Last month I quoted a stainless steel shaft — 8mm diameter, 120mm long, with threads on both ends and a bearing journal in the middle. The customer asked for milling. I told them: this part wants to be turned. If you mill it, you'll pay for 45 minutes of machine time and fight runout the whole way. If you turn it, it's 12 minutes and the concentricity is guaranteed by the process.
CNC turning is the older, less flashy sibling of CNC milling. Milling gets the attention — 5-axis videos, complex organic shapes, toolpath porn. Turning just spins the part and cuts it with a stationary tool. But for round parts, nothing beats it on speed, accuracy, or cost.
I run both mills and lathes in our shop. Here's what I've learned about which one to use, and when turning — especially Swiss-type turning — is the right answer.
Turning vs milling: the geometry decision
The fundamental difference is simple but the implications run deep:
In milling, the tool spins and the part stays still (or moves linearly). In turning, the part spins and the tool stays still (or moves in X and Z).
This one difference determines everything about the process capabilities:
| Capability | CNC Turning | CNC Milling |
|---|---|---|
| Natural geometry | Cylindrical, axisymmetric | Prismatic, complex 3D |
| Concentricity (coaxial features) | Process guarantee — all OD features share the same axis of rotation | Requires multiple setups, indicator time, and luck |
| Surface finish on OD | Ra 0.4 μm achievable in one pass | Ra 0.8 μm typical, requires extra finishing for 0.4 |
| Material utilization (from bar) | Near-net, minimal waste | Often 50-80% material removal from billet |
| Cycle time (round parts) | Fast — one chucking, all OD features | Slow — multiple setups, longer tool engagement |
| Internal features (bores) | Easy — drill, bore, ream in same setup | Possible but requires tool changes and careful approach |
| Non-round external features | Requires live tooling (mill-turn) | Native capability |
| Cross-holes, side features | Requires live tooling or second op | Native capability |
| Minimum diameter (production) | ~1mm (Swiss), ~6mm (conventional) | ~0.5mm with micro-tooling |
| Setup time | Low for bar-fed work | Higher for multi-setup parts |
For a simple shaft with threads and a bearing seat, the lathe wins on every metric. For a rectangular housing with pockets and holes on five faces, the mill wins. The question is whether your part is fundamentally round or fundamentally prismatic. If the answer is "both" — a round part with milled features — that's where mill-turn machines and live tooling come in.
Swiss-type lathes: the machine that changed medical and watch parts
The Swiss-type (sliding headstock) lathe is a different animal from a conventional lathe. On a conventional lathe, the bar stock is clamped in a chuck, the chuck spins, and the cutting tool moves along the stationary bar to cut features. On a Swiss lathe, the bar stock slides through a guide bushing while the cutting tools stay fixed at the bushing face. The bar moves in Z, not the tool.
Why does this matter? Because on a conventional lathe, an 8mm diameter shaft that sticks out 120mm from the chuck will deflect under cutting pressure. The tool pushes on the part, the part bends slightly, and the cut is shallower than programmed. The longer and thinner the part, the worse this gets.
On a Swiss machine, the cutting happens within 1-2mm of the guide bushing, so the tool is always cutting right next to where the bar is supported. The unsupported length is essentially zero. This means you can turn 1mm diameter pins 50mm long without deflection, or 0.3mm diameter needles for medical devices.
The practical differences:
| Feature | Conventional Lathe | Swiss-Type Lathe |
|---|---|---|
| Max L/D ratio (steel) | ~3:1 unsupported | 20:1 or higher |
| Min practical diameter | ~6mm | ~0.5mm (some machines go to 0.1mm) |
| Guide bushing required | No | Yes — adds setup cost, needs ground bar stock |
| Bar stock quality needed | Standard tolerance (±0.05mm) | Precision ground (±0.005mm or better) |
| Cycle time (complex small parts) | Slower, more handling | Faster — sub-spindle picks off, machines back side |
| Part complexity in one setup | Limited to turning features | Turn + mill + cross-drill + tap + broach — all in one cycle |
| Typical batch size | 1-500+ | 500-1,000,000+ |
| Machine cost | $30k-$150k | $80k-$400k+ |
| Setup complexity | Moderate | High — guide bushing, sub-spindle collet, multiple tool stations |
Swiss machines dominate for small, long, precise round parts — medical bone screws, dental implants, watch components, connector pins, catheter components, micro-fasteners. If you look at a Swiss-made part under magnification, you'll see features turned, milled, drilled, and threaded in a single handling. The part starts as bar stock, exits the sub-spindle as a finished component, and drops into the parts bin complete.
Live tooling and mill-turn: when the lathe does milling
A basic 2-axis lathe (X and Z) can only turn OD features, face ends, drill center holes, bore IDs, and cut threads. Everything is concentric to the spindle axis.
Add live tooling and you get rotating tools on the turret — end mills, drills, taps — that can operate on the part while it's held in the spindle (or sub-spindle). This gives you:
- Cross-holes drilled perpendicular to the part axis
- Flats milled on the OD
- Hex or square features on the end of a shaft
- Keyways and slots along the part
- Off-center holes on the face
- Threaded cross-holes
A full mill-turn machine with Y-axis adds the ability to move the live tool off the spindle centerline, so you can mill pockets, slots, and features at any orientation relative to the part axis. With a sub-spindle, you can transfer the part, cut off from the bar, and machine the back side — all features completed in one cycle without operator handling.
The economic argument for mill-turn: every time a part comes out of a machine and goes into another machine, you lose accuracy and add cost. A shaft that needs a cross-hole and a keyway, done conventionally: turn on lathe (1st op), transfer to mill, indicate in, drill cross-hole and cut keyway (2nd op), possibly a deburr step. Same part on a mill-turn: done in one cycle, concentricity guaranteed, no handling, no fixture cost, no indication error. The mill-turn machine costs more per hour, but the part cost is lower because you eliminate setup, handling, and scrap from misalignment.
Bar feeding economics
Bar feeding is what makes turning production-efficient. Instead of loading individual billets or saw-cut blanks, you load a 3-meter (12-foot) bar into the bar feeder. The machine pulls the bar through the spindle, machines a part, cuts it off, pulls more bar, machines the next part. The bar feeder holds the bar and feeds it forward as material is consumed.
A 3-meter bar of 10mm 316L stainless yields about 250 parts at 12mm part length. That's hours of unattended running. The operator loads a bar every few hours instead of loading individual blanks every few minutes. For production quantities (500+), bar feeding is the biggest cost lever in turning.
What bar feeding costs:
- Ground bar stock costs 15-30% more than standard tolerance bar
- Guide bushing liners for each diameter range ($200-500 each)
- Bar feeder itself adds $15k-$40k to the machine package
- Setup time for bar changeover (diameter change = 30-60 minutes)
What bar feeding saves:
- Material utilization: 90-95% vs 50-70% for billet machining (no saw kerf, minimal remnant)
- Labor: one operator tends 3-5 bar-fed machines vs one operator per machine for billet loading
- Consistency: every part gets the same bar quality, same process, same thermal conditions
The breakeven is around 300-500 parts for most geometries. Below that, billet loading with pre-cut blanks usually makes more sense. Above that, bar feeding dominates.
Guide bushing vs non-guide bushing
Swiss machines use a guide bushing. The bar slides through it, and cutting happens at the bushing face. The bushing supports the bar right where the cutting force is applied. This is what enables those 20:1 L/D ratios.
A guide bushing needs ground bar stock. The bar diameter tolerance must match the bushing clearance — typically 0.005mm or less. If the bar is undersize, it rattles in the bushing and the part chatters. If the bar is oversize, it seizes. This is why Swiss shops buy ground bar stock from specific mills and stick with proven suppliers. A batch of bar stock that's 0.01mm over tolerance can shut down a Swiss cell for a day.
Non-guide bushing lathes (sometimes called "fixed headstock" lathes) don't have this constraint. Standard tolerance bar stock works fine. The tradeoff: limited L/D ratio because the bar is only supported at the chuck, not near the cutting zone.
Some modern Swiss machines can run without the guide bushing (sometimes called "chucker mode"), giving you the flexibility to use standard-tolerance bar for short parts and ground bar for long parts on the same machine. This is useful for job shops that handle mixed work.
Speeds and feeds for turning: what I actually use
All the textbook numbers assume ideal conditions — rigid setup, perfect material, new tooling, coolant flooding the cut. Real shop conditions are messier. Here's what I actually run, day to day:
| Material | SFM (carbide) | Feed/rev (roughing) | Feed/rev (finishing) | DOC roughing | DOC finishing |
|---|---|---|---|---|---|
| 6061 Aluminum | 600-1000 | 0.15-0.30 mm | 0.05-0.10 mm | 2-4 mm | 0.25-0.50 mm |
| 7075 Aluminum | 500-800 | 0.12-0.25 mm | 0.05-0.08 mm | 1.5-3 mm | 0.20-0.40 mm |
| 303 Stainless | 180-280 | 0.10-0.20 mm | 0.04-0.08 mm | 1-3 mm | 0.15-0.30 mm |
| 304 Stainless | 120-180 | 0.08-0.15 mm | 0.03-0.06 mm | 1-2 mm | 0.10-0.25 mm |
| 316L Stainless | 100-160 | 0.08-0.12 mm | 0.03-0.05 mm | 1-1.5 mm | 0.10-0.20 mm |
| Titanium Grade 5 | 60-100 | 0.08-0.12 mm | 0.03-0.05 mm | 1-2 mm | 0.10-0.20 mm |
| C360 Brass | 600-1000 | 0.15-0.30 mm | 0.05-0.10 mm | 2-4 mm | 0.25-0.50 mm |
| C110 Copper | 300-500 | 0.10-0.20 mm | 0.04-0.08 mm | 1-2.5 mm | 0.15-0.30 mm |
| PEEK | 200-400 | 0.10-0.20 mm | 0.05-0.08 mm | 1-2 mm | 0.15-0.30 mm |
| PTFE | 250-500 | 0.08-0.15 mm | 0.05-0.10 mm | 1-3 mm | 0.10-0.25 mm |
Some things I've learned the hard way:
304 and 316L work-harden if you rub instead of cut. Keep the feed rate up — you want the tool edge under the work-hardened layer from the previous pass. A light finish pass that skims the surface work-hardens it and kills the insert in seconds.
Titanium doesn't like high speed. Run it too fast and the chip welds to the insert. Keep SFM low, feed moderate, and use sharp positive-rake inserts. Coolant is mandatory — titanium conducts heat poorly and the tool absorbs most of it.
Brass is a joy to turn. It produces short chips that break clean, surface finish comes out mirror-bright, and tool life is excellent. Copper is the opposite — gummy, strings of chips that wrap around the tool and the part, poor surface finish unless conditions are perfect. Sharp tools with high positive rake and lots of coolant help.
PEEK turns nicely but watch the heat. If the part gets too hot during turning, it can change color (brown at the surface) and lose mechanical properties. Sharp tools, moderate speeds, and coolant that actually reaches the cut zone.
When turning beats milling, by the numbers
I ran the comparison on a real part recently — a stainless 316L shaft, 12mm dia x 80mm long, with M8 threads on both ends, a 10mm bearing journal (h7 tolerance), and a cross-hole for a retaining pin:
| Cost Factor | CNC Turning (with live tooling) | CNC Milling (3-axis) |
|---|---|---|
| Machine time | 8 min | 22 min (two setups) |
| Setup time | 30 min | 45 min (two vises) |
| Tooling cost/part | $0.60 | $1.20 |
| Material (per part) | $1.10 (from bar) | $3.50 (from billet, 68% removal) |
| Labor | 2 min | 8 min (load/unload ×2 setups) |
| Inspection | Concentricity guaranteed by process | Must verify concentricity on CMM |
| Scrap rate | <0.5% | ~2% (misalignment in second setup) |
| Total cost/part (qty 100) | ~$12 | ~$38 |
The turning approach is about a third the cost, with better concentricity and lower scrap. The cross-hole requires live tooling (driven drill on the turret), which adds machine cost but pays for itself in eliminating the second operation.
This doesn't mean turning is always better. A rectangular part still wants milling. But for round parts — shafts, pins, bushings, spacers, fittings, threaded components, bearing journals — turning is usually the more economical process by a significant margin.
Common turning mistakes that cost money
Specifying milling when the part is round. I see this constantly. Engineers think "CNC" means "milling" and request a milled part when a turned part would be cheaper, faster, and more accurate. If your part is axisymmetric, ask for turning.
Over-specifying surface finish on non-functional surfaces. A turned finish of Ra 0.8 μm looks clean and professional. Specifying Ra 0.2 μm on a surface that doesn't seal, bear load, or contact anything drives up cost for no functional reason. Fine finishes take more passes, slower feeds, and sometimes grinding.
Not considering bar stock diameters. Standard metric bar stock comes in 1mm increments up to about 20mm, then 2mm increments to 50mm, then 5mm increments. If you design a shaft at 17.5mm OD with 16mm bearing journals, you'll buy 18mm bar and turn the OD down. If you design it at 18mm OD with 16mm journals, the OD might be as-supplied bar surface (no turning needed), saving cycle time. Keep standard bar diameters in mind.
Sharp internal corners on turned parts. A lathe can't cut a sharp internal corner at the bottom of a bore or at a shoulder — the insert has a nose radius, typically 0.2-0.8mm. If your part needs a sharp corner for a mating part to seat, add an undercut (relief groove). Otherwise the mating part's corner radius will interfere with the turned radius and the parts won't seat.
Thread relief oversight. External threads need a relief groove at the thread termination if the mating part needs to seat against a shoulder. Without the relief, the last thread is incomplete and the nut or mating part won't bottom out. This is standard lathe practice but I see drawings without it regularly.
Bottom line
CNC turning is not the flashy process. It doesn't make the Instagram videos that 5-axis milling does. But for round parts — and a lot of the world's precision components are round — it's faster, cheaper, and more accurate than any alternative.
If you're sending a round part out for quote, ask the shop: should this be turned or milled? A good shop will answer honestly. If the answer is turning and they're quoting it as a milling job, find another shop.
Send us your drawing and we'll tell you which process makes sense for your part. We run both mills and lathes, and we're not married to either one. The part gets the process it fits.