Advanced CNC machining: what Sweden's shops actually run
What separates advanced CNC machining from ordinary milling work. We cover 5-axis geometry, tolerance limits, material behavior, and the cases where the extra capability is not worth paying for.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
What "advanced" actually means in advanced CNC machining
The word advanced gets used loosely. In a machine shop it describes four measurable things: how many axes move at the same time, how tight the tolerance band is, how hard the material is to cut, and how much of the setup happens in one clamping. Everything else is marketing.
A three-axis mill holds a part and cuts from one direction. A simultaneous 5-axis center tilts the tool or the table while cutting, so the tool tip stays normal to a curved surface. That single difference decides whether a part can be finished in two setups or seven.
The second marker is tolerance. Standard commercial work sits around ±0.05 mm. Advanced work holds ±0.005 mm, and that number changes the whole plan. You need temperature control, sharp tooling, and a probe that measures the part on the machine before the last pass.
The third marker is material. Aluminium 6061 cuts fast and forgives small mistakes. Inconel, Ti-6Al-4V, and 17-4PH do not. They work-harden, they pull heat into the tool edge, and they need lower feed per tooth and more coolant. A shop that only runs aluminium will struggle with them.
The fourth marker is setup count. Every clamping adds error. If a part has features on five faces, a 5-axis center can reach them in one or two setups. A three-axis shop needs fixtures, and each fixture introduces its own stack-up. More setups mean more scrap risk, not just more hours.
Five-axis geometry: when simultaneous motion beats indexed work
There are two ways to use a 5-axis machine. Indexed work, often called 3+2, tilts the table to a position, locks it, then cuts. Simultaneous work moves all axes while the tool is in the material. They cost differently and they solve different problems.
Indexed 5-axis is the practical choice for most parts. A housing with pockets on four sides, a bracket with angled bosses, a manifold with holes at compound angles. The machine reaches the face, locks, and cuts like a rigid three-axis machine. You get shorter tools and better surface finish.
Simultaneous motion is for curved surfaces: impellers, turbine blades, contoured molds, orthopedic implants. The tool stays tilted so the flank of the cutter does the work instead of the tip. Cutting with the tip of a ball nose leaves a scalloped surface and burns the tool.
Simultaneous motion is slower to program and slower to verify. You need collision simulation before the first cut, because a tilted holder can swing into the table or the fixture. On a simple part, that effort buys nothing.
A useful test: if every surface of the part can be reached by a tool pointing straight down after two table tilts, indexed work is enough. If the surface normal changes continuously along a path, you need simultaneous motion.
Material behavior sets the real limits
Tolerance numbers on a drawing are a request. The material decides whether the shop can hold them. Aluminium moves with heat and cuts cleanly, so ±0.005 mm is realistic on a 100 mm aluminium part with a temperature-stable shop.
Stainless 316 and 17-4PH work-harden at the cut. If the tool rubs instead of bites, the surface gets harder and the next pass wears the edge faster. The fix is a heavier feed per tooth, not a lighter one, which surprises people who are used to aluminium.
Titanium Ti-6Al-4V has low thermal conductivity, so heat stays at the cutting edge. Tool life drops quickly above 60 m/min surface speed with carbide. Shops compensate with high-pressure coolant and shorter tool paths.
Plastics are the opposite problem. POM and PEEK cut easily but move after machining because internal stress releases. A tight-tolerance plastic part often needs a rough pass, a rest, then a finish pass. Skipping the rest means the part measures correctly on the machine and wrong the next morning.
Magnesium AZ31B and AZ91D cut fast, but the chips are a fire risk. It needs dedicated tooling and chip handling, which many general machine shops do not have.
Where advanced capability stops paying off
More axes and tighter tolerances cost money. The engineering question is whether the part needs them, and the answer is often no. A rectangular bracket with six holes does not get better because it was cut on a 5-axis center.
The first boundary is tolerance versus size. Holding ±0.005 mm on a 50 mm feature is routine for a well-maintained shop. Holding the same band across a 1,000 mm part is not. Thermal expansion of aluminium is about 23 μm per meter per degree Celsius, so a 3 °C shop drift eats the whole band.
The second boundary is surface finish. A Ra 0.2–0.8 μm finish needs a separate finishing pass, sometimes a different tool, and often hand polishing. If the drawing says Ra 3.2 μm, paying for a mirror finish is waste.
The third boundary is quantity. On one prototype, programming and fixturing dominate the price. On a 10,000-part run, cycle time dominates. The same feature can be cheap at one volume and expensive at the other, which is why quoting both is useful.
The fourth boundary is inspection. A ±0.005 mm callout without a stated measurement method is incomplete. CMM, optical, and hand gauges disagree at that level, so agree on the method before the first chips.
How to verify a shop before you send the drawing
Ask what the shop measures with, not what it claims. A ±0.005 mm claim needs a climate-controlled inspection room and a calibrated CMM. If inspection happens in the same room as the grinding, the number is aspirational.
Ask how the first article is handled. A shop that cuts the part, measures it, and sends the report before shipping is different from one that measures only when a customer complains. In-process monitoring catches a drifting tool before the last ten parts are scrap.
Ask about material certificates. For 17-4PH or Ti-6Al-4V, the mill certificate tells you the heat treatment condition. A shop that cannot show it cannot guarantee hardness or machinability.
Ask what happens when a feature cannot be held. Good shops call before cutting and propose a change. Bad shops ship the part and argue about the drawing.
Finally, ask for the setup plan. If a supplier cannot describe how many setups the part needs, they have not thought about the tolerance stack-up yet.
None of these questions requires a confidentiality risk. An NDA covers the drawing; the process conversation covers capability.
Matching the process to the part
Pick the row that matches your geometry, then check the tolerance column before you commit.
| Part feature | Right process | Typical tolerance | Why |
|---|---|---|---|
| Flat plate, holes on one face | 3-axis milling | ±0.05 mm | One setup, no fixturing cost |
| Pockets on four sides | 3+2 indexed 5-axis | ±0.02 mm | Fewer setups, shorter tools |
| Contoured blade or impeller | Simultaneous 5-axis | ±0.005 mm | Tool flank stays on surface |
| Long shaft with turned threads | Mill-turn center | ±0.01 mm | Turning and milling in one chuck |
| Thin wall under 1 mm | 3-axis with light passes | ±0.02 mm | Chatter control beats axis count |
| Hardened tool steel, 50 HRC | 3-axis with carbide | ±0.01 mm | Rigidity matters more than axes |
| Prototype, one piece | 3-axis or 3+2 | ±0.05 mm | Programming time dominates cost |
The verdict
If your part has continuously curved surfaces and a ±0.005 mm band, simultaneous 5-axis work is the only route that holds it. If it has flat faces and ordinary tolerances, a 3-axis or indexed 3+2 setup will be cheaper and just as good. Match the process to the geometry, not to the brochure.
Questions engineers ask next
Can a 3-axis machine hold ±0.005 mm?
Yes, on small features and flat geometry, with a rigid setup and temperature control. The limit is not the axis count, it is the machine condition, the tool, and the thermal environment.
Where 3-axis fails is reach. If the feature is on an angled face, you need a fixture, and each fixture adds stack-up error that eats into the band.
How do I know if my part needs simultaneous 5-axis?
Look at the surface normals. If they change continuously along a tool path, you need simultaneous motion. If they change in steps, indexed 3+2 work is enough and cheaper to program.
A second signal is tool access. If a straight tool cannot reach the surface without a long, thin extension, tilting the tool is the better answer.
Does a tighter tolerance always cost more?
Not always, but usually. The cost comes from slower feeds, more inspection, and higher scrap risk rather than from the machining itself.
The cheapest way to reduce cost is to loosen tolerances on features that do not matter functionally and keep the tight band only where it does.
What finish can I expect as-machined?
Typical as-machined surfaces land around Ra 1.6–3.2 μm. A controlled finishing pass reaches Ra 0.8–1.6 μm. Below Ra 0.8 μm usually means a separate operation, sometimes manual polishing.
Specify the finish only where it affects function, such as sealing faces or sliding surfaces.
How do you handle thin walls that chatter?
Light radial passes, a sharp tool, and support from the fixture. Sometimes the answer is to leave the wall thick, machine everything else, then take the wall down in the last operation.
Chatter is a stiffness problem, so adding axis count does not fix it by itself.
What should I send with the drawing?
The 3D model, the drawing, the material grade, the critical features, and the measurement method you expect. Naming the critical features saves a round of questions.
If you have a target quantity range, include it. Cycle time and fixturing cost trade off differently at one piece and at ten thousand.
Send the drawing, get a real process answer
We review geometry, material, and tolerance before quoting, so the price reflects the process the part actually needs.
Quote in 12 hours100% inspection before shipmentNo minimum order quantity