Vibration and chatter: how to spot them and how to get rid of them

📅 Published 28 September 2026 · A practical guide from mecanizador.com

Chatter is the most badly diagnosed fault in a workshop, because the instinctive reaction — drop the rpm and the feed — sometimes works, sometimes makes things worse, and is almost never the optimal answer. Understanding what is happening lets you fix it in two attempts instead of ten.

Two different vibrations

Forced vibration. There is an external source hitting at a fixed frequency: spindle imbalance, a chipped or badly seated insert, a damaged bearing, a pulley. You recognise it because the frequency tracks the rpm: drop the speed and the pitch falls proportionally, and the pattern on the part has a regular, repeating spacing.

Chatter (self-excited vibration). There is no external source. The cut feeds back on itself: the tool vibrates, leaves a wave in the surface, and on the next pass the edge meets that wave, which makes the chip thickness vary, which makes the cutting force vary, which increases the vibration. It is called the regenerative effect and the result is a characteristic high squeal and irregular marks that do not follow the feed pitch.

The difference matters because the fixes are opposite. Forced vibration is cured by finding and removing the source. Chatter is cured by changing the conditions to get out of the unstable zone, and sometimes that means raising the rpm, not lowering them.

How to recognise it without instruments

• Sound: a high, metallic squeal that appears suddenly on reaching a certain depth or a certain overhang. A healthy cut sounds like a steady, low hum.
• Surface: wavy marks, often diagonal or "fish-scale", that do not line up with the feed pitch. If the marks are regular and at the feed pitch, it is geometry or the edge, not chatter.
• Chip: irregular, with varying thickness and pieces of uneven length.
• Edge: chipped in several places in a short time, instead of even progressive wear.

The causes, in order of likelihood

1. Excessive overhang. The number-one cause by a distance. Rigidity falls with the cube of the free length: double the overhang and rigidity drops to an eighth. A tool sticking out 60 mm when it could stick out 35 is a chatter generator.
2. Not enough clamping on the part. A long part with no tailstock, a thin wall with no support, a vice gripping too little surface. What is vibrating may be the part, not the tool. How to improve it is in workholding.
3. Depth of cut in the unstable zone. For every combination of machine, tool and material there is a limiting depth above which the system turns unstable, and that depth depends on the rpm in a non-linear way.
4. Feed too low. Counter-intuitive but very common: at a fine feed the edge rubs instead of cutting, and the friction excites the vibration. Raising the feed sometimes kills the chatter.
5. A tool with too many teeth in contact. On thin walls or deep cuts, a four-flute cutter keeps more edges cutting at once and the forces add up; a three- or two-flute spreads the load less evenly but excites less.
6. Wear. A worn edge cuts with more force and more friction: the same pass that ran fine yesterday squeals today.

What to try, and in what order

With the machine stopped, start with what costs nothing:

1. Shorten the tool overhang as far as the part allows. If you can take 15 mm out, take it out before you touch any parameter.
2. Check the workholding. An extra support, the tailstock, more grip area, packing under thin walls.
3. Check the edge. If it is worn or nicked, change it before you diagnose anything else.

And then, on parameters:

1. Change the rpm appreciably, in both directions. Not by 5%: try 20-30% up and 20-30% down. Stable and unstable zones alternate with rpm, so going up can solve it. This is the step most people skip.
2. Raise the feed per tooth. Less rubbing, a cleaner cut.
3. Reduce the radial depth and increase the axial depth in profile milling, rather than the other way round. Narrower, deeper cuts are usually more stable than wide, shallow ones.
4. Change the tool: fewer flutes, variable helix or unequal tooth spacing. Variable-pitch cutters exist precisely to break the chatter feedback loop.

Write down what you change. A chatter problem solved and not recorded comes back on the next batch of the same part.

What does not work

Halving everything usually removes the noise and brings three new problems: the cycle time goes through the roof, the fine feed makes the finish worse in stainless through work hardening, and the edge works by rubbing, so it lasts less. It is a patch, not a solution, and in production it is expensive.

Turning up the coolant flow does not work either: coolant helps with wear and chip evacuation, but it adds no rigidity. And more aggressive is not more productive if the machine cannot sustain it; what limits you is almost always the rigidity of the set-up, not the power.

When the machine is the limit

There are parts — 1 mm walls, unavoidable overhangs, titanium — where a conventional machine will not get there however well you tune it. There the answer is anti-vibration toolholders, a stiffer spindle or a machine with better damping, and that is capital expenditure. If the job is a one-off, it usually works out cheaper to put it out to a shop that already has the right machine: the workshop directory lists the declared equipment and capabilities of each one, and the 3-axis versus 5-axis comparison explains when the geometry of the part forces a change of machine.