Chatter: how to recognise it and how to stop it

Chatter is the loud, harsh vibration that leaves lines or grooves at regular intervals on the part and can damage the tool. Ertürk, Özgüven and Budak call it a self-excited vibration met while parts are machined, and Stépán describes regenerative chatter as “a kind of self-excited vibration”, warning against the habit of identifying it as a resonance. And once it has started, as the glossary to Smith’s column in Cutting Tool Engineering puts it, it “is often self-sustaining until the problem is corrected.” Knowing what feeds it is what turns the usual trial and error into a few deliberate changes.

How chatter feeds itself

A cut is never perfectly steady. Stépán describes what happens after a disturbance in turning: “the tool starts a damped oscillation relative to the workpiece, and the surface of the workpiece becomes wavy.” One revolution later the tool reaches that wave again, so the chip thickness at the tool varies, with a delay equal to the time of one revolution. Milling does the same tooth by tooth: “The vibrating tool causes the teeth to leave behind a wavy surface,” Smith writes, and this regeneration of waviness is the main mechanism of chatter in milling. Ertürk and co-workers name it the regeneration of chip thickness, the main cause of chatter, described by Tobias and Fishwick some fifty years before their 2006 paper. Since the cutting force follows the chip thickness (cutting forces), a varying chip is a varying force on a tool that can already vibrate.

How much the chip varies depends on the phase between the two waves: the one the previous pass left (the outer side of the chip) and the one the tool cuts now (the inner side). Liao and Young put it as the time delay between two consecutive cuts causing a phase difference. In Smith’s words, “when the wave left on the surface is exactly aligned with the tool motion, the chip thickness appears as if there was no vibration.”

• previous pass (outer wave)
• this pass (inner wave)
• layer to come off
• chip thickness h at the edge

❚❚Phase ε between the two waves

90

°

Swing of the chip thickness± 1.41 × A

The phase is set by the spindle speed. Over one revolution the vibration completes some whole number of waves N plus a fraction ε/2π of a wave; ε is the phase between the present vibration and the one the previous revolution left. Tyler and Schmitz write it as fc/Ω = N + ε/2π, with fc the chatter frequency and Ω the spindle speed in revolutions per second. Change the speed and the fraction changes, which is why the same tool can chatter at one speed and cut cleanly at a slightly different one.

Is it chatter, or forced vibration?

Not every vibration is chatter. Forced vibrations, Smith explains, are “directly driven by the rotating spindle”: runout gives a force once per revolution, and each tooth that enters the cut gives a force at the tooth-passing frequency, “ft = n × m ÷ 60, where n is the spindle speed in rpm, m is the number of teeth on the tool.” They change when the speed changes. Chatter does not follow that family. Its frequency “is close to, but not equal to, one of the natural frequencies” of the machine, tool or part and “is not connected to the tooth passing frequency” (Smith). Schmitz gives the measurement version: chatter is identified “when content appears at frequencies other than the tooth passing frequency” and its multiples (harmonics).

The difference can be heard. “Stable, chatter-free milling produces a clear tone—a single, dominant frequency,” Smith writes; chatter “produces a harsh tone, often with different frequencies mixed together.” On the part, the spacing of the chatter marks depends on the frequency of the vibration.

15,000 rpm (chattering)By the rule: 12,810 rpm

Spindle speed n

15,000 rpm

Tooth-passing frequency ft = n × m ÷ 60 (m = 2 teeth)

500 Hz

Chatter frequency fc

427 Hz

New speed = fc × 60 ÷ m

427 × 60 ÷ 2 = 12,810 rpm

When the milling machine has a speed range to spare, Smith’s rule gives a direct next step: “choose a new spindle speed so the new tooth passing frequency matches the old chatter frequency,” then “adjust the feed to keep the chip load constant” (Smith). Smith calls the result, 12,810 rpm in his example, the new speed to try; Schmitz uses the same step, Ω = 60·fc/Nt, to choose the next test speed when mapping a tool’s stability. The rule is written for milling, with teeth; in turning there is one pass per revolution, and the lobe diagram below shows where the stable speeds lie.

The stability lobe diagram

The rule works because stable speeds come in islands. A stability lobe diagram maps them: in Schmitz’s words it “separates combinations of spindle speed and axial depth that produce chatter” from those that do not, and lets stable parameters be chosen “without trial and error”. Below the line is chatter-free, above it chatter (Khasawneh et al.). Stépán describes the stable and unstable domains as alternating with cutting speed, with “greater domains of stability” at higher speeds. Smith locates the largest stable zone where the tooth-passing frequency equals the natural frequency; “The peaks all occur at integer fractions of the best speed.” Davies and co-workers at NIST found the same: the most stable speeds lie approximately at integer fractions of the natural frequencies of the most flexible modes.

For turning, Tyler and Schmitz give the boundary as

$$b_{lim} = \frac{-1}{2\,K_s\,\mathrm{Re}[G]}, \qquad \frac{f_c}{\Omega} = N + \frac{\varepsilon}{2\pi}, \qquad \varepsilon = 2\pi - 2\tan^{-1}\frac{\mathrm{Re}[G]}{\mathrm{Im}[G]}$$

where G is the oriented frequency response function (FRF) of the tool point, Ks the specific cutting force, b the chip width and N = 0, 1, 2 … the lobe number, the whole waves per revolution. Only frequencies where the real part of G is negative give a limit.

Spindle speed n

3,000

rpmDepth of cut

1.20

mmGo to pocket

chatter

Waves per revolution (fc · 60 / n)

—

Limit at this speed

—

Nearest stable pocket

—

The FRF comes from a tap test: “an instrumented hammer is used to excite a structure and measure the vibration response” (Schmitz), and the purpose of the test is to identify the FRF “for the selected mechanical structure”. Schmitz and co-workers at NIST measure it for a particular tool, holder and spindle combination, and Atlar notes that the stability diagram has to be obtained for the spindle, holder and tool together. Two cautions from Schmitz: a speed and depth right at the boundary would most likely not be chosen, because some uncertainty remains, and at low spindle speeds a cut can become stable again through process damping, which “is influenced by the cutting edge design.”

Turning and boring: what to try, in order

The makers’ advice for turning and boring, in a practical order: the setup first, the speed last.

For the bar itself, its overhang limits and its deflection, see internal turning.

Milling: what to try, in order

The direction of cut and the forces it produces are explained in climb vs conventional milling.

Where the advice splits

Three remedies come with opposite advice from different makers, sometimes from the same one.

• Number of teeth. Kennametal’s Tech Tip 32 says “Reducing the number of flutes”, and Mitsubishi, for thin-walled parts or low rigidity, recommends an end mill with a small number of edges to prevent vibration; Kennametal’s Tech Tip 12 says “Select an end mill with more flutes.”
• Feed. Kennametal’s Tech Tip 12: “Increase the feed, or chipload per tooth/tool”; BIG DAISHOWA warns that cutting the feed too far for a better finish “runs the risk of chattering”. Kennametal’s Tech Tip 32 lists a smaller chip load instead, with the caution that it can harm the cutter, and Muratec writes that a lower feed reduces the cutting force.
• Speed. Mitsubishi: “Decrease the cutting speed”; BIG DAISHOWA: reduce the surface speed when it is excessive. Okuma: “our first inclination is to slow the RPM’s and this often makes the chatter worse,” and “Increasing the RPM’s may be the correct action.” Haas names running “too fast or too slow” among the causes.

Let the control vary the speed

Several controls can vary the spindle speed, or tell the operator which speed to use.

• Haas, spindle speed variation (SSV). Standard on all Haas CNC turning centres and on toolroom lathes with software 6.04 or later, it “allows the operator to specify a range within which the spindle speed will continuously vary”, suppressing chatter “by not allowing vibration to begin as a result of constantly changing frequency”. M38 turns it on and M39 off; Setting 165 sets how far the speed swings above and below the commanded value and Setting 166 the cycle. With 50 and 30 (3 s), an S1000 “will swing the spindle speed between 950 and 1050 RPM”. The spindle is not allowed below 10 rpm, and M30 or RESET turns SSV off.
• Okuma, Machining Navi. An option on the OSP control that listens through acoustic sensors in the spindle cartridge; “When chatter is detected, the function then either recommends spindle speed changes” (the L-g and M-g versions) or, in the M-i version, makes them automatically. Okuma offers it for both turning and milling machines.
• DMG MORI, CELOS Chatter Control. An acceleration sensor in the spindle detects the vibration and the software calculates a spindle speed to suppress it, based on the vibration frequency.

Writing about spindle speed variation in turning, Zelinski sums up the idea: the process “spends only an instant at any problematic speed.”