CNC Milling vs CNC Turning: How to Choose the Process

Introduction

Milling and turning solve the same problem from opposite directions. Turning spins the workpiece against a stationary single-point tool; milling spins the tool against a stationary workpiece. That single kinematic difference propagates into everything that follows: which geometries are cheap, how forces behave, what concentricity is achievable, and how many setups a part needs.

Choosing wrongly is expensive but rarely obvious from the model. A shaft with a keyway and a cross-hole is fundamentally a turned part with two milling operations, yet it is often quoted as a milled part because the CAD geometry looks prismatic in a bounding box. Conversely, a flanged housing with a bored bore and a face-seal groove is frequently milled through three setups when a single mill-turn operation would deliver better concentricity in less time.

This chapter compares the two processes on the criteria that actually decide the outcome — symmetry, concentricity, tool access, force direction, cycle time and setup count — and gives a decision sequence you can apply to a model before the RFQ goes out. It also covers where mill-turn and Swiss-type machines change the answer.

Technical Explanation

The kinematic difference and what it implies

In turning, cutting speed at the tool point depends on the instantaneous workpiece diameter. Facing to centre therefore requires either constant surface speed control, where the spindle accelerates as the tool approaches the axis, or acceptance of falling surface speed and degraded finish near the centre. In milling, the tool diameter is fixed, so surface speed is constant for a given spindle speed regardless of where the cut occurs.

Turning produces a continuous chip and a steady, largely unidirectional force. Milling loads each tooth cyclically, generating a forcing frequency equal to spindle speed multiplied by flute count. This is the root of chatter in milling and the reason stability lobe diagrams exist for milling but rarely for turning.

Concentricity, roundness and why turning wins on shafts

Any diameter turned in a single chucking shares the spindle axis. Concentricity between two such diameters is limited by spindle runout, typically 0.002-0.010 mm, not by programmed positioning. Producing the same relationship by milling requires either a rotary axis or circular interpolation, and the result inherits both axis squareness error and interpolation error. For anything with coaxial diameters, tapers, threads or face grooves, turning is the more capable and the faster route.

Roundness follows the same logic. A turned diameter is round because the workpiece rotates about a fixed axis. A circularly interpolated bore is only as round as the machine’s servo matching and reversal behaviour at the quadrant transitions, where backlash shows up as characteristic lobing.

Where milling is the only sensible answer

Milling owns everything non-rotational: pockets, planar faces at arbitrary angles, bolt patterns off the centreline, complex 3D contours, and any feature requiring tool access from multiple directions. Milling also handles asymmetric mass distribution without balancing problems, whereas an off-centre turned part must be counterweighted or run at reduced rpm.

Mill-turn and Swiss-type machines

A turning centre with driven tooling and a C axis can index or continuously rotate the spindle while a powered tool cuts. Adding a Y axis allows off-centre milling. A sub-spindle transfers the part for back-face work. The practical effect is complete parts off one machine, with the concentricity of turning and the feature freedom of milling. Swiss-type lathes with a guide bushing extend this to slender parts, supporting the bar within a few millimetres of the cut so that length-to-diameter ratios above 20:1 remain stable.

A decision sequence

Establish whether the majority of material removal is rotationally symmetric about one axis. If yes, the part is a turned part with secondary milling.
Check for coaxial tolerance callouts: concentricity, runout, or position of one diameter to another. These push strongly to turning.
Count the distinct tool-access directions needed for the remaining features.
If one or two directions suffice and driven tooling can reach them, quote mill-turn.
If features need three or more arbitrary orientations, or the part has no dominant axis, quote milling and count setups.
Compare estimated cycle time and setup count for both routes, including the tolerance stack introduced by each additional setup.
Confirm stock form: bar stock favours turning; plate, block or castings favour milling.

Engineering Parameters

Both processes share the surface speed relationship, but the diameter term means different things:

Turning:  v_c = π × D_workpiece × n / 1000   (m/min)
          MRR = v_c × a_p × f                (cm³/min; a_p mm, f mm/rev)
Milling:  v_c = π × D_tool × n / 1000        (m/min)
          v_f = f_z × z × n                  (mm/min)
          MRR = a_p × a_e × v_f              (mm³/min)

Criterion	CNC Turning	CNC Milling	Notes / Driver
Concentricity, one chucking	0.005-0.020 mm	0.02-0.05 mm	Turning limited by spindle runout only
Roundness, finish pass	0.002-0.010 mm	0.010-0.030 mm	Interpolated bores show quadrant lobing
Achievable Ra, finish pass	0.4-1.6 µm	0.8-3.2 µm	Turning benefits from continuous cut
Typical diameter tolerance	IT6-IT8	IT7-IT9	Boring closes the milling gap at added cost
Feed convention	0.05-0.5 mm/rev	0.02-0.25 mm/tooth	Different units, frequently confused
Typical depth of cut, roughing	2-6 mm	1-4 mm axial	Milling limited by tooth engagement shock
Length-to-diameter limit	3:1 unsupported, 20:1+ Swiss	4:1 tool L/D easy	Tailstock or bushing extends turning reach
Setups for a typical housing	1-2 (mill-turn)	2-4	Setup count usually dominates unit cost

These figures assume a rigid production machine and sound workholding. They degrade quickly with slender parts, long tool overhang or thin-wall geometry. Material also shifts the balance: in Ti-6Al-4V the low permissible cutting speed of 40-80 m/min makes continuous turning far more productive than interrupted milling, while in 6061-T6 running 300-1,000 m/min the difference narrows because both processes become limited by spindle power and chip evacuation rather than by tool life.

Manufacturing Considerations

Stock form drives the decision as much as geometry. Bar stock feeds automatically through a lathe spindle, enabling unattended running. Block stock must be loaded individually into a mill.
Bar-fed turning is the cheapest high-volume route for small parts. A Swiss-type machine with a bar feeder can run lights-out at cycle times of seconds per part.
Milling a rotational feature costs tolerance, not just time. Interpolated bores need a boring pass to reach H7; a turned bore reaches it directly.
Driven tooling has power limits. Live tool spindles typically deliver 2-5 kW against 15-22 kW on a machining centre, so heavy milling on a lathe is slow.
Deburring differs. Turned parts collect burrs at cross-hole intersections and thread runouts; milled parts collect them at pocket edges and face transitions. Budget for it explicitly.
Inspection method follows process. Turned features are checked on a shaft measuring system or with air gauges; milled features generally need a CMM.
Volume break-point. Below about 100 pieces, whichever route needs fewer setups usually wins. Above 1,000, cycle time and unattended running dominate, favouring bar-fed turning where geometry allows.

Common Mistakes

Mistake: Quoting a shaft as a milled part because the CAD bounding box is prismatic. — Why it fails: Milling a diameter by interpolation is slower and less round than turning it. — Fix: Identify the dominant axis of symmetry before selecting the process.
Mistake: Applying concentricity callouts across features that will be cut in separate setups. — Why it fails: Each re-chucking adds relocation error that can exceed the callout. — Fix: Group coaxial features so they are produced in one chucking, or relax the callout.
Mistake: Confusing feed per revolution with feed per tooth when moving a program between a lathe and a mill. — Why it fails: The values differ by roughly the flute count, so the feed can be several times too high. — Fix: Always state units explicitly in the tooling sheet.
Mistake: Designing a deep small bore on a lathe part without checking boring bar reach. — Why it fails: A boring bar longer than about 4 diameters chatters unless it is carbide or damped. — Fix: Keep bore depth under 4× bar diameter or allow for a tuned bar.
Mistake: Adding a single off-axis flat to an otherwise turned part and treating it as free. — Why it fails: Without driven tooling the part transfers to a mill, adding a full setup for one feature. — Fix: Confirm the supplier has C-axis live tooling, or move the flat to a turnable feature.
Mistake: Turning an unbalanced part at full rpm. — Why it fails: Centrifugal load from an off-centre mass causes vibration, poor finish and chuck wear. — Fix: Limit rpm, counterweight the fixture, or mill the part instead.
Mistake: Specifying a face groove on a milled housing. — Why it fails: Milling a narrow annular groove requires a small cutter and many passes; turning cuts it in one. — Fix: Route parts with face seal grooves through a turning or mill-turn operation.

References

ISO 286-1 — Geometrical product specifications, ISO code system for tolerances on linear sizes — defines the IT grades referenced when comparing achievable diameter tolerances.
ASM Handbook, Volume 16: Machining, ASM International — process chapters covering turning, boring and milling operations and their capability. https://www.asminternational.org
Fundamentals of Machining and Machine Tools, Boothroyd and Knight — analytical treatment of cutting forces, machine dynamics and process economics.
Machinery’s Handbook, Industrial Press — feed and speed tables in both feed-per-revolution and feed-per-tooth conventions.
ISO 3685 — Tool-life testing with single-point turning tools — the reference method behind published turning tool life data.
Manufacturing Engineering and Technology, Kalpakjian and Schmid — comparative process capability data and selection methodology.