Carbide Speeds and Feeds
The numbers behind the cut: RPM, chip load, and feed-rate math.
VIEW PAGE →Carbide does not break randomly. Every snapped end mill and chipped edge has a cause — and almost every cause is fixable. Here is the shop-floor diagnosis guide.
Work through these in order. The first two — runout and rigidity — account for the majority of unexplained carbide failures.

Runout — the tool spinning off-center — means one flute takes a far heavier cut than the others. That flute overloads and breaks while its neighbors barely cut. Causes: a dirty or damaged collet, a worn holder, a chip stuck in the collet, or a tool shank that is not fully seated. The fix is measurement: indicate the tool and keep total runout under half a thousandth for finish work. Most shops are shocked by what they find the first time they check.
Carbide is brittle — it does not flex and spring back, it fractures. Long stick-out, weak workholding, a worn spindle, or a spindly part all let the tool vibrate and hammer. The fix is stiffness: shortest possible stick-out, the tool gripped as close to the work as the job allows, solid workholding, and no cutting with a machine that needs rebuilding.
Too heavy a chip load obviously overloads the edge. But too light is the sneakier killer: the edge rubs instead of cutting, generating heat and work-hardening the material — and then the hardened skin breaks the edge on the next pass. Carbide must always be cutting. If the chips look like dust, the feed is too timid.
Plunging straight into material, slamming into a wall at full feed, or exiting a cut with the edge unsupported all shock-load the carbide. Arc entries, ramped entries, and reduced feed on entry and exit keep the shock out of the edge. In milling, climb milling generally treats the edge better than conventional — but the entry move matters more than the direction.
In interrupted cuts — milling with coolant flooding a hot edge, then air, then coolant again — the rapid heating and cooling cycles crack the carbide perpendicular to the edge. If you see fine thermal cracks, the answer is often less coolant, not more: air blast or dry cutting with the right coating survives interrupted cuts better than flood coolant.
A finishing grade in a roughing cut, a sharp positive edge in an interrupted cut, a general-purpose tool in hardened steel — the tool was never going to survive. Match the grade's toughness to the cut's violence and the geometry to the material.
| Symptom | Likely cause | Fix first |
|---|---|---|
| One flute chipped, others fine | Runout | Clean and indicate the holder; replace worn collets |
| Tool snaps at the shank or mid-flute | Deflection / excessive stick-out | Shorten stick-out; stiffen workholding |
| Edge crumbles progressively | Rubbing from too-light feed | Increase chip load until the tool is truly cutting |
| Breaks on entry into the cut | Entry shock | Arc or ramp entry; reduce feed on entry |
| Fine cracks across the edge | Thermal cracking | Reduce coolant in interrupted cuts; tougher grade |
| Instant failure in hard material | Wrong grade | Match grade toughness to the material and cut |
When a tool fails catastrophically — a full crash, not gradual wear — do the post-mortem before running the next tool. Check the program's entry move, verify the work offset, confirm the tool length offset, and look at what the tool hit. Crashes from offset errors repeat exactly until the offset is fixed, and each repeat costs another tool. Five minutes of diagnosis is always cheaper than the next snapped end mill.
Sometimes every mechanical is correct and tools still die young. Then the suspect is the program: feeds and speeds copied from a different machine, a toolpath that buries the tool in corners, or a post-processor output nobody verified. The fix is to prove the program the way you prove a setup — single-block the first run, watch the load meter, and confirm the cut matches the intent before turning it loose.
The most common causes are runout (one flute overloaded), excessive stick-out letting the tool flex, and entry shock. Measure runout, shorten the stick-out, and ease the entry — those three fix the majority of breakage.
Yes. Chatter is the tool hammering against the work thousands of times per minute, and brittle carbide fractures under that hammering. Kill chatter with rigidity, shorter stick-out, and speed or feed adjustments that break the harmonic.
Carbide is far harder but much less tough than HSS — it does not deform, it fractures. HSS bends and dulls progressively; carbide holds a perfect edge until the load exceeds its fracture point, then a piece breaks off.
In interrupted cuts, yes — flood coolant hitting a hot edge, then air, then coolant again creates thermal shock cracks. In continuous cuts, coolant is usually beneficial. Match the coolant strategy to the cut type.
Indicate the tool in the holder with a dial indicator. If total runout exceeds about half a thousandth, one flute is doing disproportionate work. Clean the collet and holder, reseat the tool, and measure again.
Yes. Broken does not mean worthless — snapped end mills are solid carbide all the way through, worth real money by the pound. There are no minimums: every bucket counts. Call 312-880-8174.
Broken does not mean worthless — snapped end mills are solid carbide all the way through. Every tool this guide could not save is still tungsten carbide, worth real money by the pound. There are no minimums: a handful of broken tools or a full drum, every bucket counts. On bigger lots, we handle the freight so you do not lift a finger. Call 312-880-8174.