How to switch to CNC parts production quickly

• 1
  Key takeaways
• 2
  What really changes when you switch to CNC parts production
• 3
  Setup and fixturing: the cost that decides whether the run is viable
• 4
  Where five-axis actually saves time in production
• 5
  Material and tooling choices that keep the run stable
• 6
  Keeping quality intact while running fast
• 7
  A practical sequence for the switch
• 8
  Which machining route fits which part
• 9
  When CNC production is the wrong answer

A prototype is a geometry question. Production is a repeatability question. The drawing may not change at all, yet everything around it does: how the blank arrives, how the part is held, how the tool wears, how the operator knows the run is still good. That is the real content of a switch to CNC parts production, and it is where most schedules slip.

The first number to look at is not spindle speed. It is setup time. On a run of 20 parts, an hour of fixture work adds three minutes per part. On a run of 2,000 parts, the same hour is almost invisible. This is why a job that felt slow as a prototype can feel fast as production, and why a rushed fixture can sink a short run.

The second number is the tolerance stack across the run, not on one part. A machine that holds ±0.005 mm on a cold morning may drift as the spindle and coolant warm up. Production planning has to absorb that drift, usually by choosing datums and locating features that keep the critical dimension on a single setup.

None of this is exotic. It is mostly the discipline of deciding, before the first chip, which features are critical, which faces can be reached in one orientation, and what the operator will measure at the machine.

Vises and soft jaws handle most prismatic parts. Cut the jaws to the blank profile, not the finished profile, and leave 0.5–1 mm of stock so the part can be flipped without losing location. A jaw that has been cut in place on the machine repeats far better than a generic jaw set, simply because the locating surface matches the spindle.

When the part has features on four or five faces, a tombstone or a trunnion becomes cheaper than four separate setups. One load, one datum, all faces reached by rotating the table. The trade-off is fixture cost, so tombstones make sense on runs where the same family repeats. For a one-off, four soft-jaw setups are usually cheaper.

For round parts, a mill-turn center removes the second operation entirely. Turning and milling happen on one spindle, so concentricity between a turned bore and a milled flat is set by the machine, not by a re-chuck. This is the single biggest accuracy gain in most production switches, and it is the reason 16 mill-turn centers sit in our own lineup.

The mistake to avoid is clamping on a finished surface. If the second operation grips a face that was already machined to tolerance, jaw pressure can distort it. Grip on stock, or use a low-pressure fixture with a supporting nest.

• 1
  Soft jaws cut in placeMatch the blank profile, not the finished part, and keep 0.5–1 mm of clamping stock.
• 2
  Tombstone over multiple vicesBetter when four or more faces need machining on a repeating part family.
• 3
  One setup per critical featureKeeps the tolerance stack inside a single work coordinate system.

Five-axis does not make a spindle cut faster. It removes setups. A part that needs five faces machined goes from five fixtures to one, and every removed fixture removes a re-datum, a re-clamp and a chance of scrap. On a 200-piece run with three saved setups at 40 minutes each, that is 40 hours of machine time returned to the schedule.

The second gain is tool access. Short, rigid cutters can reach angled features because the table tilts to meet them. Short tools deflect less, so you can hold a tighter surface finish without slowing down. A wall that needs Ra 0.8–1.6 μm is often easier to achieve with a tilted short tool than with a long reach tool on a 3-axis machine.

The third gain is on contoured surfaces. Impellers, turbine housings and organic brackets are cut with the tool axis kept near normal to the surface, which spreads the stepover evenly. The result is a finish that does not need hand blending, and hand blending is one of the slowest, least predictable steps in a production cell.

Five-axis is not automatic. Programming takes longer, simulation is mandatory on complex parts, and the machine needs a post processor that matches its kinematics. For a simple bracket, a 3-axis machine with a good fixture will beat a 5-axis machine with a weak one.

Aluminium 6061 and 7075 cut freely and tolerate aggressive parameters, but they also move. Roughing with a 12–16 mm carbide end mill at 0.15–0.25 mm per tooth, then leaving 0.3 mm for a finishing pass, keeps the heat out of the part. Skip the semi-finish and the final dimension will drift as the part cools on the bench.

Stainless 304 and 316 work-harden. A light rubbing pass destroys the surface and the next pass cuts through a harder skin. The fix is a constant feed that stays above the work-hardening threshold, usually 0.05 mm per tooth or more, with flood coolant and no dwelling in the cut. 17-4PH behaves better in the H900 condition than in the annealed state.

Titanium Ti-6Al-4V and Inconel are the opposite problem. Heat stays at the cutting edge and tool life drops fast. High-pressure coolant, conservative radial engagement around 5–8% of cutter diameter, and a fresh edge per part are normal practice, not caution. These materials rarely belong on a fast-track production line without a tool-life plan.

Plastics and carbon fibre bring their own rules. POM and PEEK cut cleanly but hold chips; air blast beats flood coolant because the material does not swell. Carbon fibre dust is abrasive and hazardous, so it needs extraction at the cut, not at the enclosure.

• 1
  AluminiumRough heavy, semi-finish, then take 0.3 mm off for the final pass.
• 2
  Stainless 304/316Never rub. Keep feed per tooth at or above 0.05 mm.
• 3
  Ti-6Al-4VHigh-pressure coolant and 5–8% radial engagement.
• 4
  POM and PEEKAir blast instead of flood coolant to avoid swelling.

Speed and tolerance are not enemies, but speed and unmeasured drift are. The parts that fail a production switch usually fail because nobody checked the trend. A single good first article proves the program, not the process. The process is proven when the fiftieth part measures the same as the first.

Thermal drift is the usual cause. A spindle that has run for two hours sits in a different geometry than one that has run for ten minutes. Shops that hold ±0.005 mm across a long run either warm the machine up before the batch or schedule the tightest features after the warm-up window. Both work. Ignoring it does not.

Tool wear is the second cause. A finishing tool that has cut 40 parts may still look fine and still be 0.01 mm undersize. Counting parts per edge and changing on schedule removes the guesswork, and the cost of one extra tool is small compared with a scrapped batch.

Inspection should be layered. A raw material check catches wrong grade or hardness before machining. In-process checks catch drift. Final inspection before shipment catches handling damage and missed features. Reports can be issued on request, which matters for aerospace, medical and automotive buyers who need traceability in their own quality file.