Machining chatter
Diagnosing unstable cutting without confusing chatter, forced vibration, tool marks or a universal speed correction.
Tags: production, machining, vibration
On this page
In the worked trial, an evenly spaced four-flute cutter at 6,000 rpm gives 400 Hz tooth passing and 1,200 mm/min feed at 0.05 mm/tooth. At 5,400 rpm, the same nominal feed per tooth needs 1,080 mm/min, and tooth passing falls to 360 Hz. Holding feed at 1,200 gives 0.05556 mm/tooth instead. A peak at tooth-passing frequency can be forced excitation. It cannot establish chatter or a stable speed.
Chatter is self-excited vibration. Altintas and coauthors distinguish regenerative and mode-coupling mechanisms in their November 2020 stability paper. Altintas and coauthors, DOI 10.1115/1.4047391
Understand the feedback before changing settings
In regenerative chatter, motion leaves waviness on a cut surface. A later cutting edge or revolution encounters that surface, changing chip thickness and cutting force. The resulting force can reinforce the motion, depending on the phase relationship and system dynamics. Stability therefore depends on more than static rigidity or spindle speed alone. The cited paper discusses structural dynamics, cutting-force coefficients, process damping and engagement as inputs to stability prediction. Chatter stability paper
The dynamic system includes the tool, holder, spindle, machine structure, fixture and workpiece together. Sandvik Coromant's milling guidance explicitly lists these possible sources of vibration limitation. A thin wall may become more flexible as stock is removed; a long tool may be limiting in another operation. The same nominal cutter and speed need not behave identically in both setups. Sandvik Coromant, reducing milling vibration
Diagnostic evidence and its limits
Chatter diagnosis depends on the material, tool geometry and wear, holder, overhang, workholding, speed, feed and engagement. Its onset can also depend on position: corner entry, increasing depth and a thinning wall change the cutting or structural conditions. Comparing several simultaneous changes cannot isolate which change affected stability. This is explanatory synthesis of the setup and engagement dependencies described by the sources below.
Sound can indicate an onset, but it is not a dimensional acceptance test. A spectral peak also needs context: sensor position, sampling and the cutting interval affect its interpretation. Periodic tooth excitation, tool runout and damaged edges can produce repeating effects without establishing a regenerative instability. Surface appearance, tool condition and vibration measurements provide different pieces of evidence.
Physical setup and dynamic stiffness
Haas's TG0100 Mill Chatter guide, last updated October 14, 2025, lists tool wear, excessive tool/holder length, unsuitable chip load, flute engagement, toolpath, coolant, drawbar hold force and machine foundation as possible contributors. Its recommended actions are machine/application guidance, not permission to alter spindle hardware without the applicable service instructions. Haas TG0100
Tool seating, clamping, support and overhang influence the installed assembly. Shorter assemblies can improve stability, subject to access and collision clearance, but shortening the tool may have limited benefit when the workpiece or fixture dominates the flexibility. A damaged holder or inadequate retention force is a maintenance condition; a cutting-data adjustment does not establish that the underlying equipment condition is acceptable.
For boring, Sandvik's Silent Tools turning guide recommends minimizing overhang and using the largest suitable bar diameter. It distinguishes ordinary, damped and carbide-reinforced solutions for different reach demands. These are product/application choices; a stated overhang range is not a guarantee for every bore, material and clamping arrangement. Sandvik Silent Tools turning guide
Engagement and cutting conditions
Haas notes that corner entry can increase tool engagement and load, and discusses reducing engagement or using a more consistent engagement toolpath. It also warns that an excessively light chip load can be a contributor. Thus “reduce feed whenever chatter appears” is not a universal corrective action. The relevant cutting-data range depends on tooling, material and local path geometry. Haas milling troubleshooting
Sandvik's Silent Tools milling guidance discusses cutter pitch, number of engaged inserts and differentially pitched cutters as ways to influence vibration behavior. Damped assemblies and unequal pitch address particular dynamic conditions; they do not eliminate the need to validate the cut. Cutter choice, engagement and the installed assembly belong in the same assessment. Sandvik Silent Tools milling guide
A stability-lobe calculation can support selection of speed and cutting depth for a characterized system. Its applicability depends on the modeled setup and engagement. A diagram for another holder, overhang or workpiece support arrangement is not a universal machine rating. Increasing or decreasing speed can change stability; a one-direction speed rule cannot replace that assessment.
Original bounded frequency and feed example
Assume an invented evenly spaced four-flute cutter runs at 6000 rpm with feed per tooth 0.05 mm/tooth. The spindle rotational frequency is 6000/60 = 100 Hz. Tooth-passing frequency is 4 × 100 = 400 Hz. Programmed feed is 6000 × 4 × 0.05 = 1200 mm/min.
If a controlled trial changes speed to 5400 rpm, maintaining the same nominal feed per tooth requires 1080 mm/min, and tooth-passing frequency becomes 360 Hz. Leaving feed at 1200 mm/min instead gives 1200/(5400 × 4) = 0.05556 mm/tooth, approximately 11.1% higher. A speed-only override has therefore also changed chip load in that scenario.
These are kinematic calculations, not recommended settings. A measured 400 Hz peak at 6000 rpm could be related to periodic tooth excitation; its presence alone does not prove chatter. Nor does choosing 5400 rpm demonstrate a stable lobe. The example omits radial chip-thinning effects, unequal pitch, runout, interrupted contact and actual dynamic measurements.
Stability and finished-part quality
A meaningful comparison of cutting trials depends on comparable geometry, engagement, tool condition and support. Finished-part dimensions, form, texture and application-specific surface integrity are distinct from acoustic behavior. A quieter cut can still leave unacceptable waviness or dimensional error. Conversely, reducing removal rate may improve one symptom while making the process uneconomic or otherwise unsuitable.
Stable behavior applies to the tested tool assembly, cutting conditions, path, material and support arrangement. Changes in reach, clamping or remaining wall geometry can alter that behavior, so a successful cut does not establish stability for every subsequent configuration. The manufacturer pages above are undated except for Haas's stated update. The paper's November 2020 issue date is printed in its PDF; sources were reviewed on 2026-11-02 PST.