Treatment of Vibrations and the Punch in CNC Turning

• 1
  What treatment of vibrations and the punch means at the spindle
• 2
  Overhang, stiffness and why ratio 3 is the line
• 3
  Lead angle, rake and nose radius in the cut
• 4
  Speed, feed and depth: where the resonance sits
• 5
  How to tell whether the fix actually worked
• 6
  Which fix to try first, by symptom

Every turning cut puts a spring between the tool and the part. The tool tip pushes in, the material pushes back, and the loop settles at some level of deflection. Treatment of vibrations and the punch is about managing that loop. If the loop returns to the same deflection after each revolution, the surface is even. If it overshoots, the next tooth or the next revolution bites deeper, and the depth of cut starts to swing.

Two things drive the swing. Regeneration is one: a wavy surface left by the previous pass changes the chip thickness on the next pass, which changes the force, which deepens the wave. Mode coupling is the other: the tool and the workpiece move in two directions at once, usually radial and tangential, so the cut pulls the tip into the part instead of pushing it away. Both feed each other.

The frequency tells you which one you have. Regenerative chatter locks onto a multiple of the spindle speed, so the pitch of the marks changes when you change rpm. Mode coupling tends to sit at the natural frequency of the boring bar, the toolholder or the part, and it barely moves when rpm changes. Read the pitch before touching a dial.

The punch is the second half of the story. It is the moment the tool enters the material, and it is a step load, not a smooth ramp. On a thin wall or a long overhang the punch rings the structure, and that ring is the seed for everything that follows. Damped entry, a lead angle or a chamfer on the blank all reduce it.

The single most useful number in turning setup is the ratio of overhang to diameter. A boring bar or a slender shaft clamped at one end behaves like a cantilever, and deflection grows with the cube of the length. Push the length from 2× diameter to 4× diameter and you lose roughly eight times the stiffness. That is why a bar that cuts cleanly at 2× howls at 4×.

In practice, keep the cutting point close to the clamping point. If the ratio crosses about 3, expect trouble in steel and stainless, and consider a different process: a shorter bar, a larger shank, or a mill-turn setup where the part is held closer to the cut. Turning a 4,000 mm shaft between centres is a different problem again, and it needs a steady or a follow rest rather than more speed.

Workholding matters as much as the bar. A three-jaw chuck gripping a thin ring lets the wall breathe; a bored soft jaw that wraps most of the circumference does not. For a box-shaped or welded sheet-metal part, the wall is the weak spring, not the tool. In those cases no tool change will help until the fixture supports the wall.

Machine condition belongs in the same conversation. A worn spindle bearing, a loose turret index or a toolholder with an uneven seat all add compliance you cannot see in a simulation. Check the obvious first: is the insert clamped properly, is the holder seated clean, is the bar extended further than the setup sheet says.

Tool geometry decides how much of the force goes sideways. A 90° lead angle pushes the load almost straight back into the bar. Drop to a 45° or 60° lead and part of that load goes into the part instead, which often quiets a slender setup. The trade is axial force on the face, so watch the part for push-off if it is only held by a light clamp.

Rake angles change the cutting force and the heat split. A positive rake lowers the force on the tip, which helps on aluminium and on light machines. A negative rake is stronger and survives interrupted cuts, but it pushes harder and asks more of the setup. On stainless and titanium, a sharp positive edge with a proper coating usually beats brute force.

Nose radius is a balancing act. A large radius spreads the load and gives a better finish at high feed, but it also increases the radial force, and radial force is what bends a boring bar. On a thin wall, a smaller radius with a slightly higher feed often holds size better than a big radius run slowly.

The relief angle must clear the material. If the flank rubs instead of cutting, the tip generates heat and a low-frequency groan that sounds like chatter but is not. Check the flank wear land at 10× before you blame the machine. A worn land of 0.2 mm on a finishing insert is enough to spoil a Ra 0.8–1.6 μm target.

Stability is not a single number. Plot depth of cut against spindle speed and you get lobes: stable pockets separated by unstable bands. The pockets repeat at multiples of the natural frequency, so moving speed 10–15% often lands you in a quieter pocket without touching anything else. Do that before you reduce the depth of cut, because a smaller bite costs cycle time.

Feed per revolution changes the chip thickness, and chip thickness is the link in the regenerative loop. Too light a feed lets the tool rub and can actually make chatter worse, because the edge cannot get under the work. Turning a finishing pass at 0.05 mm/rev on a slender bar is a common mistake; 0.10–0.15 mm/rev often cuts smoother.

Depth of cut has a hard ceiling. Below the stability limit the cut is quiet at any speed; above it, no speed helps for long. If a roughing pass at 2 mm radial depth is unstable everywhere, take two passes at 1 mm and accept the extra time. That is usually cheaper than scrapping the part.

Coolant and chip evacuation feed back into the loop. A chip recut on the flank adds a random force pulse, which can kick a marginal setup into chatter. Through-tool coolant aimed at the contact point, or a higher-pressure stream, keeps the chip moving and steadies the cut on deep bores.

Listen, then measure. A phone spectrum app or a simple accelerometer on the turret will show a peak at the chatter frequency. If the peak drops by 10 dB or more after a change, the fix is real. If it only shifts in frequency, you have moved the problem, not solved it.

Check the surface with a profilometer, not just a fingernail. Chatter shows as a periodic waveform in the trace, and the wavelength should match the mark pitch you measured. A Ra reading alone can look acceptable while the waviness is out of tolerance, so ask for both Ra and the waviness figure when the part is functional.

Inspect size across several parts, not one. Chatter often shows as scatter in diameter rather than a single out-of-tolerance part. Run five pieces at the new setting and check the spread. If the spread tightens, the loop is under control.

Finally, log what you changed. Speed, feed, depth, overhang, insert grade. The next job with a similar ratio will start from a known point instead of a guess. That log is worth more than any single fix.