CNC Cutting Vibration: Find the Cause, Pick the Fix
Chatter marks, a screaming cutter, a taper you cannot measure twice the same way. Most CNC cutting vibration comes from one weak link: the tool, the holder, the fixture, the part, or the machine. This guide maps symptoms to causes so you can decide what to change first.

CNC Cutting Vibration: Symptom, Cause, Action
Read down the first column, then confirm the cause before you touch the program.
| Symptom | Likely cause | What to do first |
|---|---|---|
| Regular marks, one per flute pass | Tool overhang too long or weak holder | Shorten overhang, switch to shrink-fit or hydraulic holder |
| High-pitched scream, sharp corners | Spindle speed at a resonance peak | Move speed 10–15% up or down, then re-cut |
| Worse as depth of cut rises | Radial engagement too high for the setup | Reduce radial width, keep axial depth for chip thinning |
| Vibration only on thin walls | Workpiece deflects under cutting force | Add supports, reduce feed, take spring passes |
| Marks fade then return in a pattern | Fixture or vise not fully seated | Reclamp, check parallels and jaw lift |
| Vibration grows with spindle speed | Spindle bearing wear or imbalance | Check runout with a dial test indicator |
| Fine finish looks torn, no marks | Wrong feed per tooth for the material | Raise feed per tooth, use a coated grade |
Start With the Symptom, Not the Fix
CNC cutting vibration rarely has one cause. The marks on the part tell you which link in the chain is weak. A pattern that repeats once per flute usually points at the tool or holder. A pattern that repeats once per spindle revolution points at the spindle or the tool imbalance. Without that distinction you will adjust the wrong parameter and lose a shift.
Look at the chip before you touch the feed override. Thin, blue, powdery chips mean the tool is rubbing instead of cutting, which loads the setup and invites chatter. Silver chips with a clean curl mean the cutting geometry is close to right and the vibration is coming from stiffness, not from the cutting edge.
Measure the part, not the sound. Chatter often leaves a wall thickness variation of 0.02 mm to 0.05 mm that a caliper will show and an ear will not. Write the number down before and after each change so you know which change actually worked.
- 1One mark per fluteTool or holder problem
- 2One mark per revolutionSpindle, imbalance, or workholding
- 3Marks grow with depthStiffness or engagement problem
Tool Overhang and Holder Choice
Overhang is the single largest lever on CNC cutting vibration. Deflection rises with the cube of the length, so pulling a Ø10 mm end mill from 60 mm overhang back to 40 mm makes the tool roughly three times stiffer. If the geometry allows it, shorten the gauge length before you change any cutting parameter.
Holder type matters as much as length. A collet chuck with a worn nut can add 0.02 mm of runout at the tool tip, and runout loads one flute harder than the others. That single loaded flute sets the vibration limit for the whole cut. Shrink-fit and hydraulic holders hold runout under 0.005 mm and are worth the cost on finishing passes in aluminium and stainless.
Tool diameter has a floor. Below Ø6 mm in steel, the tool simply cannot absorb the cutting force at normal depths, so you have to change strategy: smaller radial engagement, higher spindle speed, and shorter flute length. Do not expect a long, thin tool to behave like a stub tool no matter how you tune the speed.
- 1Shorten overhang firstStiffness scales with the cube of length
- 2Check runout at the tipKeep it under 0.005 mm for finishing
- 3Match flute length to depthLong flutes on a shallow cut add flex
Speeds, Feeds and Stability Lobes
Every spindle and holder combination has stability lobes: speed bands where the cut is quiet and bands where it howls. The bands move when you change overhang, holder mass, or tool diameter. If one speed screams and a speed 12% away is quiet, you have found a lobe and you should stay in it.
Feed per tooth controls chip thickness, and chip thickness controls whether the tool cuts or rubs. In 6061 aluminium, a good starting point for a Ø12 mm three-flute end mill is 0.05 mm to 0.08 mm per tooth with a cutting speed near 300 m/min. Too low a feed lets the edge rub, which raises force and starts chatter.
Radial engagement and axial depth trade against each other. A 10% radial stepover with full axial depth keeps the chip thin and the force direction steady, which is the usual fix for long-reach tools. A 50% stepover at the same depth loads the tool far more and will chatter on any setup that is not rigid.
- 1Find the quiet bandChange speed ±10–15% and listen
- 2Do not starve the feedRubbing raises cutting force
- 3Low radial, full axialStable for long-reach tools
Fixture and Workpiece Stiffness
A part that rings like a bell will chatter no matter how good the tool is. Thin walls, tall bosses, and unsupported overhangs deflect under cutting force and spring back, which is what you see as a wavy surface. The fix is support, not speed. Add a jack, a tailstock, or a soft jaw that wraps the part.
Clamping force direction matters too. A vise that lifts the part on the moving jaw gives you a part that moves in Z and a face that is not flat. Indicate the part after clamping, not before. If the dial moves when you tighten the vise, the setup is the problem.
For long parts, use the machine table. A 4,000 mm travel machine can hold a part down on multiple clamps instead of two, which cuts the unsupported span and raises the natural frequency of the whole setup. More clamps, less ring.
- 1Support thin wallsUse jacks, soft jaws, or tailstocks
- 2Indicate after clampingCheck the part is not lifted
- 3Add clamps on long partsShorter span, higher stiffness
When the Machine Is the Source
The machine is the last place to look, not the first. Loose leveling pads and a worn axis guide let the whole machine move under load, and the marks that result look like tool chatter. Before you blame the spindle, check the machine is level and sitting on its pads, and check that the axis moves without a jump. A worn spindle or a loose guide shows up as vibration that grows with spindle speed and does not change when you swap the tool.
Measure spindle runout with a dial test indicator on a clean taper. Above 0.01 mm at the gauge line, the spindle or the taper needs service. Below that, the tool and holder are the more likely source.
Keep a simple log. Speed, feed, depth, holder, overhang, and the result. After three or four jobs the pattern is obvious, and you stop re-testing the same wrong idea.
This is the point where a second machine earns its keep. If a part runs quiet on one machine and chatters on another at the same parameters, the difference is in the machine, and no program change will fix it.
- 1Check level firstLoose pads move the whole machine
- 2Measure runout at the taperOver 0.01 mm needs service
- 3Log every changeThree jobs reveal the pattern
A Practical Sequence for Killing Chatter
Work down the list. Stop as soon as the marks clear.
- 1Shorten the overhangPull the tool back to the shortest gauge length the part allows, ideally 3× diameter or less. Re-cut one pass and compare.
- 2Check runout at the tool tipUse a dial test indicator. Keep it under 0.005 mm for finishing. Swap the nut or holder if it is higher.
- 3Change spindle speed ±12%Move away from the resonance peak. If the sound changes, you are near a stability lobe. Fine-tune in 5% steps.
- 4Raise feed per toothFor aluminium on a Ø12 mm three-flute, try 0.05–0.08 mm per tooth. Rubbing is worse than load.
- 5Drop radial engagement, keep axial depthTry 10–20% stepover with full depth for long-reach tools. Watch the chip color and shape.
- 6Support the partAdd jacks, soft jaws, or extra clamps. Indicate the part after clamping, not before.
- 7Check machine level and runoutLevel the machine on its pads. If spindle runout exceeds 0.01 mm at the gauge line, schedule service.
Common Questions
Does a slower spindle speed always reduce CNC cutting vibration?
No. Slower is only better when you are sitting on a resonance peak. There are speed bands where a slower cut is louder and leaves worse marks.
Move the speed in steps of 10–15% in both directions and keep the band that cuts quiet. That is faster than guessing.
Can I fix chatter by taking a lighter cut?
A lighter radial cut often helps, but a lighter axial cut can make things worse because the edge starts rubbing instead of shearing.
Keep the feed per tooth in range and reduce radial engagement first. If the tool rubs, cutting force goes up, not down.
Why does the same program chatter on one machine and not another?
Spindle stiffness, bearing condition, and the machine mass are different. The same holder on a lighter machine will have a different stability limit.
If a part runs clean on one machine and chatters on another with identical parameters, the machine is the variable. Check level, guides, and spindle runout.
How much runout is acceptable for a finishing tool?
Keep total indicated runout at the tool tip under 0.005 mm for finishing work. Above that, one flute carries most of the load and sets the vibration limit.
Check the holder, the nut, and the tool shank in that order. A worn collet nut is a common cause of a problem that looks like tool chatter.
When should I stop tuning and change the process?
If three controlled changes do not clear the marks, the setup has reached its stiffness limit. More tuning wastes time.
Change the strategy instead: shorter tool, more supports, or a different operation sequence that leaves the stiff material in place until later.
Does coolant type affect chatter?
Coolant changes friction and chip evacuation more than it changes stiffness. High-pressure coolant can help clear chips from a deep pocket, which reduces recutting and force.
It will not fix a long overhang or an unsupported wall. Treat it as a supporting change, not a cure.
Send Us the Part That Chatters
Upload your model and the problem you see. We return a quotation and a free DFM analysis within 12 hours, including a note on the features that are hard to hold steady.
12-hour quoteFree DFM analysis±0.005 mm tolerance