NZ CNC machining growth: what actually drives it

• 1
  Why NZ CNC machining growth is a sourcing story, not a factory story
• 2
  What five-axis capability changes about part design
• 3
  Where the tolerance budget really goes
• 4
  Speed comes from setup planning, not from rushing the cut
• 5
  Material choice and its machining consequences
• 6
  How NZ buyers evaluate a machining partner
• 7
  Which machine class fits the part

New Zealand buys more machined parts than it used to, but the number of local machine shops has not grown at the same rate. That gap is the whole story. Design teams in Auckland, Christchurch and Wellington now release products that mix small batch runs, frequent revisions and short launch windows. A single shop has to cover prototype, pilot and low-volume production without the part leaving the country twice.

The parts themselves changed too. Housings that used to be castings are now machined from billet because the volume does not justify tooling. Sensor bodies, manifolds and brackets carry more features per part, and those features often sit on five or six faces. Every extra face is another setup, another datum, another chance to stack tolerance. That is the force behind NZ CNC machining growth: fewer parts, more geometry per part.

So when people ask about growth, the useful question is not how many spindles are running. It is whether the supply chain can quote fast, hold ±0.005 mm when the drawing asks for it, and ship in days rather than weeks. Those three things decide whether a program stays local or moves offshore.

For a NZ buyer the practical filter is simple. If a part has tight true position between features on different faces, if the material is hard to re-fixture, or if the revision cycle is faster than a month, the setup count matters more than the hourly rate.

• 1
  Fewer parts, more featuresOne housing replaces three brackets.
• 2
  Shorter revision cyclesDesigns change between runs, not after tooling.
• 3
  Setup count drives costEach new face adds a datum and a tolerance stack.

A simultaneous 5-axis center moves the tool and the workpiece at the same time. On a part with angled faces, that means one setup instead of four or five. It matters because every re-fixture introduces a new positional error. If you clamp the same part four times and each setup holds ±0.02 mm, the final true position between the first and last feature is not ±0.02 mm, it is worse.

Simultaneous motion also lets a short, stiff tool reach into pockets that a long tool would have to reach from above. Short tools chatter less. Less chatter means you can hold Ra 0.8–1.6 μm on a wall without a finishing pass on a separate machine, and it means the tool survives the run.

There is a boundary here. Five-axis work pays off when the geometry is genuinely complex, not when the part is simple and the shop just owns a fancy machine. A flat plate with holes on one face runs better and cheaper on a 3-axis mill. The right question is how many faces carry toleranced features, not how impressive the machine list looks.

Rotary table capacity is the other limit. A Ø400 mm table sets the practical envelope for parts that need to rotate through the cut. Beyond that, the part either goes on a larger machine or gets split into setups, and the setup count comes back.

• 1
  One setup, one datumAngled faces machined without re-clamping.
• 2
  Shorter toolsLess deflection, better wall finish.
• 3
  Not for flat partsSimple geometry runs faster on 3-axis.

Drawing tolerance is not the same as process capability. A ±0.005 mm callout on a single bore is routine on a good machine with a rigid setup and a warm spindle. The same callout between two bores on opposite faces is a different job, because the error comes from the fixture, the thermal drift and the probe, not from the cutter.

Thermal drift is the quiet one. A spindle that runs for two hours grows. On a part with a 300 mm span between critical features, a 2 °C change in the shop moves the metal enough to eat a meaningful share of a ±0.005 mm budget. This is why in-process probing and temperature-stable inspection matter more than the machine brochure.

Inspection closes the loop. 100% inspection before shipment means raw material check, in-process monitoring and a final pass, with reports available on request. If a feature is going to be checked on a CMM at the customer end, it should be probed the same way here, on the same datum scheme. Otherwise the two measurements disagree and the argument takes a week.

For NZ buyers, the useful move is to mark which dimensions are functional and which are reference. Send a drawing where every dimension carries ±0.005 mm and the quote reflects risk, not the work. Mark the three or four dimensions that actually matter and the shop can spend its time there.

• 1
  Single feature vs feature-to-featureSame number, very different difficulty.
• 2
  Thermal driftA 2 °C shop swing moves a 300 mm span.
• 3
  Shared datum schemeProbe the way the customer will measure.

Cycle time is rarely the bottleneck on a short run. Setup, programming and first-article inspection usually take longer than the cutting. That is why a shop can start production within 24 hours and still ship in 3–5 days: the planning happens while the material is being cut, not after.

A free DFM analysis within 12 hours of quotation changes what the customer can do with the feedback. If a 2 mm internal corner has to be a 2 mm corner, the shop needs a small tool and a slow pass. If it can be 3 mm, the same feature comes off faster and the tool lasts longer. That conversation has to happen before the program is posted.

Material availability sets a hard floor. Aluminium 6061 and 7075, stainless 303 and 316L, and titanium TC4 are stocked in common sizes. Inconel and beryllium copper are not, and neither is every temper of 17-4PH. When the alloy is unusual, the schedule bends around the mill order, not around the machine.

Surface finishing adds its own queue. Anodizing, electroless nickel, bead blasting and laser marking each run as a separate step. A part that needs hardcoat anodizing plus laser marking at 1.5 mm character height is not a one-day finish, whatever the machining time says.

• 1
  Plan while cuttingProgramming overlaps with material prep.
• 2
  DFM before programmingCorner radius changes the whole cycle.
• 3
  Unusual alloysStocked vs mill-order decides the date.

Aluminium 6061-T6 is the default for housings and brackets. It cuts fast, holds tolerance well and takes anodizing cleanly. 7075 gives higher strength but machines with more spring, so thin walls need lighter passes and more support. If a part needs stiffness and low weight, 7075 is worth the extra cycle time.

Stainless 303 is the free-machining grade and the right call for shafts and fittings that will not be welded. 316L is the choice for anything in a corrosive or medical environment, but it work-hardens, so the tool has to stay in the cut. Stopping and restarting on a 316L surface hardens the skin and dulls the next pass.

Titanium TC4 (Ti-6Al-4V) and Inconel sit at the other end. They need low surface speed, rigid setups and a lot of coolant. They are also the materials where five-axis pays for itself, because one setup avoids the re-fixture that would otherwise scrap the part.

Plastics behave differently again. POM and PEEK hold tolerance but move with temperature, so a part measured hot will not match the drawing. ABS and PC cut easily but burr, and deburring a plastic part is a manual step that shows up in the lead time. Carbon fibre needs diamond tooling and dust control.

• 1
  6061-T6Default for housings, good anodizing.
• 2
  303 vs 316LFree-machining vs corrosion resistance.
• 3
  TC4 and InconelLow speed, rigid setup, one setup matters.
• 4
  POM and PEEKMeasure at 20 °C, not off the machine.

Certification is a filter, not a ranking. ISO 9001:2015 covers the quality system. IATF 16949:2016 applies to automotive work. ISO 13485:2016 matters for medical devices. ISO 27001:2022 covers information security, which matters when the drawings are the product. A shop that holds all four can serve several industries without changing its paperwork.

Capacity is the second filter. 127 high-precision CNC machines, including 16 simultaneous 5-axis centers, 12 four-axis mills, 27 three-axis machines and 16 mill-turn centers, sets the range of work that can be absorbed at once. A shop with one 5-axis machine queues everything behind it.

Confidentiality is the third. Uploads are secure and confidential, and an NDA is available on request. For a NZ team sending a pre-launch design, that is not a formality, it is the condition for sending the file at all.

The last filter is honesty about fit. A supplier who says a flat bracket should run on a 3-axis machine is more useful than one who quotes five-axis for everything. That answer tells you the shop is reading the drawing.

• 1
  CertificationsMatch the standard to the industry served.
• 2
  Machine mixCapacity across classes, not one bottleneck.
• 3
  NDA and secure uploadCondition for sending pre-launch files.
• 4
  Honest routingThe right machine, not the most expensive one.