CNC machining Perth: what the process can and cannot hold

• 1
  How CNC machining Perth shops actually remove metal
• 2
  Reading tolerances and surface finish on a Perth drawing
• 3
  Where 3-axis stops and 5-axis earns its cost
• 4
  What local capacity in Perth does well, and what it does not
• 5
  From upload to first article: the checks that matter
• 6
  CNC machining Perth: local versus overseas sourcing

Every CNC process is the same idea in a different shape: a spinning cutter or a spinning workpiece, a controlled feed, and a path the controller follows. The controller reads G-code, moves the axes to a position, and the material that is not on the tool path stays behind as the part. Nothing about that changes whether the machine sits in Perth or in Dongguan. What changes is the machine, the fixturing, and the person deciding the order of operations.

The three variables that decide whether a feature is easy or hard are depth-to-diameter ratio, tool access, and how the part is held. A Ø6 mm end mill cutting a 30 mm deep pocket is a 5:1 ratio. Chips have to clear that pocket, and the tool deflects as it goes deeper. Push the same cutter to 10:1 and you get chatter, tapered walls, and a surface that reads Ra 3.2 μm or worse even though the drawing says Ra 1.6 μm.

Tool access matters more than most drawings admit. A slot that looks open in a 3D model may need a cutter long enough to reach the floor, and a long cutter is a flexible cutter. A 5-axis machine helps because it tilts the tool or the table instead of asking a long tool to do the work. A 3-axis machine can still cut the same slot, but it may need a second setup to reach it from the other side, and every setup adds stack-up error.

Workholding is the quiet variable. Thin walls move when the vise releases. A part clamped on 3 mm of stock will spring back after the cut. Shops that hold ±0.005 mm do it by planning the clamp sequence and often by leaving a finishing pass after the part is released and re-clamped lightly. That is a scheduling decision, not a machine specification.

A tolerance is a budget, and it should be spent where the part needs it. Blanket tolerances of ±0.1 mm across a whole drawing are cheap to hold. A single ±0.005 mm bore in the same part pushes the shop into a different process: temperature control, a finishing pass, and a CMM check instead of a caliper. Perth engineers writing drawings should mark the tight features and leave the rest loose.

Surface finish and tolerance are not the same thing. A part can hold ±0.01 mm and still read Ra 3.2 μm, because finish is set by feed per tooth, tool radius, and spindle speed, not by the position of the axis. If a sealing face needs Ra 0.8–1.6 μm, say so on the drawing. If it needs Ra 0.2–0.8 μm, expect a separate finishing operation and a longer cycle time.

Material changes the achievable numbers. Aluminium 6061 and 7075 cut cleanly and hold tight tolerances without much fuss. Stainless 316 work-hardens under a dull tool, so the same feed that works in aluminium will burn the edge and pull the wall. Titanium TC4 (Ti-6Al-4V) needs lower surface speed and more coolant. Inconel is slower again. A shop that quotes the same cycle time for all four is quoting a number it has not checked.

Thermal drift is real on long cuts. A machine that runs a 40 minute roughing cycle warms up, and a bore cut at minute 5 may not match a bore cut at minute 35. Good shops rough in the morning, let the part stabilize, then finish. On a 4,000 mm part, that discipline is the difference between a straight rail and a bowed one.

Most parts never need 5-axis. Prismatic parts with features on one or two faces run faster on a 3-axis machine with a good vise and a probe. Adding a fourth and fifth axis adds setup time and programming time, and it only pays back when the geometry demands it or when the number of setups would otherwise be three or more.

Five-axis earns its cost in three situations. First, contoured surfaces that a ball nose cannot reach without a long, flexible tool. Second, features on five faces of a part that would otherwise need four separate fixtures. Third, deep pockets with a floor that must be square to a side wall, where tilting the tool keeps the cutter short and stiff. In those cases, one setup replaces four, and the stack-up error from four re-clamps disappears.

Turn-mill centers handle a different family: shafts, bushings, and parts with both a turned diameter and milled flats. A mill-turn center machines both in one setup, so concentricity between the bore and the milled feature stays inside ±0.01 mm without a fixture. Splitting that part across a lathe and a mill usually costs more in fixtures than it saves in cycle time.

Size decides the machine too. A 4,000 × 400 × 150 mm travel handles long rails and beams. At the other end, a 500 × 500 × 450 mm machine with a Ø400 mm rotary table is the right tool for a compact housing that needs four sides cut. Matching the part envelope to the machine envelope is the first question a shop asks, and it should be the first question on your side as well.

Local capacity is strongest on short-run, high-mix work where the part is close to the design team. A prototype that needs to be in a test rig on Friday is a local job. So is a fixture built to hold a part that is still changing. The value is the conversation, not the spindle speed.

Local capacity is weakest on three things: hard-to-find material, tight finishes over large areas, and volume. A Perth shop may need to import 17-4PH or a specific titanium grade, and that shipping time lands on the schedule. Large anodized or hardcoat areas often go to a specialist plater, which adds a leg to the route. At 5,000 parts, unit cost is set by the machine cell, not by the postcode.

The practical split most engineers use is: prototype and bridge tooling locally, production overseas once the design is frozen. The risk in that split is a design that is frozen in name only. If the drawing changes after the first production run, the savings from the lower unit price go into the second setup. Freeze the drawing, then move the volume.

Confidentiality is a separate track. Uploads and drawings carry the customer's geometry, and a supplier that cannot sign an NDA or show an information security system is a liability regardless of price. GreatLight holds ISO 27001:2022 for information security alongside ISO 9001:2015, IATF 16949:2016, and ISO 13485:2016, and signs an NDA on request.

A quote is only as good as the DFM review behind it. When a file arrives, the first pass is not price, it is feasibility: can the tool reach the feature, is the wall thick enough to hold, does the tolerance sit on a surface that can be measured. GreatLight returns a quotation and a free DFM analysis within 12 hours, and production can start within 24 hours after that.

First article inspection is the checkpoint that decides whether the run continues. A shop should check raw material certificates, monitor in-process dimensions on the tight features, and inspect 100% of parts before shipment, with reports on request. If the first article is in tolerance but the fifth is not, the process was not capable, and the fix is in the process, not in the inspection bench.

Tool wear is the usual cause of drift inside a run. A Ø4 mm cutter doing a finishing pass at 12,000 rpm will wear over 300 parts, and the last 50 may sit at the top of the tolerance band. Shops that hold tight numbers replace or re-measure the tool on a fixed interval rather than waiting for a dimension to go out.

Packing is part of the process too. A thin-walled housing that survived machining can still be bent by a careless crate. For parts with a wall under 2 mm, specify the packing method and check it. It is a cheap question that prevents an expensive claim.