Machining has a fixed programming and setup cost per revision, printing does not
• Above roughly a few hundred parts per year, machining amortises its setup and starts closing the gap

Lead time and iteration

For a single revision cycle, printing usually returns a part in one to three days and machining in one to two weeks, depending on shop load. When a design is still moving, that difference compounds: three print iterations often fit into the time one machined iteration takes, and three iterations produce a better part than one.

Combining both is normal, not a compromise

A large share of the parts we produce leave with machined features. Bores get reamed to a fit, mating faces get faced flat, threads come from inserts or are cut after printing. Design for that from the start: add 0.3 to 0.5 mm of stock on the surfaces that need to be accurate and leave the rest as printed. You get machined accuracy where it matters and printed cost everywhere else.

If you are unsure which side your part falls on, send the STEP file and tell us the two dimensions that must be right. We will say plainly whether printing is the sensible route or whether you are better served by machining.