CNC turning orbital motion: preventing chatter on slender workpieces through dynamic spindle control

Content Menu

● Introduction

● What Actually Happens During Regenerative Chatter on a Long Part

● How Continuous Spindle Speed Variation Breaks the Regeneration Loop

● Waveforms and When Each One Wins

● Practical Parameter Selection Without a PhD

● Machine Capability – What Actually Works in 2025

● Real Production Numbers People Are Getting Today

● Limitations and When It Won’t Help

● Combining With Other Tricks

● Conclusion

● Q&A

 

Introduction

Slender workpieces have always been a headache on CNC lathes. Anything with an L/D ratio above 10:1 starts flexing under cutting forces, and once the ratio hits 15:1 or 20:1 the part itself dominates the stiffness of the process. Hydraulic cylinder rods, aerospace actuator shafts, oilfield pump barrels, turbine rotor tie bolts – they all share the same problem: the stability limit drops so low that you either crawl along at tiny depths of cut or you accept chatter marks that scrap the part or kill fatigue life.

For years the standard answers were steady rests, follow rests, heavier machines, or just living with multiple light passes. All of those cost time and money. Starting around 2010, more shops discovered that continuously varying the spindle speed during the cut can kill regenerative chatter without any extra fixtures. The relative motion between tool and waveform becomes “orbital” instead of purely radial, and the regeneration mechanism falls apart. The result is often a 2–5× increase in stable metal removal rate and dramatically better surface finish on parts that used to scream.

This isn’t theory anymore. The technique is in daily production on DMG Mori NLX, Okuma LB and LU series, Doosan Puma, Mazak Integrex turn-mill centers, and even some older Hardinge and Citizen machines with upgraded drives.

What Actually Happens During Regenerative Chatter on a Long Part

When you turn a flexible shaft at constant rpm the tool leaves a slight wave on the surface. On the next revolution the insert cuts into both the intended chip thickness and the wave left 60/N seconds earlier. If the phase relationship lines up, vibration grows. On a slender part the lowest natural frequencies are usually the first two or three bending modes between 60 Hz and 250 Hz, sometimes lower on really long pieces.

Example: a 32 mm diameter × 720 mm exposed length 4340 shaft has a first bending mode around 98 Hz when freshly chucked. Cut a few millimeters off the diameter and the frequency drifts down to 85–90 Hz. Traditional stability lobe diagrams drawn for a fixed frequency are almost useless because the system is time-varying.

Another common case is titanium landing gear components, 25–40 mm diameter with 500–800 mm sticking out. The combination of low stiffness and low material damping means chatter starts at depths as small as 0.4–0.6 mm unless speed is kept ridiculously low.

How Continuous Spindle Speed Variation Breaks the Regeneration Loop

Varying rpm continuously changes the time delay between successive passes. Instead of a constant τ = 60/N, the delay is now τ(t). The phase between the inner and outer modulation never settles, so energy can’t build up in one mode.

The motion of the contact point is no longer purely radial. A ±15 % speed variation at 400 rpm moves the tool circumferentially by 5–8 degrees per revolution relative to a fixed-speed wave. Combined with the radial deflection of the workpiece you get a small elliptical or orbital path – hence the term “orbital motion” you see in some papers and controller marketing literature.

Waveforms and When Each One Wins

Sinusoidal Variation

Easiest on the spindle drive, lowest acceleration. Most controllers (Siemens 840D ShopTurn, Fanuc 31i, Heidenhain iTNC 640) already have a built-in function or can run it from a simple cycle.

Typical parameters for slender shafts: Mean rpm: same as you would use constant RVA (relative variation amplitude): 0.10–0.20 Modulation frequency: 0.8–1.2 × measured chatter frequency (or workpiece natural frequency)

Shop floor result on a 38 mm × 910 mm Inconel 718 stem: constant speed stable depth 0.7 mm at 320 rpm. Sinusoidal RVA 0.18, modulation 105 Hz → stable 2.3 mm depth, Ra dropped from 3.2 μm to 0.9 μm.

Triangular / Sawtooth Variation

Higher peak acceleration but often more effective when the chatter frequency is low (<100 Hz) and the spindle can handle the torque swings. Many operators prefer it on heavy roughing cuts because the chip breaks more consistently.

Real example from a gearbox shaft shop: 115 mm × 1950 mm 18CrNiMo7-6 planet carrier shafts. Triangular variation with RVA 0.22 and 11 Hz modulation frequency let them rough at 3.8 mm depth instead of 1.1 mm.

Square-Wave or On-Off Burst Variation

Alternate 10–30 revolutions at N₀ + ΔN and N₀ – ΔN. Hard on bearings, but extremely effective when thermal issues or chip wrapping limit continuous variation. Used a lot on superalloy roughing where you want maximum disruption for short bursts.

Practical Parameter Selection Without a PhD

1. Let the part chatter once at your normal parameters – just for 5–10 seconds.
2. Record the sound with a phone or a cheap USB microphone.
3. Look at the FFT – the tallest peak between 60–300 Hz is almost always the culprit.
4. Set modulation frequency to that peak or ½ of it.
5. Start RVA at 0.10 and creep up until the noise disappears or you hit 80–90 % of drive current limit.

If the machine has an accelerometer input (most new DMG Mori and Okuma do), wire a $120 PCB sensor to the analog input and let the control ramp RVA automatically – zero operator guessing.

Machine Capability – What Actually Works in 2025

Modern built-in-motor spindles (12,000–18,000 rpm lathes) follow ±25 % variation up to 15–20 Hz modulation without breaking a sweat. Older belt or gear-head machines top out around ±10 % below 600 rpm.

Controllers that make life easy: – Siemens Sinumerik ONE / 840D: G96 Sxxx VAR=15 FREQ=120 (native command) – Fanuc 31i/32i: Custom macro or AICC+HSV option – Heidenhain TNC7 / 640: Python OEM cycle or OSC-VAR function – Okuma OSP-P500: Variable Spindle Speed Control user task

Even a 2012-vintage Mori-Seiki NLX with DCG and driven tool can do it if you write a 40-line macro.

Real Production Numbers People Are Getting Today

– Aerospace actuator housing tie rods, Ti-6-4, 22 mm × 680 mm: cycle time down 52 %, one-pass roughing instead of three + follow rest. – Oil country tubular goods crossover subs, 4145H, 73 mm × 1400 mm between centers: metal removal rate from 220 cm³/min to 680 cm³/min. – Medical knee implant femoral stems, CoCr, tapered 16→12 mm over 420 mm: eliminated all steady-rest marks, scrapped parts for chatter went from 8 % to zero. – Wind turbine main shaft, 280 mm × 3200 mm, 34CrNiMo6: roughing time per shaft dropped from 19 hours to 7.5 hours using low-frequency triangular variation.

Limitations and When It Won’t Help

– Very high-speed aluminum or brass finishing (>2500 rpm) – lobes are already wide. – Extremely brittle materials (cast iron, ceramics) where process damping already kills chatter. – Machines with weak spindle drives (old gear heads or small servo motors). – Parts that are so slender the static deflection under cutting force alone exceeds allowable tolerance – you still need support.

Combining With Other Tricks

Best results often come from stacking methods: – Fill hollow shafts with sand or lead shot + SSV – Active magnetic guides on ultra-long pump barrels + SSV – Tuned viscoelastic dampers bonded inside bore + SSV

Each one alone helps; together they can push stable depth 8–10× higher than constant speed.

Conclusion

Continuous spindle speed variation – especially the orbital-type motion you get with well-chosen amplitude and frequency – has moved from university labs to standard practice on tough slender turning jobs. Shops that have adopted it are cutting cycle times in half, eliminating secondary grinding on many parts, and taking jobs that used to be money-losers and turning them profitable.

The hardware is already in most machines built after 2015. The know-how is just measuring the chatter frequency once and turning a couple of knobs (or writing a 30-line macro). If you’re still fighting long-shaft chatter the old way in 2025, you’re leaving serious money and capacity on the table.

Q&A

Q1: How much does varying spindle speed heat up the bearings? A1: On modern liquid-cooled spindles, continuous ±20 % variation adds 3–6 °C after hours of running. Well within limits.

Q2: Does the varying surface speed mess up insert life on cermet or ceramic tools? A2: Usually improves life because peak force drops when regeneration is killed. Worst case is 5–10 % shorter on very sticky stainless.

Q3: Can I use it on Swiss-type sliding-head lathes for tiny long pins? A3: Yes and it works brilliantly – the guide bushing gives extra damping and the variation kills the 400–800 Hz modes common on 3–8 mm pins.

Q4: What about thread turning or grooving on long parts? A4: Works fine as long as you lock the variation during the actual thread pass (most controls let you switch it on/off mid-program).

Q5: Is there a simple retrofit for older machines? A5: Add a high-response servo overlay drive (Rexroth IndraDrive, Siemens Sinamics S120) and feed it a 0–10 V signal from a small PLC running the sine wave – under $15 k and transforms the machine.