CNC machining in UK racing: how parts actually get made

• 1
  What CNC machining actually does to a racing part
• 2
  Alloy choice drives the whole machining plan
• 3
  Why 5-axis setups change what is possible
• 4
  Tolerances and surface finish: what to specify
• 5
  Where CNC machining stops being the right answer
• 6
  From CAD file to a part on the car
• 7
  Which process fits which race part

A CNC machine does not make a part lighter or stronger. It removes material along a path that a CAM programmer defined, and the finished geometry is only as good as that path. In racing, the value comes from repeatability: the 40th upright must fit the same as the first, because you will not re-jig the car for it.

The cutting tool leaves a surface, and that surface is a stress feature. A sharp internal corner from a Ø6 mm end mill concentrates load. A generous radius spreads it. Two parts with identical CAD mass can behave differently on the car purely because of corner radii and tool engagement angles.

Most race parts are machined from solid billet rather than cast or forged. Billet gives you a known grain condition and no tooling cost, which matters when you build six of something. The trade is material waste and longer cycle time, so billet wins on low volume and loses on high volume.

• 1
  BilletNo tooling cost, good for 1–50 parts.
• 2
  Cast or forgedLower unit cost above a few hundred parts.
• 3
  Surface finishRa 0.8–1.6 μm is a typical as-machined target.

In UK racing classes, aluminium dominates uprights, bellhousings and brackets. 6082 and 7075 are the common picks. 6082 machines cleanly and welds, so it suits fabricated-and-machined assemblies. 7075 gives higher strength but is less tolerant of sharp corners and needs more care on thin walls.

Steel shows up where stiffness and fatigue life matter: 4130 and 4140 for suspension and driveline parts, 4340 where section size grows. These cut slower. A 4140 part that takes 20 minutes in aluminium can take two hours, so the material decision is also a scheduling decision.

Titanium sits at the top of the cost curve. Ti-6Al-4V (TC4) has low thermal conductivity, so heat stays in the cutting zone and tools wear fast. It is justified for parts where mass and temperature are both critical. For a bracket that simply needs to be stiff, aluminium does the job for a fraction of the cost.

• 1
  6082 / 7075Brackets, uprights, housings.
• 2
  4130 / 4140 / 4340Suspension, driveline, high-cycle fatigue.
• 3
  TC4 titaniumHeat-critical, mass-critical parts only.
• 4
  17-4PHCorrosion resistance plus strength.

A 3-axis machine holds the part still and moves the tool in three directions. Every new face needs a new setup, and each setup adds a datum error. For a part with features on five sides, that error stacks up. A 5-axis machine rotates the part or the spindle so more features are cut in one clamping.

One clamping is the real prize. Fewer setups mean fewer datum transfers, tighter true position between features, and less risk of a chip sitting under a locating face. It also shortens the fixture list, which matters when you are chasing a race deadline.

The limit is part size and reach. Simultaneous 5-axis work needs clearance for the tool holder, and deep pockets in tall parts can be unreachable. GreatLight runs 16 simultaneous 5-axis centers with travels up to 4,000 × 400 × 150 mm, plus a Ø400 mm rotary table for round work.

Not every part needs it. A flat plate with holes on one face is faster and cheaper on a 3-axis machine. The judgment call is feature count and orientation, not the prestige of the machine.

A tolerance is a cost instruction. Tightening a bore from ±0.05 mm to ±0.005 mm can mean a different machine, a temperature-stable room, and more inspection time. Specify tight only where the function demands it: bearing seats, spigots, dowel holes, sealing faces.

On the rest of the part, use a general tolerance block and leave it there. A bracket mounting hole at ±0.2 mm works fine. Engineers who apply ±0.01 mm across an entire drawing usually pay for it and gain nothing.

Surface finish follows the same logic. Ra 1.6–3.2 μm is a normal machined finish and fine for most structural faces. Ra 0.8–1.6 μm suits sliding and sealing surfaces. Below Ra 0.8 μm you are into fine finishing or polishing, which adds operations and handling risk on thin race parts.

GreatLight holds ±0.005 mm (±0.0002 in) where a drawing calls for it, with 100 percent inspection before shipment and reports on request. The point is not that every part gets that tolerance. The point is that the tight features are held and the loose ones are left alone.

• 1
  Bearing seatsSpecify tight, and give a datum scheme.
• 2
  Cover facesGeneral tolerance is enough.
• 3
  Sealing facesControl finish, not just size.

CNC machining is subtractive, so it cannot beat a casting on unit cost once volume climbs. If you need 500 identical housings a season, a casting with a machined interface face will usually be cheaper and lighter overall than a billet version.

Internal geometry is another boundary. A closed hollow section, a complex cooling gallery, or a lattice core cannot be cut from solid with a spinning tool. Those go to casting, additive manufacturing, or a bonded assembly.

Very thin walls are a third limit. Aluminium at 0.5 mm will deflect under cutting force and chatter. Titanium at that thickness is worse. If a design needs a 0.4 mm web, expect to redesign it or move to a different process.

Finally, size. Beyond roughly 4,000 mm in the long axis, you are into large-format or fabricated territory. Below that, a machined part usually wins on accuracy and lead time.

The workflow is short but each step can stall. A usable CAD file, native or STEP, plus a 2D drawing for critical features, is enough to quote. GreatLight returns a quotation and a free DFM analysis within 12 hours, and production can start within 24 hours of approval.

DFM feedback is where money is saved. A programmer who sees a deep pocket with a 2 mm corner radius will flag it and suggest 3 mm, because a Ø6 mm tool is stiffer and cuts faster than a Ø4 mm tool. That single change can remove hours from a cycle.

First-article parts ship in 3–5 days. For a race programme working to a test date, that window is what makes iteration possible: cut a part, run it, adjust the model, cut again.

Confidentiality matters here. Race geometry is competitive information. Uploads are handled as secure and confidential, and an NDA is available on request before files are shared.

• 1
  File formatsSTEP or native CAD, plus a PDF drawing.
• 2
  Quote turnaround12 hours, with DFM notes.
• 3
  No MOQOne prototype or a 10,000+ run.