High Precision Workpiece Processing: Where Accuracy Actually Comes From

• 1
  What high precision workpiece processing really measures
• 2
  Why 5-axis setups cut error instead of adding it
• 3
  Matching part size to machine travel
• 4
  Why the same tolerance costs more in some materials
• 5
  Inspection is part of the process, not a final gate
• 6
  Finish and tolerance are separate calls
• 7
  Which setup fits which part

A tolerance on a drawing is a promise about one feature. The shop has to keep that promise across every feature on the part, on every part in the run, for as long as the order lasts. High precision workpiece processing is the discipline of holding that promise when the numbers get small, usually ±0.005 mm on a tight callout or a few micrometres on a bore.

The number does not come from one machine. It is a stack: spindle thermal growth, tool runout, fixture deflection, material springback, and the probe that tells you what you actually cut. Change any one of them and the result moves. A shop that quotes ±0.005 mm is claiming it can hold the whole stack, not just the spindle.

This matters to you as a reader because a tight tolerance changes cost in ways that are not obvious on a quote sheet. It changes how many setups the part needs, how long it sits on the machine, and how much of the run gets measured instead of shipped.

The common assumption is that more axes mean more error. In practice the opposite holds for complex parts. A 3-axis job on a part with five angled faces needs five fixtures or five re-clamps. Every re-clamp is a new datum, and every datum shift adds to the tolerance stack.

A simultaneous 5-axis center tilts the tool and the table so the same face can be reached in one setup. One datum, one zero point, one thermal state. The error that used to come from re-fixturing simply does not appear.

That is why we run 16 simultaneous 5-axis machining centers alongside 27 three-axis machines and 12 four-axis mills. The three-axis machines are faster and cheaper for prismatic work. The five-axis centers exist for parts that would otherwise need three setups and a prayer.

There is a catch. Five-axis programming is slower to prove out, and the machine is only as good as its post-processor and its probe routine. On a simple bracket, five-axis is waste. On a housing with intersecting bores at compound angles, it is the only sane route.

• 1
  One setup, one datumFewer re-clamps means less stack-up error on angled features.
• 2
  Reach without special toolingUndercuts and deep pockets become accessible with a tilted tool.
• 3
  Not always cheaperSimple prismatic parts run faster on a 3-axis mill.

Every machine has a box it can reach, and a part that does not fit inside that box gets split, re-fixtured, or refused. Travel is the first filter in high precision workpiece processing, before tolerance even enters the conversation.

Our largest envelope is 4,000 × 400 × 150 mm, which suits long extrusions, rails, and beam-type parts that are narrow in two directions. Medium frames run 750 × 1,150 × 550 mm and 600 × 600 × 600 mm, a good match for housings and manifolds. Compact cells at 500 × 500 × 450 mm and 500 × 310 × 200 mm handle small, high-mix work where changeover speed matters more than envelope.

Rotary work adds a fourth dimension. A Ø400 mm rotary table lets a part be indexed or cut continuously around an axis, which is how we hold concentricity on cylindrical features without a second op.

The practical rule: give the shop the finished envelope plus the stock, not just the finished part. A 380 mm long part does not fit a 400 mm travel machine once you add a face mill and a fixture.

Aluminium 6061 and 7075 cut clean and hold ±0.005 mm without much drama on a rigid setup. They also move with heat, so a part that measures perfect at 22 °C can drift out of tolerance if it is inspected straight off the spindle.

Stainless 304 and 17-4PH work-harden. A light finishing pass with a dull tool raises surface hardness and pushes the next pass off the nominal. Tool changes get scheduled by time, not by wear, which is why 17-4PH parts carry longer cycle times.

Titanium Ti-6Al-4V and Inconel are the hard cases. Low thermal conductivity keeps heat in the cut, so the tool and the workpiece both grow. Finishing passes are lighter, spindle speeds lower, and the machine often needs a warm-up cycle before the first cut of the day.

Plastics sit at the other end. POM and PEEK are dimensionally stable but flexible, so clamping pressure alone can distort a thin wall past the tolerance before the tool touches it. The fix is often a soft jaw bored to the part profile, not a tighter tolerance callout.

A tolerance you cannot measure is a tolerance you cannot hold. If the drawing calls for ±0.005 mm, the shop needs a CMM or a high-resolution gauge that is itself calibrated to a fraction of that band. A pair of calipers is not evidence.

We check raw material certificates on receipt, monitor dimensions in process, and inspect 100% of parts before shipment. Reports are available on request. The point is not paperwork. It is that a drifting tool gets caught at part 12, not at part 400.

In-process probing closes the loop. The machine measures a feature, adjusts its offset, and keeps cutting. On long runs this is what holds a tight band across hours of cutting, when the spindle is warmer at the end of the shift than at the start.

For you as the buyer, the useful question is not 'do you inspect' but 'at what point in the process'. A shop that only inspects at the end is sorting good parts from bad. A shop that probes in process is preventing the bad parts from being made.

Ra and tolerance get written on the same drawing and treated as one requirement. They are not. A part can hold ±0.005 mm with a rough as-machined surface, and a part with a mirror finish can still be out of tolerance.

As-machined surfaces run around Ra 1.6–3.2 μm, which is fine for most brackets and internal parts. High-finish work sits at Ra 0.8–1.6 μm, common on sealing faces and sliding surfaces. Fine finish at Ra 0.2–0.8 μm is reserved for optical, medical, and fluid-contact features where the surface itself does the work.

Getting to Ra 0.2–0.8 μm usually means a separate finishing pass with a small stepover, sometimes a different tool, and often a polishing or lapping step after machining. It adds time, and it can move dimensions if the polishing is not controlled.

The engineering judgment: specify the finish the function needs, not the finish that looks impressive on a spec sheet. Over-specifying finish is one of the most common ways a cheap part becomes an expensive one.