Precision CNC machining in New Zealand: how the supply chain actually works

• 1
  Why CNC machining in New Zealand costs what it costs
• 2
  What actually sets the tolerance on a milled or turned part
• 3
  Where 3-axis stops and 5-axis starts to pay
• 4
  Material choice changes the process, not just the price
• 5
  Local shop or offshore: the four questions that decide it
• 6
  Machining route by part profile

New Zealand has a small population spread over two islands. A machine shop there cannot pull from a deep pool of nearby subcontractors the way a shop in Guangzhou or Chicago can. Raw stock, tooling, and heat treatment often arrive by sea or air. That inbound freight and the waiting time it creates sit inside the price of every part.

The country makes up for it with engineering depth. Marine, aerospace, medical, and agricultural equipment firms in Auckland, Wellington, and Christchurch run serious design teams. The parts they need are usually complex, low-volume, and high-value. A bracket that ships 50,000 times a year is rare; a manifold that ships 30 times a year is normal.

So the real question is not whether New Zealand can machine well. It can. The question is whether a specific part should be cut there or sent offshore. That answer depends on four things: tolerance, material, batch size, and how much of your week you can spend chasing a supplier.

One more factor is easy to miss. Because local capacity is thin, a single machine going down can push a whole order back. Any plan that assumes two suppliers are interchangeable has to account for the fact that both may be booking the same week of spindle time.

• 1
  Small batches dominateTypical runs sit between 1 and 500 parts, not tens of thousands.
• 2
  Imported inputsStock, tooling, and finishing consumables mostly arrive by freight.
• 3
  Deep design talentMarine, aerospace, and medical teams specify tight work.

Tolerance is not a setting on the control. It is the sum of several errors that either cancel or add up. Machine geometry, thermal growth, tool deflection, workholding stiffness, and the metrology used to check the result all contribute. A shop quoting ±0.005 mm is claiming it can hold that stack on the features it has agreed to inspect.

Thermal drift is the one people underestimate. Aluminium grows about 23 μm per meter per degree Celsius. A 300 mm part that warms 5 °C during roughing has moved more than the tolerance band before the finishing pass starts. Shops handle this by roughing, letting the part rest, then finishing in a controlled temperature. Skipping the rest is the most common cause of a part that measures fine on the machine and fails on the CMM.

Tool deflection scales with the cube of the stick-out length. A 6 mm end mill hanging 40 mm out of the holder will bend far more than the same tool at 20 mm. That is why deep pockets and thin walls are quoted with extra passes, not extra speed.

Workholding is the quiet one. A part clamped hard enough to stop chatter is also clamped hard enough to distort. Soft jaws, vacuum plates, and sacrificial tabs exist for exactly this reason. If a drawing has a thin floor and a tight flatness callout, expect the shop to ask about it.

• 1
  Rough, rest, finishLet the part reach room temperature between passes.
• 2
  Keep tools shortDeflection rises with the cube of stick-out length.
• 3
  Clamp gentlySoft jaws and vacuum plates reduce distortion.

A 3-axis mill cuts from one direction. Every new face means a new setup: unclamp, reposition, re-zero, reclamp. Each setup adds handling time and a fresh chance to introduce error. On a part with four machined faces, that is four chances.

A 5-axis center tilts the tool or the table so the cutter reaches angled features in one setup. The gain is not speed. The gain is that datums stop moving. A hydraulic manifold with ports on five faces can be cut without ever losing its reference, which is why the tolerance on the port-to-port relationship holds.

The trade is programming time and rigidity. Five-axis toolpaths take longer to prove out, and a tilted setup is usually less stiff than a flat one. For a simple plate with holes on one face, 3-axis wins on both cost and cycle time. Adding axes to that part buys nothing.

A practical rule: count the faces that carry a tight relationship to each other. One or two, stay on 3-axis. Three or more, price the 5-axis route. GreatLight runs 16 simultaneous 5-axis centers alongside 27 three-axis machines, so both routes are quoted on the same drawing when the answer is not obvious.

• 1
  Setup count drives errorEvery reclamp resets the datum chain.
• 2
  5-axis protects datumsAngled features cut without losing reference.
• 3
  Not always cheaperSimple single-face parts run better on 3-axis.

Aluminium 6061 and 7075 cut fast and hold good finish. They also move when you remove material, because residual stress from the mill is released as the section thins. A 7075 plate machined down to a thin web will bow. The fix is symmetric roughing and a stress-relief step, not a slower feed.

Stainless 304 and 316L work-harden. A light finishing pass with a dull insert will rub instead of cut, and the surface gets harder as you fight it. Shops run 316L with sharp tools, positive rake, and enough feed to stay under the hardened layer. Titanium TC4 (Ti-6Al-4V) behaves the same way but worse, and it also carries heat into the cutter.

Plastics are their own problem. POM and PEEK cut cleanly with sharp, polished flutes and high spindle speed, but they hold heat and expand. Measure them after they cool, not on the bed. Carbon fibre needs diamond or coated tooling and dust extraction, and the finish callout should say whether cut fibres are acceptable at the edge.

Inconel and the harder nickel alloys sit at the far end. They are machinable but slow, and the tooling cost per part is real. If a design can use 17-4PH instead of Inconel and still pass its load case, that single substitution often changes the quote more than any negotiation.

• 1
  Stress relief firstThin aluminium webs bow unless stress is released.
• 2
  Sharp tools on 316LLight passes work-harden the surface.
• 3
  Measure plastics coldThermal expansion skews in-process numbers.

Question one: how tight is the tolerance relative to the feature size? A ±0.005 mm callout on a 20 mm bore is routine for a well-equipped shop anywhere. The same callout on a 900 mm frame is a different job, and only some shops have the metrology to prove it.

Question two: how fast do you need the first article? If the answer is this week, geography matters. If the answer is next month, freight time disappears into the schedule and the deciding factor becomes capacity and price.

Question three: how much engineering support do you want? A DFM review that flags a 0.5 mm wall before the tool is cut saves more than a discount on the cutting rate. Ask what the supplier does with your STEP file before the machine starts.

Question four: what happens when something is wrong? Parts get rejected. The question is whether the replacement arrives in days or weeks, and who pays the freight. Write that into the terms before the first order, not after the first nonconformance.

• 1
  Tolerance vs sizeTight calls on large frames need real metrology.
• 2
  Schedule firstIf you need it this week, distance matters.
• 3
  DFM before cuttingEarly feedback beats a discount on rate.