Troubleshooting guide

Low Frequency Vibrations in CNC Turning: Causes and Fixes

Low frequency vibrations in CNC turning show up as a wavy surface, a drifting diameter and a chipped edge on long composite towers. This page is for machinists and process engineers who need to find the source and shut it down. You will get a symptom-to-cause table, five setup steps and the parameter ranges we use on our own lathes.

Get Instant Quote Talk to an Engineer

±0.005 mm toleranceRa 0.8–1.6 μm3–5 day shipping4,000 mm max size

CNC turning of a long composite tower prone to low frequency vibrations in CNC

On this page

1
Reading the vibration signature
2
Why low frequency vibrations in CNC turning build up on long towers
3
Workholding and support choices that remove the spring
4
Cutting parameters that shift the process off the resonance
5
Tool geometry, overhang and edge wear
6
Five steps to settle a ringing tower

Symptom → cause → action

Reading the vibration signature

Match the mark on the part to the most likely source before you change any offset.

Symptom on the part	Likely cause	First action
Even pitch marks, 3–8 per turn	Workpiece bending between centers	Add a steady rest near the cut
Waviness growing toward the tailstock	Tool overhang over 4 × shank Ø	Move to a 32 mm shank or shorten the holder
Spindle-speed-linked chatter	Dominant natural frequency of the bar	Step the speed ±10% off the peak
Taper plus a dull edge	Flank wear past 0.20 mm	Index the insert, recheck the offset
Rumble only on composite layup	Fiber pullout exciting the tool	Lower feed to 0.08–0.12 mm/rev
Marks that appear then vanish	Loose turret or worn drawbar	Check clamp force and turret bolts

Fix the setup before you rewrite the program

If you change only one thing, add support near the cut. On a long composite tower, stiffness buys more than any speed or feed adjustment, and it holds across the whole batch.

Get Instant Quote Talk to an Engineer

Mechanism

Why low frequency vibrations in CNC turning build up on long towers

Low frequency vibrations in CNC turning sit in the 20–200 Hz band. That is the range where the workpiece, the toolholder and the turret all want to move together. On a composite tower with a 6:1 or 10:1 length-to-diameter ratio, the part is the softest spring in the loop. Push it sideways with a few hundred newtons of cutting force and it bends, then springs back and cuts deeper on the next pass. The mark repeats, so you see a pattern rather than random roughness.

The force that drives it is rarely steady. In carbon fiber or glass-filled layups the fiber bundles break in bursts, so the cutting force drops and spikes several times per revolution. Each spike feeds energy into the part at its own natural frequency. Once the input and the natural frequency line up, amplitude grows until the tool lifts out of the cut. That is the point where the surface turns wavy and the insert starts to chip.

Direction matters. Force in the radial (Y) direction deflects the part the most, because the tower is usually stiffer along its axis than across it. A tool with a large lead angle pushes more of the cutting force into that soft direction. Reducing the lead angle or changing the insert geometry shifts the load back toward the stiff axis, and the amplitude drops before you touch the speed or the feed.

1
Below 200 HzStructural vibration. Fix the setup before changing the program.
2
200–1,000 HzTool or holder resonance. Shorten overhang or change the holder.
3
Above 1,000 HzEdge-level chatter. Look at insert grade, hone and edge prep.

Setup

Workholding and support choices that remove the spring

Most long-tower chatter is a support problem, not a speed problem. A steady rest placed within one diameter of the cut does more than any parameter change. On a Ø80 mm tower we keep the unsupported span under 3 × diameter and use a roller steady with light preload, about 0.02–0.05 mm of contact. Too much preload pushes the part into an oval; too little lets it ring.

When the tower has a bore, a tailstock center still leaves the middle free to sag. A plug or expanding mandrel at the free end adds stiffness without marking the finished surface. For composite parts, use a soft jaw or a urethane-lined clamp. Hard jaws on a thin composite wall crush the layup and create a high spot that shows up later as a bump in the roundness trace.

Check the spindle and turret before blaming the part. A drawbar that has lost 10% of its clamp force lets the work move under load, and the resulting mark looks exactly like a support problem. Clamp force and turret bolt torque belong on the same checklist as the steady rest adjustment.

1
Unsupported spanKeep under 3 × diameter for composite towers.
2
Steady rest preload0.02–0.05 mm contact, roller type preferred.
3
Soft jawsUse urethane or aluminum liners on thin walls.

Parameters

Cutting parameters that shift the process off the resonance

Once the setup is stiff, the fastest lever is spindle speed. Low frequency vibrations in CNC turning respond to small speed changes because the natural frequency is fixed while the tooth-passing frequency is not. If the mark repeats N times per revolution, multiply the current speed by 0.9 and by 1.1 and run two short passes. Keep the version with the lower amplitude and then fine-tune in 5% steps. Do not chase it with a single large jump of 30%; you may land on the next harmonic.

Feed rate controls the chip thickness and therefore the force per tooth. On a composite tower, a heavy feed with a light depth of cut often beats the reverse. Start at 0.08–0.12 mm/rev for roughing and 0.04–0.08 mm/rev for finishing, with depth of cut at 0.3–0.8 mm. If the surface still shows pitch marks, reduce depth first, then feed. Raising both at once just adds force.

For finishing, a wiper insert at 0.05–0.10 mm/rev can hold Ra 0.8–1.6 μm on a stable setup. On a tower that still rings, a wiper makes the pattern worse because it presses a wide contact area into a moving part. Match the insert to the stiffness you actually have, not the one in the catalog.

1
Speed trialTest 0.9 × and 1.1 × current rpm, then 5% steps.
2
Roughing feed0.08–0.12 mm/rev, depth 0.3–0.8 mm.
3
Finishing feed0.04–0.08 mm/rev with a standard radius insert.

Tooling

Tool geometry, overhang and edge wear

Overhang is the single number that decides whether a boring bar or a long turning tool will chatter. Stay under 4 × shank diameter, and if the feature forces you longer, use a carbide shank or a tuned bar. Steel bars at 6 × diameter will deflect enough to leave a taper even when they do not produce an audible squeal. The quiet ones are often the worst because nobody reacts to them.

Lead angle and rake change how much force goes into the soft radial direction. A 45° lead angle spreads the cut and pushes force sideways. A smaller lead angle, around 10–15°, directs more force along the axis, which the tower resists better. Negative rake inserts are tough but generate higher cutting forces, so on a thin composite wall they can start the vibration you are trying to remove.

Flank wear raises the force needed to cut by 20–40% before the edge looks dull to the eye. On a long tower that extra force is enough to push the process into chatter. Set a wear limit, 0.15–0.20 mm on the flank, and index on the limit rather than on the surface finish. For composites, edge chipping appears first, so inspect the corner radius under magnification every 20–30 parts.

1
OverhangUnder 4 × shank diameter; carbide above that.
2
Lead angle10–15° shifts force to the stiff axis.
3
Flank wear limitIndex at 0.15–0.20 mm, not on finish.

Field procedure

Five steps to settle a ringing tower

Run these in order. Each step is cheap; stop as soon as the mark disappears.

1
Mark the part and count the pitchCut a 30 mm test band, then count the repeating marks around the circumference. If the count is 3–8 per turn, the problem is structural. If it is higher and changes with speed, it is tool-level. Write the number down before you touch anything.
2
Add or adjust the steady restPlace a roller steady within one diameter of the cut and set preload to 0.02–0.05 mm. Re-cut the same band. If amplitude drops by half, the part was the spring and you can stop here or continue for margin.
3
Run the two-speed trialCut two 20 mm bands at 0.9 × and 1.1 × the current spindle speed, same feed and depth. Pick the quieter band, then step 5% at a time toward the best result. Avoid jumps above 15% in one move.
4
Reduce depth, then feedDrop depth of cut to 0.3 mm and feed to 0.08 mm/rev. If the marks fade, step back up in 0.1 mm and 0.01 mm/rev increments until they return, then hold one step below.
5
Index the insert and recheck the offsetFit a fresh edge, reset the tool offset, and inspect the corner radius. A worn edge is the most common reason a setup that ran clean yesterday rings today.

FAQs

Questions we get from the shop floor

Does a higher spindle speed always reduce low frequency vibrations in CNC turning?

No. Above a certain speed the process crosses into a second stability pocket, but between the pockets the amplitude gets worse. That is why we test 0.9 × and 1.1 × rather than jumping straight up.

Use the mark count per revolution as your guide. If the count stays the same when speed changes, the vibration comes from the structure, and more rpm will not fix it.

Why does the tower only chatter on composite material and not on aluminum?

Composites cut in bursts. The fiber bundles fracture, the force drops, then it spikes again several times per revolution. That pulsing input excites the part at its natural frequency.

Aluminum cuts with a steadier force, so the same setup can run clean. Lower the feed to 0.08–0.12 mm/rev and keep the depth light when you switch back to composite.

Can we hold ±0.005 mm on a long tower with a steady rest?

Yes, on a stiff setup with a supported span under 3 × diameter. The steady rest controls deflection; the machine and the tool offset control size.

On thin composite walls, thermal drift over a long run is often larger than the vibration error. Let the part reach room temperature before the final pass.

How do we tell a workholding problem from a tool problem?

Tap the part and the tool with the spindle stopped. The one that rings longer is the softer spring. Then confirm with a test cut: if the mark moves when you change the steady rest position, it is workholding.

If it moves when you change overhang or insert geometry, it is the tool side.

Is there a way to monitor chatter during the cut instead of after?

Listen and watch the load meter. A sharp rise in spindle load at constant feed means the edge is digging deeper, which is the start of a growing oscillation.

For repeat jobs, a simple accelerometer on the turret with a 200 Hz low-pass filter is enough to set a warning threshold.

What finishes can you hold once the vibration is under control?

On a stable turning setup we hold Ra 0.8–1.6 μm as a standard finish and Ra 0.2–0.8 μm where the geometry allows a wiper or a fine finishing pass.

Vibration control comes first. No insert geometry will produce a good surface on a part that is still moving in the cut.