CNC Milling Design Guidelines: What to Check Before You Send a Quote

• 1
  Key takeaways
• 2
  What to compare before you award a milling job
• 3
  CNC milling design guidelines for pockets, radii and walls
• 4
  Hole depth, thread depth and the 4:1 rule
• 5
  Tolerance, surface finish and where the money goes
• 6
  Material choice, quantity and the cost curve
• 7
  A 6-step check we run before releasing a milling quote

The first thing we look at on any drawing is the smallest internal radius. A cutter has a round profile, so an internal corner can never be sharper than the tool radius. If your pocket has a 1 mm corner and is 15 mm deep, the tool must be 2 mm in diameter with a long flute length, and it will run at a fraction of the speed of a 10 mm cutter. The practical rule is simple: make the floor radius at least 0.25 × pocket depth. A 20 mm deep pocket wants a 5 mm corner radius.

Wall thickness comes next. In aluminium 6061, walls under 0.8 mm deflect under cutting force, and the finished wall will not be straight. Keep walls at 1.0 mm or more for parts above roughly 50 mm. Small parts can go thinner because the cutting force is lower and the wall is shorter. If the function truly needs a 0.5 mm wall, say so on the drawing; we will slow the finishing pass and take lighter depths of cut.

Pocket floors and bosses should be flat where possible. A floor with a 3° draft is easy to cast but awkward to mill, because the cutter has to follow a sloping path and the finish changes across the surface. If the part will be milled, keep floors perpendicular to the tool axis. Also avoid sharp external corners where a chamfer or a small radius would do the same job. A 0.5 mm edge break removes burrs and stops the corner from chipping in handling.

One more geometry check: any feature that is not reachable from at least one of the six part faces may need a fourth or fifth axis, or a second setup. Every extra setup adds a datum transfer and a real chance of error. Ask yourself whether the feature could be reached by turning the part over instead of by buying a more expensive machine.

• 1
  Floor radius ≥ 0.25 × depthKeeps the cutter stiff and the finish consistent.
• 2
  Wall ≥ 1.0 mm in aluminiumThinner walls chatter and bow after stress relief.
• 3
  Keep floors flat, not draftedDrafted floors slow the tool path and change the finish.
• 4
  Break every edge at 0.5 mmRemoves burrs and prevents handling damage.

Hole depth is where most designs lose money. A standard twist drill is stable up to about 4 × its diameter. Past that, the drill wanders, the hole drifts off position, and the chip has nowhere to go. A Ø6 mm hole can be drilled to about 24 mm without special handling. A Ø6 mm hole that must be 60 mm deep is a different job: we peck drill in small steps, use a longer carbide drill, and often drill from both ends.

If you can, use a through hole instead of a deep blind hole. A through hole lets the coolant flush chips and lets the drill exit cleanly. A blind hole has to leave a flat or conical bottom, and the drill tip angle means a blind hole is never perfectly flat at the base. If a flat bottom is required for a dowel pin or a spring seat, call it out; we will come back with an end mill and true the floor.

Thread depth follows the same logic. A tapped hole should be at least 1.5 × the nominal diameter deep to develop full thread strength. M3 threads are the practical floor for production work, and even then a broken tap is a real risk in stainless. When a thread is smaller than M3 or the material is tough, we often suggest a threaded insert or a clearance hole with a nut on the far side.

Thread callouts also need a class. A 6H class is standard for most work. If you need a tighter fit for a precision adjustment screw, say so, because the tap and the gauge change. And keep thread depth under 3 × diameter unless you have a reason; beyond that, tap breakage and thread drift rise sharply and the cost follows.

• 1
  Stay under 4 × diameterDeeper holes need peck cycles and longer tools.
• 2
  Thread depth ≥ 1.5 × diameterBelow that, the thread never reaches full strength.
• 3
  M6 is our standard floorSmaller threads are possible but raise the risk and the price.
• 4
  Specify thread class6H is standard; tighter classes need different gauges.

A tolerance is not a decoration. Each tight tolerance adds an inspection step, and sometimes a fixturing step. Our standard milling tolerance is ±0.005 mm (±0.0002 in), but that number only applies to the dimensions you actually control. If you put ±0.005 mm on an overall length of 400 mm, the thermal expansion of the part between morning and afternoon becomes a factor. Reserve tight tolerance for the features that mate, and leave the rest at a general tolerance block.

Surface finish works the same way. As-machined surfaces land at Ra 1.6–3.2 μm. A high-quality finish is Ra 0.8–1.6 μm, and a fine finish is Ra 0.2–0.8 μm. Going from as-machined to fine finish can double the cycle time on a face because the finishing pass uses a smaller stepover and a sharper tool. If the surface is only cosmetic, bead blasting or tumbling may be cheaper than a fine milling pass.

Think about which faces are functional. A sealing face, a bearing bore or a sliding surface needs a real finish. A bracket that bolts to a frame usually does not. Mark the functional faces on the drawing and leave the rest as-machined. On one part, that decision often removes more cost than changing the material.

Finally, remember that tolerance and finish interact. You cannot hold a fine finish and a loose tolerance on the same face and expect the shop to hit both easily; the tool path that produces the finish is the tool path that holds the dimension. Tell us which one matters more when they conflict, and we can choose the right strategy.

• 1
  ±0.005 mm is our standardBut only apply it to mating features.
• 2
  As-machined is Ra 1.6–3.2 μmFine finish Ra 0.2–0.8 μm costs real cycle time.
• 3
  Mark functional facesCosmetic faces can be blasted or tumbled instead.
• 4
  Say which one winsWhen finish and tolerance conflict, we need your priority.

Material selection changes the design rules more than most engineers expect. Aluminium 6061-T6 machines fast and holds ±0.005 mm without drama. Stainless 316 and 17-4PH work-harden, so light depths of cut and constant feed are mandatory; a tool that rubs instead of cuts will harden the surface and break on the next pass. Titanium Ti-6Al-4V and Inconel need even lower cutting speeds and more coolant, and thin walls in those materials are much harder to hold.

Plastics behave differently again. POM and PEEK hold tolerance well but move with temperature, so inspection should happen after the part cools. ABS and PC scratch easily, and a bead blast that looks good on aluminium will ruin a plastic surface. If the part is a housing with cosmetic requirements, say so before we choose the finishing route.

Quantity changes the process, not just the price. A single prototype is milled from a solid billet with no fixturing. At a few hundred parts, a soft jaw or a dedicated fixture pays for itself by removing setup time. Above a few thousand parts, casting or molding may beat milling on unit cost, and we will say so even though it moves the work to another process. There is no minimum order quantity here; runs go from one part to 10,000+ parts.

The last cost driver is the drawing itself. A drawing with a clear datum scheme, defined tolerance blocks and marked functional faces quotes faster and machines more predictably. A drawing that leaves every dimension at ±0.05 mm forces us to guess, and guessing shows up later as a deviation report.

• 1
  6061-T6 is the easy pathStainless and titanium need slower, lighter cuts.
• 2
  Plastics move with heatMeasure after the part cools to room temperature.
• 3
  Fixtures pay off at a few hundred partsBelow that, solid billet milling is usually cheaper.
• 4
  No MOQOne prototype and 10,000-part runs use the same process.