Mastering Low Volume CNC: How the Process Actually Works

• 1
  How the process differs from a production run
• 2
  What part geometry suits small batches
• 3
  Material and tolerance choices that keep cost down
• 4
  Why inspection is the hidden cost driver
• 5
  Where the process stops being the right answer
• 6
  When to choose which route

Cutting metal is the same whether you make one part or ten thousand. What changes is everything around the cut. In a small batch the setup cost is spread over very few pieces, so the setup itself has to stay cheap. A long production line can justify a dedicated fixture that takes two days to build because it pays back over 20,000 cycles. Five parts cannot absorb that.

So the shop leans on standard vises, soft jaws, modular plates and probing instead of dedicated workholding. The part program is written to cut more of the feature set in one or two setups. Cutters are chosen for availability rather than minimum cycle time. All of this is a trade: slightly longer cycle time per part in exchange for far less non-recurring engineering.

That trade has a limit. Once quantities climb past a few hundred pieces, the per-part cycle time starts to dominate the total cost, and a fixture that saves 40 seconds per part wins. This crossover point is the real boundary of the process.

• 1
  Setup dominatesUnder 50 parts, setup and programming are often 40–60% of the quoted price.
• 2
  Cycle time is secondaryA 10% slower cycle is cheaper than a fixture that costs two extra days.
• 3
  Crossover near a few hundredPast that, dedicated tooling usually pays for itself.

Small batches suit parts with high value per kilogram and moderate geometric complexity. Brackets, housings, manifolds, heat sinks, impellers, surgical instruments, robot end-effectors and prototype drivetrain parts all fit this profile. They need real material properties, tight fits, or a surface that a printer cannot deliver.

The awkward cases are thin walls under 1 mm, deep pockets with an aspect ratio beyond 4:1, and features that can only be reached from five directions at once. None of these are impossible, but each adds a setup, a custom cutter, or a risk of chatter that shows up as a scrapped part. With a batch of three, one scrapped part is a 33% loss.

Undercuts, internal O-ring grooves, and cross-drilled holes that intersect at shallow angles all force extra workholding. If the design can be adjusted so that three faces are reachable from one direction, the quote usually drops by a visible margin.

• 1
  Good fitBrackets, housings, manifolds, instrument bodies, end-effectors.
• 2
  Watch outWalls under 1 mm, pockets deeper than 4× width.
• 3
  Redesign winsConsolidating features onto fewer faces cuts setup count.

Aluminium is the default for a reason. Grades 6061 and 7075 cut fast, hold a good finish, and are stocked in most plate thicknesses. Stainless 303 and 304 machine cleanly but run slower, so the same part in 316L can cost noticeably more. Titanium TC4 and Inconel are reserved for cases where the service temperature or corrosion load leaves no alternative.

Tolerance is where quotes diverge most. A general tolerance of ±0.1 mm needs no special attention. Tightening a single bore to ±0.005 mm means a finishing pass, a smaller step-over, and sometimes a temperature-stable room. Tightening every dimension on the drawing multiplies the inspection time, not just the cutting time.

A practical rule: tolerance only the features that mate with something. Mark the rest as general tolerance. That one change often removes a full inspection step from the quote.

• 1
  Start with 6061-T6Fast to cut, widely stocked, good finish.
• 2
  Tolerance where it matesCall out critical fits only; leave the rest general.
• 3
  Finish drives costRa 0.8–1.6 μm is standard; Ra 0.2–0.8 μm adds polishing time.

Cutting a part takes minutes. Verifying it can take longer. On a first article, a CMM program has to be written, datums established, and every toleranced feature probed. That work does not shrink when the batch shrinks, which is exactly why inspection is a fixed cost in small runs.

Shops handle this in different ways. Some probe the first part in the machine, then check a sample from the rest. Others run a full layout on every part when the application demands it, such as medical or aerospace hardware. The difference between those two approaches can double the price of a ten-piece order.

Ask what inspection level you are paying for. If a report is needed, say so at quoting time rather than after the parts ship. Documentation prepared after the fact costs more and delays delivery.

• 1
  First article is fixed costCMM programming does not scale down with quantity.
• 2
  Sample vs full layoutChoose deliberately; the price gap is large.
• 3
  Request reports earlyDocumentation is cheaper when planned into the run.

Machining loses to casting somewhere around 2,000 pieces for a part with a stable design. It loses to injection moulding far earlier for plastic housings, often at a few hundred units, because the tool cost is lower than people assume. It loses to sheet metal whenever the part is essentially flat and can be folded.

It also loses when the geometry is genuinely freeform and hollow. A lattice or an internal channel network that a printer builds in one shot may need four setups and a custom cutter on a mill. If the load is light and the material is not critical, printing is the cheaper path.

The honest answer is that small-batch machining wins on material properties, tolerance, and surface finish. If none of those three matter for your part, another process is probably cheaper.

• 1
  Loses to castingAround 2,000+ pieces with a frozen design.
• 2
  Loses to printingFreeform hollow geometry under light load.
• 3
  Wins on three thingsMaterial properties, tolerance, surface finish.