Machining basics: start here
This page covers how a CNC machine removes metal, what axis count changes on the floor, and where the process stops making sense. Written for design and process engineers who judge a part before sending it out. By the end you can read a drawing and tell whether milling, turning or another process is the right call.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
Key takeaways
What machining basics start with: cutting, not shaping
Machining basics start with one idea. A rotating or stationary cutting tool contacts a solid block and shears material away as chips. Nothing is cast, pressed or added. The finished part is the negative space left behind, which is why the process holds tight tolerances without a dedicated mold.
The tool edge does the work. Carbide inserts and solid end mills have a defined rake angle, and each tooth takes a chip of a set thickness. Feed per tooth, spindle speed and depth of cut decide whether that chip forms cleanly or rubs and work-hardens the surface. Get the chip load too low and the tool skates. Too high and the edge chips.
Heat is the constraint. Most cutting energy turns into heat at the shear zone, and the chip carries a large share of it away. Flood coolant, through-tool coolant or air blast all exist to control that balance. On titanium and stainless, poor chip evacuation is the usual cause of burnt edges and short tool life.
This is why material choice matters more than machine brand. Aluminium 6061 cuts at high surface speed with generous depth. 17-4PH stainless and Ti-6Al-4V need lower speeds, stiffer setups and more frequent tool changes. The same drawing can cost three times more in one alloy than another.
- 1Chip loadFeed per tooth; too light rubs, too heavy breaks edges.
- 2RigidityTool overhang and fixture stiffness set achievable finish.
- 3CoolantChip evacuation matters as much as cooling.
Cutting tool geometry and what it limits
Every cutter has a minimum internal radius. A Ø6 mm end mill cannot cut a sharper corner than its own radius, so a pocket drawn with R0.5 corners will force a smaller tool, slower feed and more passes. Designers who round internal corners to at least one third of the pocket depth usually cut cycle time without losing function.
Pocket depth also matters. A rule of thumb is that depth should stay under four times the tool diameter for a rigid cut. Deeper than that and the tool deflects, chatter appears on the wall, and the operator must reduce feed. Long-reach tooling exists, but it trades away finish and accuracy.
Threads, holes and slots each have their own limits. A Ø2 mm hole is fine in aluminium at 3,000 rpm and above. The same hole in Inconel needs peck drilling, rigid coolant and patience. Thread milling beats tapping on hard alloys and large diameters because it controls chip evacuation better.
Undercuts and side-entry features need a different approach. A T-slot cutter or a lollipop tool can reach beneath an overhang, but the tool is weak and the pass is slow. When a drawing has many undercuts, it is worth asking whether the part can be split into two pieces and assembled.
- 1Internal radiusKeep corners at least one third of pocket depth.
- 2Depth-to-diameterStay under 4:1 for a stable cut.
- 3UndercutsOften cheaper to split the part than to reach them.
Fixturing and how many setups a part needs
A setup is the operation of clamping a part in a known position. Every extra setup adds labor, adds alignment error, and adds the risk of a mark on a finished surface. Three setups and a careful operator will beat five setups and a rushed one, even on the same machine.
Soft jaws, vises and vacuum plates handle most prismatic work. For thin walls and rings, a custom fixture or a sacrificial tab holds the part without distorting it. Machining a thin wall with too much clamp pressure is a common cause of out-of-round bores and bowed faces.
Datum choice drives everything downstream. Pick a face and two edges that will be machined early, and dimension the drawing from those. When a drawing gives coordinates from three different datums, the programmer must reconcile them, and the risk of a mismatch rises.
On a 5-axis machine, one setup can reach five faces plus compound angles. The tradeoff is that the part must be rigidly held in a single orientation, and the programmer must think about tool reach across a wider work envelope. On a 3-axis machine, the same part may need four or five setups.
- 1DatumPick one primary datum and dimension from it.
- 2Thin wallsUse tabs or soft jaws to avoid clamp distortion.
- 3Setup countEach re-fixture adds error, not only time.
3-axis vs 5-axis: what actually changes
A 3-axis mill moves X, Y and Z. The tool always points straight down, so the part must be re-fixtured to reach a new face. That is fine for plates, housings and brackets where most features lie on one or two planes. It is the cheapest and fastest route for simple geometry.
A 5-axis machine adds two rotary axes, so the tool can approach from an angle. This reaches compound faces, deep pockets and sculpted surfaces without re-clamping. For parts with many angled features, one 5-axis setup can replace four or five 3-axis setups, and the accuracy improves because the part never moves.
The cost picture is not automatic. Programming a 5-axis toolpath takes longer and needs simulation to catch collisions. Short cycle parts with simple geometry are often cheaper on a 3-axis machine. The 5-axis route wins when setup reduction or reach is the real constraint.
At GreatLight we run 16 simultaneous 5-axis machining centers, 12 four-axis mills, 27 three-axis machines and 16 mill-turn centers. The right machine depends on the part, not on a preference for more axes. A mill-turn center, for example, finishes a turned shaft with cross-drilled holes in one cycle.
- 13-axisOne face per setup; lowest programming cost.
- 25-axisReaches compound angles; fewer setups, more programming.
- 3Mill-turnTurned and milled features in one cycle.
Where machining stops being the right answer
Machining is wasteful by design. Material removal can take 60 to 80 percent of the stock as chips on a complex part, and that cost scales with the volume removed. When a part is mostly a box with a few holes, a casting or a welded assembly often wins on unit cost.
Very thin walls are another boundary. Below about 0.5 mm, deflection from cutting force becomes hard to control, and the part may need stress relief or a support fixture. Some geometries can be machined, but the yield drops and the price reflects it.
Internal channels and lattice structures are difficult for a rotating cutter to reach. Additive manufacturing builds those features directly. The usual compromise is a hybrid: print the internal geometry, then machine the critical mating faces to tolerance.
Volume is the last factor. Tooling for casting or forging costs thousands, so it only pays above a certain quantity. Machining carries no tooling charge and no minimum order quantity, which makes it the default from one prototype to a few thousand parts.
- 1Material wasteChips can be most of the stock on complex parts.
- 2Thin wallsUnder 0.5 mm, deflection and yield become the issue.
- 3VolumeMachining has no tooling cost; casting needs volume to pay off.
Tolerance, finish and how they are verified
A tolerance is a promise about size, and it only holds if the machine, the tool and the fixture can all support it. A ±0.005 mm callout on a thin, unsupported wall is unenforceable. On a rigid feature measured with the right gauge, it is routine work.
Surface finish is measured in Ra. As-machined faces typically sit at Ra 1.6–3.2 μm. A fine finish at Ra 0.2–0.8 μm needs a lighter finishing pass, a sharp tool and often a different insert. Asking for a fine finish across a whole part adds cost; asking for it only on sealing faces does not.
Verification matters as much as the cut. Calipers and micrometers cover most dimensions, but a coordinate measuring machine is the right tool for true position, profile and angular callouts. GreatLight inspects 100 percent of parts before shipment, with raw material checks, in-process monitoring and a final inspection, and reports are available on request.
We hold ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022. For regulated work, those certificates mean the process is documented, not that any drawing can be met. The drawing still has to be manufacturable.
- 1ToleranceOnly valid if machine, tool and fixture support it.
- 2FinishSpecify Ra only where function requires it.
- 3Inspection100 percent before shipment; reports on request.
Choosing a process by part and quantity
Compare machining against common alternatives
| Process | Best for | Typical tolerance | Watch out for |
|---|---|---|---|
| 3-axis milling | Plates, brackets, housings | ±0.005 mm | One face per setup |
| 5-axis milling | Compound angles, sculpted faces | ±0.005 mm | Higher programming time |
| CNC turning | Round parts, shafts, bushings | ±0.005 mm | Limited to rotationally symmetric |
| Die casting | High volume, complex thin walls | ±0.05 mm | Tooling cost, porosity |
| Sheet metal | Enclosures, brackets from flat stock | ±0.1 mm | Bend radius limits |
| 3D printing | Prototypes, lattice, internal channels | ±0.1 mm | Weaker material properties |
The short verdict
If your part is prismatic with simple features, a 3-axis machine and a sensible datum will be the fastest and cheapest route. If it has compound angles, deep angled pockets or needs four faces held to one datum, go 5-axis and accept the extra programming. If it is a high-volume thin-wall housing with internal channels, look at casting or additive first and machine only the critical faces.
Questions engineers ask next
What file format do you need for a quote?
STEP and STP files carry the solid geometry and are preferred for milling and turning work. IGES and Parasolid are also accepted.
For sheet metal and simple parts, a DXF or DWG with bend lines helps. Include a PDF drawing with tolerances, finish and critical dimensions so the quote reflects the real requirements.
Can you hold ±0.005 mm on every feature?
No. That figure is achievable on rigid features with the right machine and fixture, and it is what we work to on critical dimensions.
It is not realistic across a thin unsupported wall or a very deep narrow pocket. When a drawing asks for that, we flag it during the DFM review and propose a workable callout.
How do you decide between 3-axis and 5-axis for my part?
We look at how many faces carry features and whether any of them sit on a compound angle. If two or three setups cover the part, 3-axis is usually faster and cheaper.
If the part needs four or more orientations held to one datum, or has angled holes and sculpted surfaces, one 5-axis setup usually wins on both accuracy and total cost.
What is the smallest internal corner you can machine?
It depends on depth, not on a fixed number. A corner radius must be at least the radius of the cutter that can reach that depth without chatter.
As a practical guide, keep internal corners at one third of the pocket depth or larger. That lets us use a stiffer tool and cut faster.
Do you handle small prototype runs as well as production?
Yes. There is no minimum order quantity, so a single prototype and a 10,000-part run both go through the same process.
Quotation and a free DFM analysis come back within 12 hours, and production can start within 24 hours. Parts ship in 3–5 days.
How is confidentiality handled for new designs?
Uploads are treated as secure and confidential. We hold ISO 27001:2022 for information security, and an NDA is available on request before any file is shared.
If you need a signed agreement first, contact us and we will send one back before you upload the model.
Send a drawing and get a manufacturability read
Upload your STEP file and we will return a quotation plus free DFM analysis within 12 hours, with the tolerance and setup issues called out before you commit.
12-hour quote100% inspectionNo MOQNDA on request