CNC machining Stourbridge expert: the engineering behind the quote

• 1
  What five axes actually change about a cut
• 2
  Workholding decides more than the spindle does
• 3
  How a ±0.005 mm tolerance stacks up in real parts
• 4
  Material behavior that changes the cutting plan
• 5
  Where finishing fits in the process, not after it
• 6
  Choosing between mill-turn, 5-axis and 3-axis routes
• 7
  Which route fits which part

A three-axis machine moves the tool in X, Y and Z. The part stays where the vise or fixture put it. That is fine for plates, housings and brackets where every feature can be reached from one direction. Add a fourth axis and the part can rotate around one axis, usually A or B. Add a fifth and the tool can tilt relative to the surface, so the cutting edge stays close to perpendicular as it sweeps around a curved wall.

The practical difference is not the number of axes. It is how much of the part you can reach without a second setup. Every re-clamp introduces a new datum, and each new datum carries its own position error. On a part with five faces of features, a three-axis route may need three setups; a simultaneous 5-axis route may need one. That is where the tolerance budget is won or lost.

There is a cost. Five-axis motion asks more from the post-processor, the tool holder and the operator. A tilted tool can collide with the fixture in ways a 3-axis path never would. That is why we simulate every 5-axis program before it runs, using the actual holder geometry rather than a nominal cylinder.

So the first question for any new part is not which machine is faster. It is how many setups the geometry really needs. Answer that and the machine choice usually follows.

• 1
  3-axisFlat or prismatic parts, features reachable from one or two directions.
• 2
  4-axisCylindrical parts with cross holes, slots or milled flats.
• 3
  5-axis simultaneousContoured surfaces, undercuts, deep pockets with steep walls.

A 16-ton spindle cannot help a part that shifts 0.02 mm between the roughing and finishing pass. Workholding is the quiet variable. For a first article we usually start from the largest stable face and hold it in soft jaws machined to the actual blank. For thin walls, we add a sacrificial bridge that is cut away in a later operation.

Zero-point systems matter when a part needs multiple operations. A pallet with a repeatable pull-stud interface lets us move a part from a 5-axis center to a mill-turn center and back without re-indicating the datum. Position repeatability on these systems is typically a few microns, which is what makes the multi-setup route viable at all.

For long parts, we work up to 4,000 mm of travel. Long parts deflect. A 4,000 mm aluminum extrusion will sag under its own weight if it is only supported at the ends, so we add intermediate supports or machine it in a fixture that follows the part. The 4,000 × 400 × 150 mm envelope is the largest we run; smaller envelopes like 500 × 500 × 450 mm suit compact housings.

The rule is simple. If you cannot describe how the part is held for every operation, the tolerance on the drawing is a wish, not a plan.

• 1
  Soft jawsMachined to the blank, not the nominal CAD model.
• 2
  Zero-point palletsKeep the datum across multiple machines.
• 3
  Sacrificial bridgesHold thin walls rigid, then cut them free.

A single dimension at ±0.005 mm is achievable on a rigid part in a stable setup. A chain of five dimensions each at ±0.005 mm is a different problem. The errors accumulate. If two features are located from the same datum, the stack is smaller than if each feature is located from the last one. This is why we ask for the functional datum, not just the dimension callout.

Temperature is part of the stack. Aluminum 6061 expands about 23 µm per meter per degree Celsius. A 300 mm part that warms 5 °C between the roughing and finishing pass moves roughly 35 µm, which is larger than a ±0.005 mm band. For tight work we let the part cool, or we machine in a temperature-controlled cell and measure at 20 °C.

Surface finish and tolerance are linked too. A Ra 0.2–0.8 µm finish usually means a small step-over and a finishing pass with a sharp, low-runout tool. If the tolerance is loose but the finish is fine, the cut plan changes: fewer roughing passes, more attention to tool wear. Reading both callouts together tells us what the part is for.

Our inspection reports note the actual measured value, the instrument and the temperature at measurement. A number without that context is hard to defend in a design review.

• 1
  Datum choiceLocate features from one functional datum where possible.
• 2
  Thermal driftAluminum moves ~23 µm/m per °C; let parts stabilize.
• 3
  Finish vs toleranceFine Ra needs small step-over, not just a slow feed.

The same geometry in 6061-T6 and 316L stainless is two different jobs. Aluminum cuts fast, throws chips well and holds a good finish with a sharp tool. Stainless 316L work-hardens. If the tool rubs instead of cutting, the surface gets harder under the cut and the next pass wears the insert faster. The fix is a heavier chip load and a tool that stays engaged, not a lighter one.

Titanium TC4 (Ti-6Al-4V) is worse. It conducts heat poorly, so most of the cutting heat goes into the tool edge. Speeds drop, coolant flow goes up, and tool life becomes the cost driver. Inconel is harder again. For these materials we plan the tool path around heat, not around cycle time.

Plastics behave in the opposite way. POM and PEEK cut cleanly with sharp, polished tools and high spindle speeds, but they move with temperature and can chip at a sharp corner. ABS and PC are softer and more forgiving. Carbon fibre is abrasive; it eats carbide, so we use diamond-coated tools and sealed extraction.

We keep stock of 6061, 2024, 7075, 303, 304, 316L, 17-4PH, 4130, 4140, C36000 brass, TC4 and the common plastics. That list is not a catalogue. It is the set of materials we have cutting data for, which is what makes a quote realistic.

• 1
  AluminumFast, stable, good finish. Watch thin-wall spring-back.
• 2
  Stainless 316LWork-hardens. Keep the edge cutting, not rubbing.
• 3
  Titanium TC4Heat goes into the tool. Plan for tool life.
• 4
  Carbon fibreAbrasive. Diamond coating plus extraction.

Finishing is often treated as a step that happens after machining. In practice it changes the machining plan. Anodizing adds a thin oxide layer, roughly 5–25 µm depending on the process, and it grows outward from the surface. If a bore is anodized after being machined to size, the bore gets smaller. We mask critical bores or leave stock for the coating.

Hardcoat anodizing is thicker and harder, and it can round a sharp edge. If the drawing calls for a crisp edge, we chamfer before coating and note the edge condition on the travel sheet. Electroless nickel and zinc plating behave differently: they add a more uniform layer and can reach into blind holes better than anodizing.

Bead blasting and tumbling change the surface texture, which is sometimes used to hide tool marks. Laser marking is the last step because it is hard to remove. Minimum character height is 1.5 mm; below that the mark gets fuzzy and may not survive a coating.

The point is sequencing. Tell us the finish before the first cut, and the machining plan will leave the right stock and protect the right features.

• 1
  AnodizingGrows outward 5–25 µm. Mask tight bores.
• 2
  PlatingMore uniform; better reach into blind holes.
• 3
  Laser markingLast operation. Minimum 1.5 mm character height.

A mill-turn center is the right answer when a part is mostly round but has milled features. One setup covers turning, cross drilling and milling, so the concentricity between the bore and the milled face is set by the machine, not by a re-clamp. For a shaft with a keyway and a flange, that is usually the cheapest route.

A simultaneous 5-axis center is the right answer when the part has contoured surfaces or features on many faces. Impellers, turbine blades, complex brackets and housings fall here. The machine cost per hour is higher, but the setup count is lower and the tool can reach angles a 3-axis machine cannot.

A 3-axis machine is still the right answer for flat plates, covers, manifolds and simple brackets. It is faster to program, cheaper to run and easier to fixture. Using a 5-axis center on a flat plate does not make the part better; it makes the quote higher.

We run 127 high-precision CNC machines across three plants, including 16 simultaneous 5-axis centers, 16 mill-turn centers, 12 four-axis mills and 27 three-axis machines. That mix is deliberate. It lets us put a job on the machine that fits, rather than forcing every part onto the newest one.

• 1
  Mill-turnRound parts with milled features in one setup.
• 2
  5-axis simultaneousContoured, multi-face, undercut geometry.
• 3
  3-axisFlat, prismatic parts. Cheapest correct route.