CNC machining Nottingham: how the process really works

• 1
  What CNC machining Nottingham shops actually do to metal
• 2
  Why 5-axis matters for complex geometry
• 3
  Tolerance stack-up, surface finish and the cost curve
• 4
  DFM rules that decide whether your part machines cleanly
• 5
  Material choice and when machining is the wrong process
• 6
  Choosing the right machine for the geometry

A CNC machine does one thing: it moves a rotating cutter along a path calculated in software. The metal does not care where the machine sits. What changes from shop to shop is how many axes move at once, how rigid the setup is, and how the tool path handles the corners of your part.

On a 3-axis mill the tool always points down. On a 5-axis machine the table or the spindle tilts, so the cutter can reach a face that would otherwise need a second fixture. That single difference removes a whole class of setup error. Every time you re-clamp a part, you reintroduce a small positional shift. Five-axis work removes those shifts, and that is why tight callouts survive to the inspection bench.

Cutting speed and feed rate are chosen from the material, not from habit. Aluminium 6061 runs fast and wet. Stainless 316 work-hardens if the tool rubs instead of cuts, so the feed per tooth has to stay above a floor. Titanium TC4 cuts at a lower surface speed and needs more coolant. Get the numbers wrong and you get chatter, which shows up later as a surface finish fail.

The output of all this is a part that matches the model within a stated tolerance. For most of our work that band is ±0.005 mm on critical features, with general dimensions held looser. Knowing which dimensions deserve that band is half the design job.

• 1
  Setup count drives errorEach new fixture adds positional variation.
• 2
  Feeds come from materialStainless and titanium punish rubbing tools.
• 3
  Tolerance is selectiveApply tight bands only where function needs them.

A 5-axis center adds two rotary movements to the three linear ones. The practical effect is that the cutter can stay normal to a curved surface instead of approaching it at a fixed angle. On a turbine blade profile or a deep pocket with drafted walls, that keeps the tool engagement steady and the finish even.

There is a second benefit that buyers notice sooner: fewer operations. A part with features on five faces can often be cut in one setup. That removes the re-clamp step, which removes the stacked error, which removes the argument about which operation was out of position when the part failed inspection.

Not every part needs it. A flat bracket with holes on two faces is faster and cheaper on a 3-axis machine with a vise. Five-axis programming takes longer, and the machine hour costs more. The judgment call is whether the geometry or the tolerance stack justifies it.

Our floor runs 16 simultaneous 5-axis machining centers alongside 12 four-axis mills and 27 three-axis machines. That mix exists because the right answer changes with the part, not because one machine type wins everything.

• 1
  Stays normal to the surfaceEven tool engagement on curved walls.
• 2
  One setup, five facesFewer clamps means less stacked error.
• 3
  Not always the cheaper routeSimple prismatic parts run better on 3-axis.

Every dimension on your drawing carries a tolerance, and every one of them costs money. The cost does not rise in a straight line. Moving a general dimension from ±0.1 mm to ±0.05 mm is routine. Pushing a bore to ±0.005 mm means a different inspection routine, a temperature-stable environment and possibly a finishing pass.

Surface finish follows the same curve. As-machined surfaces land around Ra 1.6–3.2 μm and suit most brackets and housings. A sealing face or a sliding bore often wants Ra 0.8–1.6 μm. Optical and medical interfaces can need Ra 0.2–0.8 μm, which usually means a separate finishing operation rather than a slower cut.

The trap is applying tight numbers everywhere. If a drawing calls out ±0.005 mm on a dimension that only locates a cover plate, you pay for precision nobody uses. A short conversation about which dimensions control fit, and which are reference, usually takes more cost out of a part than any material swap.

Inspection has to match the callout. A ±0.005 mm feature cannot be verified with a caliper. It needs a micrometer, a bore gauge or a CMM, and the report has to be traceable. We inspect 100% of parts before shipment, and dimensional reports go out on request.

• 1
  Cost is non-linearTight bands add operations, not just attention.
• 2
  Finish drives processRa 0.2–0.8 μm usually needs a second op.
• 3
  Reference vs. controlMark which dimensions actually set fit.

Most manufacturability problems are visible before any metal is cut. The common ones are deep pockets with a corner radius smaller than the cutter, holes that break into a wall at a shallow angle, and walls so thin the part deflects under clamping pressure. Each of these shows up in a DFM review within a few minutes.

Corner radii are the simplest fix. A pocket 50 mm deep cut with a Ø6 mm tool needs an internal radius of at least 3 mm, and 4 mm is friendlier. If the model calls for a 1 mm radius at that depth, the shop has to burn a small tool slowly or drop to EDM, and the price reflects it.

Wall thickness matters for a different reason. Aluminium walls below 0.8 mm tend to sing during cutting. Plastics flex even more. If the design needs a thin web, we usually recommend leaving it thick and removing material in a later operation, or adding a temporary rib that gets cut away.

Threads and text deserve a mention too. A tapped hole needs enough surrounding material for the thread to hold, and laser marking needs a minimum character height of 1.5 mm to stay legible after anodizing. Small details, but they cause rework when they are missed.

• 1
  Radius follows depthDeep pockets need a radius the tool can reach.
• 2
  Thin walls deflectUnder 0.8 mm in aluminium, expect chatter.
• 3
  Plan marking earlyLaser text needs 1.5 mm minimum height.

Material selection is usually settled by the application before the process is chosen. Aerospace brackets lean on 7075 and Ti-6Al-4V for strength-to-weight. Medical housings often use 316L or 17-4PH for corrosion resistance and cleanability. Automotive fixtures use 6061 or 4140 because they are predictable and available.

Machining handles all of these, but it is not always the economical answer. A part that will be made 50,000 times a year in the same shape is usually a die casting or an injection molding job with a machined interface. Machining wins when geometry is complex, quantities are low to medium, or the part is still changing.

That boundary is worth stating plainly. From one prototype to 10,000+ part runs, machining needs no tooling and no minimum order quantity, so design changes cost nothing but programming time. Past that volume, tooling starts to pay for itself.

We machine aluminium 6061, 2024, 5052, 5083, 6082 and 7075, stainless 303 through 17-4PH, steels 1018 to 4340, copper and brass grades, titanium, Inconel, magnesium and engineering plastics including PEEK and carbon fibre. The material list is wide, but the process recommendation still follows volume and geometry.

• 1
  Application picks materialStrength, corrosion and cleanability lead.
• 2
  Volume picks processHigh volume favors casting or molding.
• 3
  No tooling at low volumeDesign changes cost programming only.