CNC Aluminum Small Batch Manufacturing: How It Actually Works

• 1
  Key takeaways
• 2
  Why aluminum small batch manufacturing is a setup problem, not a cutting problem
• 3
  When 3-axis, 4-axis or 5-axis actually changes the outcome
• 4
  What ±0.005 mm really costs in a small batch
• 5
  Alloy choice changes cycle time more than most engineers expect
• 6
  How a small aluminum batch moves through the shop
• 7
  Route selection by part geometry and quantity

Cutting aluminum is fast. A 12 mm carbide end mill in 6061 runs at 300–500 m/min surface speed and 0.1–0.2 mm per tooth without much drama. On a 500-part run, spindle time drives cost. On a 30-part run the spindle barely matters. What decides the price of aluminum small batch manufacturing is everything around the cut: fixture design, first-article inspection, workholding changes and programming. Those hours are fixed, so they land on every part in the batch.

A simple plate with two flat faces and six holes might need one vise setup and 20 minutes of programming. The same plate with four tapped side holes and a 30° angled boss needs either a second and third setup or an indexer. That jump from one setup to three can double the quoted price even though the metal removed is identical. Buyers usually blame the machine hour rate. The real number is setup count.

There is a crossover point where setup cost stops dominating. For most 6061 parts between 20 and 200 pieces, the per-part price drops gradually, not sharply, because the fixture is reused and the operator is dialed in. Run 500 pieces and the curve flattens, because you are now paying mostly for material and spindle time. This is why a 50-piece quote and a 500-piece quote rarely differ by a factor of ten.

Material choice interacts with setup too. A 7075 bracket needs slower feeds and shallower cuts than 6061, so cycle time rises maybe 20–30%. A 2024 part may need stress relief between roughing and finishing to hold flatness. Neither changes the setup count, but both change how much time the machine holds the part.

• 1
  Count setups, not just partsEach new face or angle adds fixturing and re-datum time.
• 2
  Ask for a setup breakdownA quote that separates setup from cycle time shows where cost sits.
• 3
  Group features on fewer facesMoving a hole to an accessible face can remove an entire setup.

A 3-axis mill reaches one face of the part at a time. For a flat cover plate or a manifold block with all features on two opposite sides, that is enough, and it is the cheapest route. The part gets flipped once, re-datumed, and finished. No rotary table is involved, so there is no additional fixture to build. Simple geometry, low setup count.

A 4-axis machine adds a rotary table, usually Ø400 mm or smaller. Now the part can be indexed to four sides without re-clamping. A shaft with flats, cross-drilled holes and milled slots is the classic case: one program, one datum, four faces. Datum error from flipping disappears. Cycle time often drops because the operator is not re-clamping between operations.

A 5-axis machine adds tilting motion, which lets the tool reach an angled face without a dedicated fixture. The benefit is not speed. It is reaching features that would otherwise need a custom angle plate or a second operation on a different machine. Parts with compound angles, deep pockets with draft, or features on five sides fit this class. So do parts where position tolerance between angled faces is tight.

The limit is size and access. A Ø400 mm rotary table will not hold a part that swings wider than its envelope. Very long parts with tight tolerance on both ends may need a mill-turn center instead, which combines turning and milling in one setup. And a 5-axis machine cannot fix bad geometry: if a feature sits in a deep pocket with no tool access, no amount of axis count helps.

• 1
  3-axisFlat parts, features on two faces, lowest setup cost.
• 2
  4-axisShafts and blocks needing four-sided access without re-clamping.
• 3
  5-axisCompound angles and five-sided features that would otherwise need custom fixturing.
• 4
  Mill-turnLong parts where turning and milling must share one datum.

A tolerance callout is a cost signal. General dimensions at ±0.1 mm are free, in the sense that they follow from a normal program. Tightening a fit to ±0.005 mm means the operator must control tool wear, spindle growth and thermal drift, and inspection must verify each critical dimension. On a small batch, that inspection time is spread over few parts, so the per-piece cost climbs.

Surface finish behaves the same way. An as-machined face at Ra 1.6–3.2 μm comes straight off a sharp cutter. A sealing face at Ra 0.8–1.6 μm may need a finishing pass with a smaller stepover. Below Ra 0.8 μm the part often leaves the mill for lapping or polishing, which is a separate operation with its own handling. Every one of these steps adds a setup or a machine move.

The practical rule: put tight tolerances only where the function needs them. A mounting hole pattern for locating pins deserves ±0.005 mm. A clearance hole for a bolt does not. If the drawing shows tight tolerance on every dimension, the quote will reflect the worst case, and some of that money is wasted.

Thin walls are the other cost trap. A 1 mm wall in a 40 mm tall pocket will chatter and may spring after the vise releases. Aluminum moves when material is removed, especially in 7075. Thicker walls, or a design that lets the machinist rough, stress-relieve and finish, hold tolerance more reliably. A wall under 1 mm is possible but it is a special case, not a default.

• 1
  ±0.1 mmStandard; no special control needed.
• 2
  ±0.05 mmCommon for fits; add a finishing pass and inspection.
• 3
  ±0.005 mmCritical only; expect extra measurement and thermal control.
• 4
  Ra 0.8–1.6 μmMachined finish for seals and sliding faces.

6061-T6 is the default for a reason. It machines at high speed, holds a good finish, welds and anodizes predictably, and costs less than 7075. For brackets, housings, plates and most prototype hardware, it is the right starting point. If a part works in 6061, switching away usually adds cost without adding function.

7075-T6 is roughly twice the strength and noticeably less forgiving. It cuts with a sharper edge and lower feed, and it is more prone to distortion when a lot of material comes off one side. A 7075 part with an asymmetric pocket may need roughing, a stress-relief pause, then finishing. That pause can span a shift or a day, which shows up in lead time, not just price.

2024 is common in aerospace work for its fatigue properties. It also moves. Flat parts with tight flatness callouts often need multiple passes and sometimes a post-machining straightening step. 5052 and 5083 are softer, gummier and better suited to formed sheet than to milled structure. ADC12 is a casting alloy; machining it means porosity can appear mid-cut, so wall thickness and surface expectations need to be set accordingly.

The engineering takeaway: pick the alloy from the load case and the environment, then accept the machining consequence. Choosing 7075 for a part that sees no real stress buys nothing and costs cycle time. Choosing 6061 for a high-load bracket may force a thicker section, which changes the whole design.