Small Batch Parts CNC Machining: How Setup Cost and Tolerance Drive the Method

• 1
  What small batch parts CNC machining actually means
• 2
  Where the money goes in a small run
• 3
  When 3-axis, 4-axis and 5-axis each make sense
• 4
  Tolerance bands and what they cost at low volume
• 5
  Material behavior in short runs
• 6
  Where small batch CNC stops being the right answer
• 7
  Choosing the method for a small batch

Small batch parts CNC machining covers runs that are too large for a single prototype and too small for hard tooling. In practice that is roughly 2 to 500 pieces. Some shops stretch the term to 1,000. The upper bound is not a rule, it is where the economics of a fixture and a program start to pay for themselves.

The defining constraint is not the spindle. It is the fixed work around the cut. Programming, workholding design, first-article inspection and machine setup do not scale down with the order quantity. Ten parts carry the same setup burden as one part, spread over ten pieces instead of one.

That is why a shop with no minimum order quantity can still quote 10 pieces at a higher unit price than 1,000 pieces. The metal cost is nearly flat. The setup amortization is not.

For the engineer, the practical consequence is simple. At low volume, design decisions that reduce setup and fixturing matter more than decisions that reduce cycle time. Saving 30 seconds per part is worth almost nothing across 20 pieces. Saving one fixturing operation is worth a lot.

A quote for small batch parts CNC machining breaks into four buckets: material, programming and setup, machining time, and inspection. At 10 pieces, programming and setup often dominate. At 500 pieces, machining time usually takes over.

Programming includes CAM work, toolpath verification and, on complex geometry, a simulation pass. For a 3-axis part with simple pockets this can be under an hour. For a 5-axis part with blended surfaces and tight access, it can run several hours before a chip is cut. That cost lands once.

Workholding is the quiet cost driver. Soft jaws machined to the part profile, vacuum plates, custom clamps and sacrificial tabs all take time to design and cut. On a 10-piece order, a machined soft jaw that takes 45 minutes to make adds 4.5 minutes of cost per part. On a 500-piece order it is negligible.

Inspection is the fourth bucket. Our standard is 100% inspection before shipment, with raw material check, in-process monitoring and final inspection. First-article inspection on a new geometry is heavier than the steady-state check, and at low volume it is not diluted by quantity.

The axis count should follow the number of part faces that need machining and how the datums relate. If a part has features on three orthogonal faces, a 3-axis machine with two re-fixturings does the job. Each re-fixturing introduces a new datum stack and a new chance for position error.

A 4-axis mill adds a rotary table, typically Ø400 mm on our machines. That lets you machine around the part in one setup for cylindrical and prismatic work. It removes one or two re-fixturings compared with 3-axis, which is exactly the saving that matters in a small run.

A 5-axis machine adds tilt on top of rotation. Use it when undercut geometry, deep cavities with short tools, or blended surfaces cannot be reached without the part being moved. Five-axis also lets you keep a short, rigid tool at an angle instead of a long tool straight down, which improves surface finish and reduces chatter.

Five-axis is not automatically better. It costs more per hour and needs more programming. On a simple bracket with two flat faces, 5-axis adds cost and removes nothing. The decision is geometric, not aspirational.

Our achievable tolerance is ±0.005 mm (±0.0002 in). That is a capability limit for the right geometry, material and setup, not a default on every drawing. Calling out the tightest tolerance on all dimensions raises cost and inspection time with no functional gain.

Tolerance is a stack, not a single number. Fixture error, thermal drift, tool wear and machine positioning all add in. On a small run, thermal drift is a smaller factor than on a long run because the spindle has less time to heat the machine structure. Tool wear is a larger relative factor because a worn tool is replaced less often per part.

Surface finish follows a similar logic. As-machined is Ra 1.6–3.2 μm. A high-finish cut reaches Ra 0.8–1.6 μm. Fine finishing reaches Ra 0.2–0.8 μm and usually needs a separate finishing pass with a smaller stepover, which adds cycle time per part.

The engineering move is to split the drawing. Put tight tolerances only on the dimensions that carry function: a bearing bore, a mating face, a dowel hole. Leave clearance, cosmetic and non-critical dimensions loose. That single change often does more for small-batch cost than any machine swap.

Material choice affects small runs in a way it does not in mass production. You buy the stock for the run, and leftover bar or plate is often unusable for the next job. A small run in titanium or Inconel costs more per part partly because of that residual stock.

Aluminium is the easy case. 6061 and 6061-T6 cut fast, hold tolerance well and are widely available in the bar and plate sizes we need. 7075 and 2024 are stronger but less forgiving on thin walls and tend to move after roughing. On a 10-piece run, that movement is harder to absorb because you have fewer parts to average over.

Stainless grades 303 and 304 machine predictably. 316L and 17-4PH (SUS630) work-harden and need appropriate speeds and feeds to avoid rubbing. Inconel and titanium TC4 (Ti-6Al-4V) need lower cutting speeds and more tool changes, which raises programming and cycle cost.

Plastics are a different problem. POM and ABS cut cleanly. PEEK and carbon fibre are abrasive and expensive, and carbon fibre dust needs control. For any of these, send the grade and temper, not just the family name.

CNC is a subtractive process. If the part is a thin shell with uniform wall and no tight tolerances, vacuum casting or 3D printing may be cheaper at 20 pieces. If the geometry is a simple plate with holes, sheet metal fabrication will beat milling on cost and speed.

If the part needs a rubber-like durometer or a specific color match across a run, urethane casting or injection molding may fit better once quantity climbs. Metal die casting becomes competitive when the geometry is stable and the count is high enough to amortize the die.

There is also a geometry boundary. Very deep, narrow slots are hard to reach with a rigid tool. Long thin features deflect. Sharp internal corners need a tool radius. These are not defects of the process, they are its physics. Redesigning a 0.5 mm internal corner to a 2 mm radius can remove an EDM operation.

The right question is not whether CNC can make the part. It usually can. The right question is whether the drawing, the quantity and the tolerance callouts are aligned with a method that will not waste setup time on features nobody needs.