What Kind of CNC Machine Can Cut Stones?

• 1
  Why a CNC Machine Can Cut Stones Only With the Right Setup
• 2
  Which Machine Architecture Survives the Cut
• 3
  Spindle Speed, Feed and Diamond Tooling
• 4
  What Accuracy You Can Realistically Hold
• 5
  When to Cut Stone on a CNC and When to Stop
• 6
  Setting Up a Stone Cut Step by Step
• 7
  Machine Types for Cutting Stone: Capability Compared

Stone does not cut the way aluminum cuts. Granite and marble are brittle, so the tool does not peel a chip; it grinds a powder and fractures a shallow crack network ahead of the edge. That fracture is helpful, because it lowers the force you need. It is also dangerous, because the crack does not stop where you want it to.

The abrasive part is harder to live with. Quartz and feldspar in granite sit around 6 to 7 on the Mohs scale. A carbide end mill at that hardness wears in minutes, not hours. Diamond is the only practical cutting material, and diamond does not like heat.

So the machine has to hold three things at once: enough stiffness to stop chatter, enough damping to absorb a discontinuous cut, and enough coolant flow to keep the diamond cool. Miss any one of the three and the edge fails early, or the stone chips out along the profile.

Rigidity matters more than raw spindle power here. A 3 kW spindle on a heavy cast frame will outlast a 7 kW spindle on a light gantry, because the light frame rings and the diamond edge chips on every reversal.

Most stone work is also wet work. Dry cutting raises edge temperature fast, and above roughly 600 °C the diamond graphitizes and stops cutting. Water is not a nicety. It is part of the cutting tool.

Three machine families show up in stone work, and they are not interchangeable. The first is the stone bridge saw or stone router, built from the start with a water table, a sealed linear guide system and a low-speed, high-torque spindle. If your part is a slab or a flat architectural panel, this is the correct tool.

The second is the gantry router, the same platform used for wood and plastic. It can cut softer stone such as soapstone, slate and some marble if you slow the feed and flood the cut. It struggles with granite because the gantry flexes under the interrupted load.

The third is the metalworking VMC or 5-axis machining center. This is where the answer gets interesting. A VMC with a 40-taper spindle and a flood coolant system will cut stone, but it is the wrong purchase if stone is your only product. It is the right choice when stone is one material among many.

The deciding factor is not the number of axes. It is mass, guideway protection and chip evacuation. Stone slurry is abrasive and it destroys unprotected linear rails and ball screws within weeks.

Stone cutting runs far slower than metal cutting, and this surprises people. Peripheral speeds for diamond on granite usually land between 15 and 35 m/s. A Ø12 mm sintered diamond burr at 20 m/s turns at roughly 32,000 rpm, but most stone routers run Ø50 to Ø150 mm wheels, so the same surface speed arrives at 2,000 to 8,000 rpm.

Feed per tooth is not a useful number for grinding-type stone removal. Depth of cut per pass and stepover matter more. For granite with a Ø12 mm sintered burr, a common starting point is 0.5 to 2 mm axial depth and 30 to 50 percent stepover, then adjust by ear and by edge wear.

Diamond tooling comes in three flavors. Sintered metal bond is the workhorse for granite. Resin bond cuts faster and cooler but wears quickly, so it suits marble and finishing passes. Electroplated tools are cheap and aggressive, and they die fast on hard stone.

Water flow should be aimed at the contact point, not at the part in general. A useful target is 8 to 15 liters per minute at the cut for a Ø12 mm tool. If the slurry coming off the cut is grey and thin, flow is adequate. If the cut smokes or the slurry is paste, you are starving the edge.

Coolant also has to carry the swarf away. Recirculated water needs a settling tank and a filter, because diamond grit and stone powder recirculate and grind the machine as well as the part.

Stone moves. Marble and granite have internal stress, and a slab that is flat on the table can bow 0.1 to 0.3 mm after a few millimeters of material comes off one face. That is a material behavior, not a machine error.

On a rigid machine with a stable slab, a flat profile in granite can hold roughly ±0.05 to ±0.1 mm over 300 mm. Tighter than that is possible on small parts, but you should not design a stone assembly with a press fit.

Edge quality is the other limit. Diamond leaves a matte, slightly chipped edge on granite. If the drawing calls for a sharp cosmetic edge, plan a secondary chamfer or polish step. Marble cuts cleaner and takes a better polish.

Do not expect metal-level numbers from a stone cut. GreatLight holds ±0.005 mm (±0.0002 in) on metal parts in aluminum, stainless and titanium, with finishes down to Ra 0.2–0.8 μm. Those numbers apply to the metal side of a project, not to granite.

Stone on a CNC makes sense for one-off architectural details, inlays, small batches of countertop features, engraved panels and prototype stone-and-metal assemblies. It also makes sense when the stone part has to mate with a machined metal frame and the fit matters.

It stops making sense when the part is a large flat slab with simple straight cuts. A waterjet or a bridge saw does that faster and cheaper. A CNC adds value where the geometry is complex, the tolerance is tight, or the part is small.

It also stops making sense when the design requires a press fit into metal. The thermal expansion and the elastic behavior of stone are different enough that a stone-to-metal interference fit will crack in service. Use a compliant adhesive or a gasketed joint instead.

Mixed-material work is the strongest case for a 5-axis machining center. A single setup that machines a stone insert pocket and the surrounding aluminum housing avoids the re-fixturing error that ruins the fit. That is where 5-axis earns its hourly rate.

If stone is a small part of your product line, outsource it. Buying a stone-capable machine to run a few parts a year rarely pays back, and the slurry handling alone is a maintenance project.