Causes and repairs of CNC milling tools

Flank wear is the baseline. A uniform wear land on the primary relief face is normal and predictable. On a carbide end mill cutting 6061 at 300 m/min, a 0.15 mm land after 40 minutes is healthy. When the land reaches 0.3–0.4 mm, cutting forces climb, the surface finish drops a step, and dimensions start drifting. That is your cue, not a crash.

Crater wear sits on the rake face and looks like a shallow scoop behind the cutting edge. It comes from diffusion at high temperature, mostly on steel and titanium. If you see it on 4140 but not on 1018, the coating is losing the thermal fight.

Chipping and micro-fracture show up as a jagged edge line. Interrupted cuts, hard spots in castings, or a tool holder with 0.02 mm runout will do it. A chipped edge on a finishing pass leaves visible chatter marks and blows the Ra target.

Built-up edge is welded workpiece material on the cutting edge. Aluminium, low-carbon steel and some stainless grades form it when surface speed is too low and feed per tooth is too light. The edge looks like it grew a lump, then the lump breaks off and takes carbide with it.

Thermal cracking appears as fine comb-like cracks perpendicular to the cutting edge. It is classic in dry milling and in wet milling with interrupted coolant. The edge heats and cools hundreds of times a second.

Notch wear and depth-of-cut notching happen right at the top of the axial engagement. A hard scale, a work-hardened layer from a previous pass, or a cast skin will carve that notch in minutes. It is common on Inconel and on flame-cut plate.

• 1
  Check firstRunout at the holder, then feed per tooth, then coolant.
• 2
  Ignore lastSpindle speed alone rarely explains a chipped edge.

Cutting temperature is the biggest single driver. Carbide softens above roughly 800 °C at the edge, and cobalt binder starts diffusing into the chip. If your chips come off blue on a steel job, you are running hot. Reduce surface speed or increase feed per tooth so the heat leaves with the chip instead of soaking into the tool.

Feed per tooth is the second driver, and it is the one people get backwards. Too light a chip rubs instead of cuts. That work-hardens stainless, welds aluminium, and wears the edge faster than a heavier feed would. A 12 mm three-flute cutter in 304 usually wants 0.05–0.08 mm per tooth, not 0.02 mm.

Rigidity decides how much of that force reaches the edge. A 4,000 mm long part on a machine with 4,000 × 400 × 150 mm travel behaves differently from a compact 500 × 500 × 450 mm envelope. Long overhangs, thin floors and tall walls flex, and the edge takes a hammering instead of a cut.

Coolant chemistry matters more than volume. Straight oil and high-pressure through-spindle coolant reach the cutting zone on deep pockets. Flood coolant aimed at the outside of a 60 mm deep cavity does very little. On titanium we prefer high-pressure through-tool delivery, on aluminium a mist or air blast often beats flood.

Workpiece condition is the quiet one. Hard inclusions in ADC12, a 0.3 mm work-hardened skin on 17-4PH after a roughing pass, or abrasive filler in carbon fibre will destroy an edge that was cutting fine an hour earlier.

Uncoated fine-grain carbide is still the default for aluminium. It is sharp, cheap to regrind, and does not need a coating that will flake on a soft workpiece. Add polished flutes if you are cutting 6061 or 7075 at high speed with a lot of chip evacuation.

TiAlN-coated carbide handles steel and stainless. The coating survives the 700–900 °C range where uncoated carbide wears fast. For 4140 and 4340 in a hardened state, move to AlTiN or AlCrN, which hold up better above 900 °C.

For titanium TC4 and Inconel, sharp geometry and a smooth coating matter more than hardness. A polished AlCrN or TiSiN tool with a positive rake and generous flute space moves heat into the chip. Uncoated carbide with high cobalt content also works if you keep speeds low and feeds solid.

Diamond-like carbon suits aluminium and copper alloys when built-up edge keeps returning. CBN and ceramic inserts are for hardened steel above 45 HRC where carbide simply cannot hold an edge for long.

On plastics, use single-flute or two-flute polished carbide. PEEK and carbon fibre are abrasive, so a coated tool lasts longer, but the geometry has to clear chips aggressively or the part will melt at the edge.

Regrinding is the standard repair. A qualified shop removes the worn land, re-establishes the relief and rake angles, and recoats the tool. You lose a small amount of diameter each cycle. On a 12 mm end mill, three or four regrinds are realistic before the diameter and flute geometry drift too far from the original program.

Watch the coating. Regrinding removes it from the flank, and a partially coated tool wears unevenly. Either recoat after grinding or accept a shorter life and adjust the offsets accordingly. Never run a reground tool on a finishing pass without re-measuring length and diameter in the presetter.

Edge honing helps on interrupted cuts. A light 0.02–0.03 mm hone removes micro-chips and stops crack propagation. Too much hone on aluminium dulls the edge and creates built-up edge.

Tool holders are part of the repair. If runout is above 0.01 mm at the tool tip, the sharpest cutter will still chip. Clean the taper, check for fretting, and replace worn collets rather than tightening harder.

Stop repairing when the diameter drops below the minimum for your program, when the helix or flute form is visibly altered, or when the cost of the regrind approaches half the price of a new tool. At that point the geometry is no longer what the CAM path assumed.

On our own 5-axis work, we log tool life against material and operation. When a cutter starts failing earlier than its recorded baseline, that is a signal to check the machine, the fixture or the incoming stock before blaming the tool.

Set a wear limit and honour it. Measure the flank land with a toolmaker's microscope at fixed intervals, or let the machine monitor spindle load and cutting time. Replace or regrind at the limit instead of waiting for a bad surface finish to tell you.

Control runout at the holder. We keep runout under 0.01 mm on finishing tools. That alone can double tool life on stainless and titanium, because the load spreads across all flutes instead of one.

Match coolant to the operation. Through-spindle high pressure for deep pockets and titanium, air blast or mist for aluminium, flood for general steel. Changing coolant strategy mid-job without changing speeds usually makes things worse.

Vary the depth of cut on the finishing pass by a few tenths of a millimeter. It moves the notch wear line away from the previous pass and avoids that sharp step forming on the edge.

Keep a simple log per tool type: material, speed, feed, coolant, life in minutes, failure mode. After a few jobs the pattern is obvious, and you can set the tool change interval from data instead of from a noise or a scrapped part.