Calpol CNC machining: accurate and repeatable

On a print, accuracy usually shows up as a single tolerance block. In the shop it splits into several separate errors that stack: machine positioning, tool deflection, thermal growth, fixturing movement and the probing method used to verify the result. Holding ±0.005 mm on a feature means controlling that stack, not just buying a better spindle.

For most machined parts, ±0.005 mm is the practical floor on a well-controlled process, and it should be reserved for the features that need it. A bearing bore, a dowel pin hole, a sealing face or a mating spigot earns that tolerance. A clearance hole, a cable route or a cosmetic edge usually does not.

Tighter than ±0.005 mm gets expensive fast, and it rarely survives outside a temperature-controlled room. If a design calls for ±0.002 mm across a 300 mm part, we would rather question the requirement than quietly quote a number we cannot repeat on a Tuesday afternoon.

The imperial equivalent is ±0.0002 in. When a drawing mixes both units, we work from the tighter of the two and flag the conflict during DFM review, before any material is cut.

• 1
  Reserve tight tolerancesApply ±0.005 mm only to functional fits, not to the whole part.
• 2
  Watch the stackPositioning, deflection and thermal drift add up over a long cut.
• 3
  Fix units earlyMixed metric and imperial drawings cause more scrap than tool wear.

Every refixture adds error. A part that needs five separate operations has five chances for a locating error to creep in, and each one compounds the last. Simultaneous five-axis machining removes most of those moves by reaching the back and side faces without letting go of the part.

We run 16 simultaneous five-axis machining centers, plus 12 four-axis mills, 27 three-axis machines and 16 mill-turn centers. That mix matters for accuracy: the right call is often a three-axis job on a rigid fixture, not a five-axis job just because it is available.

Workholding is where accuracy is usually won or lost. Thin walls, long slender parts and free-form surfaces move under cutting force. We model the fixture, choose the contact points and, on thin features, adjust the toolpath to balance the load on both sides of the wall.

Once a setup is proven, it becomes the baseline. We record the offsets, the tool list and the probing routine, so the second run matches the first. That is what repeatability means in a job shop: the same numbers, not a lucky part.

Accuracy you cannot measure is a claim, not a result. Our inspection runs in three stages: incoming raw material check, in-process monitoring while the part is still on the machine, and a final dimensional inspection before shipment. Every part is inspected before it leaves, not sampled.

In-process probing is the useful one. If a critical bore is trending toward the low limit, we know while the setup is still live. Correcting a tool offset takes minutes. Discovering the drift after the parts are anodized takes a new batch.

Reports are available on request, with the dimensions and the method used for each. For regulated work under ISO 9001:2015, IATF 16949:2016 or ISO 13485:2016, we can align the report format to your quality plan. Our qualification rate across shipped parts is 99.99%.

A useful habit for engineers: mark the two or three dimensions that decide whether the part works. Those get checked first and reported in full. The other dimensions still get checked, but they do not hold up the release.

Aluminium 6061-T6 and 7075 cut cleanly and hold tight tolerances without much fuss. Stainless 17-4PH and 316L move more under heat, so roughing passes need to leave stock and the finishing pass comes after the part has settled. Titanium TC4 and Inconel are worse still: low thermal conductivity, high tool pressure, and a strong tendency to spring back.

Geometry matters as much as material. A part 4,000 mm long behaves differently from one that fits in your hand. Our largest travel is 4,000 × 400 × 150 mm; the medium platforms cover 750 × 1,150 × 550 mm and 600 × 600 × 600 mm, and the compact machines handle 500 × 500 × 450 mm and 500 × 310 × 200 mm. A Ø400 mm rotary table covers round and cylindrical work.

Thin walls below 1 mm, deep pockets with small corner radii, and long unsupported bores are the three shapes that most often force a tolerance discussion. None of them are impossible. All of them change the process, the cycle time and sometimes the order of operations. Better to raise them at DFM than at first article.

Surface finish and tolerance interact too. A Ra 0.2–0.8 μm sealing face needs a different finishing strategy than an as-machined bracket at Ra 1.6–3.2 μm, and the tight finish often needs its own setup and its own inspection step.

Tight accuracy pays off when the part has a mechanical job to do: it locates something, seals something, rotates in a bearing, or bolts to a mating part with a fixed centre distance. Aerospace brackets, EV motor housings, surgical instrument bodies, robot end-effector plates and vacuum-chamber fittings all fall into this group.

It pays off less on enclosures, covers, brackets without a critical interface and early-stage concept models. If a part only has to look right and hold together, spending the accuracy budget on it takes money from the parts that need it.

Volume changes the answer, not the tolerance. We have no minimum order quantity, so a single prototype and a 10,000-part run go through the same first-article logic. The difference is that at volume we invest more in fixture design and probing, because the setup cost is spread across more parts.

One more filter. If the design is still moving, a tight tolerance is premature. Lock the interface dimensions first, then hold them. We would rather machine a slightly loose prototype this week than a precise one next month that fits nothing.