Custom CNC Processing: How Wholesale Batches Really Work

• 1
  Why custom CNC processing is a setup problem, not a volume problem
• 2
  Which part geometry actually benefits from 5-axis custom CNC processing
• 3
  How tolerance and surface finish scale across a batch
• 4
  Where the unit price actually moves
• 5
  Inspection strategy for volume: what to check and how often
• 6
  When a part suits 3-axis, 4-axis or 5-axis volume work

A single machined part is priced by machine time. A thousand identical parts are priced by setup, fixturing and inspection. That shift is the whole story of custom CNC processing at volume. The spindle still cuts at the same feed and speed, but the minutes around the cut start to dominate the cost of the order.

Each time a part is clamped and the zero point is re-found, the machine adds positional error. On a 3-axis machine, a part with features on five faces may need four or five separate setups. Each setup carries its own datum shift and its own chance of a scrapped part. In a run, that error repeats across the batch.

Five-axis work reduces the number of setups, not the number of cuts. A trunnion table with a Ø400 mm rotary table lets the tool reach the part from any angle in one clamping. The gain is geometric consistency between parts, not a faster spindle. That is why volume work and complex geometry tend to meet in the same cell.

So when a customer asks for a cheaper unit price, the honest answer is usually about fixtures, probing and tool life, not about cutting the part faster. Cutting faster has limits. Removing a second setup does not.

• 1
  Setup countDrives cost more than cycle time once quantity passes roughly a few hundred pieces.
• 2
  Datum stabilityOne clamp means fewer stack-up errors between features.
• 3
  Inspection strategyBatch sampling only holds if the process itself is stable.

Five-axis is not automatically better. It earns its place when the part has undercuts, deep cavities, angled faces or features that would otherwise need a second or third fixture. A simple plate with holes on one face is often cheaper on a 3-axis machine with a good vise, even in large quantity.

The signal to watch is access angle. If more than two faces carry toleranced features, or if a feature sits at an angle that a standard tool cannot reach without a re-clamp, five-axis starts to pay. The same applies to deep pockets where a long thin tool would chatter: tilting the part lets a shorter, stiffer tool reach the floor.

There is a size limit to keep in mind. Five-axis trunnion machines have a swing envelope; on our compact centers that is around 500 × 500 × 450 mm and 500 × 310 × 200 mm, with medium platforms at 750 × 1,150 × 550 mm and 600 × 600 × 600 mm. For long parts we run travel up to 4,000 × 400 × 150 mm on large platforms. A part that fits a 3-axis envelope in one setup rarely needs the extra axes.

Material matters too. Aluminium 6061 and 7075 cut freely and tolerate aggressive five-axis paths. Titanium TC4 (Ti-6Al-4V) and Inconel punish tool life, so the setup savings matter more but the cycle time is harder to compress. In those alloys, reducing setups is the main lever you have.

• 1
  Good fitUndercuts, angled features, deep cavities, features on 3+ faces.
• 2
  Poor fitFlat plates, single-face hole patterns, simple turned parts.
• 3
  Hard alloysTC4 and Inconel: setup savings outweigh cycle-time gains.

Tolerance does not get easier with quantity. It gets harder, because the machine drifts as the spindle warms and the tool wears. A shop holding ±0.005 mm on a prototype is doing something different from a shop holding ±0.005 mm on part 900 of a run. The second one is managing thermal growth and tool wear, not just dialing in a cut.

Surface finish follows the same logic. Ra 0.8–1.6 μm is a normal machined finish and is stable across a run with the right cutter and stepover. Ra 0.2–0.8 μm usually means a separate finishing pass, slower feed or a secondary operation such as fine polishing. That extra step applies to every part in the batch, so it multiplies.

For most functional parts, a sensible split is to tolerance only the features that touch something else. Datum faces, bearing bores, sealing surfaces and mating holes get the tight callout. Cosmetic faces and clearance holes stay at Ra 1.6–3.2 μm as-machined. This single decision often removes more cost than any machine upgrade.

If the drawing calls ±0.005 mm on every dimension, ask whether that is a function requirement or a habit carried over from a previous supplier. Tightening a callout that nobody measures adds cost without adding value.

In a wholesale run, cost is a sum of fixed and variable parts. Fixed: programming, fixture design, first-article inspection, tooling. Variable: material, machine time, tool wear, per-part inspection. The fixed part is spread across quantity, so unit price falls steeply at first and then flattens.

That flattening point is the useful number. Below it, adding quantity cuts unit price fast. Above it, you are only buying material and machine time, and the curve is nearly flat. If a supplier quotes a steep drop between 500 and 5,000 pieces, ask what changes. Usually nothing physical does, which means the first quote was padded.

Setup reduction is the other lever. Moving a part from four setups to one can remove a fixture, a re-clamp and a manual touch-off. Those are hours of skilled labor per batch, and they are removed once, not per part. That is why the same geometry can price very differently between a 3-axis shop and a five-axis shop at the same quantity.

Inspection should scale with process stability, not with order size. If the process is stable and the first article is signed off, sampling is defensible. If the process is still being tuned, every part is a prototype and should be treated as one.

• 1
  Fixed costProgramming, fixture, first article. Spread over quantity.
• 2
  Variable costMaterial, cycle time, tool wear. Flat per part.
• 3
  Flattening pointWhere unit price stops falling. Ask where it sits.

Volume machining lives or dies on whether the process is stable. If it is, you can inspect a sample and trust the rest. If it is not, a sample tells you almost nothing. The first job of a quality plan is therefore to prove stability, not to measure parts.

A practical sequence starts with a raw material check, then a first-article inspection against the drawing, then in-process monitoring at defined intervals, then a final inspection before shipment. Reports are available on request. The interval is set by how fast the process drifts, which depends on tool life and thermal load.

For tight features, in-process probing on the machine is often better than checking after the part comes off. Probing catches a drifting tool before it produces a batch of scrap. Offline CMM checks are still needed for form and position, but they are slower and better used as confirmation.

One number worth asking about is the qualification rate. Ours is 99.99% across inspected shipments. That figure only means something alongside the tolerance band and the inspection method, so ask for both when you compare suppliers.