CNC Mirror Finishing Technology

• 1
  What a mirror surface actually is at the tool tip
• 2
  Which metals take a mirror and which fight it
• 3
  The five variables that decide the result
• 4
  Where CNC mirror finishing stops making sense
• 5
  Surface finish ranges and what produces them

A mirror surface reflects light with almost no scattering. On a machined part that means surface roughness in the low Ra 0.05 μm range and a scratch pattern smaller than the wavelength of visible light. Below that scale the eye stops seeing tool marks and starts seeing a reflection.

CNC mirror finishing technology does not start with polishing. It starts with the last cutting pass. If the tool leaves a 2 μm scallop, no amount of buffing will make that surface flat. It will only round the peaks and fill the valleys with smeared metal, which looks shiny under a lamp and turns hazy under a collimated beam.

So the first question on any mirror job is not which polish to use. It is whether the machine, the holder and the tool can hold a stable depth of cut for the whole path. A mirror finish is a machining result first and a finishing result second.

This is why two parts with the same drawing callout can arrive with very different surfaces. One shop holds the cutting conditions and skims 0.02 mm off the last pass. Another leaves 0.15 mm and tries to grind it out later. Only one of them keeps the geometry.

Aluminium is the easiest mirror substrate. 6061-T6 and 7075 cut cleanly with single-crystal diamond and hold a sharp edge, so the chip shears instead of tearing. We run these at high spindle speed with a shallow finishing pass and get a reflective surface straight off the machine on many faces.

Copper and brass behave similarly, which is why C110 and C36000 show up in reflectors and waveguide parts. The material is soft and gummy, so the risk is built-up edge. A diamond tool with a positive rake and a strong air blast usually solves it.

Stainless is a different story. 304 and 316 work-harden in front of the tool, so the cutting edge has to stay sharp and the feed per tooth has to be high enough to cut under the hardened layer. 17-4PH in the H900 condition machines more predictably than the annealed state for mirror work.

Titanium and Inconel are poor mirror candidates. Low thermal conductivity traps heat at the edge, and the surface tends to smear. We can reach Ra 0.4 μm on TC4 with careful parameters, but claiming a true mirror on Inconel would be dishonest. Plastics like PMMA and PEEK polish well but scratch if anyone looks at them wrong, so handling becomes the limit.

Tool geometry comes first. For mirror turning we use single-crystal diamond with a nose radius from 0.5 mm to 2 mm and a wiper flat where the geometry allows it. For milling, PCD or CBN inserts with a fine edge prep hold up longer than coated carbide, which tends to leave a faint orange-peel texture as the coating wears.

Spindle and holder balance sets the ceiling. Any runout above about 2 μm shows up directly as a periodic pattern in the surface. We balance holders for high-speed work and keep tool overhang as short as the part allows, because a 4× diameter overhang will chatter no matter how good the insert is.

Depth of cut on the finishing pass is small and constant, typically 0.02–0.05 mm. Feed per tooth sits in the 0.02–0.05 mm range for diamond work. Coolant choice matters too: a clean, filtered oil mist or high-pressure coolant clears chips that would otherwise drag across the surface and leave a scratch.

Finally, the part has to be rigid in the fixture. Thin walls deflect under cutting force, and a deflecting wall produces a wavy surface that no polishing step can flatten without changing the dimension. If the wall is under 1.5 mm, we often adjust the process plan before touching the finish.

Mirror finishing is slow and it is done at low material removal rates. If a feature only needs to look good in a catalog photo, bead blasting or brushing gets 90% of the visual effect at a fraction of the cost. Buyers who specify a mirror on every face usually pay for something nobody measures.

Deep cavities are the classic problem. A 4 mm ball tool cannot reach into a 60 mm deep pocket and still hold a mirror finish on the floor, because the overhang kills rigidity. In those cases we either split the part, use a larger tool on the accessible faces, or accept a fine machined finish inside and mirror only the visible face.

Internal bores and cross-holes are another boundary. Polishing a bore by hand is inconsistent, and abrasive flow machining changes dimensions. If the bore diameter tolerance is tight, mirror finishing inside it may not be worth the risk.

Sharp edges also limit the process. Diamond tools chip at a sharp corner, so we usually leave a 0.2–0.5 mm edge break. If the drawing calls for a knife edge and a mirror finish, one of those two requirements has to give.