How Do You Fix Chatter on CNC Machines?
Chatter is a self-excited vibration. It leaves marks on the part, wears the tool and can damage the spindle if you keep cutting. This page is written for machinists, programmers and shop engineers who need to find the cause fast instead of guessing at speeds. Read it and you can tell which of the five common causes is driving your instability.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
Symptom, likely cause, and what to change
Start here. Match the pattern on the part or the sound in the cut, then work the fix in the right column.
| Symptom | Likely cause | What to change |
|---|---|---|
| High-pitched ring, even pitch | Tool holder or tool overhang too long | Shorten overhang, use shrink-fit holder |
| Marks spaced evenly along the wall | Spindle speed sits on a natural frequency | Move speed 10–20%, test again |
| Loud noise only in deep pockets | Low radial engagement, thin walls | Trim radial depth, add support |
| Chatter near chuck jaws in turning | Workpiece deflection, weak grip | Shorten stick-out, check jaw pressure |
| Fine chatter on a finished face | Dull insert or wrong edge geometry | Change insert, check runout |
| Vibration that comes and goes | Inconsistent chip load or feed override | Lock feed override, check program |
| Chatter at low rpm only | Spindle bearing play or loose drawbar | Check spindle runout, call maintenance |
Fix the loop, not the symptom
Chatter comes from one weak point in the loop. Find it, change one thing, and retest. If your part is thin, long or hard to hold, send us the drawing and we will review the setup and the cut before quoting.
Why chatter is not random noise
Chatter is a self-excited vibration. The cutting force pushes the tool away from the work, the tool springs back, and the next tooth cuts a slightly thicker chip. That thicker chip pushes harder, so the cycle feeds itself. The result is a stable oscillation at a frequency set by the stiffness and mass of the whole loop: spindle, holder, tool, fixture and part.
That is why one change often kills it. If you move the tooth-passing frequency away from the natural frequency of the loop, the feedback stops. You do not need to make the machine perfect. You need to break the loop at one point.
You can hear the difference. Forced vibration from an unbalanced tool or a bad gear has a fixed pitch tied to rpm. Chatter pitch drifts with speed and gets louder as you cut deeper. If the noise changes when you drop the feed, it is chatter. If it stays the same, look at the spindle or the drive.
The fix depends on which part of the loop is weakest. A long tool in a thin wall is a stiffness problem. A dull insert on a heavy cut is a force problem. The table above and the steps below follow that order.
- 1Self-excitedThe vibration feeds itself; it does not need an outside shake.
- 2Frequency mattersIt locks onto the natural frequency of the loop.
- 3One weak linkFix the softest part first, not every part.
Chatter in milling: tool and setup checks
In milling, tool overhang is the first thing to measure. A Ø12 mm carbide end mill held 60 mm out of the holder is roughly eight times less stiff than the same tool held 20 mm out. Pull it back if the geometry allows. If the part needs the reach, move to a shrink-fit or hydraulic holder and a carbide shank tool.
Next, look at radial depth of cut (RDOC). Cutting forces scale with the chip cross-section, so dropping RDOC from 50% of the tool diameter to 15% removes most of the load that drives the vibration. This is the single most reliable milling fix, but it costs cycle time. Use it where finish and tool life matter more than minutes.
Spindle speed is the second lever. Increase or decrease rpm by 10–20% and listen. The tooth-passing frequency shifts, and the cut can move off resonance. In aluminum, going faster often helps because the cutting force drops as the material softens. In titanium and stainless, going slower with a heavier feed per tooth usually works better, because rubbing and work hardening are the real enemy.
Check runout before you touch the program. A holder with 0.02 mm runout makes one tooth do most of the work, and that unbalanced load is enough to start chatter on a light finishing pass.
- 1OverhangKeep it under 4× diameter when you can.
- 2RDOCTry 15–25% of diameter for finishing.
- 3RunoutKeep tool runout under 0.01 mm.
- 4SpeedShift rpm 10–20% and retest.
Chatter in turning: overhang and insert choice
In turning, the workpiece is often the weak side. A shaft held 4× its diameter out of the chuck will deflect under a normal turning load. Chatter appears as a regular pattern near the chuck jaws or as a ringing that grows toward the tailstock. Shorten the stick-out or add a steady rest before you change any insert.
Insert geometry matters more than grade in a turning chatter case. A positive rake, sharp edge and small nose radius cut with less force than a strong negative insert. On long slender parts, that reduction is often enough. On interrupted cuts, keep the tougher insert and fix the setup instead.
Depth of cut is the force knob. If the part is flexible, take lighter passes at a higher feed. A 0.2 mm depth at 0.25 mm/rev cuts cleaner than a 1 mm depth at 0.1 mm/rev, because the chip is thicker and the cutting edge does not rub.
Check the chuck, too. Worn jaws, low hydraulic pressure or a loose top jaw let the part move. A dial indicator on the part while you push it by hand will show movement well before you hear anything.
- 1Stick-outKeep it under 4× diameter or use a rest.
- 2InsertUse positive rake and a small nose radius.
- 3CutLighter depth, heavier feed.
- 4GripCheck jaw wear and pressure.
How the material changes the fix
Aluminum is ductile, so it deforms and springs back under the edge. That springback feeds the vibration. The usual fix is higher speed and lower radial engagement, with a sharp polished tool and good chip evacuation. Keep the flutes clear; packed chips add force and heat.
Titanium and Inconel are the opposite problem. They are strong and conduct heat poorly, so the cutting force stays high and concentrated at the edge. Lower surface speed, heavier feed per tooth and a rigid setup work better than chasing rpm. Rubbing is what kills the tool and starts the chatter.
Stainless steels like 304 and 17-4PH work harden. If the tool rubs instead of cutting, the surface gets harder and the next pass pushes even harder. Keep the feed up, never dwell, and take a depth of cut that gets under the hardened skin.
Plastics and thin-wall parts are stiffness-limited, not force-limited. Support the part with fixturing, wax or a soft jaw, and use a sharp single-flute cutter at a low feed. Reducing force helps, but supporting the wall helps more.
- 1AluminumFaster speed, light radial cut, sharp edge.
- 2TitaniumLower speed, heavier feed, rigid setup.
- 3StainlessNo rubbing; keep feed up.
- 4Thin wallsSupport the part before changing speeds.
Toolpath and program settings that suppress chatter
Trochoidal and high-efficiency milling paths keep radial engagement low and constant. That steady load avoids the sudden force spikes that trigger chatter in a corner. If you have the option, turn it on for deep pockets and hard materials.
Varying spindle speed during the cut, often called speed modulation, is a proven method for breaking the feedback loop. A change of 5–10% around the nominal speed is usually enough. Some controls have it built in; otherwise, a small programmed ramp works.
Ramping into the cut instead of plunging straight down reduces the initial shock and keeps the chip load even. On deep pockets, a helical entry with a 2–3° ramp angle is a good default.
Finally, check the feed override lock. If an operator has been riding the override, the chip load changes mid-cut and the vibration starts. Lock it for finishing passes and note the value in the setup sheet.
- 1TrochoidalLow, constant radial engagement.
- 2Speed modulationVary rpm 5–10% to break the loop.
- 3Ramp entryHelical entry at 2–3°.
- 4OverrideLock feed for finishing.
When to stop tuning and check the machine
If a cut that was stable last week now rings, the machine may have changed. Check spindle runout with a test bar. A reading above 0.01 mm at 100 mm from the gauge line points to bearing or taper wear. Stop cutting and get maintenance involved.
Check the drawbar force as well. A weak spring or low hydraulic pressure lets the holder move under load. This shows up as chatter that gets worse at higher rpm and disappears at low speed. It is a machine problem, not a program problem.
Leveling and foundation matter on large machines. A machine that has settled or lost level can show vibration in one area of the table only. Re-level and check the anchor bolts before you spend more time on parameters.
Once the machine is confirmed good, go back to the loop. In most shops, the fix is one of five things: shorter overhang, lighter radial cut, speed shift, better workholding or a sharper tool. In our experience, that covers the large majority of chatter calls.
- 1RunoutOver 0.01 mm at 100 mm needs a look.
- 2DrawbarLow force lets the holder move.
- 3LevelRe-check after any move or settlement.
Five steps to fix chatter on CNC machines
Work in this order. Stop when the noise and the marks are gone. Do not change two things at once or you will not know which one worked.
- 11. Confirm it is chatterListen for a pitch that drifts with rpm, and look for evenly spaced marks. Note the spindle speed, depth of cut, tool number and part setup. If the pitch is fixed and tied to rpm, stop and check the spindle or drive instead.
- 22. Shorten the tool or support the partPull the tool back to under 4× diameter if geometry allows. In turning, shorten the stick-out or add a steady rest. Re-cut one pass and listen. This step fixes most stiffness-limited cases.
- 33. Cut the radial loadDrop RDOC to 15–25% of the tool diameter and keep the feed per tooth constant. In turning, reduce depth to 0.2–0.5 mm and raise feed to 0.2–0.3 mm/rev. Expect a longer cycle; accept it if finish matters.
- 44. Shift the spindle speedMove rpm by 10–20% up or down. In aluminum, try faster first. In titanium and stainless, try slower with a heavier feed. If the control supports speed modulation, enable a 5–10% variation and retest.
- 55. Check the tool, holder and runoutFit a fresh insert or end mill. Measure runout at the cutting edge; keep it under 0.01 mm. Check holder taper and drawbar force. If runout is over 0.02 mm, change the holder before you touch the program again.
Chatter questions engineers ask
Does increasing spindle speed always stop chatter?
No. It shifts the tooth-passing frequency, and sometimes that moves you onto a worse resonance. Test both directions in 10–20% steps.
In aluminum, faster usually helps because the cutting force drops. In titanium and stainless, slower with a heavier feed is often the better move.
How much can I reduce radial depth of cut before it hurts?
A drop from 50% to 15% of the tool diameter removes most of the load that drives chatter. The trade-off is cycle time, which can rise by 30–50% on a roughing pass.
Use it where finish and tool life matter most, and keep the feed per tooth constant so the edge still cuts instead of rubbing.
Why does chatter appear only on the finishing pass?
Finishing passes often use a small radial engagement and a light feed. That combination lets the tool rub rather than cut, and rubbing feeds the vibration.
Raise the feed per tooth, use a sharper edge, and make sure the tool is not worn. A light pass with a dull tool is the classic finishing chatter case.
Can a fixture cause chatter?
Yes. Any looseness in the vise, clamps or soft jaws lets the part move. That movement shows up as chatter even when the tool and speeds are fine.
Push the part by hand with a dial indicator on it. If you see movement, tighten the setup before changing parameters.
Is chatter ever acceptable?
Only on a roughing pass where the surface will be removed later, and only if tool life is still acceptable. It always costs something, either in tool wear or in spindle load.
For any finished surface, treat chatter as a defect. The marks will not pass inspection and the vibration shortens tool and spindle life.
What tolerance can we hold on a part that tends to chatter?
On our machines we hold ±0.005 mm (±0.0002 in) and finishes from Ra 0.2–0.8 μm on stable setups. A part that wants to chatter will not hold that until the loop is fixed.
Once the setup is rigid and the parameters are right, 100% inspection before shipment confirms the result. Reports are available on request.
Send us the part that keeps ringing
Upload your drawing and we will return a quotation and a free DFM analysis within 12 hours. Production can start within 24 hours, and parts ship in 3–5 days.
12-hour quote100% inspectionNDA on request